A new peptide tracer let researchers see TREM2, the receptor microglia switch on when they sense brain damage, inside living mouse brains. The signal climbed 86 percent in a standard Alzheimer's mouse model and 114 percent in mice dosed with bacterial endotoxin, a sledgehammer for triggering neuroinflammation. Mice with the TREM2 gene knocked out dimmed by 42 percent, which is the cleanest specificity proof a PET tracer can give: the signal needs the target.
The work, published Tuesday ↗ in the European Journal of Nuclear Medicine and Molecular Imaging by a group led from Shenzhen, builds a peptide-based positron emission tomography (PET) tracer called [68Ga]Ga-STZL730. The molecule is a TREM2-binding peptide chemically tied to a chelator that holds onto a gallium-68 atom, the radioactive tag that PET scanners detect. The team measured a binding affinity of 274 nanomolar, a number that says how tightly the peptide grips its target. Sub-micromolar affinities (anything below 1,000 nanomolar) are roughly the floor at which a tracer becomes worth taking into clinical testing.
The validation experiments are where this gets interesting. The team imaged four kinds of mice and read the brain uptake of the tracer between 20 and 35 minutes after injection. Wild-type mice with normal TREM2 set the baseline. Knockout mice (where the TREM2 gene has been deleted) absorbed 41.6 percent less tracer, which is the cleanest sanity check for whether the tracer is actually binding its named target rather than getting stuck non-specifically in brain tissue. Mice given a dose of bacterial endotoxin, which forces microglia into a broadly activated state, took up 113.8 percent more, roughly double the baseline. APP/PS1 mice, a strain that develops amyloid plaques and activated microglia by middle age and is the standard Alzheimer's model in this kind of work, took up 86.0 percent more.
A blocking study in the Alzheimer's mice, where the team pre-injected unlabeled peptide to saturate the binding sites before delivering the radioactive version, knocked the signal down as expected. Autoradiography on brain slices and immunohistochemistry confirmed that the tracer's hot spots match the regions where activated microglia and TREM2 expression were already known to sit.
Why this matters
TREM2 has become one of the most-followed targets in Alzheimer's biology over the past five years. Carriers of certain loss-of-function variants in the TREM2 gene have roughly triple the risk of late-onset Alzheimer's, and the receptor sits at the center of how microglia (the brain's resident immune cells) respond to amyloid plaques. Several biotechs (Alector, Vigil Neuroscience, Denali) have antibody-based TREM2 agonists in clinical trials. The drugs work, in cellular assays. Whether they actually engage TREM2 in a treated patient's brain has, until now, required either lumbar-puncture cerebrospinal-fluid sampling (an invasive surrogate that reads downstream markers, not the receptor itself) or large radiolabeled antibody tracers that take days to clear non-target tissue and image poorly.
A peptide-based tracer changes the kinetics. Small peptides clear from blood and from non-target brain regions in minutes, not days, which is why the imaging window for this tracer opens at 20 to 35 minutes after injection rather than at 24 or 48 hours. Gallium-68 has a 68-minute physical half-life that matches that kinetic almost exactly. The pairing is also commercially convenient: every major hospital with a PSMA prostate-cancer PET program or a DOTATATE neuroendocrine-tumor program already has the gallium-68 supply and the chelator chemistry on hand. A TREM2 PET tracer that drops into that infrastructure does not require new isotopes, new shielding, or new regulatory pathways.
The caveats
The work is preclinical. No human imaging yet, no toxicology package, no head-to-head against the antibody-based candidates that are further along. The 274 nanomolar affinity is workable but not best-in-class for a PET tracer. Clinical-grade tracers more often sit in the low-nanomolar or sub-nanomolar range, and a tracer that binds too loosely loses signal-to-background when target expression is modest. The LPS endotoxin model is the laboratory's strongest tool for forcing microglial activation, but it is not a clean mimic of Alzheimer's microglial biology. And the 86 percent uptake increase in APP/PS1 mice is averaged across animals at a single age window. Whether the signal scales with amyloid burden across the disease course is the next question, and the paper does not answer it.
What the work does establish is that the modality can be small. A peptide, a chelator, and a routine PET isotope can read a receptor that mid-pipeline antibody drugs need to engage. The format choice is the part that matters.