The Princeton Guide to Evolution

VIII.3

Evolution of Antibiotic Resistance

Dan I. Andersson

OUTLINE

1. A medical miracle—and how to ruin it

2. Origins of antibiotics and antibiotic-resistance mechanisms

3. Transmission of resistant bacteria

4. Persistence and reversibility of resistance

5. Can resistance evolution be slowed or even stopped?

6. Will antibiotics become a footnote to medical history?

Antibiotics have revolutionized human and veterinary medicine, and over the last 60 years they have made it possible to treat efficiently most types of bacterial infections. Unfortunately, the extensive use—and frequent misuse—of antibiotics has resulted in the rapid evolution and spread of bacteria that are resistant to antibiotics. Arguably, the global use of antibiotics is one of the largest evolution experiments performed by humans, and the frightening consequence is that we are now at the brink of a postantibiotic era in which antibiotics have lost their miraculous power. This problem originates from the strong selection imposed by the extensive use of antibiotics and the resulting enrichment of resistance mutations and horizontally acquired resistance genes. Together these factors have generated high-level antibiotic resistance in the majority of significant human and veterinary pathogens. Several forces act to stabilize resistance in a population once it becomes established, and resistant bacteria may thus persist for a long time even after use of an antibiotic has been reduced. The development of new classes of antibiotics, coupled with more prudent use of antibiotics, will be required to maintain antibiotics as efficient agents for treating bacterial infections.

GLOSSARY

Antibiotic. An antibacterial compound that may have a natural or synthetic origin.

Biological Fitness Cost. The effect that a resistance mechanism has on bacterial fitness (including growth, persistence, and survival within and outside hosts) in the absence of antibiotic.

Compensatory Evolution. Reduction or elimination of the fitness cost associated with a mutation that has a deleterious side effect (e.g., a resistance mutation) by additional genetic changes (compensatory mutations).

Conjugation. Transfer of genetic material between bacterial cells mediated by direct contact between two cells.

Coselection. Process whereby a nonselected gene indirectly increases in frequency by virtue of its genetic linkage (within a genetic element or a bacterial clone) with a directly selected gene; sometimes also called genetic hitchhiking.

Horizontal Gene Transfer. A process in which a recipient organism receives and incorporates genetic material from a donor organism without being the offspring of the donor; sometimes also called lateral gene transfer.

Minimum Inhibitory Concentration (MIC). The lowest concentration of an antimicrobial drug that inhibits the growth of a bacterial population.

Nosocomial Infection. Infection contracted during treatment in a hospital or other healthcare facility.

Plasmid. A DNA molecule that is separate from and can replicate independently of the chromosomes in bacteria.

Resistome. A neologism that refers to the set of resistance genes and precursors to resistance genes that are present in all pathogenic and nonpathogenic bacterial species combined.

Transduction. Injection of foreign DNA into bacterial cells by a bacteriophage (i.e., a virus that infects bacteria).

Transformation. Uptake of exogenous DNA into a cell through the cell envelope.

1. A MEDICAL MIRACLE—AND HOW TO RUIN IT

Antibiotics represent one of the most important medical advances in modern times, and since their introduction over 60 years ago they have saved countless lives. Today we often take antibiotics for granted in the developed parts of the world, but we have to go back only to our grandparents to find a generation for which common infections such as pneumonia, meningitis, blood poisoning, and intestinal infections were potentially deadly. Charles Fletcher, a young physician who was involved in early clinical trials of penicillin in the 1940s, describes vividly how the introduction of antibiotics changed modern medicine:

It is difficult to convey the excitement of actually witnessing the amazing power of penicillin over infections for which there had previously been no effective treatment…. I did glimpse the disappearance of the chambers of horrors which seems to be the best way to describe those old septic wards … and could see that we should never again have to fear the streptococcus or the more deadly staphylococcus.

In addition to being widely used for the treatment of many common community and nosocomial (hospital-acquired) infections, antibiotics are also an essential component in the treatment and prevention of infections associated with advanced medical practices including chemotherapy of cancers, organ transplantation, implantation and replacement of medical devices and prostheses, neonatal care, and invasive surgery.

