In the version of the article initially published, there were errors in the Abstract and Discussion. In the Abstract, the sentence now reading “Since mammalian oligoadenylate synthetases also bind viral double-stranded RNA…” previously read “As the mammalian cyclase OAS1 also binds viral double-stranded RNA…”. In the first paragraph of the Discussion, the sentence now reading “A similar observation has been reported for the human antiviral 2′,5′-oligoadenylate synthetase (OAS) enzymes OAS1 and OAS3…” previously read “A similar observation has been reported for the human OAS1 and OAS3 cyclases…”. At the beginning of the final paragraph of the Discussion, “Eukaryotic synthases” previously read “Eukaryotic cyclases”. These changes have been made to the HTML and PDF versions of the article.
Antibiotics are considered one of the most important contributions to clinical medicine in the last 100 years. Due to the use and overuse of these drugs, there have been increasing frequencies of infections with resistant pathogens. One form of resistance, heteroresistance, is particularly problematic; pathogens appear sensitive to a drug by common susceptibility tests. However, upon exposure to the antibiotic, resistance rapidly ascends, and treatment fails. To quantitatively explore the processes contributing to the emergence and ascent of resistance during treatment and the waning of resistance following cessation of treatment, we develop two distinct mathematical and computer-simulations models of heteroresistance. In our analysis of the properties of these models, we consider the factors that determine the response to antibiotic-mediated selection. In one model, heteroresistance is progressive, with each resistant state sequentially generating a higher resistance level. In the other model, heteroresistance is non-progressive, with a susceptible population directly generating populations with different resistance levels. The conditions where resistance will ascend in the progressive model are narrower than those of the non-progressive model. The rates of reversion from the resistant to the sensitive states are critically dependent on the transition rates and the fitness cost of resistance. Our results demonstrate that the standard test used to identify heteroresistance is insufficient. The predictions of our models are consistent with empirical results. Our results demand a reevaluation of the definition and criteria employed to identify heteroresistance. We recommend the definition of heteroresistance should include a consideration of the rate of return to susceptibility.
There is a surfeit of bioinformatic data showing that bacteriophages abound in the enteric microbiomes of humans. What is the contribution of these viruses in shaping the bacterial strain and species composition of the gut microbiome and how are these phages maintained over time? To address these questions, we performed experiments with Escherichia coli and phages isolated from four fecal microbiota transplantation (FMT) doses as representative samples of non-dysbiotic enteric microbiota and develop and analyze the properties of a mathematical model of the population and evolutionary dynamics of bacteria and phage. Our models predict and experiments confirm that due to production of the O antigen, E. coli in the enteric microbiome are likely to be resistant to infection with co-occurring phages. Furthermore, our modeling suggests that the phages can be maintained in the population due to the high rates of host transition between resistant and sensitive states, which we call leaky resistance. Based on our observations and model predictions, we postulate that the phages found in the human gut are likely to play little role in shaping the composition of E. coli at the strain level in the enteric microbiome in healthy individuals. How general this is for other species of bacteria in the enteric flora is not yet clear, although O antigen expression is common across many taxa.
Traditionally, bacteriostatic antibiotics are agents able to arrest bacterial growth. Despite being traditionally viewed as unable to kill bacterial cells, when they are used clinically the outcome of these drugs is frequently as effective as when a bactericidal drug is used. We explore the dynamics of Escherichia coli after exposure to two ribosome-targeting bacteriostatic antibiotics, chloramphenicol and azithromycin, for thirty days. The results of our experiments provide evidence that bacteria exposed to these drugs replicate, evolve, and generate a sub-population of small colony variants (SCVs) which are resistant to multiple drugs. These SCVs contribute to the evolution of heteroresistance and rapidly revert to a susceptible state once the antibiotic is removed. Stated another way, exposure to bacteriostatic drugs selects for the evolution of heteroresistance in populations previously lacking this trait. More generally, our results question the definition of bacteriostasis as populations exposed to bacteriostatic drugs are replicating despite the lack of net growth.
Cyclic oligonucleotide-based antiphage signalling systems (CBASS) protect prokaryotes from viral (phage) attack through the production of cyclic oligonucleotides, which activate effector proteins that trigger the death of the infected host1,2. How bacterial cyclases recognize phage infection is not known. Here we show that staphylococcal phages produce a structured RNA transcribed from the terminase subunit genes, termed CBASS-activating bacteriophage RNA (cabRNA), which binds to a positively charged surface of the CdnE03 cyclase and promotes the synthesis of the cyclic dinucleotide cGAMP to activate the CBASS immune response. Phages that escape the CBASS defence harbour mutations that lead to the generation of a longer form of the cabRNA that cannot activate CdnE03. Since mammalian oligoadenylate synthetases also bind viral double-stranded RNA during the interferon response, our results reveal a conserved mechanism for the activation of innate antiviral defence pathways.
Many so-called pathogenic bacteria make their living as commensals or even symbionts of the hosts that they colonize. Bacteria such as “Neiserria Meningitidis,” “Haemophilus influenzae,” “Staphylococcus aureus (1),” “Streptococcus pneumoneae,” “Helicobacter pylori,” and “Echerichia coli” are far more likely to colonize and maintain their populations in healthy individuals, asymptomatically, than to cause disease. Moreover, the members of these otherwise benign or beneficial species that are actually responsible for diseases like meningitis and sepsis, are not transmitted to new hosts and are therefore at an ecological and evolutionary dead-end. This implies that the virulence factors responsible for the pathogenicity of these bacteria must evolve in response to selection pressures other than those for causing disease. What are these pressures? Here we consider “Neisseria meningitidis”--a common member of the commensal flora of the nasal pharyngeal passages of humans that is also responsible for sporadic and epidemic meningitis. We focus on the evolution of phase shifting--a mutational process that turns genes on and off and, in particular, genes that code for virulence determinants such as pili, lipopolysaccharide, capsular polysaccharide, and outer membrane proteins. Using mathematical models, we offer two testable hypotheses: First, within a single human host, fast phase shifting leads to virulence. And second, although virulence may be disadvantageous within the framework of a single host, fast phase shifting may evolve in response to selection operating at a multihost epidemiological level. We discuss avenues for empirically testing these hypotheses and the implications of this work for the evolution of virulence in general.
