A sick mouse’s guide to feasting and fasting

When should you feed a starving mouse and when should you just let it be?



Sick mice, especially those infected with bacteria and viruses often display an anorexic response and eat very little. More than 40 years ago it was recognized that mice sick with a bacterial infection die if you force feed them (1). Is this true for all infections? What about viruses? Should we starve a sick pet or colleague?

In a new series of experiments which explores the scientific basis for the old adage starve a fever, feed a cold, researchers have found that food makes things worse for mice with bacterial infection (such as Listeria monocytogenes) but is required for recovery from viral infections (such as influenza) (2).


When a mouse or any host is infected with a pathogen the events that follow can be resolved around 3 types of harm caused by the

i) pathogen itself – related to the number of pathogens, toxins produced by the pathogen etc.
ii) response of the body – collateral damage from the inflammatory response, immune reaction to pathogen, etc., which can often times be non-specific
iii) inability of the body or tissue to repair or take care of the damage

The authors find that it was the third kind – i.e. the ability to cope with tissue damage that ensues when mice sick with bacterial infections are fed and also when mice sick with viral infections are starved. This suggests that in the onslaught by the pathogen, there is a bystander effect upon non-immune tissues caused by host defenses that is a,critical determinant of bouncing back to health.

What did they do and find?

Mice infected with Listeria monocytogenes died when they were force-fed. The pathogen load (bacterial numbers) and defensive/ response molecules secreted by the mouse were not different between the force-fed (test) mice and mice that were not force-fed (control). The authors of the study then used a model for bacterial infection to look at why the mice are dying. In this model, the mice were challenged with a component of the outer membrane of bacteria – this is known to result in a strong inflammatory reaction – and then looked at the effect on mice upon injection of glucose, casein and olive oil. Glucose was found to be the cause of death.

This however is only one part of the story. The researchers then looked at another infection model, of influenza-infected mice, which also display an anorexic response. Here they observed the opposite – that is, if the mice were stopped from using the glucose, they died. In fact, feeding mice made them better. Viruses invoke response pathways, which are distinct from bacteria, so maybe the immune reaction was different between the fed and not-fed mice? Once again the authors ruled both pathogen numbers (viral load) as well as difference in immune responses in both groups. To understand what was causing death in these mice, the authors dissected mice that had been infected with the virus and then were given either normal saline or a molecule that made glucose unavailable to the body. Mice which were starved of glucose had lower heart rate, slightly lower respiratory rate as well as lower body temperatures about a week post infection. This was the first clue that control centres in the brain, which are responsible for these functions, may be affected. The authors extended this finding to a mouse model which cannot mount the normal immune response to viruses and challenged it with a molecular mimic of virus infection (poly I:C). In this mode, they found that when fed a molecule that made glucose unavailable, the mice died.

So why were starving mice dying in viral infections and fed mice dying in bacterial infection models? This work sheds some light on the differences. When the researchers studied glucose uptake in the brain in both models they found that there was glucose uptake in different parts of the brain during viral and bacterial infections. Viruses enter the host cells and use the sub-cellular compartments and cellular machinery to make copies of themselves. One such compartment- known as the endoplasmic reticulum – is needed both by the host cell and the virus to function normally. Infection results in a stress response in this compartment which usually signals to the cell that it should now shut-down (a particular kind of cellular suicide termed apoptosis). In this model of viral infection, glucose helps keep this compartment stress-free and therefore prevents cell death. This is particularly important for cells in the brain. What about bacterial infections then? In the brains of the mice with simulated bacterial infections and glucose injections, the authors find evidence for the accumulation of reactive oxygen species (ROS) in the brain. These molecules are also potent inducers of the cellular suicide pathways. However, the authors note that in this case, it may not be death of brain cells, but their dysfunction that may be the cause of death. This still does not explain the difference between viral and bacterial infections. To get to this, the authors analysed the starvation response. During starvation, the utilization of fats and proteins results in accumulation of ketone bodies, an important alternative fuel source during fasted states, via ketogenesis. Excessive and prolonged accumulation of ketone bodies is known to be toxic for the body. In the case of bacterial infection, this study suggests that the availability of ketone bodies may be helping cells to detoxify ROS.

Take home from this study

This study gives us a new way of thinking about infections, host response to infection (immunity) and the rest of the organs and tissues in the body, particularly the brain which must keep working normally through the pathogen-host cross-fire. There are clearly many unanswered question that this opens up, and while it demonstrates that glucose plays different roles in viral and bacterial infection of mice, the underlying mechanisms still remain to be understood in detail. It is interesting that the main difference of glucose utilization seems to be in the brain. The processes that connect what we eat, to what our body makes of it to how we feel or behave form a fascinating network with new links emerging all the time. It is not too soon to have convictions on what is good for us, our colleagues, our pets or our mice, but it is too early to really know or accept information without doubt.


