Related Experiment Video
Updated: Aug 6, 2026

Using Live Cell STED Imaging to Visualize Mitochondrial Inner Membrane Ultrastructure in Neuronal Cell Models
Published on: June 30, 2023
Mitochondrial protein alterations in vascular dementia: evidence from Mendelian randomization, transcriptomics, and a
Qian Liu1,2, Huizhong Tan2, Keke Tong1
1Graduate School of Hunan University of Chinese Medicine, Changsha, Hunan, China.
Objective:
Mitochondrial dysfunction is a key pathological feature of vascular dementia (VaD), yet the specific systemic and localized proteins involved remain unclear. We aim to utilize Mendelian randomization (MR) of plasma proteomics and Gene Expression Omnibus (GEO) datasets to identify systemic circulating biomarkers associated with VaD risk, and subsequently use in vivo models to investigate whether these systemic signals correspond to actual neuropathological mitochondrial changes in the brain.
Methods:
We first identified candidate proteins associated with VaD risk through a two-sample MR analysis of plasma proteomics and GWAS data. Candidate genes were then assessed for differential expression using the GEO dataset GSE122063. Finally, the proteins were validated in a Bilateral Common Carotid Artery Occlusion (2-VO) rat model by evaluating pathological features and measuring its expression levels via Western Blot.
Results:
Mendelian randomization analysis identified four proteins nominally linked to VaD: protective factors (AIFM1, COX5B) and risk factors (NDUFV2, NUDT5). However, cross-referencing these genetic predictions with GEO transcriptomics (GSE122063) and a 2-VO rat model revealed distinct tiers of evidentiary support. COX5B emerged as the most robust targets, demonstrating unidirectional consistency across all three analytical layers. NUDT5 showed partial consistency, supported by genetic and animal data, though its transcriptomic alteration fell short of the threshold. Conversely, AIFM1 and NDUFV2 displayed clear directional contradictions between the genetic/transcriptomic data and actual in vivo protein expression. Notably, while donepezil improved VaD pathology, it did not alter the expression of these proteins.
Conclusion:
Rather than universally validating all four genetic candidates, this rigorous multi-layer triangulation specifically pinpoints the dysregulation of COX5B as high-confidence, consistent drivers of mitochondrial impairment in VaD. Acknowledging the inconsistent complexities of AIFM1, NDUFV2, and NUDT5. Therapeutic strategies targeting the cleanly triangulated protein offer a more reliable, disease-modifying approach for VaD intervention.

