https://www.nature.com/articles/s41467-026-76728-5
The key question is how badly a cell needs to fail before the rest of the body tries to kill it off
chatGPT(5.6paid)
Overall assessment
This is a strong and potentially important mechanistic mouse study. It argues that, in SOD1^G93A^ ALS mice, activated microglia do not merely clear motor neurons after they die. Instead, microglial Axl and Mer receptor tyrosine kinases may recognize phosphatidylserine on stressed but still-living motor neurons and actively destroy them by phagocytosis—a process often termed phagoptosis.
The evidence that Axl/Mer signalling accelerates motor-neuron loss is persuasive. The stronger claim that microglia literally “eat motor neurons alive” is plausible and well supported indirectly, but not directly demonstrated.
Summary
1. The TAM phagocytic system is activated in ALS
The TAM receptor family consists of Tyro3, Axl and Mer/MERTK. Microglia express Mer constitutively and strongly increase Axl during activation.
In postmortem lumbar spinal cord from people with sporadic ALS, compared with age-matched controls, the authors report approximately:
- a threefold increase in microglial MER;
- a sixteenfold increase in microglial AXL.
However, this human analysis was small: six ALS cases and three controls.
The same pattern occurred as disease progressed in SOD1^G93A^ mice. Axl rose markedly in spinal-cord microglia, while Mer was already present and increased more modestly. Neither motor neurons nor reactive astrocytes appeared to express Axl or Mer appreciably.
2. Motor neurons provide the complementary TAM signals
TAM-mediated engulfment requires:
- Axl or Mer on the phagocyte;
- Gas6 or Protein S as a bridging ligand;
- exposed phosphatidylserine on the target cell.
The authors found that:
- GAS6 mRNA was particularly abundant in human cholinergic motor neurons and was increased in ALS;
- Gas6 protein increased in neurons and microglia in late-stage SOD1^G93A^ mice;
- phosphatidylserine was extensively exposed on the surface of motor neurons in diseased mice but not wild-type mice.
Most phosphatidylserine-positive neurons were negative for cleaved caspase-3. The authors therefore interpret them as stressed but living neurons rather than conventionally apoptotic cells.
This establishes the proposed recognition complex:
phosphatidylserine-positive motor neuron → Gas6 bridge → microglial Axl/Mer
3. Germline deletion of Axl and Mertk changes the course of disease
Deleting both Axl and Mertk produced a striking biphasic phenotype.
SOD1^G93A^;Axl^-/-^;Mertk^-/-^ mice:
- developed early symptoms about 30 days sooner;
- reached formal disease onset somewhat sooner;
- nevertheless progressed more slowly thereafter;
- reached the terminal humane endpoint approximately three weeks later than ordinary SOD1^G93A^ mice.
The early deterioration is probably related to the severe systemic phenotype of germline Axl/Mertk double-knockout mice, which develop autoimmunity, inflammation and other abnormalities. It therefore should not be interpreted as an earlier onset of motor-neuron degeneration without qualification.
At 160 days, double-mutant mice had approximately three times as many surviving lumbar ChAT-positive motor neurons as ordinary SOD1^G93A^ mice. Protection was seen across resistant and vulnerable motor-neuron subtypes rather than being confined to one population.
4. Neuromuscular junctions and muscle were preserved
At 160 days, most neuromuscular junctions in SOD1^G93A^ mice had lost their presynaptic motor-neuron terminals:
- about 90% of quadriceps junctions were empty;
- essentially none were complete.
After Axl/Mertk deletion:
- approximately 45% of quadriceps junctions were complete;
- approximately 35% of extensor digitorum longus junctions were complete.
Muscle-fibre diameter also improved markedly:
- wild type: approximately 57 μm;
- SOD1^G93A^: approximately 27 μm;
- SOD1^G93A^ with Axl/Mertk deletion: approximately 52 μm.
Thus, the preserved ChAT-positive cells were associated with retained peripheral innervation and less muscle atrophy, suggesting functional rather than merely histological preservation.
5. Conditional deletion implicates microglia-lineage cells
To reduce the confounding effects of germline deletion, the authors used tamoxifen-inducible Cx3cr1-CreER to delete floxed Axl and Mertk alleles in adults.
