Urinary Proteome Changes in Global Cerebral Ischemia-Reperfusion Injury Rat Model Using Proteomics

Background Cerebral ischemia-reperfusion (I/R) injury is the leading cause of death in severe hypotension caused by cardiac arrest, drowning, and excessive blood loss. Urine can sensitively reect pathophysiological changes in the brain even at an early stage. In this study, a rat model of global cerebral I/R injury was established via Pulsinelli’s four-vessel occlusion (4-VO) method. The proteomics techniques of data-independent acquisition (DIA) and parallel reaction monitoring (PRM) were applied to prole the urinary proteome. The differentially expressed proteins were subjected to Gene Ontology (GO) and protein-protein interaction (PPI) analysis. GO analysis showed that the acute-phase response, the ERK1 and ERK2 cascade, endopeptidase activity, blood coagulation, and angiogenesis were overrepresented. After fteen differentially expressed proteins were identied, and their expression was consistent with the DIA quantication. The abundance of FGG, COMP, TFF2, and HG2A was signicantly changed only at 12 h after I/R injury. APOE, FAIM3, FZD1, IL1R2, UROK and CD48 were upregulated only at 48 h after I/R injury. KNG1, CATZ, PTGDS, PRVA and HEPC showed an overall trend of upregulation or downregulation at 12 and 48 h after I/R injury, reecting the progression of cerebral I/R injury.


Introduction
Global cerebral ischemia-reperfusion (I/R) injury is the leading cause of death in severe hypotension caused by cardiac arrest, drowning, and excessive blood loss [1]. Because the brain is very sensitive to hypoxia, the brain tissue sustains irreversible damage 4 to 6 min after the loss of circulation, especially in highly metabolically active regions such as the hippocampus, thalamus, cerebral cortex, and striatum [2].
Global cerebral I/R injury is closely associated with neurological dysfunction, from mild cognitive impairment to a minimally conscious state or even a persistent vegetative state [3]. According to reports, up to 80% of survivors are in a vegetative state, and patients who recover well often have serious psychological complications such as depression, anxiety, and posttraumatic stress disorder, which seriously affect their quality of life [4]. To date, the diagnosis and evaluation of global cerebral I/R injury have mainly relied on clinical symptoms, neuroimaging, electrophysiology, and biochemical examination of blood or cerebrospinal uid. However, hemodynamic instability, sedation and hypothermia are common in these patients, limiting the application of these tests [5]. The main clinical treatment methods include supportive therapy, symptomatic treatment, mild hypothermia and hyperbaric oxygen therapy.
There is still a lack of speci c, effective neuroprotective strategies and drugs. Therefore, early and accurate assessment of the degree of brain injury and prognosis is critical for diagnosis and treatment.
Currently, blood and cerebrospinal uid are the main sources of samples for the study of brain injury biomarkers. Although many candidate protein biomarkers have been reported in blood and cerebrospinal uid, they have not been further applied [6,7]. There are several drawbacks to focusing on these uids: 1) An invasive sampling process is necessary to obtain blood or cerebrospinal uid, and the latter is especially di cult to obtain. 2) Blood and cerebrospinal uid are important components of the internal environment; therefore, they are strictly controlled by homeostatic mechanisms to maintain the relative stability of their components. When a certain change is introduced into the internal environment, the body will reduce this change as much as possible and eliminate it through various mechanisms to keep its composition relatively stable. Therefore, changes in biomarkers in blood and cerebrospinal uid are not su ciently sensitive. 3) Due to technical limitations, the sample size and the extent of protein identi cation in previous studies were relatively small, and the conclusions have not been extensively validated. Thus, simple, noninvasive, and sensitive biomarkers are needed.
According to the largest human urine proteome database including 6085 sequence of protein information, some of these proteins have been reported to be enriched in 32 tissues and organs, such as the brain, gastrointestinal tract, kidneys and so on [8]. Among these proteins, 1956 that were enriched in brain tissue were identi ed in urine; this number ranked rst among the 32 tissues and organs evaluated.
Several studies have shown that urine can re ect the pathophysiology of some neurological diseases, such as Alzheimer's disease, Parkinson's disease, multiple sclerosis, and neuroendocrine tumors [9]. In addition, in a glioblastoma rat model and a Walker 256 lateral ventricle inoculation rat model, the urine proteome changed signi cantly before clinical symptoms and brain histopathological changes appeared [10,11]. Overall, urine is a good source of specimens for the study of brain injury markers, and it can provide a sensitive re ection of brain pathophysiology at an early stage.
In this study, the proteomics technique of data-independent acquisition (DIA) was used to pro le the urinary proteome in a global cerebral I/R rat model, and then the altered proteins were validated by using the parallel reaction monitoring (PRM) strategy. A summary of the overall experimental approach is presented in Figure 1. This study aims to explore urinary protein biomarkers of global cerebral I/R injury and provide clues to further understand its molecular biological mechanisms.
Results And Discussion

