PerspectiveInfectious Disease

Adapting Koch's postulates

See allHide authors and affiliations

Science  15 Jan 2016:
Vol. 351, Issue 6270, pp. 224-226
DOI: 10.1126/science.aad6753

Embedded Image

Colored scanning electron micrograph of Clostridium difficile bacterial cells (yellow), a common cause of antibiotic-associated diarrhea in hospitalized patients.


In the late 19th century, Robert Koch established his famous postulates as stringent guidelines to evaluate causation in infectious disease (1). These original postulates require isolation of the putative pathogen and reinfection of a healthy host to prove causation. Over the years, Koch's postulates have been continually restated to incorporate the latest scientific findings and technologies (25). Modern molecular techniques have demonstrated that current or previous members of a microbial community can affect disease outcome, providing a nuanced view of strict causation as originally proposed by Koch. There is thus a need to incorporate microbial communities into rigorous modern guidelines for evaluating disease causation.

1 PATHOGEN = 1 DISEASE. Koch's original postulates can be summarized as follows: First, the microorganism occurs in every case of the disease; second, it is not found in healthy organisms; and third, after the microorganism has been isolated from a diseased organism and propagated in pure culture, the proposed pathogen can induce disease anew. Koch did not include the often cited fourth postulate that the microorganism must then be reisolated from the experimentally infected host, but it has come to be viewed as necessary to complete the loop asserting causation. Although revolutionary for the time, the postulates have since been a double-edged sword. For example, the third postulate was implemented to guard against misassignment of causality due to mixed cultures. However, blind adherence to this postulate would mean excluding obligate parasites and viruses as infectious agents.

Over the past decade, sequencing technologies and advanced analytic tools have enabled whole-genome sequencing of both microbial isolates and communities. These advances raise new questions of how Koch's postulates can be updated to incorporate these molecular techniques. For example, when assigning causality to an organism, can a fully sequenced genome act as a surrogate for pure culture, even when the suspected organism requires additional microbes for successful propagation? Also, how do you address the role of microbial communities in disease pathogenesis? Here we address these questions and introduce new variables into Koch's one organism = one disease equation.

1 PATHOGEN + 1 COLONIZATION RESISTOR = 0 DISEASE. A modern test of Koch's postulates is the risk to patients of becoming colonized with a pathogen while hospitalized. In this setting, it has become clear that some commensal organisms can protect the host against pathogenic enemies, a process termed “colonization resistance.” These commensal protectors defend the host either by directly inhibiting the pathogen or by enhancing host immunity (6). Recent evidence for both varieties of colonization resistance highlights how the presence of specific commensal bacteria can alter the pathology induced by Koch-verified infectious bacteria (7, 8). These studies demonstrate how microbial community sequencing can be used to differentiate when an infectious agent induces disease in some but not all hosts.

In one study, Buffie et al. (7) showed that mice treated with antibiotics exhibited varied susceptibility to infection by Clostridium difficile, a major cause of antibiotic-induced diarrhea. The authors also performed a similar analysis with a cohort of patients undergoing stem-cell transplant. Because of antibiotic treatment and compromised immune function, these patients are particularly susceptible to C. difficile infection. With microbial community sequencing and subsequent modeling of microbial interactions, the authors identified Clostridium scindens as a commensal associated with colonization resistance. This was validated when mice precolonized with a commerically available strain of C. scindens exhibited amelioration of symptoms associated with C. difficile infection. Mechanistically, it was demonstrated that C. scindens modifies endogenous bile acids to inhibit C. difficile growth (7, 9). Buffie et al. provide a wellvalidated example of how one organism can protect against a common pathogen (see the figure).

In addition to direct inhibition, a commensal organism can mediate colonization resistance through activation of the immune system. Recently, Schieber et al. demonstrated how a commensal Escherichia coli strain protects against muscle wasting associated with gut trauma and/or infection (8). By sequencing the microbial communities of mice with differential colitis severity, the authors identified an outgrowth of Escherichia species in the more resistant mice. E. coli isolate O21:H+ was subsequently isolated and administered to the susceptible mice, which were then protected from colitis-induced wasting. Notably, an unrelated commensal E. coli strain did not provide a protective effect. This strain specificity highlights the importance of using linked primary, rather than banked isolates, since strains of the same species can display extensive functional variation. Similarly, when mice were infected with Salmonella Typhimurium or Burkholderia thailandensis, precolonization with E. coli O21:H+ reduced the degree of wasting (8). With additional experiments, the authors showed that this protective effect was not due to inhibition of pathogen colonization, but rather that commensal E. coli O21:H+, acting through the innate immune system, down-regulates muscle atrophy and promotes muscle regeneration. In this example, the pathogenesis of an infectious disease is altered because of a single commensal activating the immune system rather than a commensal directly inhibiting the growth of a pathogen.

