Intellectual disability risk gene RFX4 regulates cortical neurogenesis by restraining neuronal differentiation

This study elucidates the role of the intellectual disability risk gene RFX4 in human cortical development, demonstrating that it restrains neurogenesis by cooperating with NOTCH signaling to repress neuronal differentiation and synaptic gene expression, while also revealing distinct pathogenic mechanisms for deficiency versus specific DNA-binding domain mutations.

The Gist

Imagine the developing human brain as a massive, bustling construction site. The goal is to build a complex city (the cerebral cortex) with distinct neighborhoods, proper zoning laws, and a perfect balance of different types of buildings (neurons).

For this construction to succeed, you need a strict Project Manager who knows exactly when to stop the "building" phase and when to start the "finishing" phase. If the manager is too lazy or absent, the construction crew might start building houses before the foundation is even poured, leading to a chaotic, unstable city.

This paper is about a specific Project Manager named RFX4. Scientists discovered that RFX4 is a critical gene, and when it's broken, it causes intellectual disabilities and autism. Here is what they found, translated into everyday terms:

1. The "Brake Pedal" Discovery

For a long time, scientists thought RFX4 might be the gas pedal that helps neurons grow. This paper flips that script. They found that RFX4 is actually the brake pedal.

• The Analogy: Imagine a train (the neural stem cell) that needs to stay on the tracks as a "progenitor" (a builder) for a while before it can become a "passenger" (a mature neuron).
• The Finding: When RFX4 is working, it holds the train back, ensuring it doesn't rush off the tracks too early. When RFX4 is missing (Loss of Function), the train speeds up uncontrollably. The cells differentiate into neurons too fast, too soon. This "precocious neurogenesis" means the brain builds its layers in the wrong order, leading to a messy, non-functional city.

2. The Lineage-Specific Rules (The Two Construction Crews)

The brain has two main types of neurons: Excitatory (the accelerators) and Inhibitory (the brakes). The researchers found that RFX4 plays slightly different roles for these two crews:

• For Excitatory Neurons: RFX4 acts purely as a brake, stopping them from differentiating too early.
• For Inhibitory Neurons: RFX4 is even more critical. Without it, the construction crew doesn't just rush; they stop building the foundation entirely. The "neurospheres" (clumps of cells) shrink because the cells aren't proliferating (multiplying) correctly.

3. The Teamwork: RFX4, NOTCH, and RFX3

RFX4 doesn't work alone; it's part of a management team.

• The NOTCH Connection: The paper found that RFX4 works hand-in-hand with a signaling pathway called NOTCH. Think of NOTCH as the site foreman shouting, "Hold the line!" RFX4 is the manager who actually locks the gates to enforce that order. If you block the foreman (NOTCH), the gates open, and chaos ensues.
• The RFX3 Partner: There is another manager named RFX3. The study found a fascinating dependency: RFX3 cannot do its job unless RFX4 is there to introduce it. It's like RFX4 holding the door open for RFX3 to enter the control room. Once inside, they work together to manage the "synaptic" genes (the wiring of the city). However, if RFX3 is missing, the city still functions okay, but if RFX4 is missing, the whole system collapses.

4. The "Bad Copy" vs. The "Missing Copy"

This is one of the most surprising parts of the study.

• Scenario A (Missing Copy): If you delete the RFX4 gene entirely, the brake fails, and the cells rush to become neurons.
• Scenario B (The "Typos" Mutation): The researchers also looked at a specific mutation found in patients (a typo in the DNA code, p.R79C). They expected this to act just like a missing copy.
• The Twist: While this mutation did stop RFX4 from binding to DNA (like a broken key that won't fit the lock), the resulting chaos in the cell was completely different from the missing copy. It wasn't just a "broken brake"; it was a "broken engine." The cells didn't rush; they got confused in a unique way. This suggests that treating patients with this specific mutation might require a different medical approach than treating those with a total gene deletion.

5. The Aftermath: A Messy City Layout

The researchers grew tiny, 3D "mini-brains" (organoids) in a dish to see what happens when RFX4 is broken.