Histopathological Damage in the Hippocampus
To evaluate ischemia and histological damage, H&E staining was conducted on brain sections of the hippocampal CA1 region, which is most well known as being selectively vulnerable following ischemia. H&E staining indicated that no neuron morphology abnormalities were observed in the sham group ( Fig.   2A). At reperfusion for 12 h, the number of neurons was reduced, the structure was complete, and no typical apoptotic cells were observed (Fig. 2B). With the reperfusion time prolonged to 48 h, the number of intact neurons decreased signi cantly, shrunken cell bodies and nuclear pyknosis (Fig. 2C).

Urine Proteome Changes
To preliminarily investigate how the urine proteome changes with I/R progression, twenty-one urine samples from the sham and I/R group (12 h and 48 h) were analyzed via a label-free DIA work ow.
To generate spectral library A, fractions separated with a spin column were analyzed by DDA-MS and then processed using Proteome Discoverer (version 2.3) and Spectronaut Pulsar X. The library included eight DDA analyses of fractions resulting in 1152 protein groups and 6260 peptides with at least one unique peptide and a Q value<0.01. Raw DIA-MS data les acquired with 30 re ned isolation windows from the twenty-one individual urine samples were loaded into Spectronaut Pulsar X. Overall, a total of 866 (699 on average) protein groups were identi ed from twenty-one biological replicates. All identi cation and quantitation details are listed in supporting Table S3.
One hundred and sixty-four proteins signi cantly differed in the urine samples compared to the control samples (1.5-fold change, p<0.05). There were 59 and 123 altered urinary proteins at 12 and 48 h, respectively, after I/R (Table S4- 5). Eighteen proteins changed signi cantly at both 12 and 48 h ( Table   1). Among these proteins, 7 showed an overall upregulated or downregulated trend: T-kininogen 2, Prostaglandin-H2 D-isomerase, 14-3-3 protein theta, Ig gamma-2B chain C region, Cathepsin Z, Parvalbumin alpha and Hepcidin (highlighted in red in Table 1). This may suggest that these proteins have the potential to be used for the early detection of cerebral I/R injury. "-" means a downward trend.

Function Annotation of Differentially Abundant Proteins
The functional annotation of differentially abundant proteins at 12 and 48 h consisted of sorting them into the "biological process", "cellular component" and "molecular function" categories using DAVID ( Figure 3). One hundred and sixty-four differentially abundant proteins were annotated. In the biological process category, cell-matrix adhesion, positive regulation of ERK1 and ERK2 cascade, cellular response to interleukin-6, negative regulation of endothelial cell apoptotic process, and acute-phase response were overrepresented at 12 h after 12 h I/R. At 48 h after I/R, negative regulation of endopeptidase activity, aging, negative regulation of blood coagulation, angiogenesis, and innate immune response were overrepresented (Fig. 3A).
In the molecular function category, receptor binding, serine-type endopeptidase inhibitor activity, cell adhesion molecule binding, and protein binding were overrepresented at both time points. Oxygen transporter activity was overrepresented at 12 h after I/R (Fig. 3B). In the cellular component category, most of these differentially abundant proteins were associated with extracellular exosomes, extracellular space, blood microparticles, and the external side of the plasma membrane (Fig. 3C).

Protein-Protein Interactions of Differentially Abundant Proteins
To better understand the pathogenic mechanisms in global cerebral I/R, a protein-protein interaction (PPI) network for 164 changed proteins was constructed using STRING (Fig. 4). The STRING PPI network analysis showed that the average node degree was 3.73, the average local clustering coe cient was 0.466, and the PPI enrichment p-value was less than 1.0E-16. The above results revealed that these proteins had more interactions among themselves than would be expected for a random set of proteins of similar size. Such an enrichment pattern indicates that the proteins are at least partially biologically connected as a group in I/R.

