Humanity’s ancestry has grown far clearer thanks to our ability to obtain ancient DNA. We now know that, as humans left Africa, they interbred with the groups they met there, Neanderthals and Denisovans. Evidence from the Denisovan genome also suggests that this was nothing new; the Denisovans had apparently interbred with an even earlier group. But the identity of that group remained a bit of a mystery.
Now, some evidence from ancient proteins suggests that the mystery group was Homo erectus, a species that left Africa over a million years ago and spread throughout Eurasia. And, thanks to the Denisovans, it appears that modern humans inherited some of that Homo erectus DNA.
In the teeth
Without access to all the repair enzymes made by living cells, DNA rapidly degrades. The double helix fragments, and bases change identity or fall off entirely. While cooler, drier environments slow this process, it sets a hard limit on how far back in time we can obtain DNA sequences. So far, it seems that Homo erectus remains on the far side of that time limit.
To get around these limits, people have turned to proteins. While those also degrade over time, there are a few structures, like bone and teeth, that are very robust and protect proteins from the environment. By studying proteins in tooth enamel, for example, researchers have obtained the sequence of amino acids from fragments of proteins that are 2 million years old.
Given these past successes, a group of Chinese researchers decided to use microscopic samples of Homo erectus teeth from three sites in China, each dating to about 400,000 years ago. They started by taking samples from animals at the same site and confirmed they were able to isolate and identify protein fragments from the enamel of the teeth. Once they were confident in their process, the researchers turned to the five Homo erectus samples and threw in a Denisovan from Harbin as well.
Depending on which of the six Homo erectus individuals they looked at, they obtained fragments of anywhere from six to 11 enamel proteins. The Harbin individual yielded a similar number, and a few earlier studies had obtained data from individuals elsewhere, including a Denisovan from near Taiwan and an archaic human from Spain.
The authors then searched for locations where the ancient sequences differed from the ones found in modern humans. To be confident that the difference was real, the researchers required that it show up in multiple overlapping fragments and in samples analyzed at two different locations.
A clear difference
They came up with two differences that show up in the Homo erectus proteins, but not in modern humans. Oddly, they’re both in the same protein, called ameloblastin.
One of them appears to be completely distinct to Homo erectus, as it isn’t found in any other primate we’ve looked at. But the second has an odd distribution. The Harbin Denisovan that they looked at had one copy of the version found in Homo erectus and another copy that looked the same as the one found in modern humans. The DNA change that causes this protein difference has been found in the genome of other Denisovans. And, in fact, it’s found in the genomes of many modern human populations that include DNA sequences, such as populations found in India and the Philippines.
There are two ways to interpret finding one of these two differences in these genomes. The first would be that these two changes are in the same protein, only 20 amino acids apart. That means they’re going to be relatively close together in the genome. As a consequence, the two changes are very likely to be inherited together—if you get one, you’ll get the other. While recombination between chromosomes could separate them, the probability of a recombination happening within the relatively small segment of DNA between them is small.
In this view, the Denisovans picked up one of the two mutations independently; its presence in the Denisovan genome has nothing to do with whether they and Homo erectus ever interbred.
But there’s the earlier analysis of the Denisovan genome, which had suggested they carried some DNA from a much older human relative, something that could easily be explained by interbreeding with Homo erectus. And, based on the number of changes around the site of the ameloblastin gene, there’s a good chance that some of that highly archaic DNA probably includes this region. All of which seemingly increases the probability that we’re looking at the results of an otherwise improbable recombination.
In any case, the researchers behind this work favor the latter explanation. If they’re right, then that means that many of the modern human populations that have this variant—and it’s present in over 20 percent of some populations native to the Philippines—have ended up with some Homo erectus DNA in their genomes.
Nature, 2026. DOI: 10.1038/s41586-026-10478-8 (About DOIs).
