International study identifies genes responsible for diversity of human skin colors

October 18, 2017

Human populations feature a broad palette of skin tones. But until now, few genes have been shown to contribute to normal variation in skin color, and these had primarily been discovered through studies of European populations.

Now, a study of diverse African groups led by University of Pennsylvania geneticists Sarah Tishkoff and Nicholas Crawford, with key contributions from new Arizona State University School of Life Sciences faculty Susanne Pfeifer and Jeff Jensen, has identified new genetic variants associated with skin pigmentation. Download Full Image

The findings help explain the vast range of skin color on the African continent, shed light on human evolution and inform an understanding of the genetic risk factors for conditions such as skin cancer.

“We have identified new genetic variants that contribute to the genetic basis of one of the most strikingly variable traits in modern humans,” said Sarah Tishkoff, a University of Pennsylvania professor who was the senior author of the study and led the multi-institutional, international team.

“When people think of skin color in Africa most would think of darker skin, but we show that within Africa there is a huge amount of variation, ranging from skin as light as some Asians to the darkest skin on a global level and everything in between,” Tishkoff said. “We identify genetic variants affecting these traits and show that mutations influencing light and dark skin have been around for a long time, since before the origin of modern humans.”

The new findings were recently published in the journal Science (DOI: 10.1126/science.aan8433) and featured in the national news.  

Skin color is a classic variable trait in humans, but little has been known about the genes responsible for these traits.

Both light and dark skin pigmentations, from the fair-skinned to ebony tones, confer benefits: darker skin, for example, is believed to help prevent some of the negative impacts of ultraviolet light exposure, while lighter skin promotes the synthesis of vitamin D in regions with low ultraviolet light exposure.

To objectively capture the range of skin pigmentation in Africa, the research team used a color meter to measure the light reflectance of the skin of more than 2,000 Africans from ethnically and genetically diverse populations. They took the measurement from the inner arm, where sun exposure is minimal. These measurements can be used to infer levels of the skin pigment melanin. They obtained a range of measurements; the darkest skin was observed in Nilo-Saharan pastoralist populations in eastern Africa, and the lightest skin was observed in San hunter-gatherer populations in southern Africa.

Among the international research team, ASU faculty members Susanne Pfeifer and Jeff Jensen helped analyze population data from nearly 1,600 people, examining more than 4 million single nucleotide polymorphisms across the genome, places where the DNA code differs by one “letter.”

From this data set the researchers were able to do a genome-wide association study; they identified four key areas of the genome where variation closely correlated with skin color differences.

"The main challenge in this analysis lies in teasing apart the many loci that experienced changes in allele frequency simply due to random genetic drift from those associated with natural selection, and then in asking if those naturally selected mutations are associated with differences in skin pigmentation," Pfeifer said.

The region with the strongest associations was in and around the SLC24A5 gene, one variant of which is known to play a role in light skin color in European and some southern Asian populations and is believed to have arisen more than 30,000 years ago. This variant was common in populations in Ethiopia and Tanzania that were known to have ancestry from southeast Asia and the Middle East, suggesting it was carried into Africa from those regions and, based on its frequency, may have been positively selected.

"This finding illustrates the vital role of population history in dictating evolutionary outcomes — in this instance specifically, the interplay of population size change, population splitting, as well as migration and back-migration, along with natural selection and genetic drift," Pfeifer said.

Another region, which contains the MFSD12 gene, had the second strongest association to skin pigmentation. This gene is expressed at low levels in the depigmented skin in individuals with vitiligo, a condition where the skin loses pigment in some areas.

“I still remember the ‘aha!’ moment when we saw this gene was associated with vitiligo,” said Crawford, the study's first author, and Penn postdoctoral fellow. “That’s when we knew we’d found something new and exciting.”

The team found that mutations in and around this gene that were associated with dark pigmentation were present at high frequencies in populations of Nilo-Saharan ancestry, who tend to have very dark skin, as well as across sub-Saharan populations, except the San, who tend to have lighter skin. They also identified these variants, as well as others associated with dark skin pigmentation, in South Asian Indian and Australo-Melanesian populations, who tend to have the darkest skin coloration outside of Africa.

