2025-05-26

098_ Uncovering the role of circulating tRNA-derived fragments in schizophrenia pathophysiology in the PsyCourse study

Research Question and Aims

Background and scientific rationale:
Over the past few years, the potential of circulating small non-coding RNAs (sncRNAs) as diagnostic and mechanistic biomarkers in neuropsychiatric disorders has gained significant attention. Prof. Fischer’s group, in collaboration with the PsyCourse consortium, has contributed substantially to this area through high-throughput small RNA sequencing of blood samples from a large cohort of schizophrenia (SCZ) patients and matched healthy controls. This work resulted in our recent EMBO publication (2024)1 in which we found a dysregulated miRNA; miR-99b-5p; as a main pathogenic regulator of synaptic and immunological dysfunction in schizophrenia. We showed using both human and mouse models that miR-99b-5p targets Zbp1 to orchestrate microglial engulfment of synapses and its modification in vivo generates schizophrenia-relevant behavioral and molecular abnormalities.
Driven by the coherence of our results and the high quality of the generated dataset, we now suggest to expand this effort by investigating a new and overlooked class of small non-coding RNAs: tRNA-derived fragments (tRFs). These RNA species, which arise from precisely cleaved mature or precursor tRNAs, represent a significant proportion of the small RNAome in both blood and brain and are now increasingly recognized as functional regulatory elements with implications in cell stress, immunity, and neural function2-3. Despite their prevalence, tRFs remain virtually unstudied in the context of schizophrenia, and the PsyCourse dataset offers a unique opportunity to explore this RNA class at an unprecedented scale.

Scientific objectives:
Our goal is to fully annotate and quantify tRFs, correlate their expression with disease status and phenotypic data, and find new RNA biomarkers or mechanistic regulators of schizophrenia by using the small RNA-seq dataset previously generated from PAXgene blood samples at Visit 1 of the PsyCourse cohort.

Key hypothesis:
We outline the key hypotheses that we aim to test using the PsyCourse phenotypic data, structured around three broad but interconnected dimensions of schizophrenia pathophysiology: (1) cognitive dysfunction, (2) clinical severity and symptom domains, and (3) early life stress and immunological vulnerability. Each hypothesis is grounded in both existing literature and our preliminary findings from small RNA-seq profiling of schizophrenia blood and brain samples, as detailed in our recent EMBO paper (PMID: 38528182)1 and in pilot analyses of tRF expression. Experimental design and functional validation plan of the project are provided in the figure below (see Figure 1).

Analytic Plan

Preliminary data and feasibility:
We found in a pilot study of postmortem tissue samples (prefrontal cortex A9&24) small RNA-seq dataset (n=13; SCZ, 17 controls) that tRFs significantly contribute to the small RNA content. With a majority of downregulated fragments, especially those generated from glycine and glutamate tRNAs, many of the tRFs shown marked differential expression in schizophrenia. These pilot results underline the functional complexity of tRF biogenesis and its importance in disease-relevant processes.

The specific objectives of this proposed work include:
Profiling the tRFome in schizophrenia and control blood samples
We will catalog all detectable tRF species in the ~15–40 nucleotide range using a revised bioinformatics pipeline, which includes adapter trimming, variant-tolerant mapping to GtRNAdb, and tRF-specific annotation algorithms implemented through the MINTmap tool (https://cm.jefferson.edu/mintmap/). The goal is to produce a high-confidence map of circulating tRFs and identify differentially expressed fragments between schizophrenia patients and controls, correcting for age, sex, RNA integrity, and batch effects.

Correlation of tRF expression with detailed phenotypic traits
We seek to study how individual or co-expressed tRF modules relate to cognitive function and disease severity using the extensive clinical metadata already available in PsyCourse—including PANSS, CGI, GAF, TMT-A/B, Digit Span, VLMT, and medication history. This will be carried out using co-expression network analysis (WGCNA) and principal component regression.

Cross-modal integration with miRNA and mRNA regulatory networks
We plan to integrate tRF expression patterns with our previously published miRNA data and publicly available schizophrenia mRNA datasets (e.g., PsychENCODE, CommonMind). Our preliminary results from postmortem brain samples indicate that tRFs co-cluster with gene modules associated with synaptic plasticity and neuroimmune pathways, mirroring the gene signatures disrupted in SCZ. Identifying whether these signatures are mirrored in peripheral blood will advance the field toward non-invasive biomarkers and mechanistic readouts.

Functional analysis of tRFs in schizophrenia pathways (in silico and in vitro)
Emphasizing the top candidate tRFs that is, the most dysregulated and abundant fragments, we will find their likely mRNA targets and biological functions. We will identify mRNAs with complementary sites using in silico target prediction algorithms (tailored for tRF sequences) and integrate this with mRNA expression data for inverse correlation patterns. For high-confidence targets, we will then validate direct tRF-mRNA interactions using luciferase reporter assays. Finally, we will regulate tRF levels (with synthetic tRF mimics or antisense inhibitors) utilizing neuronal and microglial cell models to observe effects on predicted target gene expression and cellular phenotypes. Our goal with these perturbation experiments is to show that tRFs play causal roles in controlling pathways related to schizophrenia (like neuroinflammatory signaling or synaptic transmission). This provides functional evidence between tRF dysfunction and schizophrenia pathogenesis (see figure 1).

