Human iPSC Models of Ganglioside Deficiency Reveal a Sialylated Lipid Requirement for Plasma-Membrane Organization and Neuronal Activity

Using human iPSC-derived cortical neurons, this study reveals that while both ST3GAL5 and B4GALNT1 deficiencies eliminate major gangliosides, only ST3GAL5 loss triggers a fatal reprogramming of the lipid repertoire that depletes plasma membrane proteins and abolishes neuronal activity, whereas B4GALNT1 deficiency is compensated by precursor sialylated lipids that preserve membrane organization and electrical function.

The Gist

The Big Picture: The Brain's "Lipid Language"

Imagine your brain's nerve cells (neurons) are like a bustling city. For this city to function, the buildings (cells) need to talk to each other. They do this by sending electrical signals, like traffic lights and phone calls, across the streets.

To make these signals work, the surface of every cell is covered in a special "skin" made of fats called lipids. Among these fats, a specific group called gangliosides acts like the signposts, road signs, and docking stations on the cell's surface. They help organize the cell's machinery, ensuring that the "traffic lights" (ion channels) and "phones" (receptors) are in the right place to send and receive messages.

This study looked at what happens when the city loses its ability to make these specific signposts. The researchers used human stem cells grown into brain cells to model two rare genetic diseases:

1. GM3SD: A severe condition causing early-life epilepsy and developmental delays.
2. HSP26: A condition causing muscle stiffness and weakness later in life.

Both diseases involve a broken "factory" that makes these gangliosides, but the researchers discovered something surprising: The city can survive if it has any signposts, but it crashes completely if it loses the specific "sialylated" ones.

---

The Two Scenarios: A Tale of Two Cities

The researchers created two types of "broken" cities in the lab to see how they reacted.

1. The HSP26 City (The "Simple Signpost" Scenario)

In this scenario, the factory stops making the complex, fancy gangliosides. However, the city still has a backup supply of simple, short signposts (called GM3 and GD3).

• The Result: The city is a bit messy, but the traffic lights still work. The neurons can still fire electrical signals and talk to each other.
• The Analogy: Imagine a city where the fancy, colorful billboards are gone, but the basic, white street signs are still there. Drivers can still find their way, even if the city looks a bit plain. This explains why patients with HSP26 have a milder disease that starts later in life; the simple signposts are doing enough work to keep the city running for a while.

2. The GM3SD City (The "Wrong Signposts" Scenario)

In this scenario, the factory stops making the gangliosides and the simple backup signposts. Instead, the cell starts making completely wrong types of fats (called globo- and o-series lipids). These are fats usually found on red blood cells or stem cells, not brain cells.

• The Result: The city goes into chaos. The electrical signals stop. The neurons cannot fire in sync, and the "traffic" grinds to a halt.
• The Analogy: Imagine the city replaces all its street signs with traffic cones, construction barriers, and random debris. The drivers (electrical signals) have no idea where to go. The city becomes gridlocked. This explains why GM3SD is so severe, causing epilepsy and developmental arrest almost immediately after birth.

The "Why": The Sialic Acid Key

The researchers asked: Why does the HSP26 city survive while the GM3SD city collapses?

They found the answer in a tiny chemical tag called sialic acid.

• The simple signposts in the HSP26 city still have this tag.
• The wrong signposts in the GM3SD city do not have this tag.

The Discovery: The "sialic acid" tag is the universal key that unlocks the cell's surface organization. As long as the cell has some fats with this key (even the simple ones), the cell's machinery stays organized. Without it, the machinery falls apart.

The Mechanism: The "Magnet" Effect

The study went deeper to see why the signals stopped in the GM3SD city. They looked at the proteins on the cell surface (the actual traffic lights and phones).