Deletion at day 40 resulted at day 160 in:
- more ChAT-positive motor neurons;
- more ChAT/NeuN-positive motor neurons;
- partial preservation of neuromuscular junctions.
Deletion at day 100 also delayed the clinical endpoint. By approximately day 168:
- none of five tamoxifen-treated mice had reached endpoint;
- roughly 70% of seven vehicle-treated mice had reached endpoint.
This substantially strengthens the attribution to CX3CR1-positive phagocytes and avoids many of the systemic problems of lifelong Axl/Mertk deficiency.
6. Microglia contained motor-neuron material
At day 160, the authors reconstructed Iba1-positive microglia and their CD68-positive lysosomes. Lysosomes in SOD1^G93A^ microglia contained abundant ChAT-positive cytoplasm.
Axl/Mertk deletion produced approximately a tenfold reduction in ChAT-positive material inside microglial lysosomes, together with reduced overall lysosomal activation.
The authors interpret this as evidence that TAM-dependent microglia engulf motor neurons. Inflammatory cytokine expression was not substantially altered at earlier stages, making conventional cytokine toxicity a less obvious explanation.
Principal novelty
1. Axl and Mer are presented as active disease executors
Axl and Mer have generally been considered beneficial or homeostatic because they promote:
- clearance of apoptotic cells;
- resolution of inflammation;
- suppression of innate immune signalling.
This study reverses that interpretation in ALS. It suggests that the normally protective clearance machinery becomes pathogenic when stressed neurons transiently or chronically expose phosphatidylserine.
2. It links a complete recognition pathway to neuronal loss
Previous work had individually implicated:
- activated microglia in ALS;
- Axl as a disease-associated microglial marker;
- phosphatidylserine as an “eat-me” signal;
- microglial phagoptosis in other settings.
The paper joins these into a specific ALS mechanism:
flowchart TD
A["Motor-neuron stress and Ca²⁺ dysregulation"] --> B["Phosphatidylserine externalisation"]
B --> C["Gas6 bridges neuron to Axl/Mer"]
C --> D["Microglial engulfment"]
D --> E["Motor-neuron and NMJ loss"]
E --> F["Faster terminal progression"]
3. It provides genetic evidence at several biological levels
The phenotype is not limited to a biomarker or cell-culture assay. Axl/Mertk removal is associated with:
- reduced neuronal material in lysosomes;
- preservation of motor-neuron cell bodies;
- preservation of neuromuscular junctions;
- less muscle atrophy;
- delayed terminal disease.
That cross-level consistency is a major strength.
4. It suggests a distinction between sick and irreversibly dead neurons
The conceptual novelty is that phosphatidylserine may function as a “kill-me” signal, not merely an indicator that a cell has already died. Microglia could therefore determine when a damaged—but potentially salvageable—motor neuron is removed.
Critique
1. Phagoptosis was not directly observed
This is the central limitation.
The study shows that:
- neurons expose phosphatidylserine;
- most are cleaved-caspase-3 negative;
- microglia contain ChAT-positive material;
- disabling Axl/Mer preserves neurons.
But it never follows an identified living neuron through microglial attack, engulfment and disappearance. Consequently, it cannot exclude the sequence:
neuron becomes irreversibly injured or dies → microglia clear its remains.
The authors acknowledge that they did not perform real-time in-vivo imaging or a direct neuron–microglia phagocytosis assay.
A decisive experiment would label living motor neurons, monitor membrane integrity, mitochondrial potential and calcium, and image whether Axl/Mer-dependent engulfment begins before irreversible loss of viability.
2. Cleaved-caspase-3 negativity is not proof that a neuron is viable
Motor neurons can die through non-canonical apoptosis, necroptosis, ferroptosis or other mechanisms without sustained cleaved-caspase-3 positivity. The marker also captures only a temporal window.
Phosphatidylserine exposure can occur reversibly on living cells, but it can also mark cells already committed to death. The paper therefore establishes “not detectably caspase-3 apoptotic” more securely than “healthy enough to survive.”
3. The conditional experiment is not entirely microglia-specific
Cx3cr1-CreER targets long-lived microglia, but also:
- border-associated macrophages;
- peripheral monocytes at least transiently;
- infiltrating CX3CR1-positive myeloid cells.
