Dieting, Stress, and the Changing Brain

Feb 09 2011 Published by under Behavioral Neuro, Uncategorized

Dieting doesn’t work. No really, it doesn’t. Much as people like to talk about how the things they are sticking to are “lifestyle changes”, etc, a lot of them (but not all) are still diets, still ways of restricting your eating, and over 80% of them will fail. By fail, scientists mean that you will gain back all the weight you lost, and then some, within five years. This is not only bad because of the weight people end up gaining, but some people believe that the “yo-yo” dieting style itself can contribute to the beginning of metabolic disorders later in life.

So why DO the diets fail? Why can’t people stick to them and why can’t they keep the weight off even after they’ve lost it? While some people may just claim weak willpower, neuroscientists know that it goes a lot deeper than that. And it turns out that going on a diet may change the way that your brain reacts to stress, changing both how you express stress and whether you start eating for comfort because of it.

Pankevitch et al. “Caloric restriction experience reprograms stress and orexigenic pathways and promotes binge eating” J. Neuroscience, 2010.

It’s time to put some mice on a diet.


In order to mimic dieting in humans, the authors of this study took a bunch of mice and restricted the amount that they could eat, to make them lose 10% of their body weight. That’s the equivalent of a 150 lb person losing 15 lbs, and is about what people on average lose when they go on a diet. They had another group of mice that ate whatever they liked. At the end of three weeks, the diet mice had lost 10% of their body weight (about 2.5 g for a mouse), while the non-dieting mice had gained weight (as mice do). They then took these two groups and looked at their stress levels.

The first thing they looked at was a response to stress.

The top figure here is a measure of corticosterone, a hormone strongly associated with stress. They took the mice here and restrained them in little tubes (not too tight, you usually just prevent them from escaping or turning around, and often people wrap them in saran wrap like little mouse burritos) for over an hour, measuring their blood levels of corticosterone. Mice don’t like being restrained, it’s stressful, and both groups had corticosterone go up. But the dieting group had a much higher response to stress than the non-dieting group.

The bottom figure shows something called the tail suspension test, which is a test usually used to detect the actions of clinically effective antidepressants. You can see that the dieting mice (shown on the right) had a higher immobility level in this test, which is sometimes associated with pro-depressive behavior. This suggests that, not only are dieting mice high stress, they may also show more depressive symptoms (in major air quotes) than non-dieting mice. But I’m not sure about that interpretation and I’ll get back to it later.

But where is this stress response coming from? The scientists looked in two areas, the first was the central amygdala, an area that has previously been shown to be involved in stress responses. They were looking for corticotropin releasing factor (CRF) a hormone that is upstream and controls corticosterone responses.

Unfortunately the Amygdala ended up looking by a big pile of nothing in terms of CRF differences. But when they looked in another area, the bed nucleus of the stria terminalis (BNST, and I have no idea why they call it the bed nucleus), they found some interesting stuff. You can see above there in B that the dieting mice (dark bar) had low levels of CRF, and those levels STAYED low even when the mice were refed on a regular diet (the grey bars).

But the question is, WHY are these levels low? To find that out, the scientists looked at DNA methylation, which is a very hot new field right now. The idea is this. You know you have DNA, and that DNA is a code which is transcribed into RNA and then translated into protein, yes? We call that the Central Dogma. Once you have a protein, you can then keep it around, break it down, etc. So when levels of a PROTEIN (like CRF) change, is that because the protein is being broken down more than before? Does it mean the protein isn’t making it as much out of the RNA stage? Or does it mean it’s never made it from the DNA stage to RNA in the first place? This last one is particularly important, because you can change whether or not DNA is available for coding by several mechanisms. Today we’re focusing on methylation. When DNA gets methylated (in this case) it gets chunked up, and is less available to be made into RNA and then into protein. So the protein levels will be lower. But the really crazy thing about this change in methylation is that it’s very long LASTING. Cells don’t methylate lightly, and once they do, the change can go on for years. In some cases, if the methylation happens to DNA in the testes, for example, that change can actually make it into your offspring!

