The Evolution of Antibiotic Resistance: Insight into the Roles of Molecular Mechanisms of Resistance and Treatment Context
Abstract: The widespread use of antibiotics has markedly improved public health over the last 60 years. However, the efficacy of antibiotic treatment is rapidly decreasing as a result of the continual spread of antibiotic resistance in pathogen populations. The evolution of antibiotic resistance is an amazingly simple example of adaptation by natural selection, and there is growing interest among evolutionary biologists in using evolutionary principles to help understand and combat the spread of resistance in pathogen populations. In this article, we review recent progress in our understanding of the underlying evolutionary forces that drive antibiotic resistance. Recent work has shown that both the mechanisms of antibiotic action and resistance, as well as the treatment context in which resistance evolves, influence the evolution of resistance in predictable ways. We argue that developing predictive models of resistance evolution that can be used to prevent the spread of resistance in pathogen populations requires integrating the treatment context and the molecular biology of resistance into the same evolutionary framework.
The classic model for the evolution of antibiotic resistance is as follows (Figure 1). When antibiotics are applied to a sensitive bacterial population, rare genotypes that acquire resistance as a result of chromosomal mutations or horizontal gene transfer enjoy a competitive advantage relative to their sensitive counterparts; namely, resistant cells have higher Darwinian fitness relative to sensitive cells, and natural selection therefore results in an increase in the frequency of resistance in the population. Resistance mechanisms are inevitably associated with a physiological cost, such that resistant strains have lower fitness than a sensitive strain in the absence of antibiotics. Once a population is dominated by resistant cells, mutants that have second-site compensatory mutations that recover the cost of resistance have higher fitness than resistant genotypes lacking compensatory mutations, and natural selection ultimately results in the fixation of compensatory mutations. Importantly, compensatory adaptation also recovers bacterial fitness in the absence of antibiotics.
Given the simplicity of this scenario, the formidable mathematical framework of population genetics (Hartl and Clark, 2007) can be harnessed to predict the dynamics of evolution of resistance (Débarre et al., 2009; Levin et al., 2000; Schulz zur Wiesch et al., 2010). The only thing needed to apply this framework is quantitative estimates of the mutation rate from sensitive to resistant and from resistant to compensated resistant, as well as the fitness of these three possible genotypes in the presence and absence of antibiotics. More complex versions of this model could also be constructed, for example, to take into account multiple antibiotics or migrations between antibiotic-free and antibiotic-containing environments, but these models would all rely on this basic framework. The key insight provided by recent work in experimental evolution has been that all of the parameters in this simple population genetic model are dependent on the molecular mechanisms of antibiotic action and resistance, as well as the treatment context in which resistance evolves (MacLean et al., 2010). Treatment context, in this case, refers to the number and identity of antibiotics that are used to combat bacteria, as well as the spatial and temporal pattern of antibiotic deployment.
In experimental evolution, populations of microorganisms are challenged with adapting to novel environments under the controlled conditions of the laboratory, such as test tubes containing culture medium supplemented with antibiotics. As a result of the large population size and short generation time of microbes, lab populations evolve very quickly in real time. Because bacteria can be cryogenically frozen in a non-evolving state, evolved populations can then be directly compared with their distant ancestors; for example, evolved bacteria can be directly competed against their ancestor to determine how fitness evolves during an experiment. Whole genome or targeted sequencing can then be used to determine the genetic basis of adaptation. A further important advantage of these experiments is that they can be highly replicated and easily repeated, allowing for powerful statistical analyses of experimental results. Further information on experimental evolution can be found in recent reviews on this topic (Buckling et al., 2009; Elena and Lenski, 2003).
Bacteria acquire antibiotic resistance as a result of spontaneous chromosomal mutations, or by acquiring plasmid-borne resistance alleles by horizontal gene transfer. There are often a number of different spontaneous mutations or horizontally-acquired resistance alleles that confer resistance to a given antibiotic, and these mutations will often have very different fitness consequences for the bacterium. For example, more than 946 unique mutations over 36 genes conferring resistance to seven antibiotics have been identified in Mycobacterium tuberculosis (Sandgren et al., 2009). Similarly, there is a massive diversity of plasmid-borne resistance mechanisms in natural environments (D’Costa et al., 2006; Perron et al., 2008b; Sommer et al., 2009). The fitness effects of these resistance alleles will depend on two key factors: how much resistance the mutation provides, and the physiological cost associated with increased resistance (Figure 2). Recent work shows that these two factors are at least partially predictable from the molecular basis of resistance.