“The origin of traits such as hair texture, skin color and stature, which are shared between some indigenous populations in Melanesia and Australia and some sub-Saharan Africans, has long been a mystery,” Tishkoff said. “Some have argued it’s because of convergent evolution, that they independently evolved these mutations, but our study finds that, at genes associated with skin color, they have the identical variants associated with dark skin as Africans. Our data are consistent with a proposed early migration event of modern humans out of Africa along the southern coast of Asia and into Australo-Melanesia.”

Also of interest was that the genetic variants at MFSD12OCA2, and HERC2 associated with light skin pigmentation were at highest frequency in the African San population, which has the oldest genetic lineages in the world, as well as in Europeans.

MFSD12 is highly expressed in melanocytes, the cells that produce melanin.

“We went beyond most genome-wide association studies to do functional assays,” Tishkoff said, “and found that knocking out MFSD12 dramatically impacted the pigmentation of fish and mice. It’s pointing to this being a very conserved trait across species.”

To verify the gene’s role in contributing to skin pigmentation, the researchers blocked expression of the gene in cells in culture and found an increase in production of eumelanin, the pigment type responsible for black and brown skin, hair and eye color. Knocking out the gene in zebrafish caused a loss of cells that produce yellow pigment. And in mice, knocking out the gene changed the color of their coat from agouti, caused by hairs with a red and yellow pigment, to a uniform gray by eliminating production of pheomelanin, a type of pigment also found in humans.

A final genetic region the researchers found to be associated with skin pigmentation included genes that play a role in ultraviolet light response and melanoma risk. The top candidate gene in the region is DDB1, involved in repairing DNA after exposure to UV light.

“Africans don’t get melanoma very often,” Tishkoff said. “The variants near these genes are highest in populations who live in areas of the highest ultraviolet light intensity, so it makes sense that they may be playing  a role in UV protection.”

The mutations identified by the team play a role in regulating expression of DDB1 and other nearby genes. 

“Though we don’t yet know the mechanism by which DDB1 is impacting pigmentation, it is of interest to note that this gene, which is highly conserved across species, also plays a role in pigmentation in plants such as tomatoes,” Tishkoff said.

The team saw evidence that this region of the genome has been a strong target of natural selection outside of Africa; mutations associated with light skin color swept to nearly 100 percent frequency in non-Africans, one of few examples of a “selective sweep” in all Eurasians; the age of the selective sweep was estimated to be around 60,000 to 80,000 years old, around the time of migration of modern humans out of Africa.

One additional takeaway from this work is a broader picture of the evolution of skin color in humans. Most of the genetic variants associated with light and dark pigmentation from the study appear to have originated more than 300,000 years ago, and some emerged roughly 1 million years ago, well before the emergence of modern humans. The older version of these variants in many cases was the one associated with lighter skin, suggesting that perhaps the ancestral state of humans was moderately pigmented rather than darkly pigmented skin.

“If you were to shave a chimp, it has light pigmentation,” Tishkoff said, “so it makes sense that skin color in the ancestors of modern humans could have been relatively light. It is likely that when we lost the hair covering our bodies and moved from forests to the open savannah, we needed darker skin. Mutations influencing both light and dark skin have continued to evolve in humans, even within the past few thousand years.”

Tishkoff noted that the work underscores the diversity of African populations and the lack of support for biological notions of race.

“Many of the genes and new genetic variants we identified to be associated with skin color may never have been found outside of Africa because they are not as highly variable,” Tishkoff said. “There is so much diversity in Africa that’s not often appreciated. There’s no such thing as an African race. We show that skin color is extremely variable on the African continent and that it is still evolving. Further, in most cases, the genetic variants associated with light skin arose in Africa.”

