KILLING the HOST
When the Microbiome Stages a Coup
by Carl Zimmer / 01/19/2015
“It’s an ugly fact of life that getting old means getting infections. Old people get attacked more by pathogens, and the damage that these germs cause can speed up the aging process, leading to even more infections. The standard explanation for this vulnerability is that the immune system falters in old age, opening an opportunity for pathogens to invade. But in the journal Biology Direct, Viktor Muller of Eotvos Unversity and his colleagues propose that something else is also going on in the aging body. Maybe the microbiome senses that its host is in bad shape and rises up in rebellion. The scientists call their idea “the Microbiome Mutiny Hypothesis.”
It may seem like a strange notion, but several lines of evidence suggest it’s worth considering. First of all, a lot of the pathogens that attack the elderly come from within. They grow quietly and harmless for decades inside people’s bodies and then switch over to causing dangerous infections later in life. The answer to why old people get infections must address why harmless bacteria turn bad in old age. To understand this turn, we have to abandon any strict division between “good” germs and “bad” ones. For the germs themselves, these are just two ends of a seamless spectrum. Depending on how they use their host, microbes may cause no harm, a little, or a lot. And the virulence of a microbe–the amount of harm it causes–can itself evolve over time. Under some conditions, natural selection may favor gentle handling. But in other situations, causing deadly disease may be the winning strategy.
A lot of factors go into determining which strategy will be a winner for a given microbe. For some microbes, the best way to multiply may be ripping open host cells and feasting on their contents. This may kill a lot of their hosts, but that may not matter to the microbes, since they can escape to a new host–say, by causing diarrhea that contaminates a water supply. But in other conditions, killing a host may be a bad long-term strategy–if, for example, the odds are low that a microbe will get from one host to the next. It’s even possible for organisms to evolve the ability to switch between these strategies, using the best strategy for different environments. In a 2013 experiment, Oxford scientists observed this switch evolve before their eyes.
They studied phages, which are viruses that infect bacteria. A phage invades a bacterium and makes new copies of itself. The bacterium eventually ruptures, spilling out the next generation of phages. The scientists reared phages under unusual conditions–they mixed together a very high concentration of phages with bacteria. As a result, each microbe tended to get infected by more than one phage. The phages would then make copies of themselves at different rates. When the microbe ruptures, out would come a mixture of phages. The faster breeders dominated over the slower ones. The scientists let the phages evolve in these conditions for 50 days. When they were done, the phages could now adjust their speed. If they found themselves alone in a host cell, they grew slowly. But if they sensed other phages in the cell, they sped up, so as to outcompete their rivals. As a result, their host died faster.
Muller and his colleagues propose that some of the microbes that live in our bodies can also switch from benign to deadly for similar reasons. While we’re healthy, they growing slowly, causing us no harm. But as we approach the end of our lives, the microbiome shifts to a more aggressive strategy. “Killing the goose that lays the golden eggs might not be such a bad idea if the goose is going to die soon, anyway,” the scientists write. There’s good evidence that microbial residents can eavesdrop on our health. A pathogen called Pseudomonas aeruginosa, for example, can sense certain molecules our brains release in response to stress. They respond by unleashing a toxin that help them grow–while also damaging our lungs.
Muller and his colleagues offer some ways to test their hypothesis. If they’re right, then infections in old age aren’t just the result of a slack immune system. Instead, bacteria and viruses sense a changed environment and respond by making new molecules, which they use to grow aggressively and cause harm. If scientists disable these molecules, then the pathogens should become tame again. It would be interesting to see the Microbiome Mutiny hypothesis put to such a test. Conceivably, scientists could someday turn the test into a treatment. Rather than blasting the elderly with broad-spectrum antibiotics, doctors could just disarm the mutiny.”
The microbiome mutiny hypothesis: can our microbiome turn against us when we are old or seriously ill?
