ABOVE: © ting pan

About a year ago, some—but not all—of the mice in Janelle Ayres’s lab at the Salk Institute for Biological Studies in La Jolla, California, got really sick. Ayres and her colleagues had infected each of the animals with the pathogenic bacterium Citrobacter rodentium, and within a few days, some of the mice began losing weight. Their colons became severely inflamed, and the animals died not long after. But other mice that were also exposed to the bacterium looked perfectly healthy.

All the mice were genetically identical. They were fed the same food, kept in the same kinds of cages, and had no notable differences in the composition of their microbiomes. “Yet half the animals died, and the other half survived,” exactly what she was aiming for, Ayres tells The Scientist.

There’s more to recovery than simply killing bugs.

—Andrew Read, Penn State...

This type of experiment is called “lethal dose 50,” the level at which the dose of a toxin or pathogen kills half of the population that’s been exposed. Ayres was running the experiment to determine what causes genetically identical mice to respond differently to the same pathogen. Scientists using the protocol typically assumed that susceptible mice had more-severe infections—that random events drove up the number of pathogenic bacteria and this caused them to die—while surviving animals got lucky and managed to keep pathogen numbers low. But when Ayres tested her mice, levels of bacteria in the gut and other tissues were the same in both groups.

To try to understand what saved the survivors, Ayres and her colleagues examined the genes turned on in the mice’s livers, an organ that plays a major role in secreting signaling molecules to maintain homeostasis in the body. Compared with mice that died from the infection, the mice that survived expressed lots of genes linked with metabolizing iron. This indicated to Ayres that iron might help the animals cope with the infection, so she and her team decided to treat mice that were on the verge of dying from the C. rodentium infection with an iron supplement. The animals recovered.

Intrigued, the team upped the stakes. They infected another set of mice with a dose of the bacterium that should kill all of the animals, not just half of them—then gave the animals iron. All of the mice survived. Still not satisfied, Ayres and her team infected a new set of mice with 1,000 times the lethal dose of the bacterium, followed by iron. “They were perfectly fine,” Ayres says, while infected mice not getting the iron supplement died within days.1

Sequencing the genomes of C. rodentium in the control and iron-fortified mice revealed that the bacterium in the mice fed iron had accumulated mutations that tamped down expression of multiple genes for proteins in a virulence pathway, disabling its ability to cause disease. The bacteria, found in the colon, were, in essence, “just part of the [mice’s] microbiome now,” Ayres says.

The results, published in the summer of 2018, support a hypothesis Ayres posited years ago: that fighting infections doesn’t have to be all-out war. Instead of trying to obliterate pathogens that have invaded the body, she proposes, organisms may give them what they want and ultimately push them to evolve into something benign, lessening the damage done by the pathogen and the immune system. This phenomenon known as disease tolerance, is something that the body can do naturally by tapping into different physiological systems, such as metabolism, to prevent illness. And though this framing is relatively new in a clinical setting, some medicines that have long been on the market encourage this strategy to improve disease outcomes.

“Anytime we take Tylenol because we have the flu and we feel terrible, that’s actually you playing with tolerance,” says Stanford University microbiologist David Schneider, Ayres’s former advisor. By quieting the immune reaction that is making you feel sick, “you’re making yourself feel better, even though you might not be affecting how much of a pathogen is in your body.”

As they come to appreciate that disease tolerance exists in animals, including humans, researchers want to tap into its mechanisms—analogous to the way they are tapping into the immune system to develop disease-fighting immunotherapies. Specific kinds of supplements, as Ayres has shown in mice, may be one solution. And bacteria that live in the body as part of its micro­biome have been shown to help mice tolerate malaria, Salmonella, and pneumonia infections.

“During infection, we all appreciate that there are these immune defenses that largely are designed to get rid of an invading pathogen, and that’s been thought to be the only or main way that we deal with infections,” says Ruslan Medzhitov, an immunologist at Yale School of Medicine. “What’s being appreciated more recently . . . is that there is also this other mechanism, so-called tolerance to infection, where instead of trying to get rid of a pathogen we change something about the body, about the physiology, and that lets us tolerate the presence of a pathogen.”

