Kininogen-1 as a protein biomarker for schizophrenia through mass spectrometry and genetic association analyses

Introduction

Schizophrenia (SCZ) is a complex psychotic disorder. It is among the top 25 mental illness in the world (Owen, Sawa & Mortensen, 2016; Vos et al., 2015). Although the prevalence of SCZ is not high (0.5–1.2%) (Huang et al., 2019), the health and economic burden is enormous for patients and their families as well as the society (Chong et al., 2014). To date, the etiology and pathogenesis of SCZ are still unknown. The diagnosis of SCZ is mainly based on descriptive and observation criteria to assess SCZ symptoms described by patients and their family members or observed by medical workers. So far, no biochemical markers or screening tools are available for SCZ diagnosis (Santa et al., 2017). Thus, it is important to identify SCZ biomarkers with high specificity and sensitivity, especially those present in body fluids which are easily obtained.

It is well known that SCZ is a genetic disorder. Candidate gene or genome-wide association studies have been applied to identify genetic factors for SCZ. Even though genes carrying SCZ-associated variants have been identified (Ghazaryan et al., 2016; Yoosefee et al., 2016), the mechanism by which how these genetic variants influence the vulnerability of individuals to SCZ is waiting to be explored. Since protein is the implementer of gene function, proteomic techniques such as mass spectrometry (MS) have been widely used in aiding in the diagnosis and identifying biomarkers of diseases including SCZ (Nascimento & Martins-De-Souza, 2015).

Mass spectrometry is an important technique for proteomics analysis. It separates and determines proteins according to the difference in mass-to-charge ratios (M/E) between different ions (Aebersold & Goodlett, 2001). A variety of MS techniques have been used in clinical application (Cho et al., 2015). For example, matrix-assisted laser desorption ionization/time-of-flight mass spectrometry (MALDI-TOF/MS), a sensitive analytical technique, has been increasingly used to evaluate the expression of specific proteins and peptides to detect biomarkers of diseases including psychiatric disorders using biological samples from patients and controls (Ding et al., 2015; Huang et al., 2017; Kelley, Perry & Bach, 2018; Lo, Shiea & Huang, 2016).

It is common that genes, proteins, and environment as well as the interaction among them lead to disease vulnerability. The exploration of disease biomarkers should conduct at multiple levels (Sokolowska et al., 2015). In the post-genomic era, genomics and proteomics studies are still two important areas that complement each other. Even though genetic variants may influence the structure and expression of proteins, variation in genes may not necessarily result in protein expression changes. However, changes in protein expression may reflect gene transcriptional. Recently, Comes et al. (2018) applied the proteomics approach in the identification of blood biomarkers for psychiatry disorders. They integrated proteomic findings from the peripheral blood and the genetic loci identified from enrichment analysis based on 30 published studies. With the availability of genomic data of a big sample the proteomic approach was demonstrated to be an essential method in discovery of biomarkers for disease including SCZ.

In present study, we initially used the MALDI-TOF/MS method to compare expression levels of serum proteins between SCZ patients and control subjects. We then screened candidate genes that code for differentially expressed proteins. Finally, we analyzed the association between polymorphisms of protein-coding gene and SCZ. Our aim was to explore etiology and diagnosis biomarkers of SCZ at both gene and protein levels.

Materials and Methods

Results

Comparison and identification of serum peptides

Peptide profiles of SCZ patients and healthy controls in training groups

Serum peptide profiles of 166 first onset SCZ patients and 201 healthy controls were analyzed. The mean age of 166 SCZ patients (67 males and 99 females) and 201 control subjects (90 males and 111 females) was 33 ± 13 and 36 ± 13, respectively. Serum peptide patterns of SCZ patients were significantly different from those of healthy controls (Fig. S1).

Selection and identification of SCZ-specific serum peptides

Through the t-test of ClinPro Tools software TM 2.1, with a P-value < 0.000001, the m/z 1,945.07 of peptides were significantly lower in patients compared to healthy controls (Fig. 1A). Using the peptide with m/z 1,945.07 to identify SCZ patients from healthy controls, the area under the curve of the receiver operating characteristic was 0.879 (Fig. 1B).

The ion of m/z 1,945.07 was confirmed to have the sequence NLGHGHKHERDQGHGHQ, which corresponded to Kininogen-1, with exactly the high molecular weight (HMW) of kininogen. The sequence map was shown in Fig. 2.

Peptide profiles of SCZ patients and healthy controls in testing groups

Class prediction analyses were conducted in another group of SCZ patients and healthy controls. There were 81 patients (35 males and 46 females) with mean age of 32 ± 13 and 103 healthy controls (45 males and 58 females) with mean age of 35 ± 10. The spectra analysis of the validation set classified 77 of the 81 SCZ patients as SCZ positive and 100 of the 103 healthy controls as SCZ negative, revealing 95.1% sensitivity and 97.1% specificity.