Microbial protectors.

(A) According to Koch's original postulates, a pathogenic organism in a host will induce disease. (B) This assumption is challenged when an organism is present that can protect against the pathogen. (C) In some cases, consortia of microbes can have an ever greater protective effect.


The idea of distinct E. coli strains conferring colonization resistance is not new. In 1917, the E. coli strain Nissle was isolated from a soldier who did not develop diarrhea during an outbreak of shigellosis (10). Since then, research on this probiotic strain has identified several mechanisms by which it outcompetes pathogens, including an efficient iron acquisition system (11). Despite this early example of colonization resistance, previous updates to Koch's postulates have not considered the overall community context in which a pathogen does or does not induce a disease. As described in (7, 8), the role of specific members of the microbial community in disease pathogenesis could only be identified with antibiotic treatment and subsequent microbial community sequencing, technical advances that Koch could not have imagined as nucleic acids and antibiotics were not discovered until years after his death.

1 PATHOGEN + 1 COMMUNITY = 0 DISEASE. The previous examples highlighted comparatively simple cases of disease causation in which single colonization resistors were important. However, multiple microbes can also have an enhanced protective effect.

As described above, Buffie et al. find that C. scindens provides colonization resistance to C. difficile (7). However, the authors present evidence that even better outcomes are achieved when C. scindens cocolonizes with three other microbes. Similarly, Lawley et al. have reported that C. difficile-infected mice become less sick and clear the pathogen more efficiently after the administration of healthy donor feces (12). To define a more tractable resistant community, Lawley et al. cultured individual isolates from the feces and combined them into phylogenetically distinct mixtures until they had found a six-member community that reproducibly reduced C. difficile infection and bacterial load.

These findings force us to consider under what circumstances a consortium of microbes can fulfill Koch's postulates. For example, do all members of the community have to be grown in pure culture and tested individually, or is it sufficient to grow and test a group culture? This is important in both scientific and translational arenas as researchers strive to create artificial communities capable of recapitulating the positive effects of fecal transplant in patients with recurrent C. difficile infections (13).

In the future, artificial communities could also be created to treat other infections associated with antibiotic-induced alterations of the microbial community. For example, women taking antibiotics are prone to develop mucosal candidiasis due to a depletion of beneficial microbes (14). Instead of traditional probiotic treatments, could a vaginal artificial community be designed to restore normal microbial community dynamics?

TOWARD DYNAMIC ADAPTATION. Combining sequencing and culturing represents a powerful way to explore and define the microorganisms affecting disease outcome. With sequencing, all organisms present in a sample can be observed without the constraints imposed by pure culture requirements. Based on conclusions drawn and hypotheses generated from sequencing results, researchers can then proceed with a more targeted culturing approach to identify organisms of interest.

In summary, updated Koch's postulates incorporating sequencing and culturing would involve the following steps: first, sequencing to classify all members of the microbial community; second, using computational models to assess microbes both necessary and sufficient for disease induction; third, targeted culturing to isolate microbes of interest from the diseased host; and fourth, testing primary isolates and consortia in relevant disease models.

In regards to steps 3 and 4, it should be emphasized that working with primary isolates cultured directly from a diseased host, rather than commercially available strains, lends scientific accuracy given the extensive genetic variability within a species. As an additional step, testing nonprimary isolates can be done to evaluate how widespread the capability is across strains of a species.

In light of recent appreciation of microbial consortia, the scientific community should consider infectious disease causation in a broader systems biology context in which host genetic variability, health status, past exposure history, and microbial strains and communities are all important. As technology advances and new scientific discoveries are made, we must dynamically adapt Koch's postulates so today's science maintains the integrity that Koch originally fostered.


Navigate This Article