New studies show that animals used in critical experiments may be out of their minds.
Originally published in Discover.
Paired twirling by lab mice, a stereotypic behavior. Video courtesy of Hanno Würbel.
IN THE EARLY 1990s A SOFT-SPOKEN doctoral candidate at Switzerland’s leading university asked a deceptively simple question: What do all those laboratory mice do after the researchers and technicians go home for the night? It wasn’t a frivolous query. In a typical animal research lab, most rodents’ lives are spent in shoebox-size enclosures containing food, water, bedding, and nothing else, all stacked from the floor to the ceiling on uniform steel racks. Hanno Würbel, the young animal behaviorist who asked the question, knew that mice living in such barren housing often develop bizarre behaviors, such as turning endless backward somersaults. But because mice are nocturnal animals, most scientists are asleep when the critters are active.
So Würbel set up a video camera to record the animals in his lab for 24-hour stretches. “I was interested in animal welfare,” he says. But as a scientist, he had an even greater concern: “If they develop abnormal behaviors, it might mean they are abnormal test subjects.”
When he reviewed the videotape, Würbel saw something reminiscent of home movies made at a psychiatric hospital. In the dark, the mice performed the same useless tasks repeatedly, with such a compulsive persistence that Würbel couldn’t help but think something had gone awry in their brains. In one sequence, a mouse climbs the stainless-steel walls of its cage, hangs from the ceiling by its forelegs while gnawing on the bars, then drops to the floor, only to repeat the process endlessly. On the other side of the cage, a second mouse performs backflips, one per second, for up to 30 minutes at a time. Animal behaviorists refer to highly regimented, repetitive activities with no apparent purpose as stereotypies. Some of Würbel’s mice exhibited stereotypic behaviors for half their waking hours.
Today Würbel teaches animal behavior and welfare at the University of Giessen in Germany and commutes once a week to do research at the school where he did his graduate work, the Swiss Federal Institute of Technology, near Zürich. From a decade of follow-up work, he has reached a disturbing conclusion: Much of the research that relies on animals could be using brain-damaged subjects, jeopardizing the validity of the data it produces. This could mean that disease modeling, pharmaceutical research, and tests of chemical toxicity are tainted. Würbel has made it his mission to alert fellow scientists that just as animals with infectious diseases make unfit research subjects, so do creatures living in impoverished lab conditions. “It took some time for scientists to realize that using ‘dirty’ animals can compromise the validity of experiments,” he wrote in the British journal Chemistry & Industry. “Today, we are about to realize that the same could hold true if we use animals with impaired welfare. It is time to improve housing conditions for scientific, if not for ethical, reasons.”
Würbel is not alone in his thinking. Scattered around the world, a small network of scientists—many of them Würbel’s generation or younger—are challenging the quality of research produced using animals that live in traditional cages. Others are questioning a wider array of lab conditions: whether animals live alone or in groups, how they’re handled by humans, even the effects of the hum of computers or the light from fluorescent fixtures. All these variables, they say, can influence animal well-being and thus undermine experimental data.
Barbara König, a professor of zoology at the University of Zürich, is not surprised that younger scientists are the ones speaking up: “If you grow up with evolutionary thinking or behavioral ecology—and this is now a more common thing—that’s a scene where you think differently about housing conditions. You don’t see animals as little machines that will always behave the same way if you put them in the same environment. You look at individuals differently, and you ask, ‘What are we doing if we ignore the evolutionary history of a species and just keep them in a clean cage with nothing else?'”
A CLEAN CAGE WITH NOTHING ELSE has been the international norm for as long as anyone can remember. Although some researchers enrich enclosures with running wheels, toilet paper rolls, or burrowing materials, the typical lab rodent lives in a sterile environment. “Some practicing scientists pay no attention to animal care,” says John Crabbe, a behavioral neuroscientist at the VA Medical Center in Portland, Oregon. “They don’t think it’s important. The animal is used to provide tissue, or it is the incubator for the disease they’re studying. Unless your particular interest is the effect of enrichment versus impoverishment, there’s no reason to throw anything else in the cage.” Crabbe himself uses standard cages because he is required to house his animals in a centralized facility at the VA.
Scientists often use the smallest enclosures that meet accreditation standards and national laws, which keeps costs low and housing standardized. Conditions like lighting and noise rarely receive much attention. “As scientists, we’re trained in a reductionist fashion, to look at main effects,” says John Capitanio, a psychologist at the University of California at Davis who has studied how housing conditions affect immune function and physiology in laboratory monkeys.
