It's one of life's special moments: a child finds a fat caterpillar, puts it in a jar with a twig and a few leaves, and awakens one day to find the caterpillar has disappeared and an elegant but apparently lifeless case now hangs from the twig.
Then, when the jar has been forgotten, soft beating against its glass walls calls attention to a new wonder: the jar now holds a fragile-winged butterfly or dusky moth with fringed antennae.
These transformations are so startling that a child's awe seems a more appropriate response than an adult's calm acceptance.
How is it, after all, that an insect can remake itself so completely that it appears to be a different creature altogether, not just once, but several times in its lifetime?
Working with fruit flies rather than butterflies, a team led by Ian and Dianne Duncan of Washington University in St. Louis provides part of the answer in the latest issue of PNAS. Ian Duncan, PhD, is professor of biology in Arts & Sciences; Dianne Duncan is a research associate and director of the Biology Imaging Facility.
The puzzling question
Fruit flies go through three main life phases: the larva, the pupa, and the adult.
Earlier work had shown that the larval and adult forms are patterned by the same "signaling systems," or chains of biochemicals that transfer a signal from receptors on the surface of cells to target genes within cell nuclei.
What scientists didn't understand was how the same signaling systems could orchestrate the formation of a larva in one case and the adult fly in the other.
The Duncans, working with collaborator Eric Baehrecke, PhD, of the University of Massachusetts Medical School and graduate student Xiaochun (Joanna) Mou were able to show that a gene expressed only in the pupal stage redirects signaling systems so that they activate a different set of target genes than in earlier stages.
This gene is itself controlled by a steroid hormone that turns on many other genes as well. So insect metamorphosis, triggered by a hormone, resembles puberty, the human analog of metamorphosis, which is also triggered by hormones.
A wholesale change
In 2011, Michael Akam and Anastasios Pavlopoulos, scientists at the University of Cambridge, published a paper in PNAS that described what happened when they artificially turned on a regulatory gene at different stages of a fly's metamorphosis.
Using microarrays that detected the products of gene activity, they found that at every stage of metamorphosis, the regulatory gene subtly increased or decreased the expression of hundreds of downstream genes. But the kicker was that at different stages of metamorphosis, different downstream genes were turned on.
"They found 870-some target genes," Ian Duncan says, "and among those 870, roughly 200 were induced in the larva, more than 400 were induced in the prepupa, and 350 were induced in the pupa, but the thing is, the genes controlled at each stage were almost completely different. So they realized there were global changes of rules from one stage to the next."
"It's as if two teams were playing soccer," Dianne Duncan says, "and at halftime the referee comes out and hands out a new set of rules. Now you've got the same players, the same field, the same goals, but the teams are playing hockey not soccer. The rules are different, so the game is different."
Akam and Pavlopoulos ended their article in PNAS by saying that more research was needed to understand how this repurposing of signaling pathways happens. The Duncans, in reply, are publishing their paper in the same journal.
Throwing two switches
The Duncans focused on a gene called E93 that is turned on by steroid pulses, but only at the pupal stage. "It is required for all patterning and production of new structures in the pupa, but it doesn't play any role in making the larva," Ian Duncan says.
To understand in detail what E93 was doing in the fly, the Duncans chose a simple and well understood patterning process: the activation of a target gene called Distal-less that makes dark spots near the fly's leg bristles called bracts.
The target gene is activated by a well-studied signaling pathway - the epidermal growth factor receptor (EGFR) pathway. "The EGFR signal pathway is used all over the place in the fly," Ian Duncan says. "It's used for different things at different times."
To demonstrate that E93 had to be activated before the EGFR pathway could turn on the target gene, the Duncans looked at flies with mutated E93 genes.
"In the mutant that doesn't have this gene, cells don't respond to the signaling pathway, and bracts fail to form," Ian Duncan says.
Further experiments showed that E93 and EGFR signaling are both needed to turn on the target gene Distal-less. E93 tells Distal-less when to turn on and EGFR signaling tells it where to turn on. Having to throw two switches ensures that the target gene is activated at only the right time and place.
Turning on bracts at the wrong time
As the "two switch" idea suggests, it isn't possible to activate Distal-less at the wrong stage by artificially turning on either E93 or EGFR signaling. Only turning on both E93 and EGFR signaling activates Distal-less.
"If both E93 and EGFR signaling are active, the Distal-less gene will turn on in the larva even though it's the wrong time," Ian Duncan says.
"This finding was important because it showed that E93 can make cells behave as though they're in a pupa even if they're not. E93 says it's the right time."
Bracts and brains
Bracts aren't very sexy, and in fact, the Duncans say, nobody knows what they do. They're just a convenient anatomical feature to study.
The Duncans plan next to study the effect of E93 on the fruit fly brain.
"We know enough now to know that E93 strongly affects brain remodeling in the pupa. Presumably E93 is doing the same thing in the nervous system that it is doing in the leg: it's affecting the responsiveness of genes," Ian Duncan says.
And there is again a tantalizing parallel to human physiology. Not only are the frontal lobes remodeled during puberty, that's also when diseases such as bipolar disorder and schizophrenia tend to manifest.
"There's so much cell death and rewiring during this period," Dianne Duncan says, "it's astonishing that we get through it as well as we do."
"There's even a human homolog of E93 called LCOR (ligand-dependent co-repressor) that is also involved in steroid signaling," she says.
Scientists don't know yet what LCOR does biologically, but odds are they will soon be looking to see.