One of the most common mechanisms of antibiotic resistance by chromosomal mutation is to modify the structure of the target enzyme to which an antibiotic binds (Walsh, 2000). In this case, the degree of resistance conferred by a mutation will depend on how successfully the mutation disrupts antibiotic-target binding, which in turn depends on the specificity of the interaction between the antibiotic and its target (MacLean and Buckling, 2009). For example, rifamycin antibiotics inhibit bacterial RNA polymerase by binding to a highly conserved pocket on RNA polymerase and blocking transcription elongation; resistance to these antibiotics arises by mutations in this domain that interfere with antibiotic-target binding. Most mutations in this domain give a large increase in fitness in the presence of rifampicin, because almost any modification to the structure of this pocket is sufficient to provide a large increase in rifampicin resistance. In contrast, the same mutations provide much lower fitness benefits in the presence of sorangicin, because the high conformational flexibility of this antibiotic means that only those mutations that directly change the handful of amino acids in the rifamycin binding pocket that sorangicin binds to provide high levels of sorangicin resistance. We highlight this example because the conformational flexibility of antibiotics can potentially be engineered (Simmons et al., 2010) to minimize selection for resistance.
All known mechanisms of antibiotic resistance are associated with costs that cause a reduction in competitive ability, transmission, and virulence in the absence of antibiotics (Andersson, 2006; Andersson and Hughes, 2010; Andersson and Levin, 1999). The cost of resistance determines the ability of resistant bacteria to compete with sensitive strains, and therefore determines likelihood that resistance will persist after antibiotic treatment is stopped. It has even been suggested that the cost of resistance is the single most important determinant of the spread and maintenance of resistance (Andersson and Hughes, 2010), and this argument is supported by the observation that the least costly resistance mutations tend to be the most common mutations in pathogen populations (Gagneux et al., 2006). Crucially, recent work suggests that costs can be explained by considering the biochemical effects of resistance mutations. First, resistance mutations will tend to reduce the thermodynamic stability of the target protein, resulting in a reduced concentration of the correctly folded and catalytically active form of the enzyme inside the cell (DePristo et al., 2005). For example, destabilization contributes to the cost of resistance due to mutations in β-lactamase (Thomas et al., 2009; Wang et al., 2002) and DNA gyrase (Blance et al., 2000) in E. coli. This cost is not specific to resistance mutations per se, and there is no reason for thinking that resistance mutations will lead to unusually large decreases in protein stability: there is simply a general tendency for random mutations to reduce stability (DePristo et al., 2005; Tokuriki et al., 2007). A second cost of resistance arises from the fact that resistance mutations have indirect effects on a wide range of the normal physiological functions of bacterial cells (Perkins and Nicholson, 2008; Ryu, 1978) — a phenomenon known as pleiotropy in evolutionary genetics. Provided that the physiology of the cell is well-adapted to its current environment, pleiotropy is expected to generate a fitness cost by perturbing the physiological state of the cell. For example, Paulander and colleagues (2009) demonstrated that variation in the cost of three streptomycin resistance mutations can be largely explained by how mutations impact the expression of a single gene, RpoS (σs), and how RpoS (σs) expression correlates with fitness under different environmental conditions.
