A team at ETH Zurich has identified clumping of a protein called GRK2 inside neurons as a distinct driver of Alzheimer’s-related brain damage, separate from the amyloid plaques that most current therapies target. The researchers also designed a small molecule, labeled Compound 10, that prevents GRK2 from aggregating and restores mitochondrial function in cell and animal tests. The peer-reviewed findings, published in Cell Reports Medicine under DOI 10.1016/j.xcrm.2026.102707, open a new front in Alzheimer’s drug development by going after damage that occurs inside brain cells rather than in the spaces between them.
Why an intracellular target changes the Alzheimer’s drug calculus
Every approved Alzheimer’s antibody, from lecanemab to donanemab, works by clearing amyloid-beta from the brain’s extracellular space. Clinical trials have shown these drugs slow cognitive decline modestly, but patients still deteriorate. That gap has pushed researchers to look for damage pathways that persist even after amyloid levels drop. The ETH Zurich group’s work on GRK2 aggregation offers one explanation for why: amyloid-beta triggers a cascade inside neurons that wrecks mitochondria, and removing amyloid from outside the cell does not automatically reverse that internal damage.
The hypothesis the new data supports is straightforward. If GRK2 clumps block the import of essential proteins into mitochondria, then a drug that prevents those clumps could protect energy production in neurons even while amyloid-beta remains elevated. Pair such a drug with an anti-amyloid antibody, and the two could produce additive cognitive benefit, one clearing the trigger and the other shielding the downstream target. No combination trial exists yet, but the mechanistic logic is now backed by mouse-model data showing that aggregated GRK2 tracks with dysfunctional mitochondria in diseased brains.
GRK2 aggregates, TOMM6, and the mitochondrial chokepoint
The study centered on Tg2576 mice, a widely used Alzheimer’s model that overproduces amyloid-beta. In aged Tg2576 animals, the proportion of aggregated GRK2, particularly its phosphorylated form phospho-S670-GRK2, was significantly higher than in age-matched controls. The aggregates accumulated on mitochondria and interfered with the organelle’s ability to import nuclear-encoded proteins it needs to generate energy. Proteomics data deposited under dataset PXD073643 mapped the specific interaction between GRK2 and TOMM6, a component of the mitochondrial outer membrane import machinery. When GRK2 clumps onto TOMM6, the import channel jams.
Compound 10, a small molecule designed by the same team, prevented GRK2 from aggregating in both cell-based assays and animal tests. With GRK2 kept in its soluble, functional state, mitochondrial import resumed and energy output recovered. The compound’s mechanism is distinct from anything in the current Alzheimer’s pipeline, which is dominated by antibodies targeting extracellular amyloid or tau. A related chemical-biology report describes how rational design and structure-guided screening can yield small molecules that modulate protein–protein interactions, a strategy that underpins the GRK2 work.
No data on Compound 10’s ability to cross the blood-brain barrier or its off-target profile has appeared in the institutional releases so far, a gap that will determine whether the molecule or a derivative can advance toward human testing. Pharmacokinetic behavior, metabolic stability, and safety margins in non-rodent species remain unknown, and all will be critical for any first-in-human study.
Convergence on intracellular pathways
The ETH Zurich findings do not exist in isolation. A separate line of research from Scripps Research, published earlier this year, identified the protein STING as another intracellular driver of Alzheimer’s pathology. That team found that S-nitrosylation of STING at cysteine 148 triggers neuroinflammation, with evidence drawn from postmortem human AD brain tissue, stem-cell-derived models, and mouse-model interventions. Together, the GRK2 and STING discoveries point to a broader shift: the field is moving beyond extracellular plaques and tangles to catalog the specific molecular events that destroy neurons from within.
In both cases, the implicated proteins sit at key control points. GRK2, long known for regulating G protein–coupled receptors, here appears as a physical blocker of mitochondrial protein import. STING, a central player in innate immunity, becomes a chronic amplifier of inflammatory signaling when chemically modified. These roles converge on two hallmarks of Alzheimer’s pathology that anti-amyloid antibodies do not fully address: energy failure and sustained neuroinflammation.
