Analyzing the relative fitness of mutations within two different species

by Delaney J.W. Ragsdale

Small mutations that confer a large increase in the relative fitness of an organism can shape the trajectory of its evolution. To investigate the relationship between natural selection and mutation, we performed two relative fitness experiments with Drosophila simulans and Escherichia coli. We witnessed phenotypic and molecular evolution occurring in a population of flies, influencing the increase in frequency of the red-eyed trait. The selection of the red-eye allele reduced the variation seen in the surrounding genes, a common pattern associated with selective sweeps. In our competition assays performed to observe the growth and biological success of rifampicin-resistant (rifR) Escherichia coli after UV exposure, we found that the relative fitness of rifR strains dramatically decreased during and after the DNA damaging event, as compared to the wildtype E. coli.

Key words: molecular evolution, relative fitness cost, antibiotic-resistance, competition assay


Evolutionary geneticists are faced with the challenge of describing the interactions among natural selection, mutation, random genetic drift, and gene flow, and how they shape the tremendous biodiversity found in every walk of life. By observing the collective action of these micro-evolutionary forces from within, and across, various populations, we can analyze their effects on genetic variation and study the consequential patterns of macro-evolution, like speciation and extinction. While gene flow and drift remain prominent aspects of study in population genetics, studies dissecting the interdependent forces of natural selection and mutation remain at the forefront of experimental and ecological genetics research. Insights leading to a deeper understanding of these tandem evolutionary processes could: 1) enhance our ability to asses allelic frequencies in a population undergoing molecular evolutionary change and 2) have profound applications in the medical industry by changing the way we treat infections and disease.

Natural selection plays a crucial role in both producing adaptation and conserving genetic states over long periods of time by acting on pre-existing mutational variation1. In this way, naturally selective pressures scrutinize the ‘fit’ between organism and environment by testing the constraints of the standing variation within a population. Although most genetic variation is neutral, some heritable mutations are responsible for increases in relative fitness, leaving an individual better suited to its environment and more likely to genetically contribute to future generations. These mutational variants become the raw materials needed for gradual adaptive evolution2. The evolutionary transition towards an advantageous trait begins at the molecular level, simply as a misplaced nucleotide, but eventually leads to phenotypic expression of the beneficial mutation at the level of the organism. The mutant genotype transcends the hierarchy of biological organization (i.e. nucleotides, genes, chromosomes, genome, cell, multicellular organism, societies) and manifests as an observable phenotype, thereby promoting the positive effect on its relative fitness and allowing natural selection to further fix the allele in the population.

While some adaptive mutations very clearly increase relative fitness, others that occur in regions encoding an essential component of biological function may actually induce a fitness deficit. While the adaptive genotype may always be present in the organism’s DNA, relative fitness is not solely defined by its success in one environment. Rather, the level of fitness remains context-dependent on the variable conditions applying selective pressure. Therefore, a mutation conferring a fitness advantage in one particular environment may actually be selected against under future environmental conditions2. One such example is the accumulation of resistance mutations in pathogenic bacteria. In general, drug resistance arises almost exclusively through the chromosomal mutation of genes encoding proteins targeted by specific antibiotics or by the enzyme required for prodrug activation3. Mutations that disrupt the targeted binding-domains or inactivate regions associated with drug toxicity are likely to confer resistance.

While these mutations weaken or even prevent the bactericidal action of the drug, they also tend to negatively impact important physiological functions that mediate cell survival. In other words, drug resistance incurs a fitness cost: in the absence of drug pressure, strains lacking resistance mutations are generally more competitive by being better able to maintain the bacterial life cycle in a variety of growth conditions3. Since antibiotics target essential cellular functions like DNA replication and cell wall biosynthesis, it is not surprising that resistance mutations in target-encoding genes will impact innate biological processes as well as pathogenesis.


In order to further examine the relationship between natural selection and the relative fitness of adaptive mutations, we subjected a Drosophila simulans parental population, consisting of ten white-eyed flies and one red-eyed fly, and its subsequent generations to a semester-long, observational experiment on evolution. Since the mutation responsible for the white eye color is associated with poor eyesight, the addition of a red-eyed male simulates the introduction of an advantageous mutation in the ancestrally white-eyed population4. We hypothesized that the number of flies with red eyes would gradually increase over many generations along with the associated decrease in proportion of white-eyed flies. If our hypothesis is correct, it would indicate that the red-eye phenotype confers a higher relative fitness and leads to favorable selection.

To gain a better understanding of the micro-evolutionary forces and their effects on genetic variation within a population, we asked the following questions: how quickly will the advantageous allele become fixed in the population? Are any other loci involved in natural selection in this sexually-reproductive organism? Are there any patterns associated with genetic variation following a selective sweep? How does recombination influence the rate of adaptive evolution?