Unfortunately, the utility of antibiotics is deteriorating at an alarming rate, and the reason for this change is easily understood in the context of Darwinian adaptive evolution. To put it simply, bacteria adapt genetically to the presence of antibiotics by acquiring various types of resistance mechanisms that prevent antibiotics from performing their inhibitory function. These resistance mechanisms allow the bacteria to grow and reproduce in the presence of antibiotics, and evolution thereby nullifies their efficacy in treating infections. The widespread use—and often the overuse—of antibiotics on a global scale (estimated currently to be several hundred thousand tons per year) for human medicine, veterinary medicine, and agriculture is the main reason for the selection and spread of resistance among both human and animal bacterial pathogens. So, although the introduction of antibiotics is often viewed as one of humankind’s greatest achievements, we are now at risk of destroying that achievement. At the very least, we are paying a high price for the increased resistance. The overuse of antibiotics reflects several factors, including poor knowledge among prescribers and patients, profits for physicians and pharmacists from the prescription and sale of antibiotics, aggressive marketing from pharmaceutical companies, and the lack of regulations and guidelines for when and how antibiotics should be properly used. Studies have shown correlations between the amount of antibiotics used and the prevalence of resistance at several levels (e.g., country, hospital), as would be expected if antibiotics select for increased bacterial resistance.

As a society, we are paying a high price for the increased levels of bacterial resistance to antibiotics: resistant bacteria limit our ability to efficiently treat bacterial infections, and they also increase the risk of complications and even death. In addition, antibiotic resistance imposes a large economic burden on the healthcare system owing to increased treatment costs as well as the costs of identifying and developing new, alternative compounds. Worldwide there are areas where bacterial infections have become untreatable as the result of antibiotic resistance, and with the recent spread of gram-negative bacteria that produce multidrug-resistant extended-spectrum beta-lactamase (ESBL), the problem is becoming even more acute. This trend toward increasing resistance, combined with diminished research and development of new antibiotics, has led to a dismal situation in which we may face a postantibiotic era.

2. ORIGINS OF ANTIBIOTICS AND ANTIBIOTIC-RESISTANCE MECHANISMS

Antibiotics are compounds that inhibit (bacteriostatic drugs) or kill (bactericidal drugs) bacteria by a specific interaction with some target in the bacterial cell. Some purists limit the definition of antibiotics to only those substances produced by a microorganism, but today all natural, semisynthetic (i.e., a combination of natural and synthetic precursors), and synthetic compounds with antibacterial activity are generally classified as antibiotics. The target for an antibiotic can be an essential enzyme or cellular process such as protein synthesis, cell wall biosynthesis, transcription, or DNA replication. Most medically relevant and industrially produced antibiotics originate in nature and are synthesized by a variety of species, mainly soil-dwelling bacteria (in particular the genus Streptomyces among the phylum Actinobacteria) and fungi. The benefits of antibiotics for microbial producers is a matter of debate; antibiotics might be used as ecological weapons to inhibit competitors, but they might have a more benevolent function as signals for cell-to-cell communication in microbial communities. In any case, the synthesis and release of antibiotic compounds in nature means that many bacteria (both producers and bystanders) have long histories of exposure to antibiotics, and as a consequence, many have evolved various resistance mechanisms. These mechanisms likely evolved to protect against self-destruction (in antibiotic producers), to defend against antibiotics produced by other species, to modulate intermicrobe communication, or to perform metabolic functions unrelated to antibiotics. This vast pool of resistance genes, known as the resistome, has the potential to be transferred within and between species, and to confer resistance to any antibiotic that might be used against human and animal pathogens. In fact, many of the resistance problems generated by the use of antibiotics since the 1940s are a consequence of the acquisition of preexisting resistance determinants by pathogens via horizontal gene transfer (HGT). Transfer mechanisms include conjugative transfer of plasmids and conjugative transposons, transduction via bacteriophages (viruses that infect bacteria), and transformation of naked DNA taken up from the environment by some species. Of these mechanisms, conjugative transfers are the most common mode of acquiring resistance, whereas bacteriophage transfers appear to be rare. Apart from HGT, resistance can also arise by mutations (including point mutations as well as rearrangements and gene amplification) in native resident genes.

