The plasminogen protein is associated with high myopia as revealed by the iTRAQ-based proteomic analysis of the aqueous humor

To explore the pathogenesis of high myopia (HM) using quantitative proteomics. The aqueous humor of patients with simple nuclear cataract and nuclear cataract complicated with HM (hereinafter referred to as “C” and “HM” groups, respectively) were collected. The isobaric tags for relative and absolute quantitation (iTRAQ)-based liquid chromatography–tandem mass spectrometry (LC–MS/MS) proteomics approach was employed to explore differentially expressed proteins (DEPs). Bioinformatics was used to interpret the proteomic results. Furthermore, the plasminogen (PLG) protein was confirmed by enzyme-linked immunosorbent assay (ELISA) as the candidate biomarker for HM through a receiver operating characteristic curve analysis. The study showed 32 upregulated and 26 downregulated proteins. The gene ontology analysis demonstrated that 58 DEPs corresponded to 325 biological processes, 33 cell components, and 45 molecular functional annotations. The Kyoto Encyclopedia of Genes and Genomes analysis showed that the upregulated DEPs were highly enriched in the coagulation and complement cascades, consistent with the gene set enrichment analysis. Our data suggested that some DEPs might be hallmarks of the development of HM. ELISA confirmed that the PLG expression levels were significantly upregulated in HM. This was a new study investigating alterations in protein levels and affected pathways in HM using iTRAQ-based quantitative proteomics. Our study provided a comprehensive dataset on overall protein changes and shed light on its potential molecular mechanism in human HM.


Scientific Reports
| (2021) 11:8789 | https://doi.org/10.1038/s41598-021-88220-9 www.nature.com/scientificreports/ expressions, which provide new theoretical bases and solutions for discovering disease mechanisms. The isobaric tags for relative and absolute quantification (iTRAQ) technology developed by AB SCIEX, USA is an accurate quantitative technique for in vitro isotopic labeling. It uses up to eight isotope reagents to label the N-terminal or lysine side chain of peptides. A high-resolution mass spectrometer can simultaneously analyze the protein expression of these eight samples. It has been a promising quantitative technique in proteomics in recent years.
In the present study, we performed the iTRAQ-based quantitative proteomic analysis of the aqueous humor of patients with nuclear cataract complicated with HM and simple nuclear cataract to identify functional proteins associated with the pathogenesis of HM. Additionally, we performed enzyme-linked immunosorbent assay (ELISA) to confirm the iTRAQ results. Our findings are likely to enhance our understanding of the mechanisms of HM in humans.

Materials and methods
Aqueous humor samples. In this prospective study, patients with simple nuclear cataract and nuclear cataract complicated with HM (hereinafter referred to as "C" and "HM" groups, respectively) who first visited the Tianjin Medical University Eye Hospital (Tianjin, China) were enrolled, and their aqueous humor samples were collected during cataract surgery. Inclusion criteria were nuclear cataract and an axis length ≥ 26 mm. Exclusion criteria were other eye and systemic diseases, such as hypertension, diabetes, coronary heart disease, and autoimmune disease. Fourteen aqueous humor samples of eleven patients, including seven samples from each group, were selected in the study. Table 1 shows patient characteristics. Informed consent was obtained from all participants of this study. The collection and preparation of the samples are performed by the same skilled surgeon as follows. 2 eye drops of oxybuprocaine hydrochloride 0.4% (Beinuoxi, Santen Pharmaceutical Co., Ltd, Osaka, Japan) were applied to the eye before rinsing twice with 0.5% povidine iodine (Anerdian, LikangDisinfection Co., Ltd, Shanghai, China). After that, the conjunctival sac was irrigated with a balanced salt solution before the aqueous humor sample collection. To avoid blood and other ocular surface contaminants, the sample was collected using a 1 mL tuberculin syringe with a 30 gauge needle at the corneal limbus prior to any other entry into the eye under the surgical microscope. Then, all samples were collected into a cryopreservation tube and stored at − 80 °C until use. This study was approved by the Human Research Ethics Committee of the Tianjin Medical University Eye Hospital and adhered to the tenets of the Declaration of Helsinki.