Despite the tendency for resistance to carry a cost, an extensive body of work shows that bacteria can evolve to reduce or even completely eliminate the cost of resistance without compromising resistance — a process known as compensatory adaptation (Andersson and Hughes, 2010; Andersson and Levin, 1999; Bjorkman et al., 2000; Maisnier-Patin and Andersson, 2004; Maisnier-Patin et al., 2002; Reynolds, 2000). The mechanistic basis of compensatory adaptation has been investigated in great detail; for example, compensation can occur via suppressor mutations (Davis et al., 2009), by amplification of resistance genes (Sandegren and Andersson, 2009), or by increases in the expression of pathways that perform the same function as the compromised resistance gene (Sherman et al., 1996). Given the incredible diversity of mechanisms of compensation, bacteria can probably compensate for the cost of any resistance mutation if given enough time, but experimental studies have shown that the potential for compensation can vary widely between resistance mutations. For example, a recent study showed that some genotypes of P. aeruginosa carrying costly rifampicin resistance mutations completely compensate for the cost of resistance in under 50 generations, while others showed little or no evidence of compensation over ~200 generations of bacterial growth (Hall et al., 2010). Intriguingly, recent work suggests that we may be able to predict the potential for compensatory adaptation from the molecular mechanisms of resistance. For example, all amino acids in a protein make some contribution to the protein’s thermodynamic stability (DePristo et al., 2005), and approximately 20% of missense mutations lead to a moderate increase in stability (Tokuriki et al., 2007); this suggests a massive potential for compensatory adaptation when the cost of resistance is due to compromised protein stability. Conversely, only a small number of amino acids directly contribute to the catalytic domains of proteins, suggesting a much more restricted potential for compensatory adaptation when the cost of resistance is due to compromised catalytic activity or pleiotropy.
Bacterial pathogens inhabit a variable environment, in which the concentration of antibiotics varies in both time and space as a result of both natural processes, such as migration, and human intervention. Recent work has shown that the interactions between antibiotics that are used to treat bacteria and the spatial context of antibiotic treatment play key roles in the evolution of resistance by modifying the costs and benefits associated with resistance mutations.
Given the ease with which bacterial populations adapt to a single antibiotic, multidrug treatment strategies (i.e., “antibiotic cocktails”) are frequently advocated as a means for treating bacterial infections. It is commonly observed that non-additive physiological interactions occur between pairs of antibiotics when they are used together to treat bacteria (Yeh, 2009), and these interactions play an important role in determining the rate of evolution of multidrug resistance (MDR). When an antagonistic interaction occurs between antibiotics (i.e., when different antibiotics mask each other’s effects), resistance mutations are associated with very small benefits, because acquiring resistance to one antibiotic unmasks the inhibitory effect of the second antibiotic, and resistance evolves slowly (Hegreness et al., 2008). Conversely, when a synergistic interaction occurs between antibiotics (i.e., antibiotics aggravate each other’s effects), resistance mutations will be associated with large benefits, because acquiring resistance to one antibiotic will eliminate both the inhibitory effect of the antibiotic and the synergistic effect resulting from the combination of antibiotics, and resistance evolves quickly (Michel et al., 2008).
Similarly, genetic interactions between resistance mutations impact the evolution of resistance in multidrug environments by modifying the cost of resistance. Positive epistasis occurs when the combined cost of carrying multiple resistance mutations is less than what would be expected if the mutations had independent (i.e., multiplicative) effects on fitness. For example, if each of two resistance mutations carries a 50% fitness cost, positive epistasis occurs if the fitness cost of carrying both mutations is less than 75%. In this example, negative epistasis would occur if the combined cost of carrying both mutations is greater than 75%. Positive epistasis promotes the evolution of multidrug resistance by minimizing its cost, and negative epistasis constrains the evolution of multidrug resistance by aggravating its cost. Recent studies have shown that epistasis between chromosomal resistance mutations tends to be positive (Rozen et al., 2007; Trindade et al., 2009; Ward et al., 2009), and in some cases the functional basis of this antagonism is well-understood (MacLean, 2010).
