Scientists aren’t sure why identical twins differ. Armadillo quads offer an answer

One of the armadillo quadruplets runs away when she hears Amanda Withnell approaching; her three siblings calmly keep their faces in their food bowls. Another, seemingly unable to summon the bravado you’d think would be standard equipment on a little oblong tank with feet, is a neurotic mess compared to his three mellow siblings, regularly running panicked into walls or jumping straight up like a jack-in-the-box — a patented armadillo move.

Withnell can keep track of which nine-banded armadillo quadruplet is the shy one or the neurotic one because the siblings often differ in appearance as much as they do in behavior. One has a touch of white on its left ear while the other three have it on the right; one has a symmetric blaze just above the nose while another’s blaze will tip left as if drunk.

“Among siblings, there are often differences in facial pigmentation and sometimes even in the number of vertebrae,” said biologist Frank Knight of the University of the Ozarks, one of the world’s leading experts on the nine-banded armadillo and Withnell’s husband. Differences from one quadruplet to another go beyond appearance. “Every year we see variations in personality,” said Withnell, an animal technician who, with Knight, raises armadillos for a government-run research center in Louisiana.

It’s no surprise that siblings, armadillo or human or otherwise, differ. But armadillo quadruplets are genetically identical, the result of a single fertilized egg splitting in half, and the two halves splitting in half again, before implanting in the uterine wall months later, a reproductive strategy unique in the animal kingdom.

So while biologists have long used armadillos in research aimed at detecting, preventing, and treating Hansen’s disease (leprosy) — they’re one of the only mammals other than humans who can harbor the bacteria that cause it — scientists at Cold Spring Harbor Laboratory had a very different thought: The long-nosed, armored creatures would be perfect for studies with a potentially greater reach, namely, probing the power of DNA.

“Uniquely among mammals, armadillos always have genetically identical quadruplets,” said CSHL computational biologist Jesse Gillis. “Mice are bred to be genetically identical, but armadillos arrive that way naturally.” Studying details of armadillo genetics beyond the basic sequence of A’s, T’s, C’s, and G’s that quadruplets share, he figured, might reveal something about whether genes are as fateful as the current infatuation with genetic testing, DNA-based personalized medicine, and personal genome sequencing seems to assume.

“I do think our research has bearing on the tendency for seeing genetic processes as deterministic,” Gillis said. “Really, our genes have a strong incentive to set things up flexibly.”

Researchers who study nine-banded armadillos (Dasypus novemcinctus) had long noticed that genetically identical quadruplets frequently look and act differently from one another. Behavioral and personality differences, they figured, reflected different life experiences; maybe one sibling had a scare as a newborn and so was jumpy forever after. Such environmental influences also explain many of the differences between identical human twins.

Physical differences, between “identical” human twins or “identical” armadillo quadruplets, were harder to explain away. They might reflect slightly different conditions in the womb, but since the quadruplets share a placenta, such differences should be slight.

Whenever scientists calculate how much of a trait is heritable and how much is environmental, they end up with a total less than 100%. That is, besides genetic and environmental influences, there is “unexplained, non-heritable ‘noise’,” Gillis said. He and his colleagues therefore set off on the trail of that noise.

Identifying that noise could prove relevant for people. Identical human twins often differ in many strongly genetic traits, such as bipolar disorder, schizophrenia, and breast cancer. For type 1 diabetes, 61% of twin pairs are “discordant” (one has it, the other doesn’t); for autism it’s 60%, for schizophrenia 58%.

Yet identical twins (from a single fertilized egg) have identical genomes. If one has a genetic variant linked to a disease then so does the other, and if one escaped being dealt a bad genetic hand then the other did, too. A few small studies had hinted that differences in which of their identical genes were activated and which were silenced caused those discordances, but the evidence was sketchy.

Enter nine-banded armadillos. “They present a unique opportunity,” Gillis said. “In people, twin studies are the gold standard for measuring heritability [of traits]. Because armadillos always produce litters of identical quadruplets, it gives you even greater statistical power” to estimate the effects of genetics, environment, and the mystery “noise” component.

