Mechanistic Hypotheses and Experimental Research Directions

How to read this page

Seven candidate mechanisms, each with the experiments that could test it. Every claim carries a label saying what kind of thing it is — and three of these hypotheses are labelled honestly as having no EEF1A2-specific support that we could find.

• Established

  A textbook-level biological mechanism. True, but not about this gene.

• EEF1A2 evidence

  Published experimental or clinical evidence in EEF1A2 itself. Note which variant, and in which system.

• Related system

  Evidence from another protein, pathway or organism. Does not transfer automatically.

• Computational

  A preliminary observation from simulation or prediction. Not an experimental result.

• Hypothesis

  Reasoned but untested. Requires experimental validation.

A · Protein instability and degradation

The question: does the D91N substitution change how much functional eEF1A2 a cell has, by altering folding, half-life or susceptibility to degradation?

Cells monitor protein folding continuously and route failures to the ubiquitin–proteasome system; protein half-life is measurable, and a shortened half-life for a mutant protein is one of the most common ways a missense variant causes disease. Established

For this gene there is a measurement that makes the hypothesis concrete and, at the same time, warns against simplifying it. In the E122K knock-in mouse, the mutant protein was relatively stable in neurons yet highly unstable in skeletal myocytes — the same substitution, the same animal, opposite stability in two tissues. The authors concluded that the homozygous phenotype was driven by loss of function in muscle. They also reported that motor abnormalities emerged far earlier in these homozygotes than in null mice, which they read as suggesting a toxic gain of function and/or a dominant-negative effect. EEF1A2 evidence

For D91N specifically, the only data we have found come from yeast. When the equivalent substitution was expressed in yeast eEF1A, protein levels were significantly reduced and fell further without the chaperone Chp1. Related system Whether this holds for human eEF1A2 in neurons is unknown. We have found no half-life measurement and no degradation assay for this variant. Hypothesis

Experiments that would test it

• Cycloheximide chase to measure mutant versus wild-type half-life — in a cell type that actually expresses eEF1A2.
• Western blot quantification of steady-state abundance, with an isoform-discriminating antibody.
• Proteasome inhibition: if the mutant is being degraded, blocking the proteasome should restore it.
• Thermal stability assays on recombinant protein, which separate a folding defect from a clearance defect.

B · Translation elongation dysfunction

The question: does altered eEF1A2 function impair elongation itself — tRNA delivery, GTP hydrolysis, elongation rate — and does that produce ribosomal pausing?

The mechanism of elongation and the consequences of a stalled ribosome — collisions, disome formation, and ribosome-associated quality control — are established cell biology. Established

Here the gene-specific evidence is comparatively strong. Functional analysis of the recurrent patient variants G70S, E122K and D252H found decreased de novo protein synthesis and a decreased elongation rate in cell culture. EEF1A2 evidence Independently, in a yeast plasmid-shuffling assay, every variant tested except the hypomorphic control impaired growth under conditions reporting on protein synthesis. EEF1A2 evidence

One further published observation sharpens the question rather than settling it: a knock-in model of E122K was reported to show transcript-selective disturbance rather than a global synthesis deficit. If elongation defects in this disorder are selective — falling hardest on particular transcripts, such as long ones — then global measures of protein synthesis are the wrong instrument. We flag this as a direction to verify in the primary literature rather than a conclusion. Hypothesis

Experiments that would test it

• Ribosome profiling, to locate pausing at codon resolution.
• Disome profiling, which specifically detects collided ribosomes.
• Polysome profiling, for the coarse question of how loaded the mRNAs are.
• Translation reporter assays, including length-matched reporters to test transcript selectivity directly.
• Biochemical GTPase and aminoacyl-tRNA binding assays on purified wild-type and D91N protein — the measurement that would tell us what this variant actually does.

C · Cellular stress responses

The question: if translation is disturbed, do stress pathways activate secondarily — and is that protective, harmful, or merely a marker?

The integrated stress response, eIF2α phosphorylation and ATF4 signalling are established. So is ribosome-associated quality control, and so is the existence of a distinct ribotoxic stress response. Established

There is real gene-specific evidence touching this, and it should be read carefully. In the cohort study, variants were described as disrupting both the protein synthesis and the integrated stress response functions of eEF1A2; concretely, in tef2Δ yeast challenged with rapamycin, all variants except the hypomorphic control failed to rescue growth, while steady-state protein levels were similar to wild type under those conditions. EEF1A2 evidence

What that is: a stress-condition growth phenotype in yeast, with abundance controlled for — which is a meaningful result, because it separates “less protein” from “protein that cannot handle stress”. What it is not: a measurement of ISR activation in a human neuron. The eIF2α/ATF4 axis has not, as far as our search reached, been measured in an EEF1A2 model. Hypothesis

For the ribotoxic stress response and for ribosome-associated quality control specifically, we found no EEF1A2-specific experimental evidence at all. Machinery that recognises and degrades eEF1A stuck on blocked ribosomes has been described in the general translation literature, which makes this a well-posed question with named molecules — but it is a question about a related system, not a finding about this disorder. Related system

Experiments that would test it

• Phospho-eIF2α and ATF4 measurement in an appropriate neuronal model, at baseline and under challenge.
• Transcriptional readout of ISR target genes.
• Markers of ribosome-associated quality control and collision-dependent signalling.
• Challenge experiments rather than resting-state measurement: if the phenotype is a loss of reserve, it will only appear under load.