PRM Validation and Analysis of Differentially Abundant Proteins
In the validation phase, 255 peptides corresponding to seventy-one proteins were nally scheduled for PRM-MS analysis in another 21 urine samples. Overall, thirty-two proteins were signi cantly altered at multiple time points (1.5-fold change, p<0.05) (Table S6). After the p-value was adjusted by the Benjamini & Hochberg method, ltering yielded fteen proteins with an adjusted p-value<0.05, including 8 increased and 7 decreased proteins ( Table 2). The expression trends of the corresponding proteins were consistent with the results from the DIA discovery quanti cation. Four proteins were changed signi cantly only at 12 h after I/R, when no obvious histopathological changes had yet appeared; these proteins included Fibrinogen gamma chain (FGG), Cartilage oligomeric matrix protein, Trefoil factor 2, and H-2 class II histocompatibility antigen gamma chain. These differentially abundant proteins may provide important clues for the early diagnosis of cerebral I/R injury. Two of these differentially abundant proteins were associated with cerebral I/R injury. Plasma brinogen was independently associated with overall ischemic stroke and all subtypes, both in the acute stage (p < 0.001) and at the three-month follow-up (p < 0.05) [12]. In a case-control study, FGG was associated with a reduced risk of ischemic stroke [13]. FGG was also highly expressed in the ischemic penumbra of focal cerebral ischemia rats [14]. Trefoil factor 2 was upregulated in the plasma of atrial brillation patients with ischemic stroke [15].
Six proteins were upregulated only at 48 h after I/R injury, when there was certain histopathological damage in the hippocampus; these proteins included Apolipoprotein E, Fas apoptotic inhibitory molecule 3, Frizzled-1, Interleukin-1 receptor type 2, Urokinase-type plasminogen activator and CD48 antigen. These differentially abundant proteins may indicate the extent of cerebral I/R injury. Two of these differentially abundant proteins were associated with cerebral I/R injury. Apolipoprotein E (APOE) is the primary apolipoprotein synthesized in the brain in response to ischemia-reperfusion injury with known neuroprotective effects exerted through antioxidant, anti-in ammatory, anti-excitotoxic, and neurotrophic mechanisms [16]. The APOE genotype showed a positive dose-response association with ischemic stroke in people of European ancestry [17]. Frz1 expression was signi cantly decreased in the brain tissue of middle cerebral artery occlusion rats [18].
Five proteins showed an overall upregulated or downregulated trend at 12 and 48 h after I/R injury, including T-kininogen 1, Cathepsin Z, Prostaglandin-H2 D-isomerase, Parvalbumin alpha and Hepcidin (Fig. 5). These differential proteins may re ect the progression of cerebral I/R injury. Three of these differential proteins were associated with cerebral I/R injury. The mRNA expression of T-kininogen 1 was upregulated in the blood of idiopathic thrombophilia [19]. Cathepsin Z was upregulated in the plasma of atrial brillation patients with ischemic stroke [15]. Parvalbumin alpha protein and mRNA were reduced in the hippocampal tissue of C57BL6 mice with unilateral right carotid ligation. Plasma/serum hepcidin levels were signi cantly higher in acute ischemic stroke patients than in the control group [21,22]. The hepcidin mRNA levels and hepcidin/prohepcidin protein levels are upregulated in the ischemic brain [23].

Conclusion
In this study, fteen differentially abundant urinary proteins were identi ed and validated in 4-VO rats.
Eight of the differential proteins were reported to be associated with cerebral I/R injury. These ndings may provide important clues for the monitoring of cerebral I/R injury and further understanding of its molecular biology mechanisms.

Animals
Forty-two male Wistar rats (200-250 g) were purchased from Charles River China (Beijing, China). All animals were maintained on a standard laboratory diet with a controlled indoor temperature (21±2℃), humidity (65-70%) and 12/12 h light-dark cycle conditions. The animal experiments were reviewed and approved by the Qingdao University Hospital Medical Ethics Committee (No.027-2019). All methods were carried out in accordance with relevant guidelines and regulations of the National Health Commission and the Ministry of Science and Technology and conformed to the guidelines for animal research.
The Wistar rats were randomly divided into two groups: a control group (n = 21) and an I/R group (n = 21). Ischemia was induced by Pulsinelli's four-vessel occlusion method [24]. Brie y, the bilateral common carotid arteries were surgically exposed and clamped shut with microclips for 10 min. The experiment was conducted in two phases; for details, see Figure 1. For the discovery phase, differentially abundant urinary proteins were identi ed by label-free DIA quanti cation in twenty-one independent samples from the control group (7 samples) and the I/R group at 12 and 48 h (7 samples per time point). For the validation phase, the 21 remaining urine samples (7 from the control group and 7 per time point from the I/R group at 12 and 48 h) were evaluated by targeted quanti cation with PRM.