• In the healthy city: The sialic acid fats act like magnets. They hold the traffic lights and phones in the right spots on the cell surface so they can work.
• In the GM3SD city: Without the sialic acid magnets, the traffic lights and phones fall off the wall. They get lost inside the cell or thrown in the trash.
  • The cell loses its ion channels (the switches that let electricity flow).
  • It loses its receptors (the antennas that catch chemical messages).
  • It loses its glue (proteins that hold synapses together).

It's not that the cell stopped making these parts; it's that without the sialic acid "glue," the parts couldn't stay attached to the surface. The cell became a ghost town of missing equipment.

The Takeaway

This paper solves a medical mystery: Why are two diseases caused by broken lipid factories so different?

• HSP26 is like a city with a shortage of fancy decorations but still has the basic infrastructure. It struggles, but it functions.
• GM3SD is like a city that lost its entire infrastructure because it replaced the essential "glue" with the wrong materials. The system collapses.

The Conclusion: The brain doesn't just need any fat; it specifically needs sialylated fats (fats with the sialic acid tag) to hold the electrical machinery in place. If you lose that tag, the brain's electrical network falls apart, leading to severe epilepsy and developmental failure. This gives scientists a clear target for future treatments: we need to find a way to restore that specific "sialic acid key" to the brain's surface.

Technical Summary

1. Problem Statement

Gangliosides are abundant glycosphingolipids (GSLs) critical for brain development and function, yet their precise roles in human neuronal physiology remain poorly defined. Loss-of-function mutations in two "gatekeeper" enzymes of ganglioside biosynthesis cause severe but clinically distinct neurological disorders:

• ST3GAL5 deficiency (GM3SD): Causes GM3 synthase deficiency, leading to ultra-rare, early-onset infantile epilepsy, profound intellectual disability, and developmental stagnation.
• B4GALNT1 deficiency (HSP26): Causes Hereditary Spastic Paraplegia 26, characterized by later-onset (childhood/adolescence) progressive spasticity and weakness.

Both mutations theoretically eliminate the four predominant neuronal gangliosides (GM1a, GD1a, GD1b, GT1b). However, existing murine models fail to recapitulate the severe human phenotypes (particularly the early-onset epilepsy in GM3SD), and human cell models (fibroblasts, HEK293) lack the specific neuronal lipid environment. The study aims to define how specific ganglioside repertoires support human neuronal function and explain the divergent clinical severity between these two disorders.

2. Methodology

The researchers utilized an isogenic human iPSC-derived neuronal system (i3N) to create precise disease models.

• Cell Models: They generated CRISPRi-mediated knockdown (KD) cell lines targeting ST3GAL5 (modeling GM3SD) and B4GALNT1 (modeling HSP26). They also created double-knockdown lines (e.g., ST3GAL5 + A4GALT or ST3GAL5 + B4GALNT1) to test the role of non-neuronal lipid accumulation.
• Differentiation: Stem cells were differentiated into mature cortical glutamatergic neurons using doxycycline-induced NGN2 expression, reaching functional maturity by day 28.
• Lipidomics: To overcome the limitations of Thin-Layer Chromatography (TLC), they employed a quantitative enzymatic approach. Lipids were extracted, ceramide tails were cleaved by endoglycoceramidase, and glycan headgroups were fluorescently labeled (2-anthranilic acid) and quantified via HPLC.
• Electrophysiology: Multi-Electrode Array (MEA) recordings were used to monitor spontaneous electrical activity, spike counts, network bursts (NWB), and synchrony (Area Under Normalized Cross-Correlation, AUNCC) from day 14 to day 35.
• Proteomics: To investigate membrane organization, they performed Plasma Membrane Proteomics (PMP) using aminooxy-biotin labeling of intact cells to enrich surface proteins, followed by Mass Spectrometry (DIA-MS). Whole-cell proteomics (WCP) was used as a control to distinguish between mislocalization and synthesis defects.

