APOE4 Genotype is Associated with Reduced Cortical VEGFR2 (KDR) Transcript Levels Independent of Endothelial Abundance: An AMP-AD RNA-seq Pilot Study

This AMP-AD pilot study reveals that APOE4 genotype is associated with a modest but significant reduction in cortical VEGFR2 (KDR) transcript levels independent of endothelial cell abundance and neuropathological burden, suggesting a specific APOE4-driven mechanism for vascular dysfunction in Alzheimer's disease.

The Gist

The Big Picture: A Genetic "Bad Apple" and a Leaky Fence

Imagine your brain is a bustling city. To keep this city running smoothly, it needs a very strong, secure fence around it called the Blood-Brain Barrier (BBB). This fence keeps the bad stuff out and the good stuff in.

The paper investigates a specific genetic trait called APOE4. Think of APOE4 as a "bad apple" in your genetic basket. If you have it, you are at a much higher risk for Alzheimer's disease. Scientists have long known that people with this "bad apple" tend to have leaky fences (a broken blood-brain barrier), but they didn't know exactly why.

This study asked a specific question: Does the "bad apple" cause the fence to fall apart because there are fewer fence-builders (cells), or because the fence-builders are just doing a worse job?

The Key Character: The Foreman (VEGFR2/KDR)

Inside the fence, there are specialized workers called endothelial cells. To keep the fence strong, these workers need a specific tool or "foreman" to tell them what to do. This foreman is a protein called VEGFR2 (encoded by the gene KDR).

• If the foreman is busy and loud: The fence is strong and tight.
• If the foreman is quiet or missing: The fence gets weak and leaky.

The Investigation: Counting the Workers vs. Checking the Tools

The researchers looked at brain tissue from 162 people (some with the "bad apple" APOE4, some without). They used a technique called RNA sequencing, which is like reading the "instruction manuals" inside the cells to see what they are trying to build.

They faced a tricky problem:

1. The "Cell Count" Confusion: Usually, if you find fewer "foreman" instructions (KDR), it might just mean there are fewer fence-builders (endothelial cells) in that area.
2. The "Efficiency" Question: Or, it could mean the fence-builders are still there, but they have stopped making the foreman tool.

To solve this, the researchers created a "Fence-Builders Score." They looked at five different markers that prove a cell is a fence-builder. This allowed them to mathematically separate "how many workers are there" from "how hard are they working."

The Findings: The Workers Are There, But They're Slacking

Here is what they discovered, broken down simply:

1. The "Bad Apple" doesn't fire the workers.
The study found that people with the APOE4 gene had the same number of fence-building cells as people without it. The "Fence-Builders Score" was identical. So, the "bad apple" isn't killing off the workers or causing them to leave the city.

2. The "Bad Apple" silences the foreman.
However, when they looked specifically at the instructions for the foreman (KDR/VEGFR2), they found something interesting. In people with APOE4, the instructions were about 10–15% quieter than in people without the gene.

3. It's a specific problem, not a general one.
The researchers checked other tools the workers use (like other proteins that hold the fence together). Most of them were fine. Only the foreman (KDR) and one other specific tool (TJP1) were quieter. This suggests the "bad apple" isn't causing a total meltdown of the construction site; it's specifically turning down the volume on the most important manager.

4. It happens regardless of how "dirty" the city is.
Alzheimer's is often associated with "trash" in the brain (amyloid plaques and tangles). The researchers checked if the "bad apple" only silenced the foreman when there was a lot of trash. They found that no, the foreman was quiet even in brains with very little trash. This means the "bad apple" causes this problem early on, before the disease gets really bad.

The Conclusion: A Broken Signal, Not a Missing Team

The Analogy:
Imagine a construction site where the crew (endothelial cells) is fully staffed. Everyone is present, and they are all wearing their hard hats. However, the site manager (VEGFR2) is whispering instead of shouting orders. Because the manager is quiet, the workers don't know how to tighten the bolts on the fence. The fence becomes leaky, not because the workers are gone, but because the signal to do their job is weak.

Why This Matters:
This changes how we think about treating Alzheimer's. Instead of trying to grow more blood vessels or replace lost cells, this study suggests we might need to find a way to turn up the volume on the VEGFR2 signal in people with the APOE4 gene. If we can help the "foreman" speak louder, we might be able to keep the brain's fence strong, even in people with this risky gene.

In short: The APOE4 gene doesn't remove the brain's protective cells; it just makes the most important manager of those cells go quiet, leading to a weaker barrier.

Technical Summary

1. Problem Statement

Apolipoprotein E4 (APOE4) is the strongest common genetic risk factor for late-onset Alzheimer's disease (AD) and is known to contribute to cerebrovascular dysfunction and blood-brain barrier (BBB) breakdown. Vascular Endothelial Growth Factor Receptor 2 (VEGFR2), encoded by the gene KDR, is a critical regulator of endothelial integrity and BBB maintenance.

While preclinical models suggest APOE4 alters vascular function, the specific relationship between APOE4 status and KDR transcript levels in human brain tissue remains unclear. A major confounding factor in bulk tissue analysis is distinguishing whether observed changes in gene expression are due to:

1. Cellular Composition: A reduction in the actual number of endothelial cells (cell loss).
2. Transcriptional Regulation: A functional downregulation of gene expression within existing endothelial cells.

This study aims to determine if APOE4 carrier status is associated with reduced KDR expression in human cortical tissue and whether this effect is independent of endothelial cell abundance.