LC-MS/MS analysis.
Samples were analyzed with the TripleTOF 6600 mass spectrometer (AB SCIEX, USA) coupled with the ekspert nano-LC 415 nanoflow LC system (AB SCIEX, USA). Fractionated peptides were loaded onto an in-house made trap (100 μm × 2 cm; particle size, 3 μm; SunChrom, USA) with mobile phase A (v/v/v, 2% acetonitrile, 0.1% formic acid, and 97.9% water) and separated with a linear gradient of mobile phase B (v/v/v, 0.1% formic acid, 97.9% acetonitrile, and 2% water) on the analytical column (150 μm × 15 cm; particle size, 1.9 μm; SunChrom, USA). The effective elution gradient was 5-35% for 1.5 h at a flow rate of 600 nL/min. The accumulation times for time-of-flight MS and product ions were 0.25 and 0.04 s, respectively. The mass ranged from 300 to 1500 m/z. The charge state ranged from + 2 to + 5, and mass tolerance was less than 50 ppm. We monitored 60 ions/cycle, and the ion exclusion time was 16 s. The fragmentation energy was adjusted using iTRAQ reagents, and a high sensitivity mode was used for scanning.
Protein identification and quantification. For each iTRAQ set, raw files from eight fractions were combined to search a human database (Swiss human database, released in March 2019, comprising 20,425 entries) with MaxQuant software version 1.6.4.0. Dynamic modifications included N-acetylation, oxidation of methionine, and cysteine carbamidomethylation as the fixed modification. All reporter ions were applied for quantification. The false discovery rate (FDR) was set at 1% for both peptide spectrum matches and proteins. At least two peptides per protein were chosen for quantification. The mass tolerance was 0.1 Da for precursor ions and 50 ppm for fragment ions.
Statistical and bioinformatic analyses. Data were normalized with quality control between two iTRAQ sets, and median normalization was applied to adjust the error among samples. Student's t-test was applied to calculate p-values. Screened proteins with a p value < 0.05 and fold change > 1.5 were considered to be significantly expressed. The gene ontology (GO) and KEGG pathway 11,12 enrichment analysis was processed through cluster Profiler, an R package. The adjusted p-and q-values were both < 0.05. Protein-protein interactions (PPIs) were analyzed with a web tool called STRING, version 11.0, and visualized with Cytoscape software version 3.7.0. Proteins were classified according to their annotated functions. The gene set enrichment analysis (GSEA) was executed with GSEA software version 3.0 (FDR < 0.25, normalized p value < 0.05). The principal component analysis was conducted with the stats package and visualized with the factoextra R package.
ELISA and receiver operating characteristic (ROC) curve analysis. The target protein plasminogen (PLG) was chosen for verification of differential expressions using ELISA. The Human Plasminogen ELISA kit (Elabscience Biotechnology Co., Ltd., E-EL-H2102c) was used. ELISA assays were performed according to the manufacturer's protocols, and samples (HM group, 14 samples; C group, 10 samples) were run in duplicate. The absorbances of the standards and samples were determined by spectrophotometry at 450 nm with a microplate reader. Statistical analyses were performed with GraphPad Prism version 6.0. Student's t-test was applied for comparisons of quantitative data, and the ROC curve analysis with GraphPad Prism version 6.0 was performed to evaluate the sensitivity and specificity of the protein PLG.

Results
Quantitative proteomic analysis. Aqueous humor proteins that showed differential accumulation between C and HM groups were identified and analyzed with LC-MS/MS. In the proteomic analysis, a total of 211 unique corresponding proteins were quantified. Among proteins showing significant changes (p < 0.05) in abundance, we used proteins with values of an abundance fold change > 1.5 to identify those with a significant difference in abundance between the C and HM groups. Figure 1 shows the experimental design of the whole study.
Bioinformatic analysis of DEPs. Data quality control. Figure 3A shows the Pearson correlation coefficients of the samples. The correlation coefficients in control and HM groups were above 0.96 and 0.87, respectively, indicating high reproducibility of the data.
GO analysis. To evaluate the major biological processes that can be perturbed by HM, the GO enrichment analysis was performed on DEPs. The results showed that 58 DEPs were involved in 403 functional annotations. KEGG analysis. To analyze the main biochemical processes and associated pathophysiological pathways affected by HM, a KEGG pathway 11,12 -based enrichment analysis was conducted to determine the DEPs in the aqueous humor. The affected pathways upregulated the proteins involved in the complement and coagulation cascades and cholesterol metabolism (Fig. 4A) and downregulated the proteins involved in glycolysis/gluconeogenesis, biosynthesis of amino acids, and HIF-1 signaling pathway (Fig. 4B). A total of eight DEPs were confirmed as key signal molecules of these pathways, including C4B (complement), PLG (plasminogen), C9 (complement component), F2 (thrombin), KNG1 (kininogen-1), CF1 (cleavage stimulation factor subunit 2), SERPINA1 (alpha-1-antitrypsin), and C4A (complement). GO showed that DEPs were mainly associated with inflammation and humoral immunity, and the major proteins involved (C4B, PLG, C9, F2, CFI, SERPINA1, and C4A) coincided with the complement coagulation cascade, implying that DEPs may participate in the process of eye damage caused by HM through the immune process involved in the complement system. Moreover, we performed GSEA (p < 0.05, FDR < 0.25) on DEPs and found that the two classical signaling pathways, the complement coagulation cascade and extracellular matrix regulator, were enriched and upregulated ( Fig. 5A-C). Consistent with the previous KEGG results, the major enriched proteins were C4B, PLG, KNG1, C9, F2, CFI, SERPINA1, and C4A. The results of Riddell Nina on HM chicken model showed that complement coagulation cascade and extracellular matrix pathway were overexpressed in HM 13 . It further suggests that the above proteins may be the key factors of pathogenicity. www.nature.com/scientificreports/ www.nature.com/scientificreports/  , indicating that there were significant differences in protein expression between the C and HM. (C) Differential protein volcano plot showed that these were distributed in two different quadrants (The logarithm of the fold change of each quantifiable protein was taken as base 2, and the negative logarithm of the P value was taken as base 10). Horizontal axis represents the fold change (log2 value) of the differential protein, and vertical axis represents the P value (− log10 value). Gray dots represent proteins with insignificant differences, red dots represent up-regulated proteins, and blue dots represent down-regulated proteins. The top10 protein was selected to display. www.nature.com/scientificreports/ PPI analysis. In order to further decipher the potential function of these DEPs, PPIs were analyzed with the STRING database and classified according to DEP functions (information from GO, UniProt, and NCBI). These interacting proteins were divided into six categories, mainly involved in the complement coagulation cascade, enzyme inhibitor activity, lipid metabolism, cell adhesion, and immune response. Figure 6A presents the results.

