The gene from Denisovan to Neanderthal to modern mucus
A “genetic sandwich” reveals how a block of DNA entered several populations successively and was affected by natural selection.
Published this week was one of the best evolutionary stories I’ve seen in a while.
On the long arm of chromosome 12 is a gene called MUC19. It’s one of a family of genes that encode proteins called mucins, which provide mucus and mucus membranes their slippery, gelatinous consistency.
Some people have a haplotype spanning part of this gene that came into present-day populations from archaic ancestors. Two of the known Neanderthal genomes, from Vindija and Chagyrskaya, each have copies of a very similar haplotype. So does the Denisova 3 genome.
The story of MUC19 has been uncovered by Fernando Villanea and coworkers, in a preprint that they released last year, and now published in Science. From the sequence of mutational and recombinational changes to the haplotype, they worked out that the gene started in Denisovans, introgressed into late Neanderthals, and from them into modern people.
It’s a game of genetic telephone.
Today this Denisovan haplotype is widespread but mostly rare. Where it is common is among the Indigenous peoples of the Americas and in groups in Latin America with Indigenous ancestry. In groups where this form of MUC19 has a high frequency, tests suggest that it increased due to natural selection. It must have some value to the survival or reproduction of peoples in this region—and maybe its uptake in the Neanderthals was a similar story.
We don’t know today what functional advantage the Denisovan haplotype may have today or in the past. By helping to form a barrier protecting sensitive tissues, MUC19 has a role in immunity. So it’s a good guess—but not certain—that one or more pathogens may have been key environmental pressures leading to selection on this piece of archaic heritage.
The nature of mucus
Mucus lubricates and protects membranes throughout the body. In the respiratory tract, eyes, alimentary tract, and urogenital tract, these membranes are the body’s line of defense against the outside environment. Mucin proteins in humans are the products of a family of 22 genes. Some of these genes make proteins that adorn the membranes of cells. Other mucin proteins are secreted from the cell where they become a free-floating scaffold for mucus.
All of them have a common structure. The two ends have functional domains that define the interactions of the protein with membranes and other proteins. These flank a central chain of many, many repeats of the same short amino acid sequences, variable in number, again and again. These short sequences, often including proline, threonine, and serine, provide sites where the cell attaches sugars, a process called glycosylation. The short sugar molecules stick out from the protein’s long strand like a bottle brush, and they are highly hydrophilic—they attract and hang onto water. So much water is bound up into these glycoproteins that it makes up around 95% of the weight of mucus.
MUC19 is one of the secreted mucins, with a backbone extending more than 8000 amino acids in length. At both ends it can link to other mucin proteins, building into a vast spiderweb of structure suspended outside of cells. This structure shared across the secreted mucins is what gives mucus its viscous, gel-like consistency.
The different members of the mucin family are expressed pretty widely across mucus-generating tissues with some concentrated more in particular parts of the body. These tissues can be studied directly by looking for MUC19 expression and indirectly by characterizing problems that happen when MUC19 goes wrong.
The eye is one place where MUC19 is especially important, with expression of the gene in the cornea, conjunctiva, and lacrimal glands. It’s there helping to lubricate the eye with every blink. The protein is also important in the trachea and respiratory tract, where it helps to trap inhaled dust and allergens, which are ultimately cleared from the body as tiny cilia on the epithelial cells beat the gunky mucus relentlessly upward. The auditory system, with the inner and middle ear, is another place with MUC19 expression.
Cells can ramp up production of MUC19 in response to signals that an infection is present. This has been noted in studies of ear infections, pneumonia, and other conditions. Studies in mice have shown that MUC19 knockouts on a sugary diet have much worse interactions with their oral microbiome, leading to development of infections and dental caries. All these effects illustrate how mucins are a key part of the innate immune system.
A Denisovan inheritance
Mucins are big. Their bloated central section is composed of repeated structures so numerous that DNA replication doesn’t always keep them straight, and the exact number matters little to the protein’s function. In evolving populations, that leads to a lot of variation in the length of the gene, a kind of variation known as variable number tandem repeat (VNTR) polymorphism. In today’s people, different alleles of MUC19 have a 30-base-pair sequence repeated between 250 and 900 times.
