Multi-omics of human plasma reveals molecular features of dysregulated inflammation and accelerated aging in schizophrenia

Abstract

Schizophrenia is a devastating psychiatric illness that detrimentally affects a significant portion of the worldwide population. Aging of schizophrenia patients is associated with reduced longevity, but the potential biological factors associated with aging in this population have not yet been investigated in a global manner. To address this gap in knowledge, the present study assesses proteomics and metabolomics profiles in the plasma of subjects afflicted with schizophrenia compared to non-psychiatric control patients over six decades of life. Global, unbiased analyses of circulating blood plasma can provide knowledge of prominently dysregulated molecular pathways and their association with schizophrenia, as well as features of aging and gender in this disease. The resulting data compiled in this study represent a compendium of molecular changes associated with schizophrenia over the human lifetime. Supporting the clinical finding of schizophrenia's association with more rapid aging, both schizophrenia diagnosis and age significantly influenced the plasma proteome in subjects assayed. Schizophrenia was broadly associated with prominent dysregulation of inflammatory and metabolic system components. Proteome changes demonstrated increased abundance of biomarkers for risk of physiologic comorbidities of schizophrenia, especially in younger individuals. These findings advance our understanding of the molecular etiology of schizophrenia and its associated comorbidities throughout the aging process.

Fig. 1. Multi-omics analyses of plasma from schizophrenia (SZ) and control non-psychiatric comparison (NC) subjects by proteomics and metabolomics to assess circulating molecular alterations.

a Multi-omics mass spectrometry-based strategy for evaluating molecular profiles in human schizophrenia. Plasma from 54 SZ and 51 NC subjects ranging in ages 28–74 years was subjected to (i) proteomics and PTM analyses of trypsin-generated peptides subjected to TMT-labeling, fractionation, and mass spectrometry-based analysis, combined with (ii) metabolomics analyses for identification of small molecules, as described in the methods. Data on patient clinical features of inflammation (assayed by hs-CRP), BMI (body mass index), diabetes (HOMA-IR), triglycerides, and cholesterol were collected. Principal component analyses (PCoA) of metavariable influence of proteome data were conducted for (b) High-sensitivity C-reactive protein (hs-CRP) levels in plasma of subjects, (c) ages of subjects, (d) body mass index (BMI) of subjects, (e) triglyceride levels in plasma of subjects, (f) homeostatic model assessment for insulin resistance (HOMA-IR), and (g) cholesterol levels in plasma. h Significance measurements for metadata variable impact on PCoA distribution of proteome data. P values for categorical variables were measured using the PERMANOVA test for significance. P values for continuous variables were measured using Adonis test. Data are represented as -Log₁₀(p value) (*p value < 0.05; **p value < 0.01; ***p value < 0.001; dotted line threshold indicates p value < 0.05).

Fig. 2. Detection of metabolic dysfunction and inflammatory signatures in schizophrenia.

a Binary comparison of proteome data by volcano plot. The Log₂(SZ/NC) ratios of relative abundance for proteins illustrates positive values indicating upregulation in SZ and negative values indicating upregulation in the NC controls. Proteins highlighted in purple or blue showed significant relative abundance alterations with p value < 0.05. b Molecular function gene ontology analysis of upregulated and downregulated proteins in schizophrenia compared to NC controls. Functional protein systems that are positively and negatively associated with SZ are illustrated by the color-coded heat map key. c String-db network of significantly altered proteins in schizophrenia patients. Network analyses conducted by String-db illustrate significantly dysregulated SZ compared to NC proteins (assessed by Log₂(SZ/NC ratios), which include protein categories of complement, insulin-like growth factor binding proteins, and apolipoproteins. SZ and NC protein abundance profiles for (d) apolipoproteins, (e) insulin-like growth factor proteins, (f) complement proteins. g Binary comparison of PTM-enabled proteome data by volcano plot. The Log₂(SZ/NC) ratios of relative abundance for PTM-proteins with positive values indicate upregulation and negative values indicate downregulation in SZ vs. NC controls. Proteins highlighted in purple or blue showed significant relative abundance alterations with p value < 0.05. h Relative abundance of 10 PTM-proteins with highest relative modification frequency in SZ (purple) or NC (blue) subjects. i Differential distribution of detected PTM types in SZ vs. NC subjects. j Binary comparison of metabolome data shown by volcano plot. Metabolomics data was assessed by Log₂(SZ/NC) and -Log₁₀(p values), illustrating SZ-associated and NC-associated metabolites. It is noted that the antipsychotic drug clozapine and olanzapine were uniquely associated with the SZ group. k Classification distribution of metabolites associated with NC and SZ subjects. Shown are differential distributions of proportions of identified lipid molecules, benzenoids, organic acids and related, organoheterocyclic compounds, oxygen compounds, and nitrogen compounds. l Dysregulated lipid molecules in schizophrenia patients assessed by GNPS spectral networks. Spectral network shown demonstrates lipid-related molecules identified by Global Natural Product Social Molecular Networking (GNPS) and their association to SZ or NC subjects. Node outline indicate significance (p value < 0.05), with significantly dysregulated proteins indicated by arrows.

Fig. 3. Identifying molecular determinants of age-related disease risk using a machine learning strategy.

a Hierarchical clustering of proteome data for age-related categories in schizophrenia and healthy control subjects revealing clustering by schizophrenia status and age. Purple: schizophrenia; blue: healthy controls; dark green: < 40; mint green: 40–60; gray: > 60. Proteins associated death in ages groups of (b) < 40 years, (c) 40–60 years, and (d) > 60 years old. Log₂(SZ/NC) ratios of proteins associated with death at different age categories are illustrated. e Age-stratification of differentially expressed protein clusters in SZ and control NC subjects. The heatmap color key shows white for relative minimum value per protein, and relative maxima are indicated by cluster-specific colors for clusters 1–8.

Fig. 4. Targetable determinants of early morbidity in schizophrenia.

a Reactome functional analysis of proteins from cluster 4 represented as a treemap. b–e Representative disease risk biomarkers identified in cluster 4. f String interaction network of proteins from cluster 4 (top 5 most interconnected proteins are highlighted in red with all interaction partners). g Trend in adiponectin protein abundance at various ages.