In this case, the jury is out on the offspring, but when the scientists looked at mice that had diets compared to those that didn’t, they saw that the methylation of the CRF gene was decreased, both in dieting mice and in mice that had previous had diets but were now being refed. This suggests that the mice that had had diets retain the changes in stress hormone expression that they had during diets. The pounds may not stay off, but the methylation will stick with you.

And it turns out that these changes will affect how the mice react to stress later on. The scientists took the mice that had had diets and those that hadn’t (AFTER they had been refed so they aren’t on diets anymore), and put them in a paradigm called chronic variable stress (CVS). This is a pile of different environmental stresses that the mice are exposed to, to give them just a lot of little stressors that get to them (kind of like grad school, only for mice). So the mice might get 15 min of being wrapped up like a burrito one day, and the next day they might get a little red ball put in their cage (mice don’t like new things), and the day after that they might get a period of being left in the dark in the middle of the day, and then they might get the radio left on, etc etc. None of them cause pain or hurt the mouse, but they are all annoying, and the mouse never knows what’s coming next, leading to a stressed out mouse.

Then they gave the now stressed out mice the opportunity to eat a high fat diet, to see how much they ate. It appears in this study they used an actual high fat diet that you can buy from a company, but a lot of labs use peanut butter chips or Crisco (mice LOVE Crisco).

You can see in the two top panels that the mice who had been on diets and mice who hadn’t generally ate the same amount of high fat food when unstressed. But when they were stressed, the former diet mice ate more high fat food than the non-diet mice did. They didn’t go whole hog or anything, but I think this effect could be higher with more stressors (they only used one stress per day, but many labs use THREE). So apparently having been on a diet makes mice more likely to eat their feelings in times of stress.

They did a bunch of other stuff in this paper (it’s a very fat paper), like looking at hormone responses to high fat diets (which are higher in mice that have been on diets), and that you can stop the high fat overeating by giving mice an MCH antagonist (MCH is a melanin response hormone and is another feeding related hormone). But the take home message is this: Dieting not only causes you to lose weight, it interacts with the way your body deals with stress, and makes it more likely for you to eat more when times of stress come around again. Over the long run, this could result in…weight gain, screwing over all you worked for.

Of course this study raises more questions than it answers. Here are a few of mine:

1) What was cort like DURING the diet? Is the dieting itself stressful? It appears that it was, as the dieting mice had higher levels of CORT at baseline (before the stress test), but it’d be interesting to see if there was a stress response to the dieting during the diet itself.

2) Does this stress effect get WORSE if you diet them again? Does it get better?

3) Why MALE MICE?!?! There are far more women than men on diets, I know that male mice are “easier” (scientists often use male mice because working out the cyclings of the females and the effects of cycling is really annoying, and ovarectomizing the females doesn’t give a picture of what’s really going on), but I think female mice would be more relevant.

4) The depression test is interesting (The tail suspension I referred to up there). Other studies with MORE caloric restriction, causing mice to lose up to 30% of their body weight, have shown increased struggle in these tests, but that could be related to the high locomotor activity seen in models of anorexia. Regardless, is there proof that dieting makes you depressed? Has anyone looked at this in humans? There are lots of possible confounds here, mice don’t care about their self-esteem and certainly don’t base their self-esteem on fitting into their skinny mouse jeans, so could this kind of effect be counteracted in humans, who might feel good about their success as they lose weight? Edit: I realize I didn't explain how I thought the tail suspension test might be working here. Tail suspension measures don't always work as a measure of "prodepressive" behavior. Rather, the measures can sometimes be interpreted as a "passive" or "active" response to a stressor, which is THEN interpreted as "prodepressive" behavior. In this case, I think that perhaps the more passive response to the stressor might be a key feature. Of course, how to get at that is a big question.

5) I notice that the dieting group had cort that went up and then down, while the non-dieting went UP the whole time, is this an adaptation to stress in the dieting model? Or did they just hit a ceiling?

6) The dieted mice do get back to weight when you feed them normally. Does this mean they are eating more normal food from the get go? Because that is also really interesting.

