The Evolution & Medicine Review

The resistance of numerous bacterial pathogens of major clinical significance to many or all relevant antibiotics has become a public health threat of such magnitude that the latest U.N. General Assembly decided to hold a one-day meeting to address the issue (1). Recently, there has been substantial coverage of this issue in the general media, including by major news organizations such as the New York Times, National Public Radio, and the Washington Post (2-4). What the stories running in these outlets often fail to adequately emphasize is that the bacterial pathogens develop resistance to antibiotics by an evolutionary process. Furthermore, the solution to the problem of effectively treating these microbial agents of human disease requires not merely producing more antibiotics but also gaining deeper insights into the evolutionary processes by which antibiotic resistance comes into being and is transmitted between bacteria. In this context, a new study (5) from the laboratory of Roy Kishony, who is affiliated with both Technion and Harvard, is a novel and welcome advance in ways to study the evolution of bacterial antibiotic resistance and potentially to communicate about this process to the broader public.

Baym et al. developed what they refer to as a microbial evolution and growth arena (MEGA)-plate. This 120 x 60 cm. acrylic dish contained several distinct regions of black agar containing progressively higher concentrations, by a factor of 10, of an antibiotic. In this study, either trimethoprim or ciprofloxacin were used as the antibiotics. Trimethoprim is an inhibitor of the enzyme dihydrofolate reductase, which reduces dihydrofolic acid to tetrahydrofolic acid (THF) (6). THF is a necessary intermediate in the synthesis of thymidine, which is a required nucleotide for DNA synthesis. Therefore, trimethoprim inhibits bacterial replication in sensitive strains. Ciprofloxacin inhibits topoisomerases II and IV (7). These gene products play important roles in bacterial DNA replication, transcription, and other genomic processes.

The zones on both ends had zero antibiotic and the central zone had an antibiotic concentration 10,000-fold that of the first antibiotic-containing zone from either end. On top was another layer of soft agar that permitted bacteria to migrate across the surface of the dark agar. In the present study, the MEGA-plate was seeded with motile E. coli in the zones containing no antibiotic.

This spatial arrangement thus permitted the investigators to watch and take time-lapse photographs as the motile E. coli used the available nutrients in one zone and spread until they encountered the boundary with the next zone, which contained a higher antibiotic concentration. Colonization of the next zone was possible only after strains acquired the genetic attributes necessary to tolerate the increased concentration of antibiotic. The authors tested how long the average lag period was between encountering and colonizing the next zone with higher antibiotic concentration as a function of the size of the jump from the current to the next area of the plate. In one experiment with trimethoprim, it took 44 hours before bacteria were able to move from the antibiotic-free zone into the first zone with three times the wild-type median inhibitory concentration of trimethoprim. In terms of maximizing average rate of evolution for the bacteria, there appeared to be an optimal range for the size of the antibiotic concentration gap between adjacent zones.

In addition, the authors were able to sample bacteria from various locations at multiple times post-inoculation. These isolates were then subjected to DNA sequencing to determine the extent and nature of mutations. As the bacterial swarms appeared as white on a dark background, with this experimental system, phylogenetic relationships of bacterial strains could be visualized on the MEGA-plate.

The isolates were classifiable into two groups: highly mutated (>60 single-nucleotide polymorphisms, insertions, or deletions; mutator phenotype) and minimally mutated (<12 single-nucleotide polymorphisms, insertions, or deletions; nonmutator phenotype). All of the isolates that were classified as highly mutated in the experiments involving trimethoprim possessed one of a few mutations in dnaQ (alternatively, mutD), which encodes DNA polymerase III, a protein with proofreading function that is important for DNA replication. As might be expected, the isolates with these dnaQ mutations did have increased rates of mutation relative to isolates without dnaQ mutations. However, these two groups of isolates did not adapt phenotypically at detectably different rates.

In the nonmutator isolates from the MEGA-plate experiments involving trimethoprim as the antibiotic, as was anticipated, the investigators determined that the most frequently mutated locus was folA, which encodes dihydrofolate reductase. Greater resistance was associated with more mutations at the locus. Of arguably greater interest, there were recurrent mutations at loci that do not encode proteins involved in folate biosynthesis but appear to contribute to resistance by less direct pathways than the mutations at folA.