Media contact: Joseph Caspermeyer, ASU Biodesign Institute, 480-258-8972, 

In addition to Tishkoff and Crawford, study co-authors included Derek Kelly, Matthew E. B. Hansen, Marcia Holsbach, Shaohua Fan, Alessia Ranciaro, Simon Thompson, Yancy Lo, Michael Campbell, William Beggs, Shanna L. Bowman, Michael Marks and Jake Haut of Penn; Ethan Jewett and Yun S. Song of the University of California, Berkeley; Susanne P. Pfeifer and Jeffrey D. Jensen of Arizona State University; Farhad Hormozdiari, Harriet Rothschild, Leonard Zon and Yi Zhou of Harvard University; Sununguko Wata Mpoloka and George Mokone of the University of Botswana; Thomas Nyambo of Muhimbili University of Health and Allied Sciences in Tanzania; Dawit Wolde Meskel and Gurja Belay of Addis Ababa University in Ethiopia; Michael A. Kovacs, Mai Xu, Tongwu Zhang, Kevin Bishop, Jason Sinclair, Cecilia Rivas, Eugene Elliot, Jiyeon Choi, Shenchao Li, Belynda Hicks, Shawn Burgess, Chrisan Abnet, Dawk E. Watkins-Chow, Kevin M. Brown, Stacie K. Loftus, William J. Pavan, Meredith Yeager and Stephan Chanock of NIH; and Elena Oceana of Brown University. The NISC Comparative Sequencing Program of the National Institutes of Health also participated in the research.

image title

Counting the dead

ASU scholar says we're due for another flu like 1918's devastating Spanish flu.
Arizona deaths from Spanish flu were underreported by as much as 1,000 percent.
ASU Library exhibit visualizes Arizona's mortality rate from 1918 pandemic.
October 18, 2017

Interactive exhibit at ASU Library explores toll of 1918 flu pandemic in Arizona, including the thousands of underreported Native deaths

So unusual was the 1918 influenza, which killed an estimated 3 to 5 percent of the world’s population, that its symptoms were often misdiagnosed, resembling cholera or typhoid. 

As opposed to juveniles, the elderly and those with weakened immune systems, this particular strain of flu virus, commonly referred to as the “Spanish flu,” took its deadliest toll on young adults who were previously healthy.

The speed with which it killed was also remarkable, claiming an estimated 25 million lives in its first 25 weeks and causing life expectancy in 1918 to drop by about 12 years in the United States, alone.

“This flu was unusually dangerous. ... We’re due for another flu like this,” says ASU scholar Jacqueline Wernimont. 

Together with Nexus Lab project manager Elizabeth Grumbach, Wernimont — an assistant professor in the Department of English and director of the Nexus Lab, an initiative within the Institute for Humanities Research — has set out to highlight the issue of pandemic with a focus on Arizona.

The two are opening up for public feedback a prototype of their multimedia installation at ASU Library. 

"Counting the Dead: Arizona and the Forgotten Pandemic" aims to re-embody Arizona’s influenza mortality data from a century ago, illustrating the ways in which illness had spread across our then-young state.

With this exhibit, the digital humanities scholars are also telling a story many have not heard — that flu deaths in Arizona, in 1918, were underreported by as much as 1,000 percent.  

“Much of what we thought we knew about the flu in Arizona is wrong,” Wernimont said.

Counting the dead

woman holding black cords

ASU Assistant Professor Jacqueline Wernimont shows the cord she has unraveled, which represents 12 influenza deaths in the exhibit "Counting the Dead: Arizona and the Forgotten Pandemic." Visitors are encouraged to unravel the cords, based off Victorian mourning braids, to distinguish the individuals killed by the 1918 flu pandemic. Photo by Deanna Dent/ASU Now

One hundred years ago, Arizona was a young state, not even a decade old — so young, in fact, that it did not yet report mortality statistics to the federal government.

When it comes to flu fatalities, however, a lack of data is not uncommon, Grumbach said.

“Underreported deaths align with what scholarship has said about the flu, but the lack of data for this flu, particularly in Arizona, is unique,” she said.