Histological Analysis
For histopathology, three rats in the I/R group and three rats in the control group were randomly sacri ced at 12 h and 48 h after I/R. The hippocampus was harvested and then quickly xed in 10% neutral buffered formalin. The formalin-xed tissues were embedded in para n, sectioned (4 mm) and stained with hematoxylin and eosin (H&E) to reveal histopathological lesions.

Urine Collection and Sample Preparation
Urine samples were collected from the control and I/R groups at 12 and 48 h after I/R. Rats were individually placed in metabolic cages for six hours. During urine collection, food was withheld from the rats to prevent the urine from being contaminated. After collection, the urine samples were immediately centrifuged at 2 000g for 30 min at 4°C and then stored at −80°C.
Urinary protein extraction: Urine samples were centrifuged at 12 000g for 30 min at 4°C. Six volumes of prechilled acetone were added after the pellets were removed, and the samples were precipitated at 4°C overnight. Then, lysis buffer (8 mol/L urea, 2 mol/L thiourea, 50 mmol/L Tris, and 25 mmol/L DTT) was used to dissolve the pellets. The protein concentration of each sample was measured by a Bradford protein assay.
Tryptic digestion: The proteins were digested with trypsin (Promega, USA) using lter-aided sample preparation methods [25]. Brie y, 100 µg of the protein sample was loaded onto a 10-kDa lter unit (Pall, USA). The protein solution was reduced with 4.5 mM DTT for 1 h at 37°C and then alkylated with 10 mM indoleacetic acid for 30 min at room temperature in the dark. The proteins were digested with trypsin (enzyme-to-protein ratio of 1:50) for 14 h at 37°C. The peptides were desalted on Oasis HLB cartridges (Waters, USA) and lyophilized for trap column fractionation and LC-MS/MS analysis.

Spin Column Separation
To generate a spectral library for DIA analysis, pooled peptide samples from all samples were fractionated using a high-pH reversed-phase peptide fractionation kit (Thermo Pierce, USA) according to the manufacturer's instructions. Brie y, 60 µg of a pooled peptide sample was loaded onto the spin column. A step gradient of increasing acetonitrile concentrations was applied to the column to elute the bound peptides. Ten different fractions were collected by centrifugation, including the ow-through fraction, the wash fraction and eight step gradient sample fractions (5, 7.5, 10, 12. An Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scienti c, Germany) was coupled with an EASY-nLC 1000 HPLC system (Thermo Scienti c, Germany). For DDA-MS and data-independent acquisition (DIA)-MS modes, the same LC settings were used for retention time stability. The digested peptides were dissolved in 0.1% formic acid and loaded onto a trap column (75 µm × 2 cm, 3 µm, C18, 100 A°). The eluent was transferred to a reversed-phase analytical column (50 µm × 250 mm, 2 µm, C18, 100 A°). The eluted gradient was 5-30% buffer B (0.1% formic acid in 80% acetonitrile; ow rate of 0.8 μL/min) for 90 min. To enable fully automated and sensitive signal processing, the calibration kit (iRT kit from Biognosys, Switzerland) reagent was spiked at a concentration of 1:20 v/v in all samples. The iRT kit reagent was spiked into the urinary peptides for spectral library generation. Additionally, before the real DIA runs, the iRT kit reagent was also spiked into all urinary samples.
For the generation of the spectral library, the ten fractions from the spin column were analyzed in DDA-MS mode. The parameters were set as follows: the full scan was acquired from 350 to 1 550 m/z at 60 000, the cycle time was set to 3 s (top speed mode), the automatic gain control (AGC) was set to 1E6, and the maximum injection time was set to 50 ms. MS/MS scans were acquired in the Orbitrap at a resolution of 15,000 with an isolation window of 2 Da and collision energy of 32% (higher-energy collisional dissociation, HCD); the AGC target was set to 5E4, and the maximum injection time was 30 ms.
For the DIA-MS method, forty individual samples were analyzed in DIA mode. For MS acquisition, the variable isolation window DIA method with 26 windows was developed (Table S1). The full scan was set at a resolution of 60,000 over an m/z range of 350 to 1,200, followed by DIA scans with a resolution of 30,000, HCD collision energy of 32%, AGC target of 1E6 and maximal injection time of 50 ms.