In order to assess the potentially negative effects of adaptive mutations on the relative fitness of organisms across environments, Escherichia coli isolates containing mutations within antibiotic-resistance genes were subjected to stressful conditions via ultraviolet (UV) irradiation. Resistance-contrived fitness costs are easily demonstrated with the use of the antibiotic, rifampicin, and any number virulent bacterial pathogens, including Mycobacterium tuberculosis (TB), Staphylococcus aureus, Streptococcal pneumoniae, Neisseria meningitidis, Haemophilus influenzae, and many more3,5,6,7. Rifampicin had been used as a first line antibiotic against TB since 1967, until an increase in rifampicin-resistant (rifR) TB was observed across the world. Despite the growing problem, rifampicin remains a cornerstone of current short-term TB treatment. Even though E. coli is generally not pursued as a therapeutic target of rifampicin, it has been used as a model species for genetic and physiological studies that focus on characterizing the rifR mutations and investigating the function of the rifampicin target, RNA Polymerase (RNAP).

After our previous lab work with the Luria-Delbruck experiment eloquently demonstrated how environmental pressure acts on pre-existing genetic variation, we began to explore the underlying mutations that confer rifampicin resistance in our E. coli isolates. To determine if these variants would also induce a fitness deficit, we designed an experiment to examine the functionality of the rifR E. coli RNAP and its ability to successfully participate in a well-studied UV repair mechanism inherent to wildtype strains. We hypothesized that the rifR E. coli strains would exhibit lower relative fitness after irradiation in the absence of rifampicin, as compared to wildtype under the same conditions.


Drosophila simulans as a model organism for witnessing molecular evolution

In order to quickly screen organisms for your gene of interest, an easily identifiable phenotype is crucial and usually must be mediated through a reporter construct or genetic marker. Drosophila simulans, a sister species to the genetic model Drosophila melanogaster, naturally encodes one of the most recognizable phenotypes in all of genetics: red eyes. Its mutant phenotype, white eyes, is in such stark contrast to wildtype that we can confidently infer which allele is present with only a quick glance. Since the white allele, associated with poor eyesight, is detrimental for feeding and reproduction and the red allele is X-linked and inherited dominantly in either sex, the eye color trait seems to be the perfect vehicle of adaptation as we watch natural selection drive molecular evolution. This semester-long experiment will allow us to observe the change in allele frequency over multiple generations and characterize the direction of selection.

To initiate the parental population, we were provided with two vials: one containing five white-eyed females and the other containing five white-eyed males and one red-eyed male. We then combined the vials by carefully pouring the males into the vial with the females. After the emergence of the first and second generations, we were given a sample of these populations to separate by sex and eye color to see if the frequency of the red-eye variant changed over time. To do this, we exposed the Drosophila to light levels of CO2 to induce their sleep and prevent them from flying away while we quantify the variation under a dissection microscope. Once we finished categorizing the flies, we put them back into their vial in order to produce another generation of offspring.

DNA extraction, PCR, and gel visualization

Almost three months after starting the parental population, we received the third-generation descendants of our original flies. We quantified the allele frequency of the red-eye variant by separating the population by sex and eye color. Seven male flies were then selected for DNA extraction: 4 with red eyes and 3 with white eyes.

Just as the original parent population was polymorphic for the white gene, they were also polymorphic at many other loci across the genome, including two markers we call “Near” and “Far,” which are named for their relative distance to the eye color locus. Each marker is polymorphic for an indel that allows us to distinguish between them. The two indels encode alleles of different lengths and correspond to the “high” and “low” variants, referring to their relative position on a gel after performing electrophoresis. The variant that is shorter in length appears as the “high” band while the longer variant appears as the “low” band. The red-eyed male parent was homozygous for the high allele whereas the white-eyed parents were heterozygous, carrying any combination of high and low variants at the “Near” and “Far” loci. Based on genotype and phenotype information, variants found at the “Near” marker should be more visible to selection based on eye color. Any form of selection may have a profound impact on the variation and polymorphism of chromosomally-linked genes observed in the surrounding genome, e.g. a decrease in variation associated with a selective sweep. To better understand the interactions between micro-evolutionary forces, we analyzed the allelic frequency spectrum characterized by our PCR.

Escherichia coli as a study system to model rifampicin resistance

E. coli undergoes the Nucleotide Excision Repair mechanism to identify and restore detrimental UV damage. UV radiation causes thymine (TT) dimers in bacterial DNA, which bends one strand of the double helix at a ~30° angle, hindering the progression of polymerase. Unlike eukaryotic cells, E. coli cannot pause replication and transcription to fix DNA damages. When RNAP approaches during transcription, the lesion gets stuck in the active site and stalls the polymerase. To begin the pathway, Mfd binds to the β subunit of RNAP, forming a complex that allows for the release of RNAP. After Mfd dissociates, additional proteins within the complex protect the unstable bend from further damage until DNA Polymerase I and ligase can repair the damage fully8.