2. Methodology

Data Source and Cohort:

• Dataset: Bulk RNA-sequencing data from the Mount Sinai Brain Bank (MSBB) cohort, part of the AMP-AD (Accelerating Medicines Partnership for Alzheimer's Disease) initiative.
• Sample Size: 162 donors providing 633 cortical samples across five regions: Parahippocampal gyrus (PHG), Frontal pole (FP), Inferior frontal gyrus (IFG), Prefrontal cortex (PFC), and Superior temporal gyrus (STG).
• Inclusion Criteria: Individuals with available APOE genotype and complete covariate data. Participants with the APOE ε2 allele were excluded to isolate ε4-specific effects.
• Groups: APOE4 carriers (ε3/ε4 or ε4/4) vs. Non-carriers (ε3/3).

Key Variables and Covariates:

• Primary Outcome: KDR (VEGFR2) transcript levels (log2 counts per million, logCPM).
• Endothelial Signal Proxy: A composite score derived from five canonical endothelial markers (PECAM1, VWF, CLDN5, FLT1, TEK) to estimate endothelial signal abundance within the bulk tissue.
• Covariates: Tissue region, age at death, sex, RNA Integrity Number (RIN), Post-Mortem Interval (PMI), and sequencing batch.
• Neuropathology: Braak stage, CERAD score, and quantitative plaque mean.

Statistical Analysis:

• Modeling: Linear mixed-effects models (LMM) using the lmer package in R.
  • Fixed Effects: APOE4 status, tissue region, endothelial composite score, and technical covariates.
  • Random Effects: Donor-level random intercepts to account for non-independence of multiple regional samples per donor.
• Interaction Analysis: Tested for interactions between APOE4 status and cortical region, as well as neuropathological burden (Braak, CERAD, plaque load).
• Candidate Panel: Extended analysis to other BBB/endothelial transcripts (CAV1, MFSD2A, CLDN5, SLC2A1, OCLN, TJP1).
• Dimensionality Reduction: Unsupervised UMAP embedding to visualize clustering by region and genotype.

3. Key Contributions

1. Disentangling Composition vs. Regulation: The study introduces a rigorous statistical approach using an endothelial composite score to distinguish between changes in endothelial cell proportion versus changes in per-cell transcriptional activity.
2. Specificity of APOE4 Effects: It demonstrates that APOE4 does not cause a global loss of endothelial cells but rather induces a selective transcriptional downregulation of specific vascular targets.
3. Independence from Pathology: The findings suggest that APOE4-driven vascular dysfunction occurs independently of amyloid and tau pathology severity.

4. Key Results

Endothelial Signal and Regional Heterogeneity:

• Bulk KDR expression varied significantly across cortical regions ($p < 2 \times 10^{-16}$).
• KDR expression was strongly correlated with the endothelial composite score ($r = 0.57, p < 2 \times 10^{-16}$), confirming that bulk signal is heavily driven by endothelial abundance.
• Crucially, APOE4 carrier status was not associated with the endothelial composite score ($p = 0.78$), indicating no difference in endothelial cell abundance between carriers and non-carriers.

APOE4 Association with KDR Expression:

• Without Endothelial Adjustment: APOE4 carriers showed a non-significant trend toward reduced KDR expression ($\beta = -0.16, p = 0.064$).
• With Endothelial Adjustment: After controlling for the endothelial composite score, APOE4 carriers showed a significant reduction in KDR expression ($\beta = -0.17, p = 0.029$).
• Magnitude: This corresponds to an estimated 10–15% decrease in KDR transcript levels in APOE4 carriers.
• Consistency: The effect was consistent across all cortical regions (no significant APOE4 $\times$ region interaction).

Neuropathology and Candidate Genes:

• Pathology Independence: No significant interactions were found between APOE4 status and Braak stage, CERAD score, or plaque mean. The KDR reduction is independent of AD neuropathological burden.
• Selectivity: Among the candidate panel of endothelial/BBB genes, only KDR and TJP1 (ZO-1) showed significant reductions in APOE4 carriers. Other markers (CAV1, MFSD2A, CLDN5, SLC2A1, OCLN) were unaffected.
• UMAP Analysis: While samples clustered strongly by cortical region, no distinct clustering by APOE4 genotype was observed after covariate adjustment, reinforcing that APOE4 does not induce a broad transcriptional shift across the entire endothelial program.

5. Significance and Conclusion

This study provides robust evidence that APOE4 contributes to cerebrovascular dysfunction through targeted endothelial transcriptional alterations rather than global endothelial cell loss.

• Mechanistic Insight: The findings support a model where APOE4 selectively downregulates VEGFR2 signaling within the endothelial compartment. Since VEGFR2 is vital for BBB maintenance and angiogenic signaling, its reduction (even without cell loss) likely compromises vascular integrity and contributes to the "leaky" BBB phenotype observed in AD.
• Therapeutic Implications: The independence of this effect from amyloid and tau pathology suggests that vascular dysfunction driven by APOE4 may be an early, independent pathway in AD pathogenesis. This positions VEGFR2 as a potential therapeutic target for vascular protection in APOE4 carriers, potentially effective even before significant neurodegeneration occurs.
• Methodological Impact: The study highlights the necessity of adjusting for cellular composition (via composite scores) in bulk RNA-seq studies of neurovascular diseases to avoid misinterpreting transcriptional regulation as cell loss.

Limitations: The study relies on bulk RNA-seq, which cannot resolve cell-type-specific signaling dynamics (e.g., pericytes vs. endothelial cells) or post-translational modifications (e.g., protein phosphorylation). Future proteomic studies are needed to validate these transcriptional findings at the protein level.