Target protein.
Pearson's correlation analysis of the proteins involved in the GSEA complement coagulation cascade revealed that proteins had a strong correlation in expression (Fig. 6B). Previous studies suggested that collagens I and XVIII played an important role in the pathogenesis of posterior scleral staphyloma caused by HM 14,15 . In the interactions among collagen I, collagen XVIII, and proteins involved in the complement coagulation cascade, SERPINA1, PLG, KNG1, F2, C4A were proteins with high interaction degree (Fig. 6C). Among the five hub DEPs, PLG had the largest expression fold change (7.78), which was equal to the sum of SERPINA1 (2.17), KNG1 (2.59), and F2 (2.99). And PLG is a member of fibrinolytic system, which plays a role in extracellular matrix remodeling and coagulation cascade (Fig. 5C). Therefore, PLG may be the most hub one among them. www.nature.com/scientificreports/ Validation with ELISA. We used the ELISA kit to verify the PLG protein expression in the aqueous humor of patients with HM. Figure 7A shows the iTRAQ quantification in the MS/MS spectrum. The expression levels of PLG were significantly upregulated in HM (Fig. 7B), which showed the same trend of change as the proteomic results.
ROC curve analysis. The areas under the ROC curves were calculated as measures of the specificity and sensitivity of the confirmed protein (Fig. 7C). The accuracy of the area under the ROC curve was assessed as follows: 0.9-1 = excellent; 0.8-0.9 = good; 0.7-0.8 = fair; and < 0.7 = not useful. The ROC analysis identified that PLG was informative (AUC: 0.8750, 95% confidence interval: 0.7280-1.022). www.nature.com/scientificreports/