LC-MS/MS Setup for PRM
In the discovery phase, seventy-one differentially abundant urinary proteins were identi ed by the labelfree DIA proteomic method. All of these proteins were evaluated by the PRM-MS method in the remaining twenty-one urine samples. LC-PRM-MS/MS data were acquired in an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scienti c, Germany) coupled with an EASY-nLC 1200 HPLC system (Thermo Scienti c, Germany).
For the generation of the PRM spectral library, pooled peptide samples were analyzed in DDA-MS mode 6 times. The peptides were loaded on a reversed-phase trap column (75 µm × 2 cm, 3 µm, C18, 100 Å, Thermo Scienti c, Germany), and the eluent was then transferred to a reversed-phase analytical column (50 µm × 250 mm, 2 µm, C18, 100 Å, Thermo Scienti c, Germany). The elution gradient consisted of 5-35% buffer B (0.1% formic acid in 80% acetonitrile; ow rate 0.8 μL/min) for 90 min. The MS parameters were set as follows: the full scan was acquired from 350 to 1 550 m/z at 60 000, the cycle time was set to 3 s (top speed mode), the AGC was set to 1E6, and the maximum injection time was set to 50 ms.
MS/MS scans were acquired using the Orbitrap at a resolution of 30 000 with an isolation window of 1.6 Da and collision energy at 30% (HCD), the AGC target was set to 5E4, and the maximum injection time was 60 ms.
For the PRM-MS method, thirty-two individual samples were analyzed in PRM mode. Ultimately, 255 peptides were scheduled, and the retention time (RT) segment was set to 8 min for each targeted peptide (Table S2). The normalized collision energy was xed ats 30%, and the quadrupole isolation window was xed at 1.6 Da. The other parameters were the same as described in the last paragraph.

Label-Free DIA Quanti cation Analysis
To generate the spectral library, the raw data les acquired for the ten fractions in DDA mode were processed using Proteome The raw DIA-MS les were imported into Spectronaut Pulsar with the default settings. In brief, a dynamic window for the XIC extraction window and a nonlinear iRT calibration strategy were used. Mass calibration was set to local mass calibration. Cross-run normalization was enabled to correct for systematic variance in LC-MS performance, and a local normalization strategy was used [27]. Protein inference, which gave rise to the protein groups, was performed on the principle of parsimony using the ID picker algorithm as implemented in Spectronaut Pulsar [28]. All results were ltered by a Q value cutoff of 0.01 (corresponding to an FDR of 1%). Peptide intensity was calculated by summing the peak areas of the respective fragment ions for MS2. Student's t-test was applied with a signi cance criterion of <0.05. A minimum of two peptides matched to a protein and a fold change >1.5 were used as the criteria for the identi cation of differentially expressed proteins.

PRM-MS Quanti cation Analysis
Skyline (version 3.6.1 10279) [8] was used to build the spectrum library and lter peptides for PRM analysis. For each targeted protein, 2-6 associated peptides were selected using the following rules: (i) identi cation in the untargeted analysis with a q value <1%, (ii) complete digestion by trypsin, (iii) containing 8-18 amino acid residues, (iv) exclusion of the rst 25 amino acids at the N-terminus of proteins, and (v) xed carbamidomethylation of cysteine. Prior to individual sample analysis, pooled peptide samples were subjected to PRM experiments to re ne the target list. Finally, forty-four proteins with 255 peptides (Table S2) were scheduled. The RT segment was set to 8 min for each targeted peptide with its expected RT in the center based on the pooled sample analysis.
All of the PRM-MS data were processed with Skyline. By comparing the same peptide across runs, the RT location and integration boundaries were adjusted manually to exclude interfering regions. Each protein's intensity was quantitated using the summation of intensities from its corresponding transitions. The Work ow of this study.

Figure 2
Page 16/18 H&E staining assessment of neuronal morphology in the hippocampal CA1 area. A: The sham group showed an orderly arrangement of neurons with complete cell structure; B: The I/R 12 h group showed a reduced number of neurons, but their structure was complete, and there were no typical apoptotic cells; C: The I/R 48 h group showed a disorderly arrangement of cells, and the cytoplasm and nucleoli were stained deeply, indicating nuclear pyknosis. Scale bars=100 µm.

Figure 3
Functional analysis of differentially abundant proteins at 12 h and 48 h after I/R. A) Biological process category; B) Molecular function category; C) Cellular component category.