The roots of tolerance

Until about a decade ago, researchers had largely overlooked the idea of disease tolerance in animals. But the physiological strategy didn’t go unnoticed among plant biologists. In research dating to the late 1800s, for example, scientists described how one variety of wheat crop infected with a fungus called leaf rust fared better and produced more grain than other infected wheat crops.2 Follow-up studies spanning the 20th century and into the 21st suggested that plants have internal ways to tolerate infections in addition to defending against them with immunity. These findings led researchers to wonder if a similar sort of tolerance exists in animals.

Researchers reported the first hints of disease tolerance in humans in 2006, when they found that people who have a type of alpha thalassemia, a blood disorder that typically reduces hemo­globin production, are somehow protected against the severe iron deficiency associated with a malarial infection. In a study published the following year, disease ecologist Andrew Read, then at the University of Edinburgh, and his former postdoc Lars Råberg found that certain strains of mice had genetic variations that boosted their tolerance to the malaria parasite Plasmodium chabaudi. Those mice had improved health, the researchers noted, but comparable numbers of P. chabaudi cells in their bodies to those in mice that weren’t as tolerant to the infection.3

What’s being appreciated more recently . . . is that there is also this other mechanism, so-called tolerance to infection, where instead of trying to get rid of a pathogen we change something about the body, about the physiology, and that lets us tolerate the presence of a pathogen.

—Ruslan Medzhitov, Yale School of Medicine

“I think that our paper resonated with some people about the need to broaden the rather narrow-minded focus of traditional immunology,” says Read, now at Penn State University. “There’s more to protection. There’s more to recovery than simply killing bugs.”

Around the same time, Ayres, working as a graduate student in Schneider’s microbiology and immunology lab at Stanford, was starting to document signs of disease tolerance in Drosophila. She had injected the fruit flies with a lethal dose of the pathogenic bacterium Listeria monocytogenes. All of the insects did eventually die, but some died more quickly than others. To understand why, Ayres and her colleagues first measured levels of Listeria in the flies’ bodies. As expected, some of the flies that died more quickly had higher amounts of pathogenic bacteria in their bodies than flies that died over the four to five days it typically took to succumb to the infection. But some of the flies that died more quickly had comparable levels of bacteria in their bodies to the flies that outlived them. Looking at the genomes of the flies, Ayres found that the ones that succumbed more quickly to infection had mutations in genes not previously tied to immunity or disease development.4

“That suggested to us that they were dying because they lacked, or they had a mutation in, a gene that [normally] promoted their health by altering some aspect of physiology without killing the pathogen,” Ayres says. “That really led us into this new area, that animals must encode some other defense strategy [besides immunity] that’s also essential for surviving interactions with pathogens.”

Ayres’s work with Schneider, published in 2008, was pivotal in bringing the idea of disease tolerance from plants to animals, says Miguel Soares, a cell biologist at the Gulbenkian Institute of Science in Oeiras, Portugal. “They found that there were genetic variations that had a tremendous impact on the [longevity of flies], but no changes in the amount of pathogen that the flies had,” he explains. “Now, if you’re not very intelligent, you just put that piece of data in the drawer, and say, ‘Oh, we cannot explain it.’ But they went back, and they picked up the right literature, and they said, ‘This is disease tolerance.’”

Mechanisms of Disease Tolerance

When we’re sick, we want to feel better—immediately. Our bodies set out to accomplish this goal by activating the immune system to vanquish the invader as quickly as possible. Vaccines help prepare the immune system for this fight, while antibiotics or antiviral drugs serve as its allies on the battlefield, targeting the pathogens and preventing them from multiplying and spreading in the body.

But often what makes us sick is our own immune response. It’s why we take pain or fever-reducing medicine such as acetaminophen, which doesn’t affect the illness-causing pathogen at all, but does quiet the inflammation that is causing symptoms. Recently, scientists have come to realize that the body can similarly work to promote health by quelling an immune response and minimizing the damage from an invading pathogen. They are beginning to piece together some of the ways disease tolerance can protect the body from damage during infection.

Sick MouseTolerant Mouse

Symbiotic bacteria

Mammalian guts teem with symbiotic bacteria, which might actually help us tolerate microbes of the pathogenic variety. Severe infection with Salmonella typhimurium can sometimes trigger muscle and fat tissue in a host to waste away. But, researchers found, when mice had a symbiotic E. coli in their guts, they didn’t suffer as much damage to their tissues. Giving E. coli to mice that lacked the symbiotic strain revealed that, upon infection with Salmonella, the resident microbe moves to the adipose tissue. There, it triggers a hormone response that protects against fat and muscle breakdown, easing the animals’ recovery (Science, 350:558–63, 2015).

