Progress Report — Final (Year 2 of 2)
Our LLF-funded project focused on developing and applying innovative methods to identify host- and microbiome-derived biomarkers of Alzheimer’s disease (AD) by characterizing host transcripts found in exfoliated epithelial cells (the exfoliome) and the host and microbial metabolome in stool samples from cognitively normal healthy, cognitively normal preclinical, and symptomatic AD individuals.
Characterizing the Host Exfoliome
We found that a targeted digital droplet PCR (ddPCR) approach provided superior quantification of host transcripts in stool compared to shotgun whole-exfoliome sequencing. We developed a pilot ddPCR panel targeting three AD-associated and three gut microbiome (GM)-associated transcripts, based on published literature and our own preliminary analyses. Applying this panel to age-, sex-, and APOE4-matched stool samples across disease stages, we observed that expression of three AD- and GM-related host genes was significantly altered:
- One AD-related marker was significantly decreased in preclinical AD, mirroring previously published observations in blood.
- A second marker — a significant AD risk factor — was overexpressed in preclinical AD stools, consistent with patterns observed in the brain.
- A marker of gut integrity showed decreased expression in preclinical AD; this marker is also associated with direct alterations in GM composition.
Identifying Metabolic Signatures of AD
We expanded our investigation to untargeted metabolomics of AD-related stool samples. Processing age-, sex-, and APOE4-matched samples, we detected over 7,000 metabolites across Healthy Tau-negative, Healthy Tau-positive, Preclinical Tau-negative, and Preclinical Tau-positive groups. Using Weighted Gene Co-expression Network Analysis (WGCNA) to cluster and correlate metabolites against amyloid and tau status, we identified 29 metabolite clusters (modules). Key findings:
- Several modules were significantly correlated with amyloid or tau status (p < 0.05); others trended toward significance (p < 0.1).
- The first amyloid-associated module (p = 0.003) contained metabolites previously reported to be depleted in AD.
- A second module — linked to metabolites shown to be increased in AD — trended toward significance (p = 0.093).
Future Directions
The exfoliome and host-GM metabolome data generated through LLF funding are now being leveraged to apply for NIH support to study a substantially larger cohort of cross-sectional and longitudinal AD-related stool samples. The promising exfoliomics results enabled independent funding to purchase a QX700E digital droplet PCR system, which automates sample processing and triples the number of probes usable in a single reaction — dramatically increasing throughput and quality.
We are collaborating with the Mass Spectrometry Access Center at the McDonnell Genome Institute (MTAC@MGI) to expand untargeted metabolic targets and develop targeted metabolomics assays. In parallel, we are integrating exfoliomic and metabolomic datasets with matched GM metagenomic and metatranscriptomic data to discover diagnostic stool biomarkers of AD using machine learning approaches.