For resistance to become a problem, the acquired or mutated resistance genes must be phenotypically expressed in clinically relevant human and animal pathogens. The evolutionary pathways leading to these outcomes are often complex and often not well understood, especially in the case of resistance acquired by HGT. Even when a potential donor of a resistance gene has been identified by genome sequence data (e.g., the CTX-M type of ESBL resistance was likely acquired from Kluyvera strains in the environment), several conditions must be fulfilled for resistance to emerge in the case of HGT: (1) a resistance mechanism must be present in a donor bacterium; (2) there must be a genetic mechanism for HGT; (3) there must be an ecological opportunity for transfer between the donor and recipient cells (e.g., in the case of conjugation direct contact is needed); (4) the transferred gene must be stably inherited, adequately expressed, and confer a resistant phenotype in the recipient; and (5) there must be strong enough selection—typically, a sufficient level of antibiotic—to favor the resistant recipient organisms, even though resistance may impose a fitness cost (as discussed in section 4). In the case of resistance that occurs by mutation in resident genes, the process is simpler and requires only a suitable resistance mutation and sufficient selection to favor the resistant mutants. Despite the relative ease by which bacteria can become resistant by mutations, HGT is the predominant route for generation of antibiotic resistance in most human and animal pathogens. The likely explanation is that pathogenic bacteria can acquire high-level resistance to a given antibiotic—and indeed, simultaneous resistance to several antibiotics—by means of a single transfer event from the relatively accessible pool of resistance genes in the microbial community’s resistome. Of special relevance here are genetic elements called integrons that can capture and express arrays of resistance genes; when integrons are transferred on a plasmid, they can convert the recipient strain from being antibiotic susceptible to multidrug resistant. In contrast, mutation-based resistance often produces lower-level resistance and may require several mutational steps to produce high-level resistance, thus requiring a longer evolutionary path to achieve a clinically resistant phenotype. A notable exception, however, is Mycobacterium tuberculosis, in which all known resistance mechanisms are the result of mutation rather than HGT, and single mutations sometimes produce high-level resistance (e.g., resistance to streptomycin and rifampicin arises from mutations in ribosomal protein S12 and RNA polymerase subunit β, respectively). Mycobacteria have conjugative plasmids and transducing bacteriophages, and it is unclear why HGT is not associated with resistance evolution in this bacterium.

Horizontally acquired genes and mutations in native genes confer resistance to bacteria by a variety of different mechanisms that either protect the normal cellular target from exposure to the antibiotic or alter the target’s structure to prevent the drug from binding. (1) The antibiotic may be enzymatically inactivated by hydrolysis (e.g., resistance to beta-lactam antibiotics conferred by beta-lactamases) or modification (e.g., acetylation, phosphorylation, or adenylylation of aminoglycosides). (2) Uptake of the antibiotic may be reduced by changes in the cell wall (e.g., mutations that confer low-level β-lactam resistance by altering channels called porins). (3) Bacteria may express efflux systems that actively pump the antibiotic out of the cell (e.g., efflux pumps that confer β-lactam or aminoglycoside resistance). (4) The target molecule may be modified such that antibiotic is prevented from binding (e.g., mutations in ribosomal proteins or rRNA that inhibit binding of aminoglycosides). (5) Resistance may result from a bypass mechanism whereby the need for the inhibited target is relieved by provision of an alternative target or pathway (e.g., resistance to peptide deformylase inhibitors by inactivation of formyl transferase). Mechanisms 1, 4, and 5 often provide high-level resistance, whereas mechanisms 2 and 3 are typically associated with lower-level resistance.

3. TRANSMISSION OF RESISTANT BACTERIA

Once resistance has evolved in a bacterial pathogen, the extent to which it becomes a medical problem depends on how rapidly and extensively the resistant type is transmitted from its place of origin into the human or animal population and the rate at which it is disseminated among the hosts. Transmission rates of resistant bacteria depend on many factors including host density, patterns of host travel and migration, various hygienic factors (e.g., in hospitals and during food preparation), host immunity (e.g., vaccination), and the intrinsic transmissibility of the resistant pathogens. In principle, humans can influence all these except the last factor.