The concentration of antibiotics in both natural and clinical environments inevitably varies in space, and bacteria migrate between areas with high and low antibiotic concentration (Kemper, 2008; Martinez, 2009). Most obviously, immigration into antibiotic treated areas will promote the evolution of resistance by introducing novel resistance alleles into antibiotic treated areas. For example, untreated, healthy patients and the hospital environment have both been shown to act as a source for antibiotic resistant bacteria that colonize susceptible patients (Sánchez-Payá et al., 2009; Sexton et al., 2006). Experimental studies show that the impact of immigration depends critically on the rate of in situ evolution of resistance in treated areas (Perron et al., 2007; Perron et al., 2008a). When resistance mutations are readily available, or lead to large increases in fitness, immigration from antibiotic-free environments leads to little, if any, acceleration in the rate of evolution of resistance. In contrast, when resistance mutations are rare, or tend to have small benefits, immigration massively accelerates the evolution of resistance. When antibiotics are withdrawn from use, immigration has the potential to eliminate resistance by forcing slow-growing resistant bacteria to compete with sensitive bacteria that have not incurred any cost of resistance (Perron and Buckling, 2010), as has been observed for macrolide resistance in Streptococcus pyogenes following cessation of antibiotic use (Seppala et al., 1997).
The ultimate goal of current research in resistance evolution is to use evolutionary principles to develop predictive models of resistance evolution that can be used to combat the spread of resistance in pathogen populations. Recent work has shown that both the mechanisms of antibiotic action and resistance and the ecological (treatment) context in which resistance evolves influence the evolution of resistance in a predictable manner (MacLean et al., 2010). We argue that, in order to develop predictive models of resistance, evolutionary biologists are going to be forced to develop evolutionary models of resistance that integrate biochemistry and ecology into the same framework. This is an important challenge, because evolutionary ecology and molecular evolution are essentially two separate fields. However, we suggest that the maturation of this integrative approach to resistance is both achievable and imminent. Applying this framework to combating resistance will require extensive collaboration between evolutionary biologists, medical microbiologists, and the medical community at large. We think that two important tools will help to facilitate this collaboration. First, detailed in vivo sampling of resistance evolution both within and between patients provides an ideal way to test the predictions of evolutionary models in a relevant clinical setting. Some excellent examples of these studies exist in the literature (Harris et al., 2010; Smith et al., 2006), but we suggest that future work in this area should be carried out with a much greater degree of replication. Second, databases that bring together information on the mechanisms and frequency of antibiotic resistance from different clinical studies, such as the TBDream database (Sandgren et al., 2009), will provide crucial data on the evolution of resistance, especially if the results of these studies are linked to information on treatment strategies.
This work was supported by funding from the Royal Society (RCM), the Leverhulme Trust (AB), and the European Research Council (AB).
Andersson DI. The biological cost of mutational antibiotic resistance: Any practical conclusions? Curr Opin Microbiol 9(5):461-5, 2006.
Andersson DI, Hughes D. Antibiotic resistance and its cost: Is it possible to reverse resistance? Nat Rev Microbiol 8(4):260-71, 2010.
Andersson DI, Levin BR. The biological cost of antibiotic resistance. Curr Opin Microbiol 2(5):489-93, 1999.
Bjorkman J, Nagaev I, Berg OG, Hughes D, Andersson DI. Effects of environment on compensatory mutations to ameliorate costs of antibiotic resistance. Science 287(5457):1479-82, 2000.
Blance SJ, Williams NL, Preston ZA, Bishara J, Smyth MS, Maxwell A. Temperature-sensitive suppressor mutations of the escherichia coli DNA gyrase b protein. Protein Sci 9(5):1035-7, 2000.
Buckling A, MacLean RC, Brockhurst M, Colegrave N. The beagle in a bottle. Nature 457:824-9, 2009.
D’costa VM, Mcgrann KM, Hughes DW, Wright GD. Sampling the antibiotic resistome. Science 311(5759):374-7, 2006.
Davis BH, Poon AFY, Whitlock MC. Compensatory mutations are repeatable and clustered within proteins. Proc R Soc Lond B Biol Sci 276(1663):1823-7, 2009.
Débarre F, Lenormand T, Gandon S. Evolutionary epidemiology of drug-resistance in space. PLoS Comput Biol 5(4):e1000337, 2009.
Depristo MA, Weinreich DM, Hartl DL. Missense meanderings in sequence space: A biophysical view of protein evolution. Nat Rev Genet 6(9):678-87, 2005.