D · Cytoskeletal dysfunction

The question: does altered eEF1A2 disturb actin organisation, and through that neuronal morphology and synaptic function?

eEF1A proteins interact with actin; the C-terminal domain III is the region implicated. Established

Gene-specific evidence exists: functional work on the recurrent variants reported altered actin bundling and altered neuronal morphology alongside the translation defect. EEF1A2 evidence And at the genotype–phenotype level, the largest cohort study reports that milder phenotypes tend to involve the actin-binding function while severe ones cluster in switch II. EEF1A2 evidence

That last point cuts against this hypothesis for D91N rather than for it. D91N lies in domain I adjacent to switch II — the severe, GTP-cycle side of that division — not in the domain III actin region. A direct actin-binding defect is therefore not the most natural expectation for this particular variant, though a downstream morphological consequence remains entirely possible. Hypothesis

Experiments that would test it

• In vitro actin bundling assays with purified wild-type and D91N protein — the direct test.
• Morphometry in differentiated neurons: dendritic arbour, spine density, growth cone behaviour.
• Synaptic assays, once a model with the right isoform background exists.

E · Mitochondrial dysfunction

The question: are mitochondrial abnormalities a secondary consequence — through energy demand, oxidative stress or proteostatic load?

Neurons are energetically expensive, protein synthesis is among the most energy-consuming cellular processes, and proteostatic stress and mitochondrial function are coupled in many systems. Established

One practical note for anyone tempted to test this in accessible tissue: because eEF1A2 expression is concentrated in brain, heart and skeletal muscle while cultured fibroblasts are dominated by the paralogue eEF1A1, a normal mitochondrial workup in fibroblasts would be falsely reassuring. That is a methodological point, not clinical advice.

Experiments that would test it

• Respirometry and ATP measurement in a cell type that expresses eEF1A2.
• Oxidative stress markers, with isogenic controls.
• Mitochondrial membrane potential and morphology.
• Before any of these: establish whether there is anything to find, since no prior evidence points here.

F · Compensatory mechanisms and isoform switching

The question: can eEF1A1 compensate for impaired eEF1A2, and could that be influenced?

The developmental handover is documented: eEF1A2 replaces the ubiquitously expressed eEF1A1 as neurons develop, becoming the predominant factor. EEF1A2 evidence And the two are close enough that cultured cells expressing eEF1A1 confound functional work on eEF1A2.

The attraction is obvious: a second gene already encodes a nearly identical protein. The difficulties are equally concrete, and they are why this is listed as a hypothesis rather than a strategy: Hypothesis

• The direction of the switch is away from eEF1A1. Maintaining or re-inducing it means opposing a developmental program, in the tissue where it is most firmly established.
• If compensation were straightforward, the disorder would be milder than it is. The severity of the published phenotypes is itself evidence that eEF1A1 does not substitute adequately in mature neurons.
• The two paralogues are not functionally interchangeable — the existence of a dedicated, developmentally timed switch argues that the difference matters.
• Mechanism-dependence. Adding more total factor helps if the problem is too little functional protein. If the problem is a harmful mutant protein, raising the total pool is at best neutral. This is the unresolved question from the homepage, and it determines the sign of this entire hypothesis.

A relevant data point for the dosage side of that question: heterozygous Eef1a2-null mice are normal. Losing one copy is tolerated in mouse — which is interesting precisely because the human missense variants are not tolerated.

Experiments that would test it

• Quantify both paralogues across neuronal maturation in a human model.
• Test whether raising eEF1A1 rescues anything, in a model with the correct isoform background.
• Compare neuronal and muscle models directly, since the E122K mouse showed these diverge.

G · eIF5A and hypusination

The question: could eIF5A-dependent translation mechanisms be relevant to the downstream consequences of altered elongation?

eIF5A is required for efficient translation of certain difficult sequence contexts, and it is the only known protein to carry a hypusine residue. Hypusination is a two-step modification by DHPS and DOHH, and it depends on spermidine as the aminobutyl donor. Established

Experiments that would test it

• Establish whether there is any interaction at all: hypusination status and eIF5A dependence in an EEF1A2-variant model.
• Genetic rather than pharmacological manipulation first — it answers the mechanistic question without the confound of a systemically active compound.
• If and only if an interaction exists: determine the sign of the effect before anything else.

What follows from all of this

Across seven hypotheses, the pattern is consistent: the EEF1A2-specific evidence that exists concerns other variants, in yeast, in mice, or in cell lines carrying the wrong isoform background — and for D91N there are yeast results and placements in structural models. The gap is not in the reasoning. It is in the measurements.

Which is why the roadmap begins where it does, and why the thing this project most needs is not another hypothesis.

If you could help test one of these See the roadmap

References

Rendered from the site bibliography. Each entry states how far the source was actually read — a database record, an abstract, or the full text — because that distinction changes how much weight a citation deserves.