4. PERSISTENCE AND REVERSIBILITY OF RESISTANCE

Whether antibiotic resistance will persist in a bacterial population after it emerges depends in general on the relative strength of several selective forces. The most obvious of these is the direct advantage to resistant bacteria caused by exposure to concentrations of antibiotics that are lethal or inhibitory to sensitive strains. An opposing force, however, is the fitness cost of resistance, that is, any effect of the resistance mechanism that reduces the ability of the pathogen to grow, persist, or spread in the host population. Such costs will impede the rise of resistant bacteria, and these costs will also affect the likelihood that resistance can be reversed or otherwise eliminated. While these fitness costs offer hope that resistance can be controlled, other forces discussed later can stabilize resistance in a bacterial population, even when the antibiotic is absent or at a low concentration.

Sub-MIC Selection

Selection clearly favors resistant strains when antibiotic concentrations are above the minimum inhibitory concentration (MIC) of the susceptible bacteria, but it remains unclear whether levels far below the MIC can also select for resistance. Direct measurements of antibiotic levels in organs and tissues of treated patients and in various natural environments indicate that bacteria are frequently exposed to sub-MIC levels of drugs. In theory, such low antibiotic levels may select for resistance if susceptible bacteria grow even slightly more slowly than resistant strains. Antibiotics can be introduced into the environment in the urine from treated humans and animals, as well as when antibiotics are used in agriculture (for example, on fruit trees). On average, roughly half of all antibiotics (the proportion varies with antibiotic class) consumed by humans and animals enter the sewage system or other environments via urine, and the amount of antibiotics released into the environment is presently on the order of several hundred thousand tons per year. Recent results have shown that the resulting environmental antibiotic concentrations may be important for both the emergence of new resistant strains and the enrichment of existing resistant strains. In competition experiments between susceptible and resistant strains, selection for resistant bacteria can occur at antibiotic concentrations even less than 1 percent of the MIC of the susceptible bacteria; similar antibiotic concentrations can be found in many natural environments.

These findings are important from a public-policy perspective because they suggest that antibiotic releases into the environment through human, veterinary, and agricultural applications contribute significantly to the emergence and persistence of antibiotic resistance. In particular, they indicate the potential benefits of reducing anthropogenically generated antibiotic pollution and avoiding treatment regimens that involve prolonged periods with low levels of antibiotics.

Coselection of Resistance Genes

A resistance gene located on a transmissible element or in a bacterial clone can increase in frequency in a population as a consequence of its genetic linkage to another resistance gene that is under selection. Such linkage and the resulting coselection is a common feature of resistance determinants that have been acquired by HGT, including plasmids, transposons, and integrons, and it can occur more generally in any multidrug-resistant clones. As a consequence, the frequency of resistance to a particular antibiotic can remain stable or even increase in environments where the antibiotic is not currently being used. The linked gene that sustains the unselected resistance gene can be any gene that increases the fitness of the bacterial strain, including another antibiotic resistance gene, a gene that encodes resistance to some heavy metal or disinfectant, or a gene that encodes some virulence-associated function.

Coselection is an important contributor to the long-term maintenance of resistance in bacterial populations, and it may explain why a reduction in the use of a particular antibiotic often has little or no effect on the frequency of resistant bacteria. For example, a recent study reported that an 85 percent reduction in the use of trimethoprim over a two-year period had only a very small effect on trimethoprim resistance in Escherichia coli. Similarly minor effects on resistance were recorded in other studies following reductions in use of sulfonamides, macrolides, and penicillin.

Cost-Free Resistance

The fitness cost of any particular resistance gene can vary depending on environmental conditions and the genetic background in which it occurs. For example, some resistance mutations impose no cost under standard laboratory conditions but have large costs in laboratory animals, and vice versa. Also, the cost of a resistance function often depends on the particular bacterial strain in which it occurs as a consequence of epistatic interactions between the resistance gene and other genes. Interestingly, some resistance genes do not appear to have any measurable fitness cost, at least in the environments and strains in which they have been tested. Of course, there may be other conditions under which these resistance genes do impose some costs, and measuring fitness costs under natural conditions is very difficult and rarely done; even in the laboratory, where genetically marked strains can be directly competed, it is difficult to measure fitness differences below about 0.3 percent per generation. It is also difficult to know what costs are relevant with respect to the persistence of an antibiotic resistance gene in a bacterial population. In principle, a fitness cost as small as 0.001 percent per generation would mean that a resistance gene would eventually be purged from the population by natural selection if the use of an antibiotic was stopped, although it might require many decades or even centuries given such small fitness costs.

