Starve a fever, feed a cold...or die

Eating during a viral infection may be beneficial while deadly during a bacterial one.

Go to the profile of Michael Chao
Sep 15, 2016
Upvote 3 Comment

The folklore of food and infections has bounced around in our cultural consciousnesses for hundreds of years, with the adage 'starve a fever, feed a cold' in the English language for around 500 years. But the science behind this idea has oscillated back and forth in the literature and in the press (see this NY Times piece or this one or this one). Does food support the immune response? Raise body temperature to kill pathogens? Take up precious energy via digestion? Scroll through the comments sections and you'll find a variety of gut (pun!) feelings and rationalizations. But everyone can agree at that though the benefits of chicken soup or food for infections is unclear, this 'treatment' has been followed myriad of times over in homes all over the world, so at the very least it clearly can't be bad advice, right?

Well, not quite. In the literature, at least, it really depends on the infection. For example, caloric supplementation in mice has been shown to enhance mortality during Listeria infection, and some studies suggest that fasting is associated with better survival in bacterial sepsis. So is eating good or bad during an infection? Is grandma trying to cure you or kill you with that bowl of chicken soup? How do we reconcile these conflicting scientific findings? That's been the larger question for decades in the absence of mechanism, and that's what Medzhitov and colleagues strove to answer in their work in the most recent issue of Cell.

The authors evaluated the impact of calories and metabolism in two lethal mouse infections--with Listeria monocytogenes or influenza. In Listeria infection, they found that either gavaging food or even injecting glucose syrup could induce 100% lethality in mice. Subsequent injection of the glucose antagonist 2DG could rescue mortality and decrease bacterial burden, but 2DG did not affect bacterial growth in vitro and cytokines associated with increased immune clearance (IFN-gamma, IL-6) were unaltered. They replicated these findings in a LPS model in which mortality is due to excessive inflammatory response--caloric supplementation again enhanced mortality, while inhibition of glucose utilization could rescue. Interestingly, this aggravating effect was only found for glucose as caloric equivalents of olive oil or casein had no effect on mortality. And again, no overt changes in major inflammatory cytokines was observed with glucose, suggesting glucose utilization does not impact the immune response per se but rather helps the host better tolerate the impact/damage from these responses.

OK, so at this point a candy bar is a bad idea for Listeria, but what about for viral infection? Repeating the experiment with influenza, they surprisingly found the opposite effect: glucose supplementation now protected against death while 2DG treatment induced 100% mortality. Viral titres were no different with glucose treatment, again suggesting modulation of host tolerance rather than induction of antiviral responses. Trying to understand why the mice died, they found that there was no exacerbation of lung damage with 2DG treatment, but treated mice did have decreased heart/respiratory rates and lower body temperature, pointing to some type of autonomic dysregulation. Interestingly, these 2DG-mediated autonomic symptoms could be reproduced with poly(I:C) treatment, which models viral inflammation, but were not observed in a bacterial lung infection model with Legionella pneumoniae.

So, why does glucose kill in one assay and protect in another? Using PET, they observed that poly(I:C) but not LPS, induced more glucose uptake in the brainstem (no differential uptake was seen in other organs). In these hindbrains, CHOP, a ER stress-induced transcription factor that mediates apoptosis was upregulated during viral inflammation. CHOP null mice were protected from poly(I:C) and influenza induced mortality and 2DG could induce CHOP and apoptosis in MEFs, which the authors speculate may mean that viral inflammation leads to neuronal cell death and potentially then mortality.

What about the glucose-mediated death in LPS treated mice? The researchers again turned to the brain: giving glucose with LPS potentiated seizures, which could be rescued using valproic acid (VA), an anti-epileptic medication. As VA is thought to participate in HDAC inhibition, and that ketone bodies (which are formed as a byproduct of lipolysis during carbohydrate fasting) may also inhibit HDACs to resist ROS damage in tissues, the authors asked if glucose supplementation was altering the body's ability to reprogram itself to deal with ROS produced during inflammation. LPS challenged mice brains had increased ROS levels and more TUNEL-positive cells with glucose challenge (though not in the hindbrain where PET differences were observed). LPS treatment in mouse strains that were unable to undergo ketogenesis lead to enhanced mortality, suggesting this metabolic program may be protective in bacterial infection. Interestingly, the same mutants were more resistant to influenza-mediated death, strengthening the centrality of glucose utilization as an axis for producing disparate immune responses to different types of infections.

Now, there still remain important caveats and unknowns. Do these conclusions hold for all bacterial infections and all viral infections? How about in more minor, self-limited infections? Is the level of caloric supplementation important (the mice were supplemented with glucose amounting to 2% of a mouse's normal caloric intake)? How do different types of (mixed) food sources impact these results? All of these questions will be important to inform the major one: Do these findings translate to humans? Obviously, more clinical analysis will be needed before we can prescribed soda for the flu and fasting (or at least a carb-free diet) for an enteric infection. It's also interesting to speculate if the anorexia we develop during infection is an evolutionary defense mechanism or merely a coincidental benefit for some infections?

So, in the end, the old advice of 'starve a fever, feed a cold' still stands, but it's clearly no longer an all purpose adage--it should really be 'starve a bacterially induced fever, feed a virally-mediated cold', I guess (but not quite the same ring, I admit). And while questions still abound, this type of work gives us a lot of food for thought (sorry, it's a compulsion, you understand). And it highlights the need for more research into the metabolic and immune changes that occur during infection so that we can better tailor our clinical management in response to specific infections and for specific human populations.

Go to the profile of Michael Chao

Michael Chao

Associate Editor, Nature Microbiology

I first developed an interest in bacterial pathogenesis while at Cornell University. I then earned my PhD in Biomedical and Biological Sciences from Harvard University in Eric Rubin’s laboratory, studying cell wall remodelling in Mycobacterium tuberculosis. From 2012-2015, I continued my training as a postdoctoral fellow in Matthew Waldor’s lab at Harvard Medical School, investigating the role of DNA methylation on regulating fundamental cellular processes in Vibrio cholerae.

Please sign in or register for FREE to comment