3. Key Results

A. Divergent Lipid Repertoires

• $\Delta$B4GALNT1 (HSP26 model): Loss of the enzyme led to the accumulation of its direct substrates, GM3 and GD3 (simple sialylated GSLs). The lipid profile remained dominated by sialylated species (>80%).
• $\Delta$ST3GAL5 (GM3SD model): Loss of the enzyme caused a massive accumulation of LacCer (lactosylceramide). This precursor was shunted into non-neuronal pathways, generating globo-series (Gb3, Gb4, Gb5) and o-series (GA2, GA1, GM1b, GD1$\alpha$) lipids. Consequently, these neurons were dominated by non-sialylated lipids, with a drastic reduction in total sialylated GSLs.

B. Electrophysiological Phenotypes

• $\Delta$ST3GAL5 Neurons: Showed a severe deficit in electrical activity. They failed to establish synchronized network bursts, had significantly reduced spike counts, and exhibited poor synchrony (low AUNCC).
• $\Delta$B4GALNT1 Neurons: Maintained near-normal electrical firing and network synchronization, despite the loss of complex gangliosides.
• Rescue Experiments: Removing the accumulated non-neuronal lipids (globo/o-series) in $\Delta$ST3GAL5 neurons via double KD did not rescue the electrical phenotype. This confirmed that the loss of sialylated lipids, rather than the gain of non-neuronal lipids, drives the dysfunction.

C. Plasma Membrane (PM) Proteome Remodeling

• $\Delta$ST3GAL5: Proteomics revealed a severe reduction in the abundance of PM proteins, including ion channels, GPCRs, and synaptic organizers. Crucially, whole-cell proteomics showed no such reduction, indicating these proteins are synthesized but mislocalized (failed to traffic to or retained at the surface).
  • Key affected proteins: Ion channels (HCN2, KCNH7, KCNA3, CLCN4, HTR3A), GPCRs (CHRM4, CXCR4, KISS1R, etc.), and adhesion molecules (NRXN3, LRRTM3).
• $\Delta$B4GALNT1: The PM proteome remained largely intact, explaining the preserved electrical function.

4. Key Contributions

1. Mechanistic Explanation for Phenotypic Divergence: The study establishes that the presence of simple sialylated GSLs (GM3/GD3) is sufficient to maintain PM organization and neuronal excitability, whereas the accumulation of non-sialylated lipids (globo/o-series) cannot compensate. This explains why GM3SD (total loss of sialylation) is catastrophic, while HSP26 (retention of GM3/GD3) is less severe.
2. Human-Specific Pathology: It highlights a critical species difference: Human GM3SD neurons accumulate non-sialylated lipids, whereas mouse models do not. This explains why mouse models fail to recapitulate the severe human epilepsy phenotype.
3. PM Organization as a Sialic Acid-Dependent Process: The paper provides evidence that sialylated lipids act as essential scaffolds for the stable localization of critical neuronal proteins (channels, receptors, adhesion molecules) at the plasma membrane.
4. Secondary Channelopathy/Synaptopathy: The authors propose that GM3SD is not just a metabolic disorder but a secondary channelopathy caused by lipid-dependent protein mislocalization.

5. Significance

• Therapeutic Insight: The findings suggest that pharmacological interventions targeting specific ion channels or GPCRs may be ineffective in GM3SD if the proteins themselves are absent from the cell surface. The only viable therapeutic strategy proposed is gene therapy to restore ST3GAL5 expression and reconstitute the sialylated ganglioside repertoire.
• Disease Modeling: The study validates human iPSC-derived neurons as superior models for studying GSL disorders compared to mouse models, particularly for understanding species-specific metabolic pathways and severe neurodevelopmental phenotypes.
• Broader Implications: The mechanism of sialic acid-dependent protein retention may be relevant to other neurodegenerative conditions (e.g., Alzheimer's, Parkinson's) where membrane microdomain organization is disrupted.

In summary, this paper demonstrates that sialylation is the critical determinant for neuronal membrane organization and electrical activity. The loss of sialylated lipids leads to a collapse of the plasma membrane proteome, resulting in the severe neurological phenotype observed in GM3SD patients.