Six teeth from Homo erectus individuals who lived roughly 400,000 years ago in China have yielded enamel proteins carrying an amino-acid variant in the AMBN gene, the same variant found in Denisovans, a far younger branch of the human family tree. The finding, reported in Nature in 2026, raises a pointed question: did two archaic human species separated by hundreds of thousands of years of evolution exchange genetic material through direct contact, or does the shared variant reflect an independent mutation that happened to land on the same spot twice?
Ancient enamel proteins link two distant human species
The answer matters because it rewrites assumptions about how isolated early human populations really were. If the AMBN variant spread from Homo erectus into Denisovans through interbreeding, the window for gene flow between archaic humans stretches back far earlier than the Neanderthal–Denisovan mixing events already documented through ancient DNA. That would mean reproductive contact between human lineages was not a late, unusual accident but a recurring feature of hominin life across Asia for at least several hundred thousand years.
Denisovans are known primarily from fragmentary fossils in Siberia and Tibet, with their genetic footprint surviving today in modern populations across Southeast Asia and Oceania. Homo erectus, by contrast, is one of the longest-lived human species, occupying parts of Africa and Asia from roughly 1.9 million years ago until perhaps 100,000 years ago. The two species overlapped in time and geography across eastern and southeastern Asia, which makes low-level interbreeding biologically plausible even if hard to prove.
A testable prediction follows from the new data. If the AMBN variant traveled through admixture corridors along the southern arc of Eurasia rather than spreading in a single mixing event somewhere in central or northern Asia, then additional Homo erectus teeth from Southeast Asian sites dated between 300,000 and 500,000 years ago should also carry the variant. Targeted enamel-protein sampling of those specimens could distinguish a broad geographic pattern of gene flow from a one-time, localized event.
Mass spectrometry and the AMBN variant in 400,000-year-old teeth
The research team used mass spectrometry to sequence proteins preserved in the tooth enamel of six Homo erectus individuals from China, as described in a recent Nature study. Enamel proteins survive far longer than DNA in warm, humid environments, which makes them one of the few molecular tools available for studying human ancestors this old. Among the proteins recovered was ameloblastin, encoded by the AMBN gene, which plays a direct role in building tooth enamel during development. The amino-acid variants identified in the erectus specimens were then compared against known Denisovan sequences, and the overlap in AMBN stood out.
This approach builds on earlier work that established enamel proteomics as a reliable method for placing ancient hominins on the evolutionary tree. Researchers had previously extracted enamel-protein sequences from Homo antecessor at Atapuerca in Spain using similar techniques, work reported in a 2020 analysis, and from Homo erectus at Dmanisi in Georgia. Those studies showed that enamel proteins could be used not only for phylogenetic placement but also for determining the sex of individual specimens, all from teeth hundreds of thousands of years old. The Chinese erectus data extend that record deeper into East Asia and add a new dimension: possible evidence of gene flow between species.
The protein variants were interpreted as possible gene flow between Homo erectus and Denisovans, an idea highlighted in a news feature on the findings, though the researchers acknowledged that convergent mutation, where the same amino-acid change arises independently in two lineages, cannot yet be ruled out. The distinction is not trivial. Gene flow implies physical contact and successful reproduction between members of the two species. Convergent mutation implies no contact at all, just similar selective pressures acting on the same gene.
What the protein data cannot yet settle
Several gaps limit how far the current evidence can reach. The raw mass spectra and variant calls from the six Chinese specimens are archived in the PRIDE proteomics database, but no public sequence alignments or statistical significance thresholds have been released in the available reporting. Without those, independent researchers cannot yet evaluate how strongly the AMBN variant in erectus matches the Denisovan version versus modern humans or Neanderthals.
No direct statements from the study’s lead authors address whether the AMBN variant frequency differs meaningfully across other hominin lineages. If the variant also appears at low frequency in Neanderthals or early modern humans, the case for specific erectus-to-Denisovan gene flow weakens. If it is absent from those groups, the case strengthens considerably. The authentication criteria used to validate the antecessor benchmark sequences have not been reproduced in detail for the new erectus samples in the publicly available summaries, leaving open questions about contamination controls and protein degradation patterns.
Geographic evidence presents its own challenge. No direct fossil record documents a specific time and place where Homo erectus and Denisovans lived side by side. The overlap is inferred from the broad ranges of both species across Asia, but inference is not the same as proof. The Denisovan fossil record is still sparse and concentrated in a few caves, while Homo erectus remains are scattered across China, Indonesia, and other parts of Eurasia. Without a site that clearly preserves both lineages in the same stratigraphic layers, the spatial context for potential interbreeding remains circumstantial.
Chronology adds another layer of uncertainty. The Chinese Homo erectus teeth sampled for enamel proteins date to around 400,000 years ago, whereas the best-characterized Denisovan genomes come from individuals who lived roughly 200,000 to 50,000 years ago. Even if the AMBN variant did pass from one lineage to the other, the timing of that transfer could fall anywhere within a broad window. It might reflect early contact between ancestral populations that later gave rise to Denisovans, or later contact between a lingering Homo erectus population and already distinct Denisovans.
There is also no clear functional story yet for why the AMBN variant would have been favored by natural selection, if selection played any role at all. Ameloblastin is crucial for enamel formation, but the specific amino-acid change shared by Homo erectus and Denisovans has not been tied to measurable differences in tooth hardness, wear resistance, or resistance to caries. Without a functional link, it is difficult to decide whether the variant spread because it conferred an advantage, drifted neutrally through small populations, or simply persisted by chance in both lineages.
Next steps for testing the gene-flow hypothesis
To move beyond speculation, researchers will need more data from both molecules and fossils. On the molecular side, expanding enamel-protein sampling to additional Homo erectus specimens across Asia, especially in Southeast Asia and along proposed migration corridors, would reveal whether the AMBN variant is rare and localized or widespread. If the variant appears in multiple erectus populations separated by thousands of kilometers and hundreds of thousands of years, it becomes more plausible that it was an established feature of the species rather than a one-off mutation.
Parallel work on Denisovan material could also help. Although DNA remains the primary tool for studying Denisovans, enamel proteomics might extend their molecular record to warmer regions where DNA preservation is poor. If Denisovan or Denisovan-like teeth from such environments also carry the AMBN variant, and if those teeth can be securely dated and placed geographically, they could anchor a more precise map of where the shared variant occurred.
On the fossil side, targeted excavations in regions where Homo erectus and Denisovans are most likely to have overlapped could eventually produce the kind of mixed assemblages that are currently missing. Even fragmentary remains, if tied to robust dating and environmental reconstructions, would help constrain where and when contact zones may have existed. Combined with climate models and reconstructions of ancient habitats, these data could indicate whether shifting ecosystems funneled different hominin groups into the same refugia during glacial and interglacial cycles.
For now, the AMBN variant in 400,000-year-old Chinese teeth stands as a tantalizing signal rather than a definitive answer. It shows that enamel proteins can preserve evolutionary information across extraordinary spans of time and that molecular echoes of contact between human species may still be hiding in old collections. Whether those echoes ultimately reveal a deep history of interbreeding between Homo erectus and Denisovans, or a striking example of evolutionary coincidence, will depend on how quickly researchers can fill the many gaps that still surround this shared molecular fingerprint.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