by Lajos Rózsa, Péter Apari, Viktor Müller / 14 January 2015
“From an evolutionary point of view, virulence is defined as the reduction in the lifetime reproductive success of host individuals owing to infection with a symbiont/parasite. A full continuum exists from lethal pathogens to obligate mutualists, and the lines between parasitism and mutualism are often blurred ,; some species are even able to adjust their virulence reacting to a changing environment ,. Differences in the speed and extent of the symbionts’ reproduction can directly contribute to the variability of their virulence. Slower growth and multiplication of symbionts is associated with a lighter metabolic burden and smaller costs of immunity and collateral damage. However, as expressed in the ‘trade-off hypothesis of virulence’ , slower replication is also likely to result in reduced transmissibility of the symbiont over a unit time, due to lower densities in the infected individual. General theoretical considerations suggest that symbionts evolve towards an optimal virulence that maximizes their transmission over the entire life cycle of infection ,. Optimal virulence depends on the life history traits of both the host and the symbiont species, and both highly lethal, and benign or mutualistic (“negative virulence”) interactions can be evolutionarily stable strategies. The symbionts adapted to long-term persistence in their hosts are characterized by low (or negative) virulence to allow the hosts to survive through longer periods ,. Virulence is not only subjected to quick evolutionary changes, but some pathogens are also capable of exhibiting phenotypic plasticity in virulence traits . For example, some parasites exhibit low virulence when facing “cooperative” hosts (that mount weak defense) and high virulence in “non-cooperative” hosts (that mount vigorous defense) .
The human microbiome is a taxonomically diverse mass of bacteria (and to a lesser extent, also archaea and fungi) living in and on both healthy and diseased humans . This microbial community is widespread across all human body surfaces, namely the skin, nasal and oral cavities, genitals, lungs, and is particularly abundant within the intestines. In healthy humans, most members of the microbiome exhibit zero or negative virulence thus comprising neutral or mutualistic relationships, respectively. However, this mode of benign coexistence likely evolved under the conditions of a healthy host—representing the majority of the lifespan of humans. Human mortality exhibits a pattern of low mortality rates during most of the lifespan, but steep increase at old age, and episodes of serious injury or illness can substantially raise mortality also at younger age. Evolutionary theory suggests that increasing mortality might increase the optimal virulence of a symbiont ,: killing the goose that lays the golden eggs might not be such a bad idea if the goose is going to die soon, anyway.
Presentation of the hypothesis
Extending earlier work ,, we propose that symbionts of the human microbiome might shift to higher virulence (or from mutualism/commensalism to parasitism/pathogenesis) as background mortality increases steeply at the end of human lifespan or due to serious injury or illness. As the option for long-term persistence becomes increasingly limited, the symbionts are likely to benefit from increasing host exploitation rates so as to maximize chances of immediate transmission. A recent survey of the human microbiome found that a large number (>50 species) of opportunistic pathogens are widely prevalent in the microbiota of healthy individuals , suggesting that many species of the microbiome (sometimes referred to as ‘pathobionts’ ) are indeed able to switch facultatively to higher virulence. Switches to higher virulence might also occur if mortality increases due to other reasons, e.g., severe injury or infection, provided the symbionts are able to detect such changes in the host’s condition (which is not unreasonable to assume: see below). Observations consistent with our hypothesis include the increased incidence of diarrhea in old age; the reactivation of herpesviruses in aging  or after helminthic co-infection ; and the increased risk of common infections (including those with opportunistic pathogens) in aging individuals  or in patients with type 1 and type 2 diabetes mellitus . Of note, aging has the strongest impact on the incidence of pneumonia and urinary infections , which are often caused by opportunistic pathogens and which both provide a simple mechanism of increased shedding of the infectious microorganisms (coughing and bacteriuria) when the replication of the pathogens accelerates. Admittedly, many of these observations could also be explained by compromised immune control over the (potential/opportunistic) pathogens in aging or diseased individuals, rather than a change in the behavior of the symbionts/commensals. We further argue that virulence shifts across the microbiome might occur in a synchronized fashion. Whenever certain members of the microbiome switch to higher levels of virulence, the expected lifespan of the host decreases further. Sensing either the further decline in the health of the host, or directly the increased replication or virulence of the other symbionts co-inhabiting the same person might trigger further microorganisms to switch to increased virulence. As a consequence, the switching of some major components of the microbiome to higher host exploitation rates likely provokes a chain reaction of virulence shift across the whole community, somewhat analogous to a ‘regime shift’ (sudden switch between alternative states) in ecosystems .