© mesa schumacher


When pathogens invade a host, they need sugar along with essential nutrients such as iron to survive. Many pathogens break down the blood protein hemoglobin to get iron, which leads to extra molecules of heme byproduct in the blood. That excess heme tamps down the activity of an enzyme called glucose 6-phosphatase (G6Pase) that is essential for producing glucose in the liver. Without glucose production, sugar levels in the blood drop, sometimes so low that the host dies. In mice with high levels of ferritin, an iron-storage protein, glucose production doesn’t slow down, allowing the body to tolerate infection by multiple pathogens. Supplementing with a protein that grabs iron to make ferritin saved mice that were succumbing to the infection (Cell, 169:1263–75.e14, 2017).

© mesa schumacher

Pathogen mutations

In one experiment, researchers showed that when a mouse’s small intestine produces glucose during a Citrobacter rodentium infection, sugar molecules are absorbed into epithelial cells and then released into the bloodstream. This can feed the pathogen, heighten the immune response to the invader, and lead to the host’s demise. However, when infected mice were fed an iron-rich diet, glucose wasn’t as easily absorbed from the intestine, leaving it available for the pathogen to consume. Having access to the glucose led to mutations in the virulence genes of the bacteria, rendering them less pathogenic and allowing the bacteria to live peacefully in the body (Cell, 175:146–58.e15, 2018).

See full infographic: WEB | PDF
© mesa schumacher

Bacterial partners

Intrigued by the idea of disease tolerance, Ayres wanted to identify exactly how it helped defend the body against illness. The microbes of the body have many of the same molecular features that the body’s immune system recognizes on pathogens, so she wondered if these resident microbes could somehow shape the way the body responds to infections. “I thought that our interactions with beneficial microbes could be important,” she says.

Ayres was on to something. As a postdoc in Russell Vance’s innate immunity lab at the University of California, Berkeley, she and colleagues reported that treatment with antibiotics made mice susceptible to severe inflammation of the colon sparked by a multidrug-resistant strain of E. coli that naturally lived in the animals’ gut. The antibiotics appeared to kill other bacteria that somehow keep the E. coli strain from alarming the immune system. But with the healthy gut bacteria gone, the E. coli triggers an innate immune response—specifically, the NLRC4 inflammasome, part of a multiprotein complex that detects pathogenic microbes in host cells and pumps out proinflammatory cytokines in response. The mice eventually died as a result of the overactive immune response, similar to what happens in humans who develop sepsis.5

In 2013, Ayres launched her own lab at Salk, where she started to tie together her work from graduate studies and her postdoc. In one of her first papers as an independent researcher, she looked at cachexia, the loss of skeletal muscle and fat tissue that can occur during infection, cancer, and other diseases in subsets of patients. Ayres wondered if the microbiome somehow dictated who did and did not suffer from cachexia.

Anytime we take Tylenol because we have the flu and we feel terrible, that’s actually you playing with tolerance.

—David Schneider, Stanford University  

So she and her colleagues tested how genetically identical mice inoculated with Salmonella typhimurium, a common foodborne bacterium, or with Burkholderia thailandensis, a pathogen that can cause pneumonia-like symptoms, responded to infection. They found that, whether or not the animals suffered severe muscle loss, the levels of the pathogens were the same. The mice’s microbiomes, on the other hand, were different. Specifically, in mice that did not suffer muscle and fat loss, the team identified a benign strain of E. coli that tended to be missing from those that developed cachexia. When the researchers gave that E. coli strain to mice before infecting them with one of the two pathogens, the mice were completely protected from muscle and fat loss. (See infographic.)

Additional experiments showed that when given to healthy mice, the E. coli stayed in the intestine, but in mice with bacterial pneumonia, typhoid fever, or colitis, the E. coli migrated to the animal’s fat. There, the bacteria activated the NLRC4 inflammasome, triggering the production of insulin-like growth factor-1 (IGF-1) in the adipose tissue. Once secreted into the bloodstream, IGF-1 acted as a messenger, signaling muscle not to deteriorate.6

The results revealed exactly how the E. coli helped the mice tolerate the bacterial infections, Ayres says. “That was exciting for us, because we showed that commensal microbes have evolved ways to turn on this defense strategy in their hosts.”

