Interactome mapping in human excitatory neurons reveals novel risk genes and pathways in Alzheimer's disease

This study introduces ADNeuronNet, a comprehensive, neuron-specific protein-protein interactome map derived from human iPSC-derived excitatory neurons, which reveals novel AD risk gene interactions and uncovers a previously unknown BIN1-APC/C-APOE regulatory axis that modulates Tau aggregation.

The Gist

The Big Picture: Finding the Missing Puzzle Pieces

Imagine Alzheimer's disease as a massive, broken machine. Scientists have spent years finding the broken gears (genes) that cause the machine to fail. They have a list of about 100 "suspect" genes that are linked to Alzheimer's.

However, knowing which gears are broken doesn't tell us how the machine breaks down. A gear doesn't work in isolation; it turns other gears, pulls levers, and connects to belts. To understand the disease, we need to see how these broken gears interact with the rest of the machine.

The Problem: Until now, scientists have only been able to map these interactions in "generic" cells (like factory workers in a standard office) or in yeast (a simple single-celled organism). But Alzheimer's happens in the human brain, specifically in a type of cell called an excitatory neuron. These cells are like the VIPs of the brain—they do the heavy lifting for thinking and memory. The interactions happening inside them are unique and were previously invisible to scientists.

The Solution: This paper introduces a new map called ADNeuronNet. It is the first detailed "social network" map of how Alzheimer's risk genes talk to each other specifically inside human neurons.

---

How They Did It: The "VIP Club" Analogy

Think of the human brain as a giant, bustling city. Most previous studies tried to understand the city's traffic by looking at a generic town or a simulation.

1. Growing the VIPs: The researchers took stem cells (blank slate cells) and turned them into human excitatory neurons. Think of this as growing a specific, high-security VIP club where only the brain's most important workers hang out.
2. The "Bait" Strategy: They took 57 of the known Alzheimer's risk genes and gave them a glowing "GFP tag" (like a neon name badge).
3. The Fishing Trip: They dropped these tagged genes into the neuron club. When a tagged gene (the bait) reached out to grab a friend (a protein it interacts with), they caught the whole group.
4. The Snapshot: They used a high-tech microscope (Mass Spectrometry) to take a snapshot of everyone who was holding hands with the bait.

The result? A map of 1,767 interactions between 1,189 proteins. Crucially, 1,375 of these interactions were brand new discoveries that no one had ever seen before because they only happen in neurons.

---

The Big Discoveries: Three New Secrets

The map didn't just show old connections; it revealed three major new secrets about how Alzheimer's works.

1. The "Neuron-Only" Friend (RIN2)

One of the biggest suspects in Alzheimer's is a gene called BIN1.

• The Old View: Scientists knew BIN1 had a friend named RIN3 that helped it move around inside the cell.
• The New Discovery: In this new map, they found BIN1 has a different friend named RIN2.
• The Catch: RIN2 is like a celebrity who only shows up to VIP parties (neurons). It doesn't exist in the generic cells used in previous studies. That's why no one found it before!
• What it does: The researchers showed that RIN2 acts like a magnet, pulling BIN1 to a specific spot in the cell (the early endosome) where it helps process Amyloid Beta (the sticky plaque in Alzheimer's). This suggests RIN2 is a new key player in the disease.

2. The "Switch" That Breaks the Machine (BIN1 and APC/C)

The map showed that BIN1 also connects to a massive machine called the APC/C complex.

• The Analogy: Imagine the APC/C is a security guard that decides which proteins get thrown out of the cell and which stay.
• The Discovery: When the researchers broke the APC/C connection (by turning it off), the cell started making too much APOE.
• The Chain Reaction: APOE is the most famous genetic risk factor for Alzheimer's. The study found that when the APC/C guard is missing, APOE goes wild, which causes Tau tangles (the other major hallmark of Alzheimer's) to clump together.
• The Takeaway: They found a new highway: BIN1 → APC/C → APOE → Tau Tangles. This explains how a problem with BIN1 can lead to Tau tangles, linking two major parts of the disease.

