One Electron Beam, Two Images: Multicolor Electron Microscopy Arrives

For decades, biologists have faced an uncomfortable trade-off when imaging cells. Fluorescence microscopy tells you where specific proteins are, lit up in vivid color by glowing tags - but resolution tops out around 250 to 300 nanometers, too coarse to see individual proteins clearly and too limited to reveal the surrounding cellular architecture. Electron microscopy delivers exquisite structural detail down to a few nanometers but has traditionally been colorblind, unable to identify specific molecules within the ultrastructural landscape it reveals.

Scientists have tried combining the two by taking separate images with each method and then computationally aligning them. In small samples that works tolerably well. In large, complex tissues - like brain sections - precise alignment becomes nearly impossible.

A team at Harvard University has developed a single technique that bypasses the alignment problem entirely.

One Beam, Two Datasets

The approach, called multicolor electron microscopy, sends a single electron beam through a biological sample. The beam does two things at once. It scatters off cellular structures, producing the standard high-resolution structural image that electron microscopy is known for. And it excites fluorescent probes attached to specific proteins, causing them to emit visible light through a process called cathodoluminescence - emission of light triggered by electrons rather than by incoming photons.

Both signals are captured simultaneously from the same scan, so structural information and molecular identity data are inherently co-registered. No alignment required.

"We're not sending in light - we're sending an electron beam," explained Debsankar Saha Roy, a postdoctoral fellow in the laboratory of Maxim Prigozhin at Harvard. "We have probes that you can attach to a protein that emit visible light when excited by electrons. So from the same electron beam, you get two sets of information: the colored signal from the probes, and also the detailed structural image from the electrons."

Standard Dyes Work Too

The team had previously developed specialized lanthanide nanoparticles as cathodoluminescence probes. But a more striking finding emerged when they placed ordinary fluorescent dyes - the kind already used in fluorescence microscopy labs worldwide - into the electron microscope.

"The most surprising thing we observed was that standard dyes used in fluorescence microscopy also emit visible light when you excite them with electrons," Roy said. "That had never been seen before. And these dyes - and their protein labelling methods - are already developed and available. You don't have to create anything new."

This is practically significant. Researchers who have spent years developing labeling protocols with conventional fluorescent dyes can port those directly to multicolor electron microscopy without redesigning their experimental toolkits.

Demonstrated in Living-System Tissues

The technique has been tested in mammalian cells and in biological tissues, including fungus-infected fly specimens, demonstrating that it functions in intact, complex biological material - not just simplified model systems. The researchers can simultaneously visualize the structural context of a cellular compartment and identify which proteins are present within it, at nanometer resolution, in a single imaging session.

The current implementation produces two-dimensional images. This is a meaningful limitation - cellular biology is three-dimensional, and many questions about protein organization and cell architecture require 3D data to answer. Single planes of information can be misleading when structures are distributed across the depth of a cell.

The Next Step: Into Three Dimensions

The team's stated next target is adapting multicolor electron microscopy for use in cryo-electron microscopy workflows. In cryo-EM, samples are flash-frozen in vitreous ice, preserving them in a near-native state without the chemical fixation artifacts that can distort structure in conventional electron microscopy. Images are taken from multiple tilt angles and computationally combined to produce 3D reconstructions at atomic or near-atomic resolution.

Integrating the cathodoluminescence approach into cryo-EM would combine the structural fidelity of frozen samples with the molecular identification capability of multicolor imaging - and extend it into three dimensions. "We want to extend this multicolor electron microscopy approach to 3D," Roy said. "To get there, we aim to implement this technique in ultrathin sections of cell embedded matrices and/or in cryo-electron microscopy."

That integration has not yet been demonstrated and represents a substantial technical challenge. Cryo-EM samples are held at cryogenic temperatures and are highly sensitive to radiation damage from the electron beam - the same beam that would need to generate cathodoluminescence signals. Finding probe chemistries and imaging conditions that work within those constraints is the central open problem.

If solved, the result would be a tool capable of showing, in full three-dimensional context and at nanometer resolution, exactly which proteins are where inside a cell - and in what structural environment they sit. The research was presented at the 70th Biophysical Society Annual Meeting in San Francisco in February 2026.