A link with metabolism

Ayres’s recent study documenting the benefits of iron supplementation in mice infected with Citrobacter rodentium suggests that metabolism might also play a key role in disease tolerance. The extra iron reduced glucose absorption in the intestines of the ill mice, giving the invading bacteria access to more food. And that, in turn, made the pathogen less virulent.

Soares’s work at the Gulbenkian Institute of Science similarly points to metabolism as a likely mediator of disease tolerance. He’d been studying sepsis in mice and came to understand that un­fettered inflammation explained only part of the pathogenesis. A breakdown in glucose production, ultimately leading to organ failure, was another big reason why the disease was so often fatal, at least in mice. If this metabolic issue could be resolved, Soares found, the animals could survive the sepsis-causing infection.

Typically during infection, a host will sequester iron in its cells so an invading bacterium can’t get it. But pathogens can go after iron-containing hemoglobin in red blood cells. The bacteria break down red blood cells, releasing hemoglobin, which is further broken down into iron-rich heme. This heme is toxic to intact red blood cells, leading to the breakdown of more of them and the release of more heme, some of which is ferried into the liver, where it can release its iron. Working with Sebastian Weis and Ana Rita Carlos at Gulbenkian, Soares and colleagues showed that high levels of iron from heme in mice can suppress the expression of genes involved in glucose production in the liver, the body’s main distributor of the sugar. Low glucose levels, in turn, can lead to multiorgan shutdown and death.

There is an emerging theme that controlling metabolism is one of the things that you need to do to survive an infection.

—Miguel Soares, Gulbenkian Institute of Science

But the researchers discovered that if the animals’ livers had high levels of a protein called ferritin, which sequesters iron, the decreased expression of the metabolic genes was averted.7 When the researchers gave the mice apoferritin, a protein that binds to free iron in the liver to produce ferritin, the animals maintained liver glucose production during sepsis, preventing the condition from becoming severe.

Soares says that the finding fits nicely with Ayres’s recent work, which shows how iron is linked with genes that regulate glucose absorption in the intestine during a gut infection. Both studies, he says, seem to show “an emerging theme that controlling metabolism is one of the things that you need to do to survive an infection.”  

Ayres notes, however, that her team has identified a different mechanism by which iron relates to glucose metabolism: insulin resistance, a condition in which tissues fail to grab glucose out of the bloodstream when signaled by the hormone to do so. In turn, the glucose stays in the intestines, where invading pathogens can feed on the sugar. That form of insulin resistance may result from increased levels of an enzyme called heme oxygenase-1 (HO-1). High levels of HO-1 in the liver and fat are associated with insulin resistance in metabolic disease, such as diabetes.8 And, though they didn’t show signs of metabolic disease, Ayres’s mice that survived the Citrobacter infection also had elevated levels of the enzyme in their livers.

Soares has also shown that giving HO-1 as a supplement to mice that have developed sepsis or to mice infected with P. falciparum induces disease tolerance.9 HO-1 may therefore promote insulin resistance, maintaining glucose production in the gut and helping to keep the animals healthy during infection, Ayres says. Acute insulin resistance, she notes, is a common response triggered during infectious diseases, which suggests it evolved to help the body defend itself during infection.

“We say insulin resistance is bad, because we think about it in the context of metabolic disease. Of course, in those scenarios it is bad,” Ayres says. “We evolved that response for a reason, so it must be serving some beneficial function in certain contexts.”

One beneficial function may be connecting disease tolerance and immunity. Metabolic pathways have a tremendous effect on immune cells, and glucose is at center stage in all of that. Macrophages and many other immune cells require glucose to kill viral and bacterial invaders, while glucose also drives certain disease tolerance pathways. It seems that metabolism, Soares says, may serve as a link between disease tolerance and the immune system, increasing the body’s chances of surviving an infection.

Questioning tolerance

If organisms have evolved to tolerate some infections, rather than eradicate them, the phenomenon could have population-level effects. In Ayres’s iron-fed mice, for instance, the animals were healthy, but still maintained a population of the Citrobacter in their gut, which they excreted when they defecated, potentially giving the pathogen an opportunity to spread. This raises the question of whether meddling with a pathogen through disease tolerance mechanisms might make it easier for it to spread through populations and possibly regain its virulence.