3. The "Mutation" Effect (How Tiny Changes Break the Network)

The researchers also tested what happens when these genes have tiny typos (mutations) found in patients.

• The Analogy: Imagine a handshake between two people. If one person changes their grip slightly (a mutation), the handshake might become too tight, too loose, or fall apart completely.
• The Discovery: They found that specific mutations in genes like RIN3 or different versions of BIN1 (isoforms) completely rewired who they held hands with.
  • For example, a specific mutation in RIN3 made it drop the hand of a protein called CD2AP. This drop breaks the chain of events needed to clear out bad proteins, potentially speeding up the disease.

---

Why This Matters

Think of this paper as upgrading from a black-and-white sketch of a city to a 3D, color-coded, real-time map of the actual city where the crime is happening.

• Before: We knew the suspects (genes) but didn't know who they were meeting or what they were planning.
• Now: We have a map of their secret meetings inside the actual crime scene (human neurons).

This map helps scientists:

1. Find new drug targets (like RIN2 or the APC/C complex) that they didn't know existed.
2. Understand why certain mutations make the disease worse.
3. Finally connect the dots between the genetic "suspects" and the actual "crime" (brain cell death).

In short, by looking at the right place (human neurons) with the right tools, they found the missing links that explain how Alzheimer's actually destroys the brain.

Technical Summary

1. Problem Statement

Alzheimer's Disease (AD) is a complex neurodegenerative disorder driven by amyloid-beta plaques and Tau tangles. While Genome-Wide Association Studies (GWAS) and sequencing projects have identified ~100 genetic risk loci, the molecular mechanisms linking these risk genes to disease pathology remain poorly understood.

• The Gap: Existing large-scale human protein-protein interaction (PPI) datasets (e.g., BioPlex, OpenCell) were generated in generic cell lines (HEK293T, HCT116) or yeast. These datasets lack cell-type specificity, missing critical neuronal interactions that occur exclusively in the brain.
• The Challenge: AD risk genes often function in a polygenic, context-dependent manner. Without a neuron-specific interactome, it is impossible to map how risk proteins converge on specific pathways (e.g., synaptic dysfunction, Tau propagation) within the actual cell type most vulnerable to AD: glutamatergic excitatory neurons.

2. Methodology

The authors developed a highly scalable, high-throughput pipeline to map the interactome specifically in human induced pluripotent stem cell (iPSC)-derived excitatory neurons.

• Bait Selection:
  • Compiled 137 high-confidence AD risk genes (from ADSP, GWAS, and ExWAS).
  • Filtered for neuronal expression (97 genes) and successfully cloned 57 into lentiviral vectors with GFP tags.
  • Included specific AD-associated isoforms (e.g., neuronal BIN1 isoform 1 vs. ubiquitous isoform 9) and patient-derived mutations (e.g., APOE alleles, RIN3 mutations).
• Cell Model & Differentiation:
  • Used the i3N iPSC line (doxycycline-inducible Neurogenin-2) to rapidly differentiate into functional glutamatergic cortical neurons.
  • Employed lentiviral transduction followed by Fluorescence-Activated Cell Sorting (FACS) to generate stable, GFP-positive neuronal lines for each bait protein.
• Mass Spectrometry Workflow:
  • Immunoprecipitation (IP): Performed on GFP-tagged baits in neurons (quadruplicates).
  • Quantification: Utilized a Data-Independent Acquisition (DIA) with Label-Free Quantification (LFQ) workflow on a Bruker timsTOF HT mass spectrometer. This was optimized to provide >3x proteome coverage compared to previous TMT-based methods, using <50% of input material.
• Data Analysis:
  • Applied an empirical Bayes statistical framework (limma) to identify high-confidence interactors (Log2FC > 1, adjusted p < 0.1).
  • Generated atomic-resolution structural models for all interactions using AlphaFold3 (AF3) to predict direct binding interfaces.
  • Conducted "perturbation analysis" to quantify interaction changes caused by specific mutations/isoforms.

