The worm’s tail wriggles, a micrometer-scale twitch. A scanner captures the new posture. Software recognizes the motion. Life goes on in the Lifespan Machine, a new system devised in the lab of Walter Fontana that, essentially, counts dead worms.
“Why take five years to make a machine for counting dead things? Because it’s time to become quantitative if we want to understand aging,” says Fontana, HMS professor of systems biology.
The Lifespan Machine was described in online edition of Nature Methods. This tool will enable researchers to make highly accurate calculations of lifespan curves that reveal not only the average lifespan of a colony of a genetically unique strain of worms, but also how soon they begin to die, how long they are capable of living and even when they become sedentary.
The worm, Caenorhabditis elegans, has been used for decades to study aging. Since nematodes live for just a few weeks, entire lifespans can be easily observed. In addition, worms are self-fertilizing, so they create colonies of genetically identical individuals. This makes it possible to study the lifespans of large populations in genetically and environmentally controlled conditions.
In the past, measuring the lifespan of a population of worms required a rather rudimentary and repetitive process—namely, poking each individual worm daily to determine whether it was alive. While it’s relatively easy to tell if that dark mark on a petri dish is still breathing, a researcher can study only about 500 worms once a day. “To do more would be a huge waste of time,” says Fontana. “You need to automate the process.”
The Lifespan Machine does just that. Conceived and built by the paper’s first author and systems biology graduate student Nicholas Stroustrup, the machine consists of 50 off-the-shelf scanners— the same kind you can purchase at any office supplies store— each rigged to scan a spread of 16 petri dishes once per hour. At about 36 worms per dish, the current prototype machine monitors 30,000 animals across 800 dishes. The scanners capture images at 3,200 dots per inch, a resolution high enough to detect movements of eight micrometers, or about 12 percent of the width of an average worm.
Stroustrup has coded software that learns to distinguish worms from, say, lint, and to detect changes in movement from the frenetic, dish-crossing wriggles of youth to the sedentary twitches of old age. He also included a worm browser, a web interface that allows a researcher “to look over the machine’s shoulder” and validate worm identification and death decisions, says Fontana. This feature is key since experiments in the machine cannot be disturbed once started.
“It looks simple on the surface but making it was fraught with dead ends,” says Fontana. “We wondered if it would ever work. But when we had beautiful data and a reason to believe them, we realized, wow, now we can start doing interesting experiments.”
Earlier methods were powerful enough to determine an average lifespan of the worm, but the Lifespan Machine allows the collection of enough data to accurately determine a lifespan distribution curve. This curve is rich with information, such as whether most worms die at the same age or whether their lifespans vary widely. “The entire distribution matters. It’s possible that the average won’t be affected by a genetic or environmental change, but the breadth will, or the onset of death, or the longest possible lifespan,” says Fontana.
In addition, the high-throughput capabilities of the Lifespan Machine make it possible to produce lifespan data for genetically different strains under multiple environmental conditions. “The combined ability to generate accurate quantitative data for many test cases is what makes the method so useful,” says Fontana.
Once Fontana, Stroustrup, and co-senior author Javier Apfeld, instructor in systems biology, determined the lifespan distribution curve of wild-type worms, they introduced single genetic mutations and environmental variations. For instance, they validated the machine using point mutations that had been discovered previously, correctly measuring the lifespan-extending properties of mutations in the age-1 gene and the lifespan-shortening effects of alterations in the daf-16 gene.
Because the Lifespan Machine is so automated, the method is also more reproducible than manual measurements. “One day, another lab could pool its data with my lab’s. We can compare lifespan curves,” says Fontana. “Right now, you always have to run the controls in your own lab. But with the machine, data become more tradable.”
This work was funded by the US National Institutes of Health.