Fitness-Enhancing Resistance

Although antibiotic resistance often has a fitness cost, in some cases it can actually be advantageous, even in drug-free environments. Interesting examples of such fitness-increasing effects of resistance functions have recently been demonstrated in two bacteria for the fluoroquinolone class of antibiotics. In E. coli, fluoroquinolone resistance commonly evolves by a multistep process involving mutations that alter efflux mechanisms and the proteins targeted by the drug. Each resistance mutation alone provides only a small increase in the MIC, so that clinically relevant levels of resistance require the accumulation of several mutations. In laboratory selection experiments, the accumulation of several resistance mutations typically led to reduced fitness in the absence of the antibiotic, but in a few cases an increase in resistance produced higher fitness. Campylobacter jejuni provides an interesting example of the background dependence of fitness effects associated with fluoroquinolone resistance. A single mutation in the gene encoding DNA gyrase enhanced the fitness of the resistant strain in a chicken-infection model, but when that same mutation was transferred into a different strain of C. jejuni, it imposed a fitness cost. The disturbing implication of these findings is that selection for improved growth may sometimes favor increased resistance even in the absence of drug selection.

Compensatory Evolution That Reduces Fitness Costs

Resistance to an antibiotic may impose a fitness cost because it disrupts the balanced growth of a bacterial cell that has been finely tuned to express genes and functions at levels that maximize fitness. A common process that stabilizes resistance is thus compensatory evolution, in which selection favors mutations that restore the cell’s balance and thereby reduce or eliminate the cost of resistance, often without any significant loss of resistance. Indeed, several laboratory and animal studies have demonstrated the evolution of mutations that restore fitness and, as a consequence, stabilize resistant populations. Whether adaptation in the absence of antibiotic occurs by compensatory mutations or by reversion (loss of resistance) will depend on several factors particular to any given case, including the mutation rates and fitness effects for compensatory and reversion mutations as well as population size. The genetic mechanisms of compensation vary depending on the particular drug and microbe involved. These mechanisms may include mutations in the resistance gene itself as well as mutations that alter the expression of the resistance gene or other genes in ways that restore the appropriately balanced gene expression.

Plasmid Persistence

Plasmids typically carry genes that are nonessential and beneficial only under specific environmental conditions. Hence, they are often expendable, and their persistence requires either ongoing selection (e.g., for resistance genes) or other mechanisms that assure their continued carriage. The various selective processes discussed earlier can promote the maintenance of both chromosomal and plasmid-encoded resistance functions; there are also several mechanisms that can promote plasmid persistence even without selection for antibiotic resistance. For example, some plasmids enhance bacterial growth even in the absence of antibiotic. Many plasmids encode resolution and partitioning systems that prevent spontaneous plasmid loss during cell division, and some plasmids even have toxin-antitoxin systems that kill cells that lose the plasmid. Also, plasmids can be maintained in bacterial populations by their conjugation-mediated horizontal transfer between cells even if they impose a fitness cost.

5. CAN RESISTANCE EVOLUTION BE SLOWED OR EVEN STOPPED?

A pressing question is whether society can reduce the rate at which antibiotic resistance emerges and spreads. Various approaches have been suggested in the literature, but only a few are known to work. One approach—perhaps the most obvious but still difficult to implement—is to reduce the use of antibiotics, thereby reducing the strength of selection that favors both the emergence and spread of resistant bacteria. The efficacy of this approach follows from basic evolutionary principles and is also supported by numerous studies showing that the frequency of resistance is correlated with the volume of antibiotics used at various levels including individual hospitals, communities, and countries. Global restraint in antibiotic use can be achieved only by concerted action and will require the implementation of several strategies, including (1) avoidance of antibiotic use when none is needed (e.g., when the infection is caused by a virus); (2) discontinuance of the use of antibiotics as growth promoters in animal husbandry; (3) discontinuance of the use of antibiotics in the production of crops and in aquaculture; (4) avoidance of economic situations in which the prescription of antibiotics is profitable for the prescriber; (5) appropriate control and regulation of antibiotic marketing by the pharmaceutical industry (in which prescribers, pharmacists, and consumers are targeted); and (6) prohibition of the sale of antibiotics to the public via the Internet or from pharmacies or other outlets without the need for a prescription.