Discussion

Even though there have been some studies reported that Kininogen-1 was significantly reduced in SCZ patients, our study was conducted in northern Han Chinese population, and we explored if variants in its coding gene were associated with SCZ. We confirmed that Kininogen-1 was a serum protein biomarker of SCZ, and it was lowly expressed in SCZ patients. SNP rs2983640 in CST9, one of the coding genes of Kininogen-1, was significantly associated with SCZ. Additionally, the frequency of haplotypes carrying CST9 variants was significantly different between SCZ patients and healthy controls.

There are two types of Kininogens, low molecular weight kininogen and HMW kininogen (Lalmanach et al., 2010), in mammals. In 1999, Irina firstly reported the association between SCZ and the kallikrein–kinin system (KKS). He found that plasma KKS system activity was increased in SCZ patients, and kallikrein in the brain of patients was significantly reduced. This finding suggested that KKS might play a role in blood–brain barrier and brain tissue damage, leading to the development of SCZ (Shcherbakova et al., 1999). Kininogen-1 is an important component of KKS. Together with factor XII, kallikreins, kinins, and kininases, it is involved in the regulation of multiple organs and diverse pathophysiological processes such as cardiovascular, renal, nervous system, tumor, thrombosis, atherosclerosis and cell proliferation, inflammation, and apoptosis (Albert-Weissenberger, Siren & Kleinschnitz, 2013; Puchades et al., 2003; Regoli & Gobeil, 2015; Rhaleb, Yang & Carretero, 2011).

High molecular weight is a multifunctional glycoprotein in the plasma. It is synthesized by hepatocytes, and can be degraded to bradykinin (BK) under the action of kallikrein. BK receptors are also distributed in the central nervous system and play a role in diseases of the central nervous system (Chen et al., 2000). BK induces the synthesis and release of other inflammatory mediators in the central nervous system and the glial tissue, causing brain tissue damages or prolonged disturbance of blood–brain barrier function, which in turn can lead to brain damage (Ding-Zhou et al., 2002).

The production of kinins is influenced by the expression of kininogen. There was evidence about the mediatory role of kinins in the inflammatory response associated with different neurological disorders (Guevara-Lora, 2012). In neurological diseases, such as Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis, proteins involved in kinin generation or kinin receptor function were overexpressed (Khandelwal, Herman & Moussa, 2011; Lisak et al., 2011).

Kininogen is a member of the cystatin superfamily, which consists of type 1 cystatins (stefins), type 2 cystatins, and the kininogens. The results of this study indicated that compared with healthy controls, HMW was significantly reduced in SCZ patients. The deficiency of kininogen may lead the balance between proteases and protease inhibitors to be broken down in SCZ patients, resulting in the increase of homocysteine (Hcy) levels in the serum. Since Hcy is neurotoxic can affect the development of the central nervous system, it may be involved in the pathogenesis of SCZ. Several studies have shown that Hcy is significantly higher in SCZ patients than in healthy controls (Kale et al., 2010; Moustafa et al., 2014; Numata et al., 2015).

Kininogen is coded by the cystatin 9 gene (CST9), and cystatin 9 belongs to the Cystatins superfamily. It is one of 14 members of type 2 cystatins, which were found in human body compartments and fluids (Ochieng & Chaudhuri, 2010). In this study, we explored the association between three tagSNPs in CST9 and SCZ. Since age and gender of patients and controls were not well matched, we adjusted their confounding effect in the analysis. In the HWE test, SNP rs2983640 in the patient group did not meet the norm of HWE. This may be due to that rs2983640 is associated with SCZ. Findings from previous studies also indicate that SNPs that are not in HWE may contribute to complex diseases (Mayo, 2008). For rs2983640, the genotype distribution was significantly different between patients and controls under the genetic model of codominant, recessive, or overdominant. Haplotypes CG (rs2983639–rs2983640) and TCG (rs13037490–rs2983639–rs2983640) were found to be associated with SCZ. All these results imply that polymorphisms of CST9 were associated with the risk of SCZ. Our genetic association studies not only indicated the association between CST9 variants and SCZ, but also confirmed that the Kininogen-1 protein could be a biomarker for the diagnosis of SCZ.

However, there is still follow-up work waiting to be done in the future. For example, we need to conduct the study with a larger sample to ensure having sufficient statistical power. Subjects in this study were recruited in the northeast of China, and they were all Han Chinese. We need to conduct studies in subjects from other areas or ethnicity groups. We also need to study the effect of CST9 interaction with other genetic or environmental factors on the occurrence of SCZ. Finally, candidate biomarkers should be validated in the absence of interference from other diseases.

Conclusions

In summary, our study showed that using the MALDI-TOF/MS technique, Kininogen-1 could be a biomarker to distinguish SCZ patients from healthy controls with a high sensitivity and specificity. We also confirmed this at the gene level, that is, variants in CST9, the coding gene of Kininogen-1, were associated with the risk of SCZ. Both protein and genetic association studies demonstrated that Kininogen-1 could be a biomarker of SCZ. Definitely, our findings should be validated in a larger sample and replicated in different populations.

Supplemental Information