Circulating neuron-derived cfDNA for blood-based detection of Alzheimer's and other neurodegenerative conditions.

Blood-based biomarkers for neurodegenerative diseases are improving early detection and staging, but current assays primarily reflect aggregate neuropathology or generalized neuronal injury and do not resolve the specific neuronal populations affected. Circulating cell-free DNA (cfDNA) retains stable DNA methylation patterns reflective of tissue and cellular origin, making it a promising substrate for cell-of-origin analysis. However, conventional methylation approaches are limited by bisulfite-associated DNA damage and amplification-related bias, hindering the detection of neuron-derived cfDNA, a small fraction of total circulating cfDNA. Here, we present proof-of-concept evidence that native nanopore sequencing can support both brain methylation atlas generation and downstream cfDNA cell-of-origin classifier development in neurodegenerative disease. By directly profiling endogenous DNA methylation without bisulfite conversion or PCR amplification, nanopore sequencing preserves native molecules, reduces processing-related bias, and enables flexible, genome-wide methylation profiling that can be iteratively expanded as additional reference cell types are incorporated. Using whole-genome native nanopore sequencing, we generated a methylation reference atlas from six primary human neural cell populations-cortical neurons, dopaminergic neurons, spinal motor neurons, astrocytes, Schwann cells, and microglia-and developed cell-type-informed cfDNA classifiers. Classifier performance was assessed in silico using dilution series designed to model physiologic admixture. The framework was then applied to 137 blood plasma samples from individuals with mild cognitive impairment (MCI), Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), and healthy controls. Elevated circulating cfDNA fragments exhibited methylation patterns similar to reference profiles from selectively vulnerable neuronal populations, including cortical neuron-like signatures in AD and progressive MCI, dopaminergic neuron-like signatures in PD, and spinal motor neuron-like signatures in ALS. Multivariate integration of neuronal signatures improved the separation of diagnostic groups within this cohort (AUC > 0.85). Although the reported atlas is limited and additional validation in larger and independent cohorts will be required, these results support the feasibility of native cfDNA nanopore methylation sequencing as a flexible platform for brain-derived cfDNA analysis and more cell-type-informed investigation of neurodegeneration from peripheral blood.