Despite all these questions, I think this paper is very interesting. It not only shows THAT diets don’t work, it provides a lot of insight into HOW. And research in this area could help scientists study how we can change those stress reactions and stop stress responses to diets. After all, when are people today NOT stressed? And if dieters eat more when stressed, and they are ALWAYS stressed, take makes them doubly doomed to failure.

Pankevich, D., Teegarden, S., Hedin, A., Jensen, C., & Bale, T. (2010). Caloric Restriction Experience Reprograms Stress and Orexigenic Pathways and Promotes Binge Eating Journal of Neuroscience, 30 (48), 16399-16407 DOI: 10.1523/JNEUROSCI.1955-10.2010

25 responses so far

  • Fab Kate says:

    I read this study last week, and I was pretty upset about it, because it would seem that we're DOOMED to failure. The thing is, we're not mice, and we have different motivations, which must be able to overcome stress at times, since some people DO keep the weight off. At the same time, I'd guess we humans undergo more stress than your typical lab mouse. I know in a recent study mice would accept electric shocks to get at high calorie foods, and that how we eat changes our metabolism and brain chemistry, but whether or not it's merely hope or some sort of dieter's delusion, I have to believe that early eating issues don't preclude ever achieving long term fitness later in life.

    • scicurious says:

      I don't think we're doomed, honestly. I think that it means that we need to look at the way that we respond to stress, and work with keeping people relatively unstressed. Also, as you mentioned, there are MANY other factors that go into what humans eat and why. And this may also change the way we work with eating and changing eating habits to get people to lose weight, maybe putting more focus on exercise and less on food restriction, or something (though interestingly, exercise isn't as good for aging as food restriction is).

  • lcoole says:

    Stress is part of the world we live in and helps us respond appropriately to situations. Reducing stress is a smart thing when it can be done, unfortunately, there's often little we can do about the ongoing stress of relationships, work, family and friends. As long as it's under control, its okay.
    But what this study seems to stress is how diet can increase stress response or help moderate it. If we choose lifestyles based on healthy eating, we're much more capable of dealing with all the other problems in our life. When diet is formulated only for weight loss, it is unsustainable. For our friends out there that want to lose the pounds... choose a diet that emphasizes healthy foods - FOR THE REST OF YOUR LIFE. It's not something you have to be scared of and it doesn't mean you have to eat bland food either. We evolved to consume fruits, vegetables, grains and meats - simply prepared and those are the foods that help our bodies function at their best.
    Let's not forget the power of exercise, as well, to reduce weight AND lower stress. The answers seem simple to me.

    • scicurious says:

      The answers are not simple to everyone, though. We may have evolved EATING whole grains and fruit and meat and all. But we have also evolved to WANT fats and sugars, which provide far more bang for the buck, and which are much more pleasurable. I don't think this study shows that diet moderates a stress response at all.

      And I'll go ahead and say it: eating healthy for the rest of my life is an option, but is it WORTH IT?! Not to me. I think there are ways we can work with how we deal with stress, and I also think there are ways we can work with nutrition and exercise to combat overeating and weight gain without undergoing the kind of restriction that has been shown here quite clearly to be stressful.

  • Jeanne garbarino says:

    My biggest beef with this paper is that I'm worried the general public will interpret this as a green light to eat whatever they want. Obesity is a huge epidemic and, even though this study has it's flaws, perhaps it can help to point us in the right direction regarding how people with metabolic disorders (obesity, CVD, diabetes, etc) can implement the necessary "lifestyle changes" in a more permanent manner.

    • scicurious says:

      I don't think the public will interpret it that way, we all know that diets don't work in the long run (even as society continues to recommend diets). But there's certainly nothing in this paper that says to go ahead and eat twinkies.

  • Katie says:

    Is there a difference between dieting and just reducing your caloric intake back to maintenance levels? For example, people gaining weight are eating more calories than is needed to maintain their weight so most recommendations are to eat 500 calories less than maintenance per day to lose the weight. What if the people just went back to maintenance calories? Is that still dieting?