Elena SF, Lenski RE. Evolution experiments with microorganisms: The dynamics and genetic bases of adaptation. Nat Rev Genet 4(6):457-69, 2003.
Gagneux S, Long CD, Small PM, Van T, Schoolnik GK, Bohannan BJM. The competitive cost of antibiotic resistance in mycobacterium tuberculosis. Science 312(5782):1944-6, 2006.
Hall AR, Griffiths VF, Maclean RC, Colegrave N. Mutational neighbourhood and mutation supply rate constrain adaptation in pseudomonas aeruginosa. Proc R Soc Lond B Biol Sci 277(1681):643-50, 2010.
Harris SR, Feil EJ, Holden MTG, Quail MA, Nickerson EK, Chantratita N, Gardete S, Tavares A, Day N, Lindsay JA, Edgeworth JD, De Lencastre H, Parkhill J, Peacock SJ, Bentley SD. Evolution of mrsa during hospital transmission and intercontinental spread. Science 327(5964):469-74, 2010.
Hartl D, Clark A. Principles of Population Genetics, 4th edition. Sinauer, Sunderland, MA, USA, 2007.
Hegreness M, Shoresh N, Damian D, Hartl D, Kishony R. Accelerated evolution of resistance in multidrug environments. Proc Natl Acad Sci U S A 105(37):13977-81, 2008.
Kemper N. Veterinary antibiotics in the aquatic and terrestrial environment. Ecol Indic 8(1):1-13, 2008.
Levin BR, Perrot V, Walker N. Compensatory mutations, antibiotic resistance and the population genetics of adaptive evolution in bacteria. Genetics 154(3):985-97, 2000.
Maclean RC. Predicting epistasis: An experimental test of metabolic control theory with bacterial transcription and translation. J Evol Biol 23(3):488-93, 2010.
Maclean RC, Buckling A. The distribution of fitness effects of beneficial mutations in pseudomonas aeruginosa. PLoS Genet 5(3):e1000406, 2009.
Maclean RC, Hall AR, Perron GG, Buckling A. The population genetics of antibiotic resistance: Integrating molecular mechanisms and treatment contexts. Nat Rev Genet 11(6):405-14, 2010.
Maisnier-Patin S, Andersson DI. Adaptation to the deleterious effects of antimicrobial drug resistance mutations by compensatory evolution. Res Microbiol 155(5):360-9, 2004.
Maisnier-Patin S, Berg OG, Liljas L, Andersson DI. Compensatory adaptation to the deleterious effect of antibiotic resistance in salmonella typhimurium. Mol Microbiol 46(2):355-66, 2002.
Martinez JL. The role of natural environments in the evolution of resistance traits in pathogenic bacteria. Proc R Soc Lond B Biol Sci 276(1667):2521-30, 2009.
Michel JB, Yeh PJ, Chait R, Moellering RC, Kishony R. Drug interactions modulate the potential for evolution of resistance. Proc Natl Acad Sci U S A 105(39):14918-23, 2008.
Paulander W, Maisnier-Patin S, Andersson DI. The fitness cost of streptomycin resistance depends on rpsL mutation, carbon source and RpoS (sigmaS). Genetics 183(2):539-46, 2009.
Perkins AE, Nicholson WL. Uncovering new metabolic capabilities of bacillus subtilis using phenotype profiling of rifampin-resistant RpoB mutants. J Bacteriol 190(3):807-14, 2008.
Perron GG, Buckling A. Hypermutability and compensatory adaptation in antibiotic-resistant bacteria. Am Nat, in press, 2010.
Perron GG, Gonzalez A, Buckling A. Source-sink dynamics shape the evolution of antibiotic resistance and its pleiotropic fitness cost. Proc R Soc Lond B Biol Sci 274(1623):2351-6, 2007.
Perron GG, Gonzalez A, Buckling A. The rate of environmental change drives adaptation to an antibiotic sink. J Evol Biol 21(6):1724-31, 2008a.
Perron GG, Quessy S, Bell G. A reservoir of drug-resistant pathogenic bacteria in asymptomatic hosts. PLoS ONE 3(11):e3749, 2008b.