Nonetheless, scientists can’t say they haven’t been warned about the dangers of this issue. The first building blocks for the argument were laid in the late 1950s, when Mark Rosenzweig, a biological psychologist at the University of California at Berkeley, found that an animal’s living environment does affect the development of its brain. While conducting learning experiments on rats, Rosenzweig discovered that animals living in larger cages, enriched with various stimulants such as mazes, ladders, and sponges, had higher levels of acetylcholinesterase, an enzyme used as an indicator of neural activity, in their brain tissue. “Then we had a further surprise,” Rosenzweig recalls. “We realized that the weights of standard samples of cerebral cortex were also varying with experience.” Teaming up with neuroanatomist Marian Diamond, Rosenzweig found that different housing environments change the size of neuron cell bodies and the number of synaptic contacts. The implications, Rosenzweig says, were staggering. “With all sorts of measures—behavioral, neurochemical, neuroanatomical—the researchers might be led astray by using animals from impoverished environments,” he says. “I think a good deal of pharmaceutical research may miss some points because it uses deprived animals.”
In the 47 years since Rosenzweig reported his pioneering work, scientists have come up with more anecdotal evidence that keeping animals under different conditions can dramatically alter research outcomes. For example, lead-contaminated drinking water damages the brains of impoverished mice but not enriched ones. Rats can tolerate 60 times more uranium if they’re allowed time to acclimate to new cages, and even dim light in the lab at night speeds up tumor growth by inhibiting production of the hormone melatonin. “The literature is pockmarked by these little things people have found,” says Capitanio.
The genetics revolution of the 1990s brought the issue into sharper relief. Now scientists can add or remove a single gene in their animals to help them understand the cellular origins of diabetes, depression, alcoholism, anxiety, and cardiovascular disease. Demand for specialty mice for biomedical and behavioral neuroscience research has skyrocketed. Charles River Laboratories, a Massachusetts company that supplies animals to universities and pharmaceutical firms, saw its animal sales increase from $178 million in 2000 to $224 million in 2002, largely because of the development of rodents designed to mimic human medical conditions like diabetes and kidney disease.
Research results spawned by this new technology are, however, made murky by the fact that animals are not tabulae rasae. “A fundamental assumption that underlies this genetic work is that when you change one gene, that’s all you’ve changed,” says Joseph Garner, an ethologist at the University of California at Irvine. “In fact, a tiny change in the environment can change how the gene is expressed.” In 1999, for example, a team headed by Joe Z. Tsien, a molecular biologist at Princeton University, took a group of mice and deleted a gene associated with the N-methyl-D-aspartate receptor in the hippocampus, a receptor considered critical for transferring short-term memories into long-term ones. The genetically altered mice had trouble remembering the smells of cinnamon and cocoa, which were used to scent their food. They also seemed to forget that entering a certain chamber led to a foot shock. These findings were consistent with what scientists had come to believe about how this receptor works.
Then, in a twist of protocol, the biologists altered the animals’ housing environments. For three hours a day, some of the rodents visited enriched cages equipped with running wheels, playhouses, and an ever-changing assortment of toys. After two months, the enriched mice developed the ability to form memories—despite the missing receptors—while the impoverished mice didn’t. “That was a surprise,” says team member Ya-Ping Tang, now an associate professor of psychiatry at the University of Chicago. “Generally, we think the brain cannot compensate so fully for a deficit.” By changing lab conditions, researchers gained an insight into brain function that they would have missed otherwise.
If the only danger of sterile housing were misunderstanding how rodent brains work, the issue could be dismissed as esoteric. However, “most people are not using mice and rats because they are interested in mice and rats,” says Fred Gage, a neurobiologist at the Salk Institute for Biological Studies in La Jolla, California. “They’re a model for human behavior.” But how suitable are they? That, says Gage, may depend on whether researchers provide their animals with environments that allow the creatures to express natural behaviors.
Emma Hockly, a behavioral pharmacologist at King’s College in London, set out to study this in 2001 after she heard that the British government might start requiring enriched cages for lab animals. She wondered how this would affect her experiments with the transgenic mice she uses to study Huntington’s disease. So she divided a batch of rodents into three groups. Some were housed in impoverished cages. Others received cardboard tubes to hide in. Some lived together in much larger enclosures equipped with seesaws, ladders, and cardboard boxes.
Hockly’s mice had been genetically manipulated to mimic human Huntington’s patients with symptoms that include loss of coordination, dementia, and premature death. As the rodents’ health declined, she gave them tests, including a RotaRod analysis, which measures muscle coordination by forcing the mice to keep their balance while standing on a spinning horizontal cylinder. The impoverished mice showed a dramatically faster decline on the RotaRod test than their enriched peers—a decline, she says, that might not accurately mirror the progression of the disease in human Huntington’s patients. “What we’re interested in is a model that has higher predictive power of whether a therapeutic regime works for humans,” Hockly says. “Humans have a very complex environment. If the mice are sitting around twiddling their thumbs like they’re in prison, you probably don’t have a good model.”