The two ends of mucin proteins are much less variable. The sequence of any protein has a direction defined by the linkage of each peptide bond between the amino group (NH2) of one amino acid and the carboxyl group (COOH) of the next, so that the beginning of the protein is known as N-terminal, and the end as C-terminal. The N-terminal section of MUC19 has three functional units known as von Willebrand factor type D (VWD) domains. These determine the protein’s ability to connect with other mucins.
Villanea and coworkers found that a 72 kilobase sequence introgressed from Denisovans, overlapping with two of the VWD domains. People who carry this haplotype share a set of at least nine Denisovan single nucleotide polymorphism (SNP) alleles in the MUC19 sequence in this part of the gene. Based on the way that they change the amino acid sequence, some of these SNP alleles are predicted to make a difference to the gene’s function, but none have yet been studied in cell cultures to see what difference they make.
The Chagyrskaya and Vindija high-coverage Neanderthal genomes each have one copy of this same 72-kb Denisovan haplotype. Each of these individuals also have another haplotype that is more similar to the Altai Neanderthal (Denisova 5) genome. When Villanea and collaborators looked at sequences of living people, they focused on the whole-genome sequences from the 1000 Genomes Project, which represent many populations around the world. The SNPs that define the 72-kb Denisovan haplotype are shared by many people in the Americas.

Villanea and coworkers show that the Denisovan haplotype is most common today in the 1000 Genomes Project sample from Mexico City, followed by Lima, Peru, and less common in other present-day samples from Colombia and Puerto Rico. They also looked at data from ancient genomes across the Americas. They found that these ancient genomes also have the 72-kb haplotype at a comparable or slightly higher frequency than in the Mexico 1000 Genomes Project sample.
Then the researchers looked at a wider window. People who have the 72-kb Denisovan haplotype today have that segment embedded within a wider 742 kilobase haplotype that looks like Neanderthal genomes. The Denisova 3 genome itself lacks that wider Neanderthal-like flanking haplotype.
“In sum, our analyses suggest that some non-Africans carry a mosaic region of archaic ancestry: a small Denisovan-like haplotype (72-kb) embedded in a larger Neanderthal haplotype (742-kb), that was inherited through Neanderthals who themselves acquired Denisovan ancestry from an earlier introgression event.”—Fernando Villanea and collaborators
It’s a genetic sandwich with a thin Denisovan filling surrounded by thick Neanderthal bread.
All the detective work to figure out the introgression pathway is what makes this a great story. Today’s people received this sequence from Neanderthal ancestors, who themselves had inherited it from Denisovan ancestors.
The wide dispersal of the Denisovan haplotype into Neanderthals is pretty interesting. We knew already that there was a lot of gene flow between Neanderthals and Denisovans in central Asia. But so far there has been little evidence of Denisovan mixture much further to the west in European Neanderthals. I’ve suggested that this may reflect a long-term population sink in central Asia. People flowed in, but genes didn’t flow out.
But maybe instead the lack of evidence about Denisovan mixture is just a lack of looking with good statistics. There are very few Neanderthal genomes and there has been little analysis of their haplotype structure. The genome-scale analyses that were done ten years ago using f4 statistics brought to light a greater similarity of Neanderthals and African groups when compared to Denisovans. That broad-scale pattern might be keeping folks from looking closer at windows of Denisovan introgression.
Why the long haplotypes?
There is a remaining mystery about MUC19. Villanea and coworkers characterize the variation of the gene in the Americas and other populations, finding that the protein is much larger in people today who have the Denisovan haplotype. Those people have more VNTR copies—double or more in many cases—which makes the bottle-brush glycosylated region much longer.
But the Denisova 3 genome itself does not have a very long VNTR region. The 72-kb haplotype doesn’t overlap with the VNTR region of MUC19. Neither the Chagyrskaya nor Vindija Neanderthal genomes have the very long VNTR region, either, even though both have one copy of the Denisovan 72-kb haplotype.
One more thing: Living people across Eurasia who have the Denisovan 72-kb haplotype do have a little bit longer VNTR regions than average. But people in the Americas have vastly more copies of the repeat—in some cases more than 800 compared to an average closer to 350.