Reynolds MG. Compensatory evolution in rifampin-resistant escherichia coli. Genetics 156(4):1471-81, 2000.
Rozen DE, Mcgee L, Levin BR, Klugman KP. Fitness costs of fluoroquinolone resistance in streptococcus pneumoniae. Antimicrob Agents Chemother 51(2):412-6, 2007.
Ryu JI. Pleiotropic effect of a rifampin-resistant mutation in bacillus-subtilis. J Bacteriol 135(2):408-14, 1978.
Sánchez-Payá J, Bischofberger C, Lizan M, Lozano J, Platán EM, Navarro J, Paz J, Vicente JA. Nosocomial infection surveillance and control: Current situation in spanish hospitals. J Hosp Infec 72(1):50-6, 2009.
Sandegren L, Andersson DI. Bacterial gene amplification: Implications for the evolution of antibiotic resistance. Nat Rev Microbiol 7(8):578-88, 2009.
Sandgren A, Strong M, Muthukrishnan P, Weiner BK, Church GM, Murray MB. Tuberculosis drug resistance mutation database. PLoS Med 6(2):e2, 2009.
Schulz ZUR Wiesch P, Engelstadter J, Bonhoeffer S. Compensation of fitness costs and reversibility of antibiotic resistance mutations. Antimicrob Agents Chemother 54(5):2085-95, 2010.
Seppala H, Klaukka T, Vuopio-Varkila J, Muotiala A, Helenius H, Lager K, Huovinen P. The effect of changes in the consumption of macrolide antibiotics on erythromycin resistance in group a streptococci in finland. Finnish study group for antimicrobial resistance. N Engl J Med 337(7):441-6, 1997.
Sexton T, Clarke P, O’neill E, Dillane T, Humphreys H. Environmental reservoirs of methicillin-resistant staphylococcus aureus in isolation rooms: Correlation with patient isolates and implications for hospital hygiene. J Hosp Infect 62(2):187-94, 2006.
Sherman DR, Mdluli K, Hickey MJ, Arain TM, Morris SL, Barry CE, Stover CK. Compensatory ahpC gene expression in isoniazid-resistant mycobacterium tuberculosis. Science 272(5268):1641-3, 1996.
Simmons KJ, Chopra I, Fishwick CWG. Structure-based discovery of antibacterial drugs. Nat Rev Microbiol 8(7):501-10, 2010.
Smith EE, Buckley DG, Wu Z, Saenphimmachak C, Hoffman LR, D’Argenio DA, Miller SI, Ramsey BW, Speert DP, Moskowitz SM, Burns JL, Kaul R, Olson MV. Genetic adaptation by pseudomonas aeruginosa to the airways of cystic fibrosis patients. Proc Natl Acad Sci U S A 103(22):8487-92, 2006.
Sommer MO, Dantas G, Church GM. Functional characterization of the antibiotic resistance reservoir in the human microflora. Science 325(5944):1128-31, 2009.
Thomas VL, Mcreynolds AC, Shoichet BK. Structural bases for stability-function tradeoffs in antibiotic resistance. J Mol Biol 396(1):47-59, 2010.
Tokuriki N, Stricher F, Schymkowitz J, Serrano L, Tawfik DS. The stability effects of protein mutations appear to be universally distributed. J Mol Biol 369(5):1318-32, 2007.
Trindade S, Sousa A, Xavier KB, Dionisio F, Ferreira MG, Gordo I. Positive epistasis drives the acquisition of multidrug resistance. PLoS Genet 5(7):e1000578, 2009.
Walsh C. Molecular mechanisms that confer antibacterial drug resistance. Nature 406(6797):775-81, 2000.
Wang XJ, Minasov G, Shoichet BK. Evolution of an antibiotic resistance enzyme constrained by stability and activity trade-offs. J Mol Biol 320(1):85-95, 2002.
Ward H, Perron G, Maclean RC. The fitness cost of multiple drug resistance in pseudomonas aeruginosa. J Evol Biol 22:997-1003, 2009.
[Discovery Medicine; ISSN: 1539-6509; eISSN: 1944-7930. Discov Med 10(51):112-118, August 2010.]