MEANWHILE, WÜRBEL WAS WORRYING about cutting-edge research performed with disabled mice. “My first concern was that the more these people examined complex brain function, the more it really started to matter whether these animals were normal or not,” he says, sitting in the office of his Swiss research lab, a suburban Euromodern building with corrugated tin exterior walls and crimson window frames. “If you’re studying mechanisms of learning and memory, then it matters whether your animal has a serious behavioral disorder and is seriously cognitively handicapped.”
Würbel knew that mice housed under standard conditions sometimes spend hours doing backflips, gnawing cage bars, and twirling in place—stereotypies that seem to start out as functional activities (trying to escape the cage, for example) but soon morph into ritualized behaviors. When he added even the slightest enrichment—cardboard tubes, specifically—stereotypic behavior dropped by more than 40 percent.
In the past, many scientists have argued that animals perform these repeated actions as a natural way of coping with the tedium of their environments. Würbel was particularly intrigued by one hypothesis, which argues that stereotypic behavior, like all physical exercise, releases painkillers called endogenous opioids in the brain, allowing animals “to drug themselves away from this adverse world.” He decided to test the hypothesis, reasoning that if an animal cannot perform some natural coping activity, its long-term stress level will increase as a result. He rigged cages to prevent the mice from gnawing the bars, then measured the stress hormone corticosterone in their blood plasma. After three days, the animals that couldn’t chew the bars registered hormone levels identical to those that could chew the bars. To Würbel, that suggested the stereotypies had no useful coping function.
That left a more chilling possibility. Perhaps, he says, “the environment was impinging on them in a way that drives them nuts.” After all, humans with certain psychiatric disorders engage in stereotypic behavior too: Autistic children flap their hands or rock back and forth, while schizophrenics repeat the same inappropriate words. Do unadorned cages produce mentally ill mice? Elsewhere, two of Würbel’s colleagues were asking the same question.
In California, Joseph Garner was also studying stereotypic behavior in animals. “What struck me,” says the ethologist, “was that a lot of people had raised the possibility that these behaviors were related to behaviors in human mental illness, but no one had done an experiment to see if this is the case.” Garner knew the human disorders have been correlated with dysfunction in the brain’s basal ganglia, which regulate the sequencing of movements, among other things. He also knew that when an animal has damage to that part of the brain, it tends to repeat the same actions inappropriately.
To see how well the basal ganglia of stereotyping animals functioned, Garner devised some tests. He took a group of voles that had developed different levels of stereotypy, ranging from 3.5 to 28.1 percent of their activity. When the rodents were hungry, he ran them through a maze with two corridors. At the end of one corridor—always the same corridor for a particular animal—the animals found a glass nozzle that dispensed sugar water. Once they consistently chose the correct corridor, Garner removed the reward. He expected, based on prior studies of the basal ganglia, that a healthy vole would quickly revert to running the maze randomly, while a more stereotypic one would choose the same route over and over, reward or no reward. As Garner predicted, the least stereotypic vole started running the maze randomly after 26 trials, while the most stereotypic animal took 244 trials.
Garner and his former doctoral adviser, Georgia Mason, a behavioral biologist at the University of Oxford in England, have seen similar results with marsh tits that develop severe stereotypies while living in barren cages. The birds repeatedly store seed in the same hideaway even when the seed is consistently stolen from that spot by researchers. Perhaps they know this habit is counterproductive, Garner speculates, but their damaged brains can’t translate cognition into new behavior. That disability could mightily skew the data from a learning experiment.
Corner flips by a caged bird, a stereotypic behavior. Video courtesy of Hanno Würbel.
Likewise, one of Mason’s doctoral candidates, Sophie Vickery, has replicated the experiment with Malaysian sun bears and Asiatic black bears housed in isolation at a government rescue center in Thailand. The bears that paced most repetitively were also the ones who kept returning to the same food dispenser long after it stopped producing fruit treats. Mason has seen similar results in many species. “You can say, ‘Oh well, they’re stereotyping because they want to get out,'” Mason says. “But it doesn’t explain why they might do this hour after hour, day after day, month after month. It doesn’t explain why a vole jumps up and down 45,000 times in a night. It must know it can’t get out. Behaviors should be suppressed if they stop working. They’re persisting in the fruitless attempts for way too long. If captivity alters animals to make them abnormal in their behavior, that raises the question: Are mad mice good human models?”