Data access:
We generated this sncRNAs sequencing dataset from PAXgene-collected whole blood samples prepared with NEBNext protocols. Originally described for characterizing circulating miRNAs, the dataset is also naturally well-suited for in-depth study of tRFs. Comprehensive quality control, normalizing, and miRNA-focused differential expression analyses performed in our lab have validated the integrity of the dataset and its relevance for more general small RNA discovery projects. We are now requesting access to the raw and processed small RNA sequencing data for Visit 1 of the PsyCourse cohort, encompassing all available 244 schizophrenia patients and 331 matched healthy controls.
To enable replication and validation of the discovered tRF signatures, we also plan to include, where relevant, data from the second longitudinal small RNA-seq dataset and the upcoming total RNA-seq dataset (n=352), which includes significant sample overlap with the current selection.

Resources needed

v1_id
v1_stat
v1_id
v1_stat
v1_school
v1_prof_dgr
v1_ed_status
v1_cntr_brth
v1_outpat_psy_trm
v1_age_1st_out_trm
v1_daypat_inpat_trm
v1_age_1st_inpat_trm
v1_dur_illness
v1_1st_ep
v1_tms_daypat_outpat_trm
v1_cat_daypat_outpat_trm
v1_fam_hist
v1_height
v1_chol_trig
v1_hyperten
v1_ang_pec
v1_heart_att
v1_stroke
v1_diabetes
v1_hyperthy
v1_hypothy
v1_osteopor
v1_asthma
v1_copd
v1_allerg
v1_neuroder
v1_psoriasis
v1_autoimm
v1_cancer
v1_stom_ulc
v1_kid_fail
v1_stone
v1_epilepsy
v1_migraine
v1_parkinson
v1_ever_smkd
v1_age_smk
v1_alc_pst12_mths
v1_lftm_alc_dep
v1_evr_ill_drg
v1_scid_dsm_dx
v1_scid_dsm_dx_cat
v1_scid_age_MDE
v1_scid_no_MDE
v1_scid_age_mania
v1_scid_no_mania
v1_scid_age_hypomania
v1_scid_no_hypomania
v1_scid_ever_halls
v1_scid_ever_delus
v1_scid_ever_psyc
v1_scid_evr_suic_ide
v1_suic_attmpt
v1_scid_no_suic_attmpt
v1_nrpsy_mwtb
v3_cts_1
v3_cts_2
v3_cts_3
v3_cts_4
v3_cts_5
v3_cts_els_dic
v3_cts_1_dic
v3_cts_2_dic
v3_cts_3_dic
v3_cts_4_dic
v3_cts_5_dic
visit
age
Antidepressants
Antipsychotics
bdi2_sum
bmi
gaf
idsc_sum
marital_stat
Mood_stabilizers
no_suic_attmpt
nrpsy_com
nrpsy_dg_sym
nrpsy_dgt_sp_bck
nrpsy_dgt_sp_frw
nrpsy_tmt_A_err
nrpsy_tmt_A_rt
nrpsy_tmt_B_err
nrpsy_tmt_B_rt
nrpsy_vlmt_check
nrpsy_vlmt_corr
nrpsy_vlmt_lss_d
nrpsy_vlmt_lss_t
nrpsy_vlmt_rec
panss_sum_gen
panss_sum_neg
panss_sum_pos
panss_sum_tot
smRNAome_id
weight

Variables such as:
TMT-A/B,
Digit Span (forward/backward),
Digit Symbol Test,
Intelligence score (v1_nrpsy_mwtb)
Rationale:
Cognitive impairment is a core feature of schizophrenia. Distinct tRF expression patterns in blood reflect underlying synaptic dysfunction in schizophrenia, manifesting as correlations with patients’ cognitive deficits and negative symptom severity. In particular, we hypothesize that differentially expressed tRFs (DE-tRFs) (or clusters of tRFs) will be associated with worse performance on neuropsychological tests of cognition (e.g. attention and executive function on the Trail Making Test, working memory on Digit Span, or verbal learning on the VLMT), possibly reflecting tRF-mediated regulation of synaptic plasticity genes. Prior studies have shown that small RNAs, including miRNAs, influence neuronal maturation and synaptic integrity1. Emerging evidence suggests tRFs also play roles in translational repression and synaptic homeostasis.

Variables such as:
PANSS positive/negative/general/total scores
Global Assessment of Functioning (GAF)
Disease duration
Medication use (antipsychotics, antidepressants, mood stabilizers)
Rationale:
We hypothesize that patients with more severe illness, as indicated by higher total symptom burden (PANSS) or lower global functioning (GAF scores), will display distinct multi-tRF expression patterns that could serve as biomarker fingerprints of disease state. These tRF signatures are expected to correlate with cross-sectional clinical ratings (e.g. PANSS total) and possibly with longitudinal outcomes. The premise is that circulating tRFs not only mirror specific pathways (e.g. immune and synaptic) but also converge to reflect the integrated clinical phenotype of a patient. If certain tRFs track core pathology, their collective dysregulation should be greatest in those with the most pronounced illness manifestations. We will identify tRF modules (via co-expression network analysis or principal component analysis) and test their associations with global clinical measures (symptom severity and psychosocial functioning). Also, we will account for medication and physical health variables to ensure these associations are driven by disease processes rather than extrinsic factors. Antipsychotic medications can influence gene expression and immune-metabolic profiles; thus, we will include patients’ medication history and metabolic health metrics in our models. For example, we might find a specific cluster of tRFs (e.g. downregulated tRFs related to neuronal/synaptic genes coupled with upregulated immune-related tRFs) that is strongly associated with poor global functioning regardless of treatment status. Such a result would underscore the potential of tRF profiles as stable peripheral indicators of central pathology.