That convergence is reflected in a broader wave of mechanistic studies, including recent coverage highlighting how intracellular stress responses, protein misfolding, and mitochondrial dysfunction may interact. Rather than treating amyloid as the sole cause, these efforts frame it as one node in a network of intracellular insults that accumulate over decades.
Gaps between mouse data and a human treatment
Several questions stand between the current findings and a therapy patients could use. The GRK2 aggregation data comes entirely from Tg2576 mice and cell models. No postmortem human brain tissue or iPSC-derived neuron data confirming GRK2–TOMM6 aggregates in sporadic Alzheimer’s cases has been reported. Sporadic Alzheimer’s accounts for the vast majority of diagnoses, and mouse models that rely on familial-mutation transgenes do not always predict what happens in those patients.
The published record also lacks long-term behavioral or survival outcomes in treated mice. Showing that Compound 10 restores mitochondrial function in cells is a necessary first step, but clinicians and regulators will need evidence that improved bioenergetics translates into preserved cognition, delayed symptom onset, or extended lifespan in animal models. That means months-long dosing studies, standardized memory tests such as Morris water maze or novel-object recognition, and neuropathological readouts comparing treated and untreated cohorts.
Safety is another open question. GRK2 participates in multiple signaling pathways, including those governing cardiovascular function. Completely blocking its activity could therefore introduce unacceptable side effects. The ETH Zurich work suggests Compound 10 modulates aggregation rather than global kinase function, but a detailed kinome profile, cardiotoxicity assays, and off-target screens will be needed to confirm that selectivity. Dose-ranging studies will also have to balance sufficient exposure in the brain with tolerability in peripheral organs.
Translationally, the blood-brain barrier looms as a practical hurdle. Many small molecules with promising in vitro effects fail because they do not reach adequate concentrations in brain tissue. Medicinal chemistry optimization for lipophilicity, polar surface area, and efflux transporter avoidance may be required to turn Compound 10 into a drug candidate, and each modification will need to preserve the anti-aggregation activity that underlies its appeal.
What an eventual clinical program might look like
If GRK2 aggregation is confirmed in human Alzheimer’s brains, an early clinical program would likely start with a Phase 1 study in healthy volunteers to establish safety, pharmacokinetics, and any obvious dose-limiting toxicities. Parallel biomarker work could track whether the compound affects mitochondrial function markers in blood or cerebrospinal fluid, though identifying specific, sensitive readouts for GRK2–TOMM6 disruption will be challenging.
Subsequent Phase 2 trials might enroll patients with early symptomatic Alzheimer’s, testing Compound 10 or a successor both as monotherapy and in combination with an anti-amyloid antibody. Endpoints would probably combine cognitive scales with imaging and fluid biomarkers of neurodegeneration and mitochondrial health. Because intracellular pathways may respond more slowly than plaque clearance, trial durations may need to be longer than the 18-month windows that have become standard for antibody studies.
Regulators and payers will also scrutinize how an intracellular-targeting drug fits into an already complex treatment landscape. If the benefit is incremental but the cost and monitoring burden are high, adoption could be slow. On the other hand, a therapy that meaningfully preserves function for several additional years, especially in combination with existing antibodies, would address a major unmet need for patients and caregivers.
A new layer in the Alzheimer’s puzzle
The ETH Zurich group’s focus on GRK2 aggregation adds a fresh layer to the Alzheimer’s puzzle: damage within neurons may be as crucial as the visible plaques that have dominated the field’s attention. By tying a specific intracellular event to mitochondrial failure, and by demonstrating that a rationally designed small molecule can reverse that failure in preclinical models, the work sketches a plausible path toward therapies that shore up the brain’s energy supply.
Yet the distance from Tg2576 mice to human patients remains considerable. Confirming that GRK2–TOMM6 aggregates occur in sporadic Alzheimer’s, establishing safety and brain penetration for any drug candidate, and proving that mitochondrial rescue changes the trajectory of cognitive decline will all be necessary steps. As parallel efforts on STING and other intracellular targets advance, the picture that emerges is not of a single “magic bullet” but of a multi-front campaign against the many ways neurons can fail. In that context, Compound 10 is less a finished solution than a proof of concept that intracellular aggregation can be drugged-a concept that could reshape how researchers think about treating Alzheimer’s disease.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