• The Result: The mini-brains tried to build their layers, but because the cells rushed the process, the layers got mixed up. The "deep layer" buildings were built on top of the "surface layer" ones.
• The Takeaway: Even after the cells finally became neurons, the damage was done. The "memory" of that rushed construction stayed with them, leading to neurons that couldn't form proper connections (synapses) or organize themselves correctly.

Summary

In simple terms, this paper tells us that RFX4 is the strict supervisor that ensures brain cells wait their turn.

1. It stops cells from becoming neurons too early.
2. It works with other managers (NOTCH, RFX3) to keep the construction site organized.
3. If RFX4 is missing, the city is built too fast and in the wrong order.
4. If RFX4 has a specific "typo" mutation, it breaks the system in a weird, unique way that isn't just about being "missing."

This discovery is huge because it moves us from just knowing which gene causes these disorders to understanding how it causes them. This is the first step toward designing drugs that can either fix the "brake" or help the construction crew get back on track.

Technical Summary

1. Problem Statement

The development of the human cerebral cortex is a tightly regulated process, and its disruption leads to neurodevelopmental disorders (NDDs) such as intellectual disability (ID) and autism spectrum disorder (ASD). A central hypothesis for NDD etiology involves an imbalance between excitatory (cEX) and inhibitory (cIN) neuronal activity, often stemming from dysregulated cortical neurogenesis.

While RFX4 was recently identified as a risk gene for NDDs, its specific role in human cortical development remained unclear. Previous studies offered conflicting evidence: some suggested RFX4 overexpression stabilizes progenitors, while others suggested it induces differentiation. Furthermore, it was unknown whether RFX4 functions differently in excitatory versus inhibitory lineages, how it interacts with other signaling pathways (like NOTCH) or transcription factors (like RFX3), and whether pathogenic missense mutations act via simple haploinsufficiency or distinct mechanisms.

2. Methodology

The authors utilized a comprehensive suite of human pluripotent stem cell (hPSC) models to dissect RFX4 function:

• Genetic Models:
  • Loss of Function (LOF): Generation of heterozygous (HET) and homozygous (HOM) RFX4 knockout hPSC lines (using CRISPR-Cas9) in both embryonic stem cells (WA09) and induced pluripotent stem cells (AN1.1).
  • Pathogenic Mutation: Engineering of a patient-specific missense mutation (p.R79C) in the DNA-binding domain of RFX4 (HET and HOM models).
  • Control Models: Generation of RFX3 HOM knockout lines to study interplay between RFX family members.
• Differentiation Protocols:
  • Lineage Specification: Differentiation of hPSCs into dorsal telencephalic progenitors (D-NPCs, modeling excitatory neurons) and ventral telencephalic progenitors (V-NPCs, modeling inhibitory neurons).
  • Direct Differentiation: Generation of induced glutamatergic-like neurons (iGluts) via Neurogenin-2 (NGN2) overexpression to bypass the progenitor stage.
  • 3D Organoids: Generation of dorsal-patterned cortical organoids (D-ORGs) to assess tissue-level stratification.
• Omics and Molecular Assays:
  • CUT&Tag: To map endogenous RFX4 and RFX3 genome-wide binding sites.
  • RNA-Sequencing (RNA-seq): To profile transcriptomic changes in WT vs. mutant models across different lineages and timepoints.
  • ChIP/Epigenetics: Analysis of histone modifications (H3K4me, H3K27me3) and NOTCH inhibition effects.
  • Functional Assays: Luciferase reporter assays for cis-regulatory element (CRE) activity, siRNA knockdown of interacting factors (SOX2, HES5), and unbiased drug screens (using $\gamma$-secretase inhibitors).
  • Phenotyping: Immunocytochemistry for proliferation (KI67), progenitor markers (SOX2, PAX6), and neuronal markers (ASCL1, TBR1, CTIP2, TUJ1, SYN1, PSD95).