    Another question: the tail suspension test is usually used to assess the effects of anti-depressants. Is it commonly used as it is in this study as well: as a baseline measure of depression?

    • scicurious says:

      Yes, the tail suspension IS commonly used for measures of "pro-depressive" behavior, though they usually require other methods (forced swim, novelty induced hypophagia, sucrose intake) as well to really talk about "pro-depressive" behavior.

      As for maintenance calories...I don't know. I think it might have a lot to do with how people view maintenance calories. Is it still "restriction" to them?

  • Katie says:

    That's what I thought about TST but a reviewer recently said otherwise.

    Good point on the maintenance calories. It seems like it would feel like restriction in many cases. For example, if they've been eating over for a long time (even going only slightly over) or going way over for even a short time. But I don't study diet so I'm just speculating.

    Katie aka katiesci on Twitter

  • [...] This post was mentioned on Twitter by Scientific American, KentBottles, Kathleen O'Neil, ClinPsy, LindaVansteenwinckel and others. LindaVansteenwinckel said: Dieting, Stress, and the Changing Brain | Neurotic Physiology blog by @scicurious [...]

  • AK says:

    Don't forget that rodents have been evolving separately from primates for at least 50 million years, perhaps as much as 80MY. Quite a few behavioral mechanisms may have evolved differently, coming to depend on (potentially) different neurotransmitters, in (potentially) different parts of the brain.

    When it comes to dieting, can we be sure that low food consumption due to no food has the same effect as when its due to self-restraint? Mice are social mammals (although their sociality has been evolving separately from primates for 50-80MY and may even have appeared independently), so why are we assuming that low food consumption due to not enough food (alone in a cage) will have the same effect as if it were due to social conditions. Ideally experiments should be done with mice in social conditions where some are forced to lose weight because they're denied enough food due to competition.

    And for humans, does dieting count as not enough food, or due to social conditions?

  • Hi Scicurious -

    Very neat post. Thank you.

    It might be nice to see what the shifts look like in calorie restriction models; which seem to show increases in lifespan. It would seem that if dieting could manipulate your stress response maladaptively, it seems possible this would have shown up in longevity measures in CR studies. (?)

    The piece regarding the persistence of effect even after diets have stopped was especially cool. Extending this out to humans is really tough; I'd bet that all kinds of things jingle up our methylation in different ways. What does the profile look like for the dieting guy who also takes a martini and a bonghit once a night?

    And more importantly, can you please craft another review I can use as evidence that the very foundation of my DNA transcription operating at peak efficiency involves us having more sex. I keep telling her science is on my side, but an impartial reviewer would help a lot.


    - pD

    - pD

  • Errr , should read:

    And more importantly, can you please craft another review I can use as evidence that the very foundation of my DNA transcription operating at peak efficiency involves my wife and I having more sex. I keep telling her science is on my side, but an impartial reviewer would help a lot.

    Note: I do not know scicurious. [Sorry]
    - pD

  • becca says:

    Yeah, are we sure this isn't an effect of dieting because of direct mechanisms but the effect of chronic stress on stress responses? It sucks to not have enough food. However, it's different if you *have* food but *choose* not to eat it. No way I can see to test with mice, but it's kind of a different thing.

  • olegt says:

    Hi scicurious,

    Long-time fan here.

    Might you be interested in covering this? Death rattle of a decapitated brain at New Scientist.

  • Lotharloo says:

    It's a very interesting post but have these results been replicated in humans? If not, shouldn't we be cautious about generalizing them? Does caloric restriction have the same effect on stress in humans? What are the effects of different nutritions on stress levels? Does exercising reduce stress? And besides, one can be overweight and healthy at the same time ( ) so shifting from an unhealthy diet to a healthy one is beneficial even if it the latter does not result in weight loss.

