A single-molecule imaging platform from the Broad Institute and MIT has run continuously for more than 16 minutes at 100-millisecond resolution — and it does not photobleach. The instrument, built around upconverting nanoparticles (UCNPs) doped with ytterbium, erbium and thulium, tracked EGFR, HER2 and HER3 simultaneously in three colors on live cells. Conventional dyes lose signal within seconds.

The biology it delivered is published in Cell, in a paper entitled "ErbB family receptor dimerization dynamics and dysregulation via long-term single-molecule imaging." The headline finding: wild-type HER3 forms unexpectedly stable homodimers, and cancer mutations destabilize them.

For R&D managers weighing imaging platforms, the spec matters. The probe can run for hours; the team stopped at 16 minutes only because they had already captured the long EGFR homodimers they were chasing.

"If you're trying to understand a conversation between two people… if all of a sudden I just cut off this conversation, then you have to guess my answer based on our previous two-minute interaction, which is often impossible to do," said Sam Peng, comparing conventional short-duration imaging to eavesdropping.

What did the HER3 result actually show?

HER3's kinase domain is too weak to signal on its own, so the prevailing model treats it as a heterodimer partner — a receptor that pairs with others. Peng's team expected HER3 to serve as a negative control for dimer detection. It should have come back empty.

"The first time we saw HER3 homodimers, we were really puzzled," Peng told GEN. "We thought that maybe this was some experimental artifact."

The team reran the experiment with different labels and probes. The result held every time: wild-type HER3 homodimers proved remarkably stable, far outlasting unstimulated EGFR homodimers. The working model is that these homodimers form a signaling-inactive pool that sequesters HER3, limiting its availability to pair with other receptors and trigger cancer signaling.

Cancer mutations support the model, but EGFR and HER3 moved in opposite directions:

• Mutations like the exon-19 deletion make EGFR homodimers more stable, driving signaling that tracks with clinical aggressiveness — the more stable the dimer, the more signaling it triggers.
• The HER3 mutations studied destabilize the homodimer, potentially freeing HER3 to form signaling-active heterodimers.
• HER2 mutations only modestly enhance homodimer stability, consistent with HER2 cancers being driven primarily by gene amplification.

These findings could help shape new cancer therapies, though the therapeutic payoff remains a projection; the dimer dynamics are measured results.

Where do the rare earths come from?

The platform depends on heavy rare earth elements, and the supply chain is concentrated. China controlled 85% of total rare earth oxide production in 2025, including an estimated 99% of dysprosium oxide and terbium oxide, according to Benchmark Mineral Intelligence. In 2025, dysprosium oxide in North America cost on average 4.4 times the Chinese price.

China introduced export controls on seven rare earth elements in April 2025. An October 2025 expansion added ytterbium, erbium, holmium, thulium and europium — including the elements in Peng's probes. That expansion is currently suspended under a U.S.–China trade truce, which officials said has been extended from November to January.

At bench scale, none of this registers. Peng's lab works in milligrams to grams and buys rare earth salts in vials from a U.S. distributor. "They come in as a form of powder, like salt, just like your kitchen salt," he said. Erbium salts arrive in pink. "Pretty beautiful."

Peng's lab does not track where the material was mined; it likely comes from China. Heavy rare earth elements are not scarce in the earth's crust — they are rarely found in concentrations worth mining, and separating them is difficult.

Is the bottleneck supply or expertise?

Currently, expertise. Peng estimates roughly a dozen groups worldwide work at the intersection of UCNP imaging and single-molecule biology, because the work demands nanomaterial chemistry, molecular labeling, advanced optics and computation in one lab. "It's the integration of this entire pipeline that makes it challenging," Peng said.

Rare earth elements already appear in TR-FRET, some forms of NIR-II imaging, mass cytometry and UCNP tracking — each using different elements and probe chemistries. Labs buy the salts the way they buy antibodies. For now, the rare earth salts are the easy part; the optics, and the people who can build them, are what's scarce.

With the U.S.–China truce extended into January, Peng's lab and the roughly dozen others in this niche have a window to keep pushing the technique — and to test whether HER3's stable homodimer pool can be exploited therapeutically.