The experiment supports a role for CX3CR1-lineage phagocytes but does not prove that parenchymal microglia alone are responsible. More microglia-specific drivers or fate-mapping of recombined cells would improve the attribution.
4. Tamoxifen is a potentially active comparator
Conditional animals received tamoxifen or corn oil. Tamoxifen has estrogen-receptor-dependent and independent effects, including reported anti-inflammatory and neuroprotective actions.
A stronger control would include tamoxifen-treated SOD1^G93A^ mice lacking Cre or lacking floxed alleles. Without that control, some portion of the conditional phenotype could theoretically be caused by tamoxifen rather than gene deletion.
5. The conditional survival result is preliminary
The survival groups were only:
- five tamoxifen-treated mice;
- seven vehicle-treated mice.
Moreover, the graph stops while all treated mice remain alive, rather than reporting their complete terminal survival distribution. It demonstrates delayed endpoint entry over the displayed interval, but the statement that conditional deletion produced a “comparable extension of lifespan” to germline deletion is stronger than the available completed survival data justify.
6. The SOD1 model limits generalisation
The model massively overexpresses mutant human SOD1—approximately 20 transgene copies per haploid genome. It is reliable but biologically extreme and represents only a minority genetic form of human ALS.
The mechanism needs testing in:
- lower-copy SOD1 models;
- C9orf72, TDP-43 and FUS models;
- sporadic ALS-derived human systems;
- ideally human motor-neuron–microglia co-cultures or organoids.
The human tissue confirms TAM-system activation, but not TAM-dependent phagoptosis.
7. Reduced mutant SOD1 aggregation offers an alternative or upstream explanation
The high-molecular-weight mutant SOD1 species was reduced four- to fivefold in double-mutant mice. This is potentially very important but underexplored.
If Axl/Mertk deletion alters:
- transgene expression;
- SOD1 aggregation;
- proteostasis;
- turnover of damaged cells;
- tissue composition,
then reduced intrinsic SOD1 toxicity could contribute to neuronal preservation independently of direct phagoptosis. Establishing unchanged transgene copy number, monomer expression within motor neurons and cell-autonomous toxicity would help distinguish cause from consequence.
8. Histological sample sizes are often small
Several important analyses used only three to five mice per group. Figure 5 includes many individual microglia, but the true independent biological sample is the mouse, leaving only three mice per genotype for the engulfment conclusion.
The effects are large, but replication in an independent cohort and blinded automated analysis would increase confidence.
9. Preserved neurons are not necessarily normally functioning neurons
NMJ and muscle preservation provide good evidence of retained function, but the study does not report:
- grip strength;
- rotarod performance;
- electrophysiological motor-unit function;
- respiratory motor function;
- motor-neuron firing or conduction.
The intervention could retain severely dysfunctional neurons while delaying terminal denervation. Functional measurements would clarify the quality of the rescue.
10. Systemic TAM inhibition would be hazardous
Axl and Mer have important roles in:
- apoptotic-cell clearance;
- immune restraint;
- retinal homeostasis;
- infection responses;
- tissue repair.
The severe phenotype of germline double-knockout mice illustrates this directly. Broad long-term TAM inhibition could replace neurodegeneration with autoimmunity, inflammation or retinal disease. The authors appropriately refrain from presenting systemic TAM inhibitors as immediately viable ALS treatments.
A more plausible therapeutic strategy might involve:
- transient or CNS-restricted inhibition;
- blocking the Gas6–phosphatidylserine interaction;
- protecting neuronal phosphatidylserine asymmetry;
- inhibiting the relevant neuronal scramblase;
- identifying which phosphatidylserine-positive neurons remain recoverable.
Bottom line
The paper convincingly establishes that Axl/Mer activity in CX3CR1-positive phagocytes accelerates motor-neuron, neuromuscular-junction and muscle loss in SOD1^G93A^ mice.
It makes a compelling case—but does not conclusively prove—that the mechanism is microglial phagoptosis of still-viable motor neurons. Its most important implication is not necessarily that TAM receptors should be systemically inhibited, but that microglial clearance may be an active checkpoint determining whether a stressed motor neuron is rescued or irreversibly eliminated.