The NY Times headline on Nov 18, 2009, read: “Why Exercise Makes you Less Anxious

I exercise (training for the Paris marathon next April. Woo hoo!) and I’ve been known to dabble in anxiety, so I linked on it for a read (but with due anxiety: what if I find I’ve been exercising wrong? what if I learn that exercise makes people more anxious?). But the article wasn’t about anxiously exercising people, at least not directly. It was about laboratory rodents.

And it’s already made a splash in the blogosphere, despite the scant details in the NYT. So, despite the scant details, I’ll weigh in too, not on what this says about human health, but about how and why this study would be done with animals. NYT reports the “remarkable discovery” that some rat neurons respond differently to stress in exercised versus unexercised rats. Others have chimed in that now there’s proof that exercise is a good thing.

Is this new info? No, not on the grand scale: A lengthy review article out last year (Greenwood and Fleshner, NeuroMolecular Medicine, 10(2): 81-98) begins: “It is now well accepted that habitual physical activity significantly reduces the incidence and severity of stress-related mood disorders, such as depression and anxiety.”  What the Princeton group did is to look at which specific genes are turning on in a particular part of the brain when faced with a major stressor, and found that brain cells that had grown  in response to several weeks of exercise were the ones most resistant to the stress of a forced cold-water swim.

And so, a question you might ask: How do you know when a rat is anxious?

In this study, rats were plunged into cold-water, probably for about 10 minutes. We actually have several ways of testing anxiety in rodents, and drug companies do use them to develop anxiolytic drugs (like Valium/benzodiazepine and its anxiety-busting competitors). You can suspend a mouse by his tail and measure how long he struggles to escape before he gives up. You can put mice or rats on an elevated platform, some of it with walls and some of it a bare gang-plank and measure how much time they spend in the safe vs. the scary part. Wikipedia (of course) has pictures of these standardized platforms, called “Elevated Plus Mazes,” though they’re really not much of a maze. You can simply put your animals in a bare low-walled box and record how much time they hug the walls versus how much they explore the Open Field (a scary place, if you’re a tasty little mouse). Or measure how much time they spend exploring or avoiding a novel object or an unfamiliar individual introduced into their cage. When you run these tests, you find that some breeds of mouse and rat are inherently more anxious/fearful than others, and you can experiment on how genes, environment, drugs affect an animal’s anxiety (or more precisely, her anxiety-related behaviors).  In fact, when you’re shopping for mice the catalog will even tell you what’s known about a given strain’s performance on these sorts of tests (go to and search for “anxiety”).

Scientists believe these anxiety tests are relevant to human anxiety, and just as no human psychologist would administer a single test of anxiety to a patient, they will often use a battery of these rodent anxiety tests to more confidently make their diagnosis. Response to therapy can be a useful part of diagnosis, and we do find that valium decreases what we score as anxious-rodent behaviors. So, these behaviors look like anxiety, and respond to anxiety-busting drugs, bolstering our confidence that we truly are measuring in rodents something similar to valium human anxiety.

Which test(s) to use? Strictly from a rat-vet point of view, and all things being equal, I prefer to see scientists use tests that let the animal choose how much to expose himself to anxiety-inducing situations. Afraid you might fall off the Elevated Plus Maze platform? You can stay in the protected part.  Afraid an owl might eat you in the Open Field? Huddle safely along the wall. Unnerved by a marble placed in your cage? Don’t approach and explore it. Nobody is forcing the mouse to go beyond her comfort zone in these tests. As I read the Princeton abstract, cold-water was chosen so they could confidently elicit a stress-induced genetic event. They wanted to look at something that happens during a significantly anxiety-provoking experience. It’s not ideal, but choosing an anxiety test with an “opt out” clause for the animal subjects would not have worked. To get approval to do this study, I hope they explained to their campus animal committee why they needed this test, how they knew they’d cause just enough stress to get the effect they’re studying – how many minutes in what temperature water – without over-taxing the animals.

Next, a question you should ask: Couldn’t this work be done with human volunteers? Are anxious people in such short supply we can’t get half of them in a study to exercise and half to stay sedentary and ask them about their anxiety levels weeks or months later? Excellent question, and yes, plenty of studies have been done and it seems pretty well established through human studies that exercise is “significantly related to a reduction in anxiety.” So, the <?xml:namespace prefix = st1 ns = "urn:schemas-microsoft-com:office:smarttags" />Princeton scientists plunged rats into cold water and the next day, killed them and removed their brains not to “prove” that exercise can reduce anxiety but to explore how it might do that.

Granted, but you can also ask lots of “how” questions with human volunteers. Improvements in neuroimaging are expanding what can be studied directly in people, using fMRI, for instance, to explore brain regions related to anxiety, and changes in brain activity in response to exercise. Resolution to the level of individual cells, answering whether those individual hippocampal cells that grew in response to exercise differ from other cells, is presently not possible with fMRI or PET. And that cell-level question was the focus of the Princeton study. 

Something you might not think to ask, but that I get paid to ask: Why rats and not mice? I think the NYTimes got it wrong (!) --- the abstract of the Princeton paper says they used mice, not rats, and that’s a smart move. More and more, science that’s established at the physiological level leads to questions at the genetic level, and for genetic-level studies, that usually means using mice. With mice, you can add genes, knock genes out, alter the timing of when they’re turned on or off; that technology is not really available in rats.  Rats were once the mainstay of psychological research, but if you want to look at the genetics underlying the psychology, you’ll need to repeat much of the rat work in mice, verify an effect occurs in mice as in rats, before moving on to look at the genetics. So if you find that 3 weeks of a certain training regimen induces anxiety-resistance in rats, you need to verify the timing, establish the training regimen, work out details of the cold-water stress in mice before you finally get around to being able to do the experiments you want to do, looking at the role of genes. It makes sense to bypass rats, and I could see this Princeton work opening up questions (answerable more in mice than in rats, since we can manipulate mouse genetics so much better) like how all this anxiety-exercise interaction works in animals who have defects in insulin production, in aerobic capacity, or any of dozens of other metabolic conditions.

One final point a rodent vet needs to think about: Studies of exercise in rodents sometimes force rodents to exercise, while others allow them to exercise. The NYT says the Princeton group “allowed” exercise, though other papers from that group force the exercise. If you think I’ll fuss about forcing animals to exercise, you’re wrong. Yes, there can be the occasional study that forces exercise to the point of exhaustion and distress or injury. But studies like the Princeton one, looking at psychological benefits, aren’t likely to go to those extremes, and rodents seem to adapt to enforced moderate exercise pretty well. If the forced exercise (15 minutes a day on a treadmill would be a typical regimen) leads to emotionally healthier animals, I can’t get too worked up about it.

What concerns me more are the “allowed to exercise” studies, for what they tell us about how we treat laboratory animals in general. For every group of experimental rats “allowed” to exercise, there’s a control group forced to remain sedentary. Most American labs house mice in cages that just barely meet the standards of the NIH’s Guide for the Care and Use of Laboratory Animals. That book, last updated in 1996 and under revision for an updated edition now, sets a standard that “Animals should have opportunities to exhibit species-typical activity patterns.” Wheel-running is a “species-typical activity pattern” in rodents (no, they don’t do it in the wild, but give them a wheel in their cage, and they’ll run for miles every night). Not only are most lab rodents housed without a running wheel, now we see one more piece of evidence that if they were, they’d possibly be less anxious all around. They are the "not 'allowed' to exercise" control group, and they number in the millions. We see one more piece of evidence that simple questions of how we house our animals can have some pretty profound effects on the research results we get.