Ayres also had that question, and to address it, she and her team ran experiments housing infected mice with naive, uninfected animals. When infected mice were fed iron-fortified chow for 14 days, then three days later cohoused with naive mice, all of the naive mice died. When the researchers waited 10 days after feeding them iron-fortified food before housing the infected mice with naive mice, only 40 percent of the naive mice died. And when iron-fortified, infected mice weren’t housed with naive mice until 45 days later, all of the mice survived. As the bacteria evolved to be less virulent, even mice that didn’t eat iron could tolerate them. “That’s really exciting because . . . at least in our studies we’ve reduced the threat of the infection at the population level by pushing [the pathogen] to evolve into something benign,” Ayres says.

I think this idea of disease tolerance has been a completely neglected area in immunology and host-microbe interaction.

—Janelle Ayres, Salk Institute for Biological Studies

Because of the promise of disease tolerance, researchers are looking to tap into it to treat infections and other disease. What might make this tricky, however, is that the mechanisms of disease tolerance at play likely depend on the infecting pathogen, says Dan Littman of New York University Langone Medical Center. For example, Yale’s Medzhitov showed that mice that didn’t eat while sick with bacterial sepsis survived the infection, but when mice had viral sepsis and didn’t eat, they died.10 The effect was linked to glucose meta­bolism and the release of different immune signaling proteins during the two types of infection.11

Such variation might make it hard to generate general disease tolerance-based treatments for illnesses, some researchers suggest. But Medzhitov remains optimistic, explaining that there may well be commonalities among disease tolerance pathways that can be exploited, especially for superbugs and viruses without vaccines. “That’s where [disease tolerance] would be our main hope.”

Alexander Rudensky, an immunologist at Memorial Sloan Kettering Cancer Center, agrees the whole field is exciting, calling disease tolerance “a largely untapped area that could have applications not only to infectious disease, but also other clinical conditions.” It could even apply to treating cancer, Ayres explains, by helping to stave off a loss of muscle and other pathologies that develop as a result of the disease and its treatment.

“I think this [idea of disease tolerance] has been a completely neglected area in immunology and host-microbe interactions,” Ayres says. It’s a research field that she thinks could have saved her father’s life in 2015 when he lost a short battle against sepsis. “I really think we’re onto something,” she says, “and it could really make a difference.”


  1. K.K. Sanchez et al., “Cooperative metabolic adaptations in the host can favor asymptomatic infection and select for attenuated virulence in an enteric pathogen,” Cell, 175:146–58.E15, 2018.
  2. N.A. Cobb, “Contributions to an economic knowledge of Australian rusts (Uredineae),” Agric Gaz NSW, 3:44–48, 1892.
  3. L. Råberg et al., “Disentangling genetic variation for resistance and tolerance to infectious diseases in animals,” Science, 318:812–14, 2007.
  4. J.S. Ayres et al., “Identification of Drosophila mutants altering defense of and endurance to Listeria monocytogenes infection,” Genetics, 178:1807–15, 2008.
  5. J.S. Ayres et al., “Lethal inflammasome activation by a multi-drug resistant pathobiont upon antibiotic disruption of the microbiota,” Nat Med, 18:799–806, 2012.
  6. A.M. Palaferri Schieber et al., “Disease tolerance mediated by microbiome E. coli involves inflammasome and IGF-1 signaling,” Science, 350:558–63, 2015.
  7. S. Weis et al., “Metabolic adaptation establishes disease tolerance to sepsis,” Cell, 169:1263–75.E14, 2017.
  8. A. Jais et al., “Heme oxygenase-1 drives metaflammation and insulin resistance in mouse and man,” Cell, 158:25–40, 2014.
  9. R. Larsen et al., “A central role for free heme in the pathogenesis of severe sepsis,” Sci Trans Med, 2:51ra71, 2010.
  10. A. Wang et al., “Opposing effects of fasting metabolism on tissue tolerance in bacterial and viral inflammation,” Cell, 166:1512–25.E12, 2016.
  11. A. Wang et al., “Glucose metabolism mediates disease tolerance in cerebral malaria,” Cell, 115:11042–47, 2018.

Interested in reading more?

living with bacteria 2019 the scientist june issue

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!