Discussion
The development of HM is associated with changes in the collagen fibers of the sclera, as demonstrated in both human and animal models 9,14 . Studies have identified numerous proteins that are involved 9,16,17 . However, to date, most insights regarding HM have come from studies on animal models, which typically focused on the retina, choroid, and sclera 9 . Although changes in protein levels of the aqueous humor in myopic eyes have been extensively studied, few studies have focused on HM. Therefore, the aim of this study was to investigate the changes in aqueous humor protein levels in HM. The results of proteomics might offer new insights to elucidate the mechanism of HM and identify new potential targets for the treatment.
In this study, a total of 211 proteins were identified. The iTRAQ technique revealed 58 DEPs in the aqueous humor of patients with HM, which indicated that the protein expression in the aqueous humor changes in HM. However, the origin and causal relationship with those proteins are unclear. In the study investigating the proteome of the aqueous humor in patients with HM versus controls using iTRAQ-based quantitative proteomics by Xiang M and colleagues, they compared proteins in the aqueous humor of 12 patients undergoing cataract www.nature.com/scientificreports/ surgery (6 patients with HM, 6 controls) and found 445 proteins including146 DEPs 18 , with no mention of PLG among the significant proteins probably because of sample differences. Duan et al. characterized the aqueous humor of five patients with HM and five age-matched patients with cataract without HM undergoing routine senile cataract surgery and identified that a total of six spots were significantly increased in the two-dimensional gel analysis in HM, with the proteins being albumin, transthyretin, and a vitamin D-binding protein 9 . Transthyretin and the vitamin D-binding protein were included in the 211 proteins we identified but not in the 58 DEPs. This may be caused by differences in protein identification techniques. The upregulated DEPs were highly concentrated in the coagulation and complement cascade (GSEA and KEGG analysis). A meta-analysis of transcriptome datasets in chick myopia and hyperopia models provided the evidence for involvement of the complement pathway in hyperopia 13 . Some studies showed that the complement system might promote extracellular matrix (ECM) remodeling in the sclera 13,19 .
PLG can be activated by the tissue or urokinase-type plasminogen activator as active plasmin. Plasmin degrades fibrin and converts potential matrix metalloproteinases (pro-MMPs) into active MMPs, which can destroy the ECM 20 . Moreover, plasmin plays an important role in many physiological and pathological processes which may be involved in the pathogenesis of high myopia, such as tissue remodeling, angiogenesis [21][22][23][24] . www.nature.com/scientificreports/ Additionally, some studies have shown that plasmin is closely related to inflammation, autoimmunity, tumor formation, and neurodegeneration [25][26][27] . Thus, PLG is a multifunctional factor associated with the pathogenesis of HM that warrants further investigations. Posterior scleral staphyloma is a basic lesion in a series of interrelated degenerative changes in HM and significantly associated with macular holes, retinal detachment, and myopic foveoschisis; therefore, it is recognized as a hallmark of HM 28 . Previous studies pointed out that the overall scleral thinning observed in highly myopic human eyes were associated with a reduction in size of the individual collagen fibrils with a preponderance of unusually small diameter fibrils averaging below 60-70 nm 29 . After 12 days of experimental myopia in the tree shrew (equivalent to 12-14 D myopia), the thickness of the posterior sclera had been reduced by 20% and showed a slow, progressive thinning subsequently. Simultaneously, the dry weights of the sclera and the posterior sclera were reduced by 7% and 17%, respectively 29 . These ultramicroscopic alterations in the sclera of patients with HM suggested that the disorder of the scleral collagen fibrils may be due to the decreased synthesis or increased accentuated breakdown. The overexpression of the PLG protein in HM activates it outside the blood vessels, dissolves the fibers, and makes the sclera thinner, which accelerates the occurrence of posterior scleral staphyloma. According to the website of UniProt (https:// www. unipr ot. org/), KNG1 plays a role in increasing vascular permeability. Therefore, the upregulated KNG1 can damage the blood-retina barrier and increase vascular permeability by regulating the tight junction with the vascular endothelium, which in turn causes inflammatory mediators to enter the aqueous or vitreous humor. Furthermore, the enriched ECM regulatory factor signaling pathway indicates that the activated PLG may reduce scleral hardness by affecting the composition of the scleral connective tissue. Additionally, the classical signaling pathways of the ECM regulator suggest that excessive activation of PLG may reduce the scleral stiffness by affecting the components of scleral connective tissue. The composition of the scleral ECM is the key to maintain its inherent stiffness, strength, and elasticity.
Apart from its potential role in the mechanism of posterior scleral staphyloma caused by HM, PLG plays a role in the pathogenesis of other complications. First, glaucomatous optic neuropathy (GON) is a common cause of irreversible vision loss worldwide. Several hospital-and population-based investigations showed that the prevalence of GON was higher in highly myopic eyes compared to emmetropic eyes [30][31][32][33][34][35][36] . Enlargement of the optic disc, including the stretching and thinning of the sieve plate, may be a factor causing susceptibility to GON 37 . Therefore, we speculated that fibrinolysis induced by PLG activation may be the main factor of optic disc stretching and cribriform plate thinning in HM, which may also be the reason why patients with HM are prone to normal-tension glaucoma 38,39 . Additionally, liquefaction of the vitreous humor begins at a relatively young age in patients with HM 40,41 , which may directly relate to PLG activation. Harocopos and Filas et al. suggested that loss of the vitreous gel with aging increases the risk of nuclear cataracts 42,43 , which are strongly associated with HM 44,45 . Therefore, the activation of PLG may be the cause of patients with HM suffering from nuclear cataracts.
Our study had some limitations. Frist, the sample size was relatively small because of the stringent patient recruitment criteria. Second, we have not conducted further animal and cellular studies on the role of PLG in the pathogenesis of posterior scleral staphyloma. However, the proteomic results were verified using an independent cohort with ELISA (supplementary table).