Male researchers might be skewing their own experiments without even knowing it — if they work with mice.
A paper published April 28th in Nature Methods shows that mice get stressed out by the scent of men. If so, that could be a big deal: one of the most fundamental things about experimental science is trying to control for all the variables you can. If it turns out that the sex of the researcher is an important variable, it could raise questions about previous research. Some effects previously seen in rodents could simply be because of the scent of men.
Scientists in Jeffrey Mogil's lab at McGill University had been seeing some weird effects where mice seemed to be in less pain when they were in the presence of an experimenter. So they teamed up with some other researchers and looked into it. (Full disclosure: one of them was a college professor of mine.)
"We found out very quickly that it was half true," Mogil told me. They learned that the presence of a male experimenter or his dirty T-shirt or some molecules associated with men's sweat stressed out both male and female mice. But female experimenters didn't produce the same effects.
When under the influence of these kinds of cues, the animals showed more signs of stress, including having more of the stress hormone corticosterone in their blood and leaving more poops ("fecal boli") in their cages. The stress level was the same as being physically restrained for 15 minutes and wore off after about half an hour. The mice were also more anxious from the scents according to a standard anxiety test.
But wait, why would mice have evolved a stress response to human males? Most likely, it has far more to do with other animals — those smelly male molecules are similar across many different species. The researchers showed that mice also got stressed out when near bedding from a male cats, dogs, rats, guinea pigs, and unfamiliar mice. "Males are up to no good," Mogil says. "They’re hunting, or they’re defending territory. I think it’s reasonable to evolve a stress response until you can determine that there’s no actual threat. "
Does this mean past experiments with mice are contaminated?
Possibly. Mogil looked at archival data from his own lab's pain research in mice and found that the experimenter's sex had significant effects. Although those effects weren't large enough to invalidate any of his previous conclusions, they are still troubling. Here's why.
Stress affects all kinds of things in rodents and people. For example, when stress goes up, pain goes down. Now think about all those laboratories trying to test out new pain drugs on mice. Maybe a lab is having trouble replicating its findings, never realizing that differing odors of its researchers is the reason why.
Stress is also closely linked to many other health problems, including heart disease, diabetes, and depression. So it's possible that this phenomenon could affect a wide variety of studies, possibly even with animal cells that are now in a Petri dish. After all, those animals were smelling whoever was there at the moment they were killed. "That of course is speculative," Mogil says, "but I wouldn’t be surprised if that turns out to be true, in at least some cases."
The good news is that after researchers are aware of a possible male odor effect in their experiments, they should be able to measure and account for it in their calculations, if they want to. Other options include giving the mice time to acclimate to men or simply reporting the sex of the experimenter in research papers.
Susannah Locke: What was it about the men that was freaking mice out? What’s wrong with them?
Jeffrey Mogil: Well, I mean nothing’s wrong. The men are producing stress in the animals. And we figured out that this was olfactory. So we can get the same results by using T-shirts as we can by live humans.
And ultimately at the end of the day, we were able to get the same results by using the individual chemicals themselves that we think are responsible. And so these are chemosignals that are in humans, releasing from the armpit. Other animals of course would release them in other ways. And the chemosignals that produce the effect were ones where men are known to have a higher concentration of them coming from their armpit than women.
But again we don’t think this has anything to do with people, specifically. Because we’re able to also get exactly the same effect by using bedding from any number of species including other mice, as long as the subjects didn’t know them. So strange mice or rats or guinea pigs or cats and dogs. As long as they were a stranger to the subject, male, and had testosterone.
Susannah Locke: Who were the lucky people who got to wear these T-shirts?
Jeffrey Mogil: They were people in the lab — grad students, undergrads. And again we just asked them to wear whatever T-shirt they were going to wear to bed, something that came out of the laundry. They wore it one night. In the morning, they stuck it in a plastic bag. And then we simply took it out of the plastic bag and stuck it on the chair, the same chair where the experimenters would [otherwise] be sitting.
Susannah Locke: And the three ways that you were looking at stress were [the stress hormone] corticosterone ...
Jeffrey Mogil: Plasma corticosterone, fecal boli, and handling hyperthermia. So when you go to measure a mouse’s body temperature, the fact that you’re picking it up and restraining it to get its body temperature, increases its body temperature. And that increase was bigger if a man was taking the temperature than if a woman was.
Susannah Locke: And fecal boli is a fancy way of saying they were pooping?
Jeffrey Mogil: Poop. Yeah, just how many individual poops there are in the bottom of the cage. Simple as that.
Susannah Locke: Were the effects different for male and female animals?
Jeffrey Mogil: Both male and female rats and mice show the effect. Females show it a little bit stronger than males do, but not in any major way.
Susannah Locke: So, what is the thought about why this might be going on?
Jeffrey Mogil: I don’t think it’s too much of a stretch to imagine that if a mouse smells "male," — and it’s important that it has to smell just male. So in one important experiment that was actually forced by a reviewer, we found that a female T-shirt counteracts the effects of a male T-shirt. So it’s not just male they’re smelling. They have to smell just male.
And we think that if you smell just male, well, males are up to no good — they’re hunting, or they’re defending territory. I think it’s reasonable to evolve a stress response until you can determine that there’s no actual threat. And then the stress can go away. And in fact, what we found was that it’s a very robust effect, but it’s not particularly long-lasting. It starts to diminish within minutes, and it’s completely gone somewhere between 30 and 45 minutes later.