SOME OF THE SCIENTISTS WORKING on lab-condition issues were drawn by their concern for the animals. Garner, for example, was influenced by zoologist Marian Stamp Dawkins, who writes about animal consciousness and suffering. “If I lived in the time of Star Trek,” he says, “I’d like to be the biologist on the spaceship who learns how to talk to all the animals he meets.”
Würbel is different. “I don’t deny there’s an emotional component involved,” he says. “But my primary motivation is not improving the welfare of the animals. I’m basically doing science, and I want to understand certain things, like how environmental conditions impinge on behavioral development at the border between normal and abnormal. That’s a purely scientific question. I’m fascinated by doing this science in a context where we can examine all the issues, including animal welfare, validity of research, and how to develop useful animal models for human diseases.”
Because these issues, by definition, affect every animal researcher, Würbel believes his work is most useful when it sparks a conversation in the scientific community. Over the past two years, he has launched something of a crusade for better laboratory housing, arguing that scientists risk their data every time they use brain-impaired animals. Last year, for example, the journal Genes, Brain and Behavior kicked off its inaugural issue with a commentary by Würbel arguing that the interaction of genetic and environmental factors determines a creature’s behavior. Put an animal in a wholly unnatural setting, he declared, and its genes will not express themselves normally. “Behavioral genomics is part of a huge endeavor to unravel no less than the biological principles underlying the most complicated manifestations of life,” he wrote. “In such an endeavor, we might have to start to think bigger and join forces to develop the concepts and tools that are needed to address the complexities of life, rather than ignoring them for pragmatic reasons.”
Needless to say, in a community where small, empty cages are the norm, not everyone has been quick to embrace Würbel’s message. “What is discouraging,” says Crabbe, the Oregon neuroscientist, “is the value-laden tone of his arguments for enrichment, which do everything but state that we are mistreating our mice and rats by providing them only with water, three square meals, and clean bedding. There are differences in behavior between mice raised in standard versus enriched housing, but which are ‘better’ or ‘normal’ cannot be straightforwardly answered. Mus musculus, the house mouse, has been raised in ‘barren’ laboratory cages for hundreds of generations, where most breed quite well, and it should at least be considered that this caging condition is, in some sense, their natural habitat.”
Equally bothersome to some scientists is the sweep of Würbel’s argument: his contention that poor lab conditions could potentially corrupt almost any kind of research—not just learning and behavior studies but also experiments that measure physiological and anatomic changes. Last September, Würbel found himself sparring over this issue with Peter Maier, a toxicologist who works for the pharmaceutical company Novartis, during a symposium sponsored by a Swiss animal protection organization. “I’m disappointed by the work of these behaviorists,” says Maier. “They apply conclusions from behavioral experiments directly to biological effects. It’s pure speculation. Of course you can say, ‘The immune system is affected by stress, and stress affects tumor growth.’ But nobody has shown me so far.”
Würbel insists there’s ample evidence backing up his ideas. “The point that the environment might change behavior, but it doesn’t change biology, is ridiculous,” he says. “Every behavior has a physiological background. So, if behavior changes, some physiological changes need to take place.”
Ultimately, Würbel says, his crusade is about more than cardboard tubes and running wheels. On a rainy Saturday morning, sitting in the apartment he shares with his architect wife and 7-year-old son in a leafy Zürich neighborhood, Würbel lets himself imagine a radical new approach to laboratory science. In his ideal world, “the animal becomes less of an instrument to measure something and more of an organism, a creature, an animal that we’re studying for its own sake. I think it involves a paradigm shift: taking these animals and developing a deep interest in how they function and how they work, including their whole evolutionary background, and trying to derive knowledge about these animals for its own sake. I think that will probably provide us with much better and more relevant information, which we might then exploit for ourselves, rather than trying to use these animals in a way that just doesn’t suit them very well.”
When scientists test medicines on humans, Würbel points out, the patients are not isolated in empty plastic cages. They are free to live their lives, enabling researchers to measure how effectively the drugs work in the real world. Würbel believes that if this principle were applied to laboratory animals, it would produce better data. “Somehow,” he says, “I have this vision that there will be a time where we will have natural-like, although heavily managed, populations of rats or mice, maybe in big enclosures, representing whole populations. Depending on the needs of the study, we can then choose our study population, as we do in human trials. I don’t expect this to become the common standard within the next couple of years. But I think it’s an interesting vision to keep in mind—because in the end, this is what in fact we should do.”
Then he pauses. “But you know what the problem is with this? If we get to the stage where we think that we need to treat the animals this way, experimenting on them will probably become impossible—because that would mean they would almost achieve the same status that we have.”