To see how neurons wire together, you usually have to pick a side. Electron microscopy shows you the exquisite architecture of brain tissue but tells you almost nothing about which specific proteins sit where. Fluorescence microscopy lights up the molecules you tagged in advance and leaves everything else invisible. Combining both, in the same sample, has been a long-standing headache.
A team at Yale School of Medicine now reports a way around the tradeoff. Their method, called pan-ExM-t, physically enlarges a piece of brain tissue and then images it on an ordinary confocal microscope. The work appeared in Nature Biotechnology on November 26, 2025.
Blow up the sample, not the microscope
The core idea sounds almost too simple. Instead of building a more powerful lens, you make the specimen bigger. Expansion microscopy embeds tissue in a water-absorbing gel, then floods it with water so the whole structure swells and pulls molecules apart. Features that were once packed below the resolution limit of light get physically separated until a normal microscope can tell them apart.
pan-ExM-t pushes that expansion to roughly 16 to 24 times the original linear size. At that scale, the authors report they can resolve presynaptic and postsynaptic densities, the dense protein zones on either side of a synapse where signals cross from one neuron to the next. They can also see the fine tangle of neuropil and individual cellular organelles.
What makes it more than a bigger picture is the staining strategy. The team applies fluorescent pan-stains that bind broadly to proteins and lipids, which paints in an anatomical background reminiscent of what electron microscopy produces. On top of that they add antibody labels for specific molecules. So you get the ultrastructural context and the molecular identity in one image, all read out with light.
What it can actually see
The researchers put the method through several settings. They imaged synaptic and cell-type-specific antibodies inside dissociated neuron cultures and in sections of mouse brain, resolving synaptic densities, organelles, and the surrounding neuropil in three dimensions. They also went further than static snapshots. From pan-ExM-t image volumes, they traced neuronal circuitry, following processes through the tissue the way connectomics researchers do with electron microscopy data.
The authors frame the accessibility point plainly. Because the whole thing runs on a confocal microscope, they suggest any laboratory with one can now localize specific molecules within nanoscale cellular and circuit contexts. That matters because electron microscopy setups and the expertise to run them are scarce, while confocal microscopes are standard equipment in biology departments everywhere.
The fine print
This is a demonstration of a new tool, and a few things are worth keeping in mind before treating it as a finished workhorse. The paper reports circuit tracing as something the method makes possible, not as a large validated reconstruction on the scale of published connectomes. Expansion methods also depend on the gel swelling evenly. If it does not, distances between molecules can distort, so the anatomical faithfulness of any given sample still needs checking against known structures. And all of this so far is neurons and mouse brain tissue. How the approach behaves in denser or more heterogeneous tissues is an open question the study does not settle.
There is also a broader caution. Antibody labeling is only as good as the antibodies. The molecular channel inherits whatever specificity problems the chosen reagents carry, which is a familiar limitation across fluorescence imaging rather than something unique here.
Even with those caveats, the appeal is easy to state. A great deal of what we want to know about the brain sits at the junction of anatomy and molecular identity, and until now getting both usually meant expensive instruments and specialist pipelines. Swelling the sample and reaching for a microscope most labs already own is a genuinely different bet on how to get there.
Comments