Susannah Locke: I thought one of the most interesting things in your paper was you then retrospectively went through years of your own research data and found out that men versus women were affecting it.
Jeffrey Mogil: Yeah, that’s crazy, isn’t it? And it’s a really big difference, too. These are really, really robust differences in what the apparent baseline sensitivity to pain was, based nothing more than on who was doing the testing.
Susannah Locke: Did you find that you’re going to have to retract any old studies because of this? I mean, what is the effect on your research?
Jeffrey Mogil: Well, look, I mean, remember that those are only the baselines of the studies. The study ultimately was about whether a drug or any other sort of manipulation changes that baseline. So it doesn’t invalidate anything really, but there are of course situations where the baseline if it’s higher or lower can change the likelihood of seeing an effect or not seeing an effect. It’s also true that if the baseline is artificially high because of stress and then that stress wears away, the baseline can come down just because of the passage of time. And the drug might not have anything to do with it.
So how much of a confound is it? I don’t know. What I’m really excited about is now that this is about to be published, I’m sort of expecting that I’m going to start to get emails from people telling me that this has explained the mystery in their lab about how they got one effect with one post-doc but then that postdoc left the lab and a new postdoc took it over and suddenly things weren’t the same anymore. So I’m hoping that people are going to sort of reanalyze old data and come up with examples where this is the thing that made the difference. But, you know, we’ll see.
Susannah Locke: This seems like a huge confounding variable that could affect basically every study ever done in a rat or mouse, which is a lot.
Jeffrey Mogil: Well you said it, I didn’t, but yeah. And not only behavioral studies, non-behavioral studies as well. So for example let’s say you’re doing an experiment with liver cells, okay, on some sort of liver physiology. Well, where did those liver cells come from? Well, it turns out they came from a rat that was sacrificed [killed] and at the moment of sacrifice, its plasma corticosterone levels were either high or low, depending on the gender of the person who sacrificed the rat. And it’s highly likely that liver cell physiology is affected by stress hormone levels.
And so it’s possible that this confound can even be there even when you’re doing studies that don’t involve live animals. Of course, I don’t know that for a fact. But I wouldn’t be surprised if people start coming up with evidence that that’s true.
Susannah Locke: So you could be doing research on cells in a dish and when that animal was killed like a week ago, the sex of the person who did that could possibly be affecting your results.
Jeffrey Mogil: Exactly. And that of course is speculative, but I wouldn’t be surprised if that turns out to be true, in at least some cases. For any phenomenon affected by stress, which is most of them.
Susannah Locke: And I guess that reproducibility might also be an issue.
Jeffrey Mogil: I’m glad you brought that up, actually, because I think that these findings have important implications for the issue of reproducibility. So, as you know there’s been a lot of wringing of hands and gnashing of teeth over the idea that animal studies don’t replicate well.
And when people say that usually they lay the blame on validity. They think the problem is that there’s false positive findings that are getting into the literature too easily. Well, I think these data suggest that it’s not that at all — or at least it’s not only that.
It’s possible that when two labs get a different result, it’s not that one of them is right and one of them is wrong or both of them are wrong. They could be both right. That is to say the biology under study could interact with these sort of laboratory environment factors, and both labs are right in their own particular environmental context. We just don’t know always what the environmental factors are.
In my career, I’ve personally been involved in showing that this is true for a number of different things. People used to think that it didn’t matter what mouse strain you use, and now we all know better. And people used to think it didn’t matter what sex you used, and that’s not true either. It really does matter. And now we’ve shown that experimenter gender matters. And God knows what else matters. Maybe the tap water matters, right?
And so what that means is that when people get different results, that doesn’t mean that it was a false positive. It doesn’t mean they’re wrong. They could both be right. It just it’s complicated, and we haven’t quite figured out all of the complexity yet.
Susannah Locke: One of the things that I found interesting when this paper came out was that it reminded me of some other findings we’ve been seeing recently with people examining what the life of a research rodent is like and if that’s representative of people.
Jeffrey Mogil: Or another question is is it even representative of mice?
Susannah Locke: Yes.
Jeffrey Mogil: And it’s a good question. And obviously this is one of these situations where you can choose to see the glass as half empty or half full. Laboratory mice are not like wild mice. For one thing, they’re almost always inbred whereas real mice aren’t. They’re kept in cages of four instead of roaming the wild. There’s lots of reasons to believe that there are differences. But then on the other hand, are those differences differences that affect the thing you’re studying? And sometimes they might, and sometimes they probably don’t. It’s really, really hard to know.
Susannah Locke: Some of the interesting things that I think have come out are particularly about metabolism. I think there’s a line of research questioning "Maybe these animals all just kind of have obesity problems and metabolic issues?"Jeffrey Mogil: It’s absolutely true. And the question then becomes is it close enough. Because obviously there’s not much we can do about it. As a simple matter of economics, we kind of have to use what we have. We can’t go trapping mice and using them because they’re better research subjects. It would greatly increase the cost. Much less research would get done. We’re not going to go and develop different mouse models that we’ve spent the last 50 years developing. So yes there are limitations of mice and rats as research subjects, but the question is could anyone do any better? And I’m not sure it’s clear that we could.
This interview has been edited and condensed. Photo courtesy Jeffrey Mogil.