Also, by increasing use of various hygienic and infection control measures, society can reduce the transmission of pathogenic bacteria and thereby reduce the use of antibiotics. The extent to which these measures will work depends on the pathogen and its mode of transmission among hosts. Pathogens for which hygienic measures have been shown to be particularly successful include various food-borne pathogens (e.g., Salmonella) and nosocomial infections such as methicillin-resistant Staphylococcus aureus (MRSA). For MRSA infections, screening strategies to track and isolate affected patients, coupled with improved hospital hygiene, have been successful in reducing the transmission of these dangerous bacteria.

Other approaches that have been proposed to reduce the rate at which resistance evolves include changes in dosing regimens and use of antibiotic combinations that reduce selection for resistant mutants without affecting treatment efficacy or safety. The use of drug combinations has been shown to be effective in treating many HIV (the virus that causes AIDS) infections because a mutant that becomes resistant to one drug is nonetheless susceptible to others that are provided at the same time. In addition, drugs and drug targets might be chosen during research and development such that the risk of resistance is minimized. For example, new antibiotics might be developed such that (1) resistance is difficult to acquire by mutation or HGT; (2) the resistance mechanism confers a high fitness cost; and (3) the opportunities for compensatory adaptation are limited. It is interesting to note that no clinical cases of resistance have been reported for certain combinations of drugs and bacteria even after decades of use. For example, penicillin has been used successfully to treat Streptococcus pyogenes infections for 60 years. An understanding of the reasons for the lack of resistance evolution might allow more rational choice and design of drugs and drug targets.

In addition to limiting the rates at which resistance emerges and spreads, it might even be possible to reverse the existing problem of resistance by reducing the use of antibiotics. Whether this strategy will be successful depends on the strength of the forces driving reversibility. At the levels of the individual and community, the fitness cost of resistance in the absence of antibiotic is probably the main force pushing toward increased sensitivity, whereas in hospitals the main driving force is probably the continuous influx of patients with susceptible bacteria. In hospitals, mathematical modeling and correlative studies suggest that changes in antibiotic use can cause rapid changes in the frequency of resistance. However, when the fitness cost of resistance drives reversibility, the rate of change is expected to be much slower. The main reasons for this are that in addition to the factors described earlier that can stabilize resistance in bacterial populations, the intrinsic dynamics of reversal is expected to be slow because the strength of selection for sensitivity in the absence of antibiotic is generally much weaker than selection for resistance when antibiotics are used. This inference is supported by clinical intervention studies, performed at both the individual and community levels, in which it has been observed that resistant clones are remarkably stable and persistent even when antibiotic use is reduced.

6. WILL ANTIBIOTICS BECOME A FOOTNOTE TO MEDICAL HISTORY?

How will future generations view our ongoing experiment with antibiotics? Will antibiotics retain their therapeutic value for generations to come? Or will antibiotics be viewed as a failed experiment, one that becomes a mere footnote in the history of medicine? The answers to these questions will depend on many factors, of which two challenges are of particular importance. The first is whether society—including medical practitioners, patients, and the pharmaceutical industry—will have the resolve to use antibiotics in a more restrictive and medically responsible way that will slow the emergence and spread of resistance. Success will require global implementation of changes in healthcare systems and practices that are specifically aimed at reducing the overall use of antibiotics that selectively favors resistant bacteria. Many international resolutions to this effect have been put forward, but so far little has been done to implement any global strategies. What is needed now is leadership and coordination that will allow these recommendations to be put into action. If we fail to implement these recommendations, it is certain that resistance will continue evolving to existing antibiotics as well as to any new ones that are discovered.

The second major challenge is that the pharmaceutical industry has largely abandoned the development of new antibiotics, mainly for economic reasons; as a consequence, few new classes of antibiotics have been introduced for clinical use in recent decades. It is essential that this industry be recommitted to antibiotic discovery and the development of novel drugs. Potential ways forward might include new business models for collaboration between industry and public sectors, including new regulatory rules and funding schemes. Of course, there are real scientific challenges in finding new drugs, including antimicrobials. However, increased knowledge of structural biology, bacterial physiology and metabolism, medicinal chemistry, genomics, and systems biology provide new opportunities for the discovery of novel antibiotics, including ones that might inhibit new targets such that the evolution of resistance is impeded.