The proposed hypothesis of “microbiome mutiny” depends on the assumption that symbionts are able to obtain reliable information on the health and life expectancy of the host that they inhabit. This might occur, e.g., by sensing increasing oxidative stress  or other molecular or physiological markers of senescence . Impaired health due to injury or disease might be sensed based on systemic markers of inflammation , while an increased level of heat shock proteins can indicate stress and reduced health in both aging and disease . Remarkably, stress-induced host signaling molecules have been shown to induce a high-virulence phenotype in experimental Pseudomonas aeruginosa infection of mice . In addition, symbionts might detect direct or indirect clues of the presence and activity of other (potentially) disease-causing organisms, sensing secreted diffusible molecules in the frame of a microbial community-wide quorum-sensing system ,, or host immune status manipulated by co-infections ; these and other mechanisms of sensing signals from both the host and other members of the microbiota are reviewed in . The ability to adjust virulence in response to external stimuli does not seem to be particularly difficult to evolve: even a simple bacteriophage could quickly be adapted to conditional virulence under experimental conditions. Finally, we note that the evolution of host-health dependent virulence may depend on some aspects of the life history of the hosts. For example, both healthy and ailing hosts must be sufficiently frequent to create alternating selection regimes that favor low and high virulence in the symbiont, respectively. If the symbiont spends very little time in ailing hosts, then this rare selection regime may be insufficient for the emergence and/or maintenance of adaptive virulence mechanisms. Furthermore, the population structure of the microbiome across host individuals may also be important: if host individuals live in close-knit groups that share the same microbiome, then a temporary increase in transmissibility from old or ailing individuals may bring little benefit for a symbiont that is already present in the other members of the group. However, while such a close-knit structure may have characterized hunter-gatherers, increasing population density and mixing in recent human history has created conditions where a huge inter-individual diversity of microbiomes can now be readily exchanged. The composition of microbiomes is still more similar within than between families, but family membership accounts for only about 20% of the compositional variation in fecal samples of the gut microbiome . We summarize the conditions required for the evolution of mechanisms of “microbiome mutiny” in Table 1.
In turn, humans (and other affected host species) might have evolved counter-adaptations to reduce the risk of “microbiome mutiny”, either by interfering with the microbial sensing of declining host health, or by creating conditions that favor symbionts that are not prone to virulence shifts but might suppress potentially pathogenic species . Remarkably, systemic exposure to bacterial products triggers in mice the production of fucosylated proteins that can be metabolized by the gut microbiome and that have been shown to down-regulate virulence genes in gut bacteria . Selection to oppose a microbiome mutiny in old age is likely to act in humans in particular, where living a long and healthy life even after the reproductive phase of life is adaptive because elder individuals provide valuable altruistic help to their relatives. In prehistoric ages, elderly people probably played an outstanding role in gathering, storing and distributing knowledge through the community. Moreover, grandparents can provide care specifically for their grandchildren—a social system unique to humankind .
Testing the hypothesis
Our hypothesis predicts a secondary increase in host mortality due to “microbiome mutiny” when an independent cause (old age or illness) has generated a primary increase from the healthy baseline. Because we are typically unable to predict the primary increase quantitatively, it is not possible to assess the existence of a secondary increase solely on the basis of observed mortality rates. Even “per pathogen pathogenicity” (introduced as “per parasite pathogenicity” in ) is not a reliable marker, because the sensitivity (tolerance) of an immune compromised host to the same level of symbiont /pathogen load may also change independent of a virulence change in the symbiont/pathogen. The validation of the hypothesis will therefore have to rely on the discovery (and possible targeted manipulation) of the microbial mechanisms of “microbiome mutiny”. Direct tests of the hypothesis should include comparisons of the expression of virulence factors and in vitro measures of virulence between isolates of the same species of the microbiome obtained from healthy/young and ailing/old individuals. Our hypothesis predicts that increased virulence is not merely a consequence of increased replication due to relaxed immune control, but is also associated with the upregulation of virulence factors. We note, however, that the two mechanisms are not mutually exclusive and might, in fact, act in synergy: relaxed immune control might allow the replication of the microorganisms to higher levels, which might then turn on virulence mechanisms by quorum sensing ,. Once putative mechanisms of microbiome mutiny have been identified, strong validation for the hypothesis can be derived from targeted inhibition of these mechanisms, which should then result in the maintenance of the low-virulence phenotype and reduced mortality, without directly interfering with the primary cause of increased background mortality or the growth of the targeted facultative pathogen. The feasibility of this approach has already been demonstrated in an animal model of P. aeruginosa infection displaying facultative virulence .