1. Anorexia of infection as a mechanism of host defense.” M J Murray and A B Murray , Am J Clin Nutr. 1979

2. Opposing Effects of Fasting Metabolism on Tissue Tolerance in Bacterial and Viral Inflammation,Andrew Wang et al., Cell. 2016

An interview with L.Harding, S.Huen and A.Wang

Q. The idea that the there is tissue tolerance to injury caused by a pathogen-host battle seems reasonable, can you tell us more about the evolution of this idea and its implications for how people now view disease? Are there biomarkers of tissue tolerance?

The idea evolved from the recognition that oftentimes in sepsis, the immune response is more detrimental to the host than the damage incurred by the pathogen. The robustness of a tissue’s ability to tolerate inflammatory challenge can be measured by the ability of tissues to perform their function during inflammatory challenge. Clinically, physicians use plasma biomarkers of tissue dysfunction—for example, troponins for cardiac dysfunction, creatinine for kidney function, transaminases for liver function—as surrogates for tissue function.

Q. How easy or hard is it to distinguish between bacterial and viral infections in a clinical setting – in humans? Are there good diagnostic tests for this?

It is currently very difficult to distinguish the type of infection at the time of admission. This is an area of active research. Currently, clinicians rely on biomarkers such as procalcitonin, which have poor specificity for infection type, and/or detection of the pathogen itself, which often takes many hours if not days to verify, if at all.

Q. What about mixed infections? How do mice respond to a mixed Listeria and Influenza infections? Your group has explored this co-infection model previously, do you understand it better now?

Historically, it has been observed that mixed infections are worse for the host than either of the infections separately. The most famous example is influenza infection followed by a staphylococcus aureus infection. We have previously looked at influenza followed by listeria monocytogenes, and then at influenza followed by legionella pneumophilia. Generally, it appears that viral “priming” potentiates severe disease from otherwise sublethal challenges with bacteria. The mechanisms operating in these different infection pairs was different, but we are trying to understand if there are more general principles that could make this specific sequence of virus then bacteria more lethal.

Q. Do you plan to study this in humans? If yes, then how would you control for cultural variables, the availability of food and the process of habitual eating that many human beings now live by?

We do plan on studying this in humans. The setting where much of this can be best controlled is the intensive care unit (ICU). In patients admitted to the ICU, many are unconscious for one reason or another. Currently, these patients are fed by tube feeding very shortly after they are admitted. The goal of our initial studies will be to see if restricting glucose in feeds delivered to individuals with documented infections would be better for their outcome compared to standard formula feeds.

Q. Do you suspect that there is a strong genetic component to tissue tolerance, set-points or points of no return?

There is likely a strong genetic component to tissue tolerance. Since the immune response has been subject to great selective pressure, it should follow that tissue response to inflammatory signals generated by the immune response would also be under the same selective pressures, especially because it is ultimately tissue dysfunction that leads to death and thus the inability to transmit genetic material. However, because the field of tissue tolerance is relatively unstudied, no studies that try to identify those genetic components exist.

Q. Is the brain the most vulnerable organ – as opposed to say the kidneys which flush out toxins from the body, in terms of coping with damage from an infection? Did this finding surprise you?

In any injury, there is usually an organ or small set of organs, which, if dysfunctional, becomes limiting for the organism’s survival. The limiting organ in turn depends on the type of insult. In general, if the heart, lungs, or brain fail, it is rapidly lethal for the host in the absence of medical intervention. There is a lot of precedence for central nervous system dysfunction in bacterial sepsis, but we were surprised to find that the brain also appeared to be limiting in our influenza model, which is primarily a lung-injury model.

Q. For some bacterial diseases, tuberculosis is a case in point, we know that malnutrition makes the condition worse. How do you reconcile these observations with your finding?

There is a big difference between acute infection and chronic infection. What we were studying was the response to acute self-limited infections. In chronic infection, the persistence of the inflammatory response, persistence of the pathogen, and the changes that this dynamic imposes on the host is very different than the acute phase response. So, it is likely that the metabolic requirements of chronic infections are very different from the metabolic requirements of acute infections. Also, even in the acute setting, bacteria have co-evolved with their hosts and in the process may have developed mechanisms that interfere with the tissue tolerance mechanisms that we have described here. Therefore, our current work may not be generalizable to the full spectrum of bacterial and viral infections.