Mutations at three loci not typically connected to trimethoprim resistance and encoding a phosphate transporter, a kinase, and a regulator of a pathway involved in resistance to a different antibiotic also occurred in multiple isolates. Deleting these genes in the ancestral bacterial strain demonstrated that these mutations did indeed contribute to resistance.

Resistance-inducing mutations were frequently but not always associated with decreased rates of bacterial proliferation. These effects on replication were in some instances countered by further compensatory mutations.

In the experiments with ciprofloxacin, it was found that the isolates with both resistance-eliciting mutations and mutations that restored “normal” proliferative rate were nevertheless frequently not found at the advancing front of the bacterial population. When spatially confined to regions behind the bacteria leading the extension into new territory, they were not able to have a major impact on the continuing evolutionary trajectory of the population.

The authors interpret the preceding findings to mean that fitness alone, as measured by the magnitude of antibiotic resistance, is not the sole determinant of ultimate evolutionary dominance. Perhaps another way to look at these results, if an isolate’s position is viewed as determined by the extent of motility and not by the randomness of motility, is that fitness depends on more than the extent of antibiotic resistance. From this perspective, on the MEGA-plate, fitness is a composite variable including both resistance level and the degree of motility. The authors do not address the possibility the variation in the motility of different lineages contributes to evolutionary trajectory.

Kishony and colleagues do note at the end of the article that their new experimental system permits exploration of different selection pressures in a way that can be readily visualized. They expect these attributes of their experimental system to facilitate the formulation and testing of hypotheses pertaining to evolutionary mechanisms and pathways.

MEGA-plate experiments could be used to study the influence on evolutionary dynamics of variation for many different variables: 1) using different bacterial species, 2) plates of different sizes or layouts, 3) sizes of antibiotic concentration steps from zone to zone (already preliminarily studied in this publication), 4) starting genotype of bacteria, 5) starting gene expression pattern of bacteria, 6) presence of competing strains of the same or other species, 7) varying amounts or types of nutrients, and 8) the presence or absence of host mechanisms for immunity to bacteria, such as antibodies and complement or phagocytes. This system could also be used to study the simultaneous evolution of resistance to two different antibiotics. In the future, MEGA-plate experiments thus offer promise to illuminate many aspects of the bacterial evolution of antibiotic resistance.

A reasonable hypothesis is that such studies will demonstrate that accumulation of multiple antibiotic resistance mechanisms will necessarily impose trade-offs even if there are compensatory mutations that appear to restore growth rate to “normal” levels. I am suggesting that while there are universal fitness-reducing mutations that undermine one or more crucial cellular functions, there are unlikely to be universal fitness-enhancing mutations that improve fitness under all environments. MEGA-plate experiments under enough different conditions might provide evidence relevant for assessing this hypothesis.


“*” used to replace “u” so as not to add searchable instances of a term for bacteria resistant to many antibiotics that I regard as misleading and preferably eliminated from news coverage.

  1. S*perb*g explosion triggers U.N. General Assembly meeting.*perb*g-explosion-triggers-u-n-general-assembly-meeting/. September 7, 2016.
  1. Tavernise, S. and Grady, D. Infection raises the specter of s*perb*gs resistant to all antibiotics. May 26, 2016.*perb*gs-resistant-to-all-antibiotics.html?ref=topics. (last accessed on 9/27/16)
  1. On Point. Antibiotic resistant s*perb*g arrives in America.*perb*g. May 31, 2016. (last accessed on 9/27/16)
  1. Emanuel, E. Want to win $2 billion? Create the next antibiotic. Washington Post, May 30, 2016.*perb*gs/2016/05/30/38e94370-25a9-11e6-ae4a-3cdd5fe74204_story.html?tid=a_inl&utm_term=.f0de21a95c5d. (last accessed on 9/27/16)
  1. Baym M, Lieberman TD, Kelsic ED, Chait R, Gross R, Yelin I, Kishony R.Spatiotemporal microbial evolution on antibiotic landscapes. Science. 2016 Sep9;353(6304):1147-51. doi: 10.1126/science.aag0822. PubMed PMID: 27609891.

6. Trimethoprim. (last accessed on 9/27/16)

7. Ciprofloxacin. (last accessed on 9/27/16)