What the researchers discovered while digging through data from the Arizona State Board of Health — and many, many death certificates — were a lot of holes, specifically in the reporting of Native American deaths.

Working with documents from the Bureau of Indian Affairs and the work of scholars researching flu cases on the Navajo reservation, Wernimont and Grumbach arrived at a much higher total number of deaths in Arizona than had been previously reported.

“The data was messy and incomplete, so getting the numbers right became a much larger part of the project than we had planned,” Wernimont said. “We had started out thinking naively that there was existing data on this, but once we got through all the death records, it became clear there was a disparity between what was reported and the numbers we were uncovering.”

While the State Board of Health reported just 519 influenza-related deaths in 1918, Wernimont and her team have identified 2,228 influenza-related deaths in official death certificates. They estimate nearly 6,000 Arizonans died in the pandemic in 1918 alone. (The full scope of the pandemic extended through 1920 in some places.)

Drawing on the work of flu scholars Benjamin Brady and Howard Bahr, which uses oral history and ethnographic accounts of influenza mortality in Navajo communities, Wernimont estimates that nearly 4,000 Native American deaths were missed in the official influenza mortality statistics.

Between the death certificates and estimates, the ASU team’s work suggests that the official mortality numbers for Arizona are as much as 1,100 percent underreported.

Unfolding of flu in time and space

Video by Deanna Dent/ASU Now

How we experience data is at the heart of Grumbach and Wernimont’s exhibit prototypeIt is located in the upper concourse of Hayden Library (room C2)., which explores alternative ways of understanding — through touch and sound — the unfolding of influenza in time and space. 

“We spent a lot of time thinking about how to best memorialize human lives,” Grumbach said.

Wernimont and Grumbach are building on the work Wernimont has executed as part of the Human Security Collaborative, which has taken real-time data from cellphones and presented it as sound and tactile feedback for audiences. This historical project is an extension of that work, Wernimont said.

“We’re interested in how you can take numbers and tell stories in ways that encourage affective and sensorial reactions,” Grumbach said. “How can you re-embody the data — take the numbers out of the table and return it to bodies in order to give it a lasting impact.”

In the same way that a quilt resonated deeply with a nation just waking up to the AIDS epidemic, “data represented by feminized labors, such as textile work, can offer powerful narratives that restore humanity, through acts of care, and spur new thinking around complex issues,” Wernimont added.

Grumbach and Wernimont’s “acts of care” can be seen in almost 500 ropes they’ve cut, tied and suspended from above in Hayden Library. Each rope is a 12-strand braid, with each strand representing one person who died from the flu.

Joining the tactile elements, Wernimont has hand-crafted sonification of the data, giving people a sense of how the flu spread across the year. In addition to experiencing what has already been created, Wernimont and Grumbach are encouraging visitors to participate in the acts of care.

“People are welcome to sit with the mortal ropes and unbraid them, in effect, helping to bring individuals even further out of the aggregated data,” Wernimont said. 

“Counting the Dead” will be open to visitors through Nov. 12. A self-guided tour of experiential data in three-dimensional space, the installation prototype is set to coincide with the Society for Literature, Science, and the Arts (SLSA) annual conference, taking place Nov. 7–12 on ASU’s Tempe campus, with the theme “Out of Time.”

“This year’s SLSA conference is about time and memory and spaces,” Wernimont said, “so this fits well with our exhibit. We’re trying to re-embody a forgotten memory.”

Top photo: ASU researchers Elizabeth Grumbach (left) and Jacqueline Wernimont (right) pose with their work on the interactive exhibit, "Counting the Dead: Arizona and the Forgotten Pandemic," on display at Hayden Library through Nov. 12. Each cord represents 12 deaths from the 1918 worldwide influenza pandemic with black cords representing death certificates and gray cords representing oral histories within statistical likelihoods. Photo by Deanna Dent/ASU Now

Britt Lewis

Communications Specialist , ASU Library