  • Mike B says:

    Honestly I've half-suspected at least part of the source of the obesity epidemic in has been increasing social stress; be interesting to see if the problem has worsened with the current economic climate, though I'm not sure how you'd control for the "unhealthy foods are cheaper" issue. The fact that serotonin is IIRC linked to both appetite and obviously depression (most of it is actually made in the gut as I remember) is rather suggestive of a link between mental stress and appetite level, and evolutionarily I could see there being a possible selective benefit to encouraging stressed individuals (because of, say, living in a predator-rich environment) to "eat and run" as it were. Very interesting area of research.

  • JEQP says:

    The methylation factor is interesting. It's been known for quite a while that people going on diets get cravings to eat when they're not hungry, and one of the things that starts this is stress. This could be (maybe one of) the mechanisms.

    That's why a good diet will teach you to recognise cravings and how to deal with them. It certainly is a lot more complex than "eat less, move more".

  • [...] recently covered a mouse paper on dieting and subsequent high fat eating in mice, and then she found THIS paper, on stress reactions in overweight humans! And I like how the two [...]

  • Jake says:

    One more case made for the paleo diet, in my mind. Or really any diet that isnt calorie restrictive, but more focused on which kinds of foods we eat.

  • Chris Cline says:

    OK, so I have a simple question, and having looked briefly at your CV, I'm guessing (happily) that you can answer it: how can corticosterone levels in dieting mice be HIGHER if the CRF levels are LOWER? It would seem that if the thing that stimulates the release of corticosterone is down-regulated (putatively by methylation of the genes and therefore decreased transcription), then corticosterone levels would also be decreased. Is there another transcription pathway that takes over, or that is hyperstimulated? Also, the whole CVS thing looks interesting, but not all that impressive. I'm assuming that because there are asterisks over the histogram bars that the differences between control and treated are significant, but they don't look "significantly significant." Which is not to say "not interesting." Very interesting, but I think that my stress associated with the fact that I'm 25 lbs overweight and that I don't have any pants that fit, in combination with the physiological stress on my knees, heart, pancreas, etc., probably outweigh the stress I feel associated with not having enough to eat. And hopefully my knowingly taking on a diet (vs. being a basically healthy mouse whose diet is suddenly and capriciously reduced) will be less stressful. But thanks for the good interpretation of the forest of neuroscience for the lay person. Good job.

    • scicurious says:

      Ok, so CRF stimulates ACTH, which stimulates corticosterone. But here's the deal, you can have lower levels of CRF and thus lower levels of BASELINE cort, while the STIMULATED response of cort to STRESS is high. One was baseline, and the other stimulated under stressful conditions, and I would bet that makes all the difference here. What I would hypothesize is that the receptors for ACTH in the arenal cortex (or the receptors in the HPA that release ACTH itself), which produces cort in response to stress, are probably MORE SENSITIVE than normal in these animals. This would mean that even though they had less CRF at baseline, the system is more responsive to CHANGES in CRF, and thus the cort response might well be bigger even though there are lower baseline levels of CRF present. You could maybe look for this by grinding up the adrenal cortex of the mice and looking for ACTH receptors, or by looking at ACTH levels and release from the pituitary, OR by looking for CRF receptors in the pituitary.

      Does that make sense? If it doesn't I can try again. :)

  • Chris Cline says:

    Also (follow-up to my post above)-- aren't there genes that code for proteins that go along the DNA strand and do repairs, specifically demethylation of DNA? I thought that there were a lot of things that goober up DNA, and so it's been evolutionarily advantageous to have a mechanism to fix it... ??

    • scicurious says:

      Ok, Good questions! I'll get this one first cause it's the quickest. There are indeed things that de-goober (I shall start using this word in the lab) your DNA. BUT, DNA methylation is not necessarily a bad thing and usually is intentional. The things that degoober your DNA are usually to fix mutations, etc, rather than fixing up methylation. There are obviously also proteins that specifically demethylate DNA, but they usually react in response to a stimulus. So basically the idea here is that the methylation or lack thereof of the DNA is an intentional response to a stimulus, and makes SENSE for the cell to do that, just, well, it's not so hot for you in the long term. It makes SENSE for your body to want to hold on to excess weight in times of stress, and the methylation allows it to keep doing that. Does that help?

Leave a Reply