It’s possible that an initial expansion in repeat copies happened in Neanderthals that were closer to us than the Chagyrskaya and Vindija genomes. The introgressed haplotype is linked to somewhat higher repeat numbers in every Eurasian sample where it is present. This makes it likely that either it introgressed with higher repeat numbers to begin with, or rapidly evolved higher repeat numbers in parallel in the ancestors of modern groups that share it.
The high repeat numbers in the Americas are another story. These must have evolved in the common ancestors of the American populations. At the same time, there is a high degree of variation even within the individuals who share the Denisovan haplotype. Clearly expansion and contraction of repeat numbers were common events, due either to mutation or recombination with unequal crossing-over in the ancestors of today’s people. Across time, the longer haplotypes were favored in this part of the world.
One possibility is that longer MUC19 proteins were actually the target of selection in these populations. The doubling of the glycosylated repeat domains may have made a big difference.
If so, then the MUC19 example would be a case where introgression from Neanderthals and ultimately Denisovans may not have directly advantaged their descendants until further changes evolved within some modern populations.
Bottom line
There is a lot left to do on MUC19 to understand its evolution. In today’s people, when MUC19 is not regulated correctly, it causes a wide range of problems, many of them autoimmune such as Sjögren’s syndrome affecting the eyes, Crohn’s disease affecting the digestive tract, and asthma in the respiratory tract. The effects of normal genetic variation on these and other conditions remain unclear—especially within understudied human populations like those of the Americas.
The MUC19 story highlights the potential of studying archaic introgression in understudied populations. Working out the ways that different lines of ancestry weave into function will benefit from more and more examples. In the Americas, as in several other parts of the world, admixture has been important at different time depths within the populations’ histories. This kind of work will lead to better understanding of the evolution of immune systems, ancient environments relevant to health, and the interactions of ancient groups.
References
Ongaro, L., & Huerta-Sanchez, E. (2024). A history of multiple Denisovan introgression events in modern humans. Nature Genetics, 56(12), 2612–2622. https://doi.org/10.1038/s41588-024-01960-y
Patin, E., & Quintana-Murci, L. (2025). Tracing the Evolution of Human Immunity Through Ancient DNA. Annual Review of Immunology, 43(Volume 43, 2025), 57–82. https://doi.org/10.1146/annurev-immunol-082323-024638
Peyrégne, S., Slon, V., & Kelso, J. (2024). More than a decade of genetic research on the Denisovans. Nature Reviews Genetics, 25(2), 83–103. https://doi.org/10.1038/s41576-023-00643-4
Vespasiani, D. M., Jacobs, G. S., Cook, L. E., Brucato, N., Leavesley, M., Kinipi, C., Ricaut, F.-X., Cox, M. P., & Romero, I. G. (2022). Denisovan introgression has shaped the immune system of present-day Papuans. PLOS Genetics, 18(12), e1010470. https://doi.org/10.1371/journal.pgen.1010470
Villanea, F. A., Peede, D., Kaufman, E. J., Añorve-Garibay, V., Chevy, E. T., Villa-Islas, V., Witt, K. E., Zeloni, R., Marnetto, D., Moorjani, P., Jay, F., Valdmanis, P. N., Ávila-Arcos, M. C., & Huerta-Sánchez, E. (2025). The MUC19 gene: An evolutionary history of recurrent introgression and natural selection. Science, 389(6762), eadl0882. https://doi.org/10.1126/science.adl0882
Thank you for highlighting the new Denisovan article in Science. I found the short "News" review on the paper hoping to find a source for the stunning image you posted at the top. But no, "DNA from ancient bones reveals how Indigenous Americans got their mucus" shows a ginormous jaw with teeth. Do you have a source for illustration of the handsome fellow you posted?
This is a really great exposition on the article (itself wonderfully interesting)! Upon initially reading the article I was struck by how inaccurate the Editor's Summary was.
It's just wrong to say that "this introgression occurred in Indigenous Americans during their migration to the Americas".
Depending on which introgression is referred to (i.e., D into N, or the NDN "sandwich" into OoA AMHs), that statement is wrong by tens to hundreds of thousands of years.
We expect better from ScienceMag. The next time you chat with Razib Khan I hope you guys scold ScienceMag.
(I also very much enjoyed your recent chat with him!)
Lastly, the old editor in me has to report a "72-bp" typo in your superb write-up.