A method is presented to evaluate in vitro the efficacy of antibiotics to treat bacteria growing as discrete colonies on surfaces and the contribution of the colony structure to the antibiotic susceptibility of bacteria. Using this method, we explored the relative efficacy of six bactericidal and three bacteriostatic antibiotics to inhibit the growth and kill Staphylococcus aureus colonies of different sizes, densities and ages. As measured by the reduction in viable cell density relative to untreated controls, of the bactericidal drugs tested ciprofloxacin and gentamicin were most effective. By this criteria, ampicillin was more effective than oxacillin. Daptomycin and vancomycin were virtually ineffective for treating S. aureus growing as colonies. The bacteriostatic antibiotic tested, tetracycline, linezolid and erythromycin were all able to prevent the growth of S. aureus colonies and did so even more effectively than daptomycin, which is highly bactericidal in liquid culture. The results of these experiments and other observations suggest that relative inefficacy of oxacillin, vancomycin and daptomycin to kill S. aureus in colonies is due to the density and physiological state of the bacteria rather than the inability of these drugs to penetrate the colonies. The methods developed here are general and can be used to explore the efficacy of antibiotics to treat bacteria growing in biofilms as well as discrete colonies.
Bacteriophages are deemed either lytic (virulent) or temperate, respectively depending on whether their genomes are transmitted solely horizontally, or both horizontally and vertically. To elucidate the ecological and evolutionary conditions under which natural selection will favor the evolution and maintenance of lytic or temperate modes of phage replication and transmission, we use a comprehensive mathematical model of the dynamics of temperate and virulent phage in populations of bacteria sensitive and resistant to these viruses. For our numerical analysis of the properties of this model, we use parameters estimated with the temperate bacteriophage Lambda, λ, it’s clear and virulent mutants, and E. coli sensitive and resistant - refractory to these phages. Using batch and serial transfer population dynamic and reconstruction experiments, we test the hypotheses generated from this theoretical analysis. Based on the results of this jointly theoretical and experimental study, we postulate the conditions under which natural selection will favor the evolution and maintenance of lytic and temperate modes of phage replication and transmission. A compelling and novel prediction this in silico, in vitro, and in plastico study makes is lysogenic bacteria from natural populations will be resistant-refractory to the phage for which they are lysogenic as well as lytic phage sharing the same receptors as these temperate viruses.
The most significant difference between bacteriophages functionally and ecologically is whether they are purely lytic (virulent) or temperate. Virulent phages can only be transmitted horizontally by infection, most commonly with the death of their hosts. Temperate phages can also be transmitted horizontally, but upon infection of susceptible bacteria, their genomes can be incorporated into that of their host’s as a prophage and be transmitted vertically in the course of cell division by their lysogenic hosts. From what we know from studies with the temperate phage Lambda and other temperate phages, in laboratory culture, lysogenic bacteria are protected from killing by the phage coded for by their prophage by immunity; where upon infecting lysogens, the free temperate phage coded by their prophage are lost. Why are lysogens not also resistant as well as immune to the phage coded by their prophage since immunity does not confer protection against virulent phages? To address this question, we used a mathematical model and performed experiments with temperate and virulent mutants of the phage Lambda in laboratory culture. Our models predict and experiments confirm that selection would favor the evolution of resistant as well as immune lysogens, particularly if the environment includes virulent phage that share the same receptors as the temperate. To explore the validity and generality of this prediction, we examined ten lysogenic Escherichia coli from natural populations. All ten were capable of forming immune lysogens but their original hosts were resistant to the phage coded by their prophage.
The minimum inhibitory concentration (MIC) of an antibiotic required to prevent replication is used both as a measure of the susceptibility/resistance of bacteria to that drug and as the single pharmacodynamic parameter for the rational design of antibiotic treatment regimes. MICs are estimated in vitro under conditions optimal for the action of the antibiotic. However, bacteria rarely grow in these optimal conditions. Using a mathematical model of the pharmacodynamics of antibiotics, we make predictions about the nutrient dependency of bacterial growth in the presence of antibiotics. We test these predictions with experiments in a rich media and a glucose-limited minimal media with Escherichia coli and eight different antibiotics. Our experiments uncover properties that question the sufficiency of using MICs and simple pharmacodynamic functions as measures of the pharmacodynamics of antibiotics under the nutritional conditions of infected tissues. To an extent that varies among drugs: (i) The estimated MICs obtained in rich media are greater than those estimated in minimal media. (ii) Exposure to these drugs increases the time before logarithmic growth starts, their lag. (iii) The stationary phase density of E. coli populations declines with greater sub-MIC antibiotic concentrations. We postulate a mechanism to account for the relationship between the sub-MIC concentration of antibiotics and the stationary phase density of bacteria and provide evidence in support of this hypothesis. We discuss the implications of these results to our understanding of the MIC as the unique pharmacodynamic parameter used to design protocols for antibiotic treatment.