Scientists at the Hansen’s disease center in Baton Rouge shipped samples of blood and genetic material from five sets of quadruplets, taken at three time points, to Gillis’ lab. He and his colleagues sequenced the animals’ DNA and its downstream cousin RNA — a first for a full set of quadruplets — and ran a series of molecular analyses, focused on changes to gene expression over time, comparing siblings to each other.

What they found is that differences in gene expression start early, when an armadillo embryo consists of only about 25 cells. That’s when females, who carry two X chromosomes in every cell, begin silencing an X in each one. Whether the X inherited from mom or the X inherited from dad is silenced is completely random — a coin flip — but determines whether maternal or paternal X-linked traits are expressed. X inactivation is therefore the first source of trait differences in genetically identical siblings, Gillis said: “It creates permanent variability between individuals.”

In their other 62 chromosomes (armadillos have 32 pairs, compared to humans’ 23 pairs), quadruplets had an average of 700 genes that differed in whether they were expressed or not, Gillis and his colleagues report in a preprint posted to bioRxiv. The source of that variance is random silencing of genes, or epigenetic changes. By chance alone, some genes get clobbered by clusters of atoms called methyl groups, which sit atop the DNA like a soundproof ceiling and keep it silenced — the trait it codes for unexpressed.

By tracing back cell lineages, the scientists estimated that the epigenetic silencers hopscotched around the genome when the embryos each consisted of a few hundred cells.

To put those 700 genes in perspective, armadillos have about 20,000 genes, as humans do. For identical siblings to differ in 700 is approximately equivalent to differing by fully half an X chromosome’s worth of genes. The set of 700 that differed between siblings was unique to each set of quadruplets, underlining the randomness of the silencing.

It’s certainly possible that armadillos have different individual experiences growing up, but for the most part, those in captivity experience the same environment. Their genomes are identical, too, of course. The epigenetic differences therefore stand out as an important source of individuality, Gillis said. “We estimate that individuality is encoded by a small fraction of the genome,” he and his colleagues wrote.

All told, they estimate, the randomness with which genes are silenced or not at the very dawn of embryonic development accounts for about 10% of individual variability.

Similar calculations have been done for animals much more distantly related to humans, including crayfish and fruit flies, where scientists estimate that as many as 25% of genes in genetically identical animals raised in identical (lab) environments are differentially expressed. If not the genome and not environment, then the source of variation must be randomness.

That’s a crafty evolutionary play, said evolutionary biologist Benjamin de Bivort of Harvard University. “Introducing random [genomic] changes can be strategic,” he said. “It’s another mechanism by which individuals can differ, increasing the odds that some will survive even if threats like predators and pathogens, or environmental conditions, change. When you don’t know what the future has in store, it’s adaptive to have a way to survive [as a species] under all contingencies.”

In ‘dillos, some of the traits that differ due to random gene expression or repression are visible to the naked eye, while others are detectable only through molecular testing. One armadillo sibling often differs from another in size as much as it does from an unrelated ‘dillo, for instance, while other siblings differ in their immune system, cardiac muscle growth, and number of scales.

Random genetic events can even shape the animals’ eponymous nine bands. In one sibling, two bands might fuse, Knight said, giving a nine-banded armadillo eight bands.

The widespread genetic randomness present from birth in armadillos, a fellow mammal, should spur more research on whether the same is true in people, Gillis said. A few studies have found DNA-expression differences between identical twins, which can be as high as 20%, including in regions that affect the risk of developing lupus, bipolar disorder, and schizophrenia.

That underlines how having a particular genetic variant is not necessarily synonymous with having a trait, including a disease risk. DNA testing, even genome sequencing, can’t tell which variants are silenced and which are expressed. Yet many DNA-testing customers believe that their fate lies in their genes. “Genetic testing is the most prominent current example of genetic determinism,” Gillis said.

Scientists who study armadillos as something more than collections of genes have wondered when molecular biologists would get around to appreciating all ‘dillos have to offer. “It’s an obvious model for [studying] heritability, and underutilized,” Knight said. “It seems like there’s a lot more” geneticists could learn from these four-footed tanks.