From the host side, our hypothesis likely applies to long-lived species that have a steep increase in mortality at the end of their life span. Large mammals and some birds fall in this category, while long-lived reptiles and plants tend to have flat or even decreasing mortality rates at higher ages . From the symbionts’ side, the hypothesis applies to species that are able to establish long-term persistence in the host, and that have the capacity to increase their replication rate and transmissibility from the level attained in a healthy host. This might include normally harmless or beneficial members of the microbiome, with a probable dominant role of the gut microbiome that harbors the greatest microbial mass and taxonomic diversity, and that has an “easy route” to higher immediate transmissibility and virulence in the form of diarrhea. In addition, the hypothesis might also apply to persistent and/or latent infections with herpesviruses, hepatotropic viruses, spirochaetes (Lyme disease, syphilis), parasitic protozoa (malaria, toxoplasmosis), and even parasitic worms. Finally, we note that increasing virulence in aged or chronically ill individuals might also occur by genetic changes (adaptive evolution) in infections with fast replicating organisms ,. However, within-host evolution will play a role only if increased virulence confers a selective advantage to the symbiont/pathogen within the host, while our argument is based on changes in the degree of virulence that maximizes transmission, i.e. on selection at the between-host level. The two levels of selection may nonetheless be connected in some cases: in addition to conditional phenotypic plasticity, the capacity for microbial mutiny can also emerge by the evolution of “enabling constraints” that create a high-probability trajectory of evolution towards increasing virulence within the host when the appropriate conditions occur. The direct causes of the virulence shifts (phenotypic plasticity/facultative virulence vs. adaptive evolution/genetic changes) can be distinguished by screening for causative mutations.
Implications of the hypothesis
Our hypothesis posits that at the moments when our health is most fragile due to old age or severe illness, our condition can be further exacerbated by the “treachery” of our previously benign microbiome. In terms of failing health, “whosoever hath not, from him shall be taken away even that he hath” (New Testament, Matthew 13:12). However, even if this phenomenon is widespread, the effect of shifting virulence might easily have gone unnoticed, because the declining health of the individuals could be attributed to the original condition that triggered the change in the microbiome. Elucidating this potential mechanism might open up new possibilities for the clinical management of age related health issues and critical injuries or disease. Targeted prophylaxis against the microbes capable of virulence shifts could break the harmful feedback loop between deteriorating health and the “mutiny” of the microbiome. In this regard, the widespread emergency practice of applying preventive antibiotics even in the apparent absence of known pathogens, though criticized as a source of selection for antibiotic resistance, might not be such a bad idea . A more detailed knowledge of the virulence shift mechanisms might in the future allow us to target directly the mechanisms responsible for the virulence switch. If virulence shifts occur in a chain reaction of several species of the microbiome, then the altered regime of increased virulence might remain stable or might take a long time to revert even if the injury or disease that triggered the switch is cured. The negative effect of the symbionts already in “virulence mode” might suffice to keep each of them in this state, which might prolong convalescence. Of note, patients who survive to hospital discharge after sepsis remain at increased risk for death in the following months and years . This state of ‘chronic critical illness’ can also be triggered by other acute episodes of illness (e.g., acute lung injury), and its maintenance seems related to persistent systemic inflammation , which might act by maintaining the microbiome in the altered regime of increased virulence. Targeted interventions might be able to break this state and quickly return the microbiome to the healthy regime of benign coexistence. The hypothesis also predicts that symbionts transmitted from old or seriously ill individuals might display higher initial virulence in the new host until virulence is “reset” to the baseline level optimal in healthy individuals.
One of the direct causes of mortality in cancer is wasting, and microbiome-induced inflammation has been implicated as a possible action mechanism . This raises the possibility that “microbiome mutiny”, triggered by the primary condition of cancer, may contribute also to cancer mortality. Virulence shifts might also play a role in the increased mortality observed after the loss of a lifelong partner, which occurs sometimes without specific pathological reasons . Abruptly increasing levels of stress and associated psychosomatic changes  might trigger the shift to higher virulence in the microbiome, although other pathological processes, e.g., takotsubo cardiomyopathy (“broken heart syndrome”) , might also play an important role in the mechanism of this process. Remarkably, a recent study found that increased levels of norepinephrine (indicating emotional of physical stress) can induce the dispersion of P. aeruginosa biofilms , which might trigger acute cardiovascular disease: bacteria might even have to do with a broken heart. Mortality might occur without a specific apparent cause also in the elderly: this is referred to as ‘debility not otherwise specified (NOS)’ or ‘failure to thrive (FTT)’ ; in colloquial Hungarian a specific expression exists, spelled as ‘végelgyengülés’ and loosely translated as “terminal frailty”, for this generic cause of death. Virulence shifts of the microbiome might contribute also to this gradual nonspecific weakening at the end of life. Finally, the analysis of the microbiome from old but healthy individuals might help us identify microbial species that are less prone to virulence shifts, or host factors that can prevent the shifts. In contrast, the transmission of symbionts from individuals of failing health should be prevented, as these symbionts are more likely to have switched to higher virulence.”