3. Key Contributions

• ADNeuronNet: The creation of the first comprehensive, neuron-specific PPI map for AD risk proteins, containing 1,767 high-confidence interactions among 1,189 proteins.
• Neuron-Specific Discovery: The dataset is the only one significantly enriched for genes expressed in neurons but absent in standard cell lines (HEK293T/HCT116), validating the necessity of cell-type-specific mapping.
• Structural Validation: Integration of AlphaFold3 modeling to distinguish between co-complex associations and direct physical bindings, providing a structural basis for novel interactions.
• Mechanistic Insights: Uncovered specific regulatory axes linking major AD risk genes (BIN1, APOE) to Tau pathology through novel interactors (RIN2, APC/C complex).

4. Key Results

A. Quality and Specificity of ADNeuronNet

• Enrichment: ~95% of interactors are neuron-expressed. Crucially, ADNeuronNet is the only dataset significantly enriched for neuron-specific genes (not found in HEK293T/HCT116).
• Disease Relevance: Interactors showed higher overlap with CRISPRi hits associated with Tau pathology and upregulated genes in AD patient brains (ROSMAP snRNA-seq data) compared to literature-based interactomes.
• Structural Confidence: AF3 modeling showed ADNeuronNet interactions had significantly higher interface TM-scores (ipTM) than random pairs or OpenCell interactions, confirming high-quality predictions.

B. Novel Neuron-Specific Interactions

• BIN1 and RIN2: Identified RIN2 (a RAB5 GEF) as a novel interactor of the neuronal BIN1 isoform.
  • Significance: RIN2 is not expressed in HEK293T cells, explaining its absence in previous studies.
  • Function: Co-expression of BIN1 and RIN2 recruits BIN1 to RAB5-positive early endosomes, a key site for APP processing and Aβ generation.
• BIN1 and APC/C: Identified interactions between BIN1 and subunits of the Anaphase-Promoting Complex/Cyclosome (APC/C) (e.g., ANAPC1, ANAPC2).
  • Validation: Confirmed via endogenous co-IP in human neurons.

C. Isoform and Mutation Perturbation

• BIN1 Isoforms: The neuronal isoform (BIN1v1) uniquely interacts with the AP-2 complex and Clathrin (via the CLAP domain) and mitochondrial TIM/TOM complexes, whereas the ubiquitous isoform (BIN1v9) does not.
• RIN3 Mutation (W63C): A pathogenic mutation in RIN3 was found to significantly weaken its interaction with CD2AP (a key AD protein) while maintaining BIN1 binding. This disruption likely impairs APP sorting and degradation.

D. The BIN1-APC/C-APOE Regulatory Axis (Major Discovery)

• Mechanism: The study revealed that loss of APC/C function (via knockdown of ANAPC1/2) leads to:
  1. Upregulation of APOE (specifically neuronal APOE).
  2. Increased Tau aggregation (seeding-induced MC1 inclusions).
• Rescue: Depleting APOE in APC/C-deficient neurons rescued the Tau aggregation phenotype.
• Conclusion: This establishes a novel BIN1–APC/C–APOE axis where APC/C dysfunction drives Tau pathology via aberrant APOE induction, linking two of the most significant AD risk genes (BIN1 and APOE).

5. Significance

• Paradigm Shift: Demonstrates that studying AD mechanisms in generic cell lines misses critical, disease-relevant interactions that only occur in neurons.
• Therapeutic Targets: Identifies new candidates (e.g., RIN2, APC/C subunits) and pathways (APOE regulation) that could be targeted to halt Tau propagation.
• Resource: ADNeuronNet serves as a foundational resource for the community to integrate genetic data with functional networks, moving beyond statistical associations to mechanistic understanding.
• Scalability: The pipeline proves that high-throughput interactome mapping in human neurons is feasible, paving the way for mapping other brain cell types (e.g., microglia) and multicellular organoid systems.

In summary, this paper bridges the gap between AD genetics and molecular pathology by providing a cell-type-specific interactome that reveals how risk genes like BIN1 and APOE converge on Tau pathology through previously unknown neuronal mechanisms.