3. Key Contributions and Results

A. RFX4 Restrains Neurogenesis in a Dosage-Dependent Manner

• Phenotype: RFX4 LOF (HET and HOM) caused a significant reduction in neurosphere outgrowth and a dose-dependent increase in neurogenesis markers (ASCL1, DCX) in both D-NPCs and V-NPCs.
• Lineage Specificity: While RFX4 LOF reduced proliferation in V-NPCs (inhibitory lineage), proliferation remained unchanged in D-NPCs (excitatory lineage), indicating lineage-specific roles.
• Mechanism: RFX4 acts as a transcriptional repressor. LOF led to the upregulation of pro-neuronal and synaptic genes (e.g., SYT5, CAMK2A) and downregulation of early developmental genes.

B. Interaction with NOTCH Signaling and Transcription Factors

• NOTCH Pathway: RFX4 functions downstream of NOTCH signaling. Inhibition of NOTCH (via DAPT) phenocopied RFX4 LOF but also induced terminal differentiation, whereas RFX4 LOF alone did not.
• Co-factors: RFX4 cooperates with SOX2 and HES5 (NOTCH effectors) to restrain neurogenesis. Knockdown of HES5 upregulated deep-layer neuron markers (TBR1, CTIP2) in WT but not in RFX4 LOF models, suggesting RFX4 and HES5 function in the same pathway to prevent premature differentiation.
• RFX3 Interplay: RFX3 binding is contingent on RFX4. In RFX4 LOF models, RFX3 genome-wide binding was lost, particularly at synaptic gene promoters. Conversely, RFX3 LOF did not affect RFX4 binding. While they co-occupy synaptic enhancers, RFX4 represses these genes in progenitors, while RFX3 appears to activate them later in differentiation.

C. Pathogenic Missense Mutation (p.R79C) Acts Distinctly from LOF

• Binding: The p.R79C mutation (in the DNA-binding domain) caused a near-complete loss of RFX4 genome binding, similar to the HOM knockout.
• Transcriptomic Divergence: Despite similar binding loss, the p.R79C mutation caused a unique transcriptomic profile distinct from the LOF model.
  • LOF: Upregulated neurogenic/synaptic genes; increased ASCL1+ cells.
  • p.R79C: Downregulated neuronal/synaptic genes; no increase in ASCL1+ cells or neurosphere outgrowth.
• Conclusion: The pathogenicity of p.R79C is not due to simple haploinsufficiency but likely involves dominant-negative effects or altered protein-protein interactions (e.g., heterodimerization) that disrupt transcriptional regulation differently than total loss of the protein.

D. Persistent Effects on Cortical Stratification

• Progenitor Requirement: Transcriptomic disruptions caused by RFX4 LOF in D-NPCs persisted into mature cortical excitatory neurons (cEXs) but were absent in iGluts (which bypass the progenitor stage). This confirms the defect originates in the progenitor stage.
• Organoid Defects: In 3D cortical organoids, RFX4 LOF led to precocious neurogenesis (increased TUJ1+ area) and disrupted cortical layering. Specifically, there was an overproduction of early-born deep-layer neurons (TBR1+, CTIP2+) and a failure to properly stratify later layers, mimicking the E/I imbalance seen in NDDs.

4. Significance

• Mechanistic Insight: This study redefines RFX4 as a critical repressor of neurogenesis that maintains the progenitor state by cooperating with NOTCH signaling effectors (HES5, SOX2).
• Lineage Specificity: It highlights that RFX4 has distinct roles in excitatory vs. inhibitory lineage development, particularly regarding proliferation control in inhibitory progenitors.
• Pathogenic Complexity: The work challenges the assumption that all RFX4 mutations act via haploinsufficiency. The p.R79C mutation demonstrates that DNA-binding domain mutations can cause unique, non-haploinsufficiency mechanisms of disease, necessitating mutation-specific therapeutic strategies.
• Developmental Etiology: The findings support the hypothesis that NDDs associated with RFX4 arise from a "developmental window" error where unconstrained neurogenesis leads to persistent transcriptional dysregulation and cortical malformation, rather than just functional deficits in mature neurons.

This research provides a foundational framework for understanding RFX4-associated disorders and suggests that restoring the balance of neurogenesis (rather than just replacing the protein) may be a viable therapeutic avenue.