2024-04-24

083_ Exploring the regulatory potential of non-coding RNAs on the plasma lipidome

Research Question and Aims

Lipidomic profiling is essential for elucidating the lipid variations in the blood plasma of schizophrenia (SCZ) patients, potentially paving the way for the early identification of disease markers (Tkachev et al., 2023). Non-coding RNAs related to severe mental health disorders like SCZ have also been described in, for example, mouse models (Kaurani et al., 2024). Interestingly, recent advancements indicate that certain microRNAs, like miR-122 and miR-133, act as regulators of lipid metabolism (Elmén et al., 2008; Goedeke et al., 2015). Additionally, genetic studies have linked variations in the proximity of miRNA genes with unusual lipid levels in the bloodstream (Goedeke et al., 2015). For instance, miR-148a and miR-128-1 influence important regulatory proteins involved in lipid processing, including the receptors for low-density lipoprotein (LDLR) and the ATP-binding cassette A1 (ABCA1) (Wagschal et al., 2015; Shi et al., 2015). Multicohort studies indicate that variations in genes related to lipid metabolism coincide with shared genetic markers for SCZ, major depressive disorder (MDD), and bipolar disorder (BPD), and suggest a genetic connection to lipid-related phenotypes (Tabassum et al., 2019; Tkachev et al., 2023). However, how these profiles are brought about is not yet understood. The influence of non-coding RNAs (ncRNAs) on lipid metabolism in healthy individuals but also individuals with severe mental health disorders is largely unclear (Zhang et al., 2020). This study aims to understand the effect of circulating ncRNAs on the plasma lipidome. In a secondary analysis, it will attempt to understand the genetic determinants of any ncRNA-regulatory effects of the plasma lipidome identified and how and if potential lipid-regulatory effects related to ncRNAs are related to lipidomic changes observed in mental health disorders. With regard to severe mental health disorders like SCZ or BPD, the identification of potential underpinnings of the reported alteration of lipidomic profiles in these disorders, could also provide valuable insight into its pathogenesis and potential paths for diagnosis and treatment.

Analytic Plan

According to our hypothesis, circulating ncRNAs can serve as regulatory factors for the plasma lipidome. This level of biological regulation could also be related to the altered plasma lipidomic profiles present in individuals with SCZ or BPD. To investigate the association between miRNA expression and lipid metabolism in the PsyCourse study, we will integrate the existing plasma lipidomic data of 623 PsyCourse individuals (controls, SCZ, and BPD) with existing ncRNAome sequencing data from the same individuals. All analyses will be performed using corresponding lipidomic and ncRNAomeSeq data from the same (mostly the initial) visit. The main objective of this approach is to elucidate ncRNA-regulatory effects on lipid metabolism.

Stage 1: Data Preparation and Quality Control for ncRNAome Sequencing data: We will first extract ncRNA data from .fastq files of the PsyCourse dataset corresponding to the same individuals and the same visits as are present in the lipidomic dataset. Objectives of this phase are quality assurance in order to precisely determine count of ncRNAs sequences as well as put up a robust foundation for subsequent analyses. We will implement miRDeep2 pipeline (Friedländer et al., 2012) to extract the reads from .fastq files followed by quality control, mapping the reads to the reference genome, and ncRNA differential expression analysis with each diagnostic group and controls.
Should count matrices of qc-ed data be available by other means by the start of the project (e.g. from other PsyCourse proposals like “Assessment of miRNA-eQTLs in the PsyCourse Study”) these will be used to ensure a homogeneous database for all analyses within PsyCourse that use the same data. QC-ed, log2 transformed and normalized plasma lipidomic data already exist for all individuals.

Stage 2: ncRNA and Lipidomics Integration: Clinical and demographic data will be used to identify the potential covariates that affect the relationship between ncRNA expression and lipid levels. Linear regression models will be applied to correlate ncRNA expression with lipids by incorporating the identified covariates to adjust the model (Timmons et al., 2018). Spearman correlation function will be beneficial to compute the strength of association between ncRNAs and lipid species (Tan et al., 2023). Further, we will utilize Bonferroni or Benjamini-Hochberg correction methods to address the multiple testing issues inherent in data analysis (Mengelkoch et al., 2023). The functional role of ncRNAs will be assessed using the Viper tool by linking their regulatory potential on lipid metabolism and mental health disorders (Carceller et al., 2023). This integrative approach will help us to understand the interactive effects between the plasma lipidome and the ncRNAome. It also holds potential to reveal biological markers for future study and contribute to understand the molecular mechanisms of SCZ and BPD.

Stage 3: Understanding the genetic determinants underlying ncRNA-related lipid alterations: Should ncRNAs that drive lipid species levels be identified, we will try to determine underlying genetic determinants in the form of ncRNA-QTLs. We will use the full visit1 ncRNAome sequencing dataset from PsyCourse (not just those that also have lipidomic data) to perform a genome-wide association studies (GWAS) for expression level of the ncRNAs of interest as a continuous trait. Should existing ncRNAQTL data already be available from the full PsyCourse data by the time we reach this point, we will, for the sake of homogeneity in data analysis, use existing ncRNA-QTL information. LipidQTLs have already been calculated for PsyCourse participants as part of another project and will be used herein. We will conduct mediation analysis to evaluate whether lipidQTLs mediate the ncRNAs regulatory effects on lipid profiles. By identifying overlapping and non-overlapping QTL and lipidQTLs signals we will understand shared genetic determinants of ncRNA and lipid levels in the participating individuals in general, independent of disease status. Considering the limited power, the known batch-effects in the currently existing ncRNASeq data as well as the limited disease specificity of other QTLs (e.g. Yang C et al, 2021), the identification of mental health-related QTLs is not the primary aim of the project depicted herein. Stage 4: PheWAS: Upon successful identification of genomic loci in Stage 3, we will perform a phenome-wide association study (PheWAS) to explore the possible connections between these loci and specific traits or diseases. This comprehensive exploration will be facilitated by the PheWAS catalog (https://phewascatalog.org), which will improve the interpretation of the identified QTLs and their potential implications for mental health disorders. This approach will contextualize the genetic findings within the broader spectrum of phenotypic outcomes, which enhances our understanding on the genetic foundation of lipidomics and ncRNAomics in health and disease.

Resources needed

v1_id
v1_stat
v1_sex
v1_age
v1_yob
v1_cur_psy_trm
v1_outpat_psy_trm
v1_age_1st_out_trm
v1_age_1st_inpat_trm
v1_dur_illness
v1_fam_hist
v1_bmi
v1_ever_smkd
v1_scid_dsm_dx
v1_scid_dsm_dx_cat
v1_med_pst_wk
gsa_id
v1_smRNAome_id
v2_smRNAome_id
v3_smRNAome_id
v4_smRNAome_id