The antibiotic rifampicin functions by targeting the β subunit of RNAP and disrupting the active site that binds to the template strand of DNA5. Our objective was to investigate the nature of these rifR mutations and a potential subsequent fitness cost based on the dysfunction of the β subunit during rifampicin treatment as well as its role during UV repair.

Exposing E. coli to ultraviolet (UV) light

After first subjecting E. coli subcultures to a wide range of UV doses, we concluded that 1 mJ/cm2 was the appropriate dose to be used in all of our experiments. To determine if rifR E. coli experienced a disadvantage in UV-stressed environments, we performed competition assays comparing relative growth to a wildtype E. coli strain. Three rifR strains, known as rifA, rifB, and rifC, were used to determine if different mutations confer different growth effects. Throughout the experiment, we plated each trial on both DIFCO agar and DIFCO agar containing rifampicin to compare the number of CFUs per plate.

After mixing the initial 1:1 co-cultures (rifA:wt, rifB:wt, rifC:wt), we suspended the bacteria in water and then exposed the co-culture to 1 mJ/cm2 UV in a GS Gene Linker UV chamber. During irradiation, it is important to suspend the bacteria in water due to the wavelength filtering effects of DIFCO liquid media. We immediately plated the irradiated co-culture to compare the ability of each strain to survive UV exposure. To compare the ability of each strain to recover from the stressful environment, the remaining irradiated co-culture was resuspended in DIFCO then incubated overnight and plated ~24 hours after UV exposure.


Increased Frequency of Red Eyes in Drosophila

To further examine the relationship between natural selection and relative fitness of an advantageous trait, we calculated the allele frequency of red eyes in each generation of D. simulans by counting the number of individuals with the associated phenotype. One group did not observe any red-eyed flies in any of their generations, similar to the control population (Table 1). An analysis of their results will be omitted in this section, but will be reviewed in the discussion section below.

In the first generation, we saw a relatively low frequency of red eyes across the entire population (Table 1). The red-eye phenotype did not appear in any males, confirming its X-linked inheritance. In the second generation, we observed a large jump in frequency across each group with about two-thirds of the populations exhibiting the red-eye phenotype. One group observed more red-eyed males than red-eyed females (group 1), however all the other groups experienced almost 9 times more red-eyed females. The third generation had more variable results with the red-eye phenotype ranging from complete saturation to only 33% of the population. In both the groups reporting 100%, their population size consisted of only a single red-eyed female by the third generation. Another group had a total of three flies, only having a single red-eyed male. Our group (group 5) had the largest population with 86% of flies having red eyes. In our population, the red-eye phenotype appeared in twice as many females than males (Table 1).

After performing gel electrophoresis, we identified which genetic variants our sample of males had at both the “Near” and “Far” loci. We found that red-eyed males only had the high variant at the “Near” site while white-eyed males showed both variants (Fig. 1A). At the “Far” site, both red- and white-eyed males had both variants (Fig. 1B). This is relatively consistent with the data collected from the rest of the class (not shown).

Figure 1
Figure 1: Gel electrophoresis showing the DNA variants at the a) Near and b) Far loci. Four red-eyed males (R) and three white-eyed males (W) were analyzed. One arrow identifies the high variant and a second arrow identifies the low variant. The DNA ladder was added on either the left or right side of the DNA samples taken.

Table 1: Summary of the number of white-eyed or red-eyed males and females in each generation. The frequency of the total number of red-eyed flies in each population was calculated and the value is shown in the right column. The data collected from each generation corresponds to the seven rows below the generation title.  Each population corresponds to the group listed in the left column.

Lower ability to survive and recover from UV exposure in rifR strains

We determined the effect that UV exposure had on rifR E. coli by counting the number of colonies grown from the co-culture on either regular media (no antibiotic) or rifampicin (antibiotic) plates,. Overall, we found a massive decrease in ability for rifR E. coli to survive and recover from UV exposure as compared to wildtype (Fig. 2). While the ratio of rifR strains generally increased in the first trial, the second trial showed no ability for the resistant strains to recover.

In our first trial, rifA and rifB did not grow on rifampicin plates in our control group and these results may be inconsistent with the rest of our data. However, we observed a large decrease in the ratio of rifC compared to wildtype after UV exposure (Fig. 3A), suggesting a large fitness cost of rifR in rifC. The co-culture group that was not exposed to UV consisted of almost 8 times the amount of rifC compared to wildtype E. coli. Interestingly, rifC made up only a fourth of the bacteria that survived UV exposure. This fraction went up to about a third after being incubated in liquid media, showing a slight increase in its ability to recover.

In the second trial, the ratio of resistant bacteria to wildtype bacteria dropped slightly just after UV irradiation, then significant dropped after allowing growth in culture (Fig. 3B). RifA and rifC had very similar survival and recovery results which may suggest they contain the same rifampicin-resistant mutation. The results from rifC were slightly different in that there was almost no difference in ratio immediately after UV exposure, and a significant drop in ratio after allowing growth.

To further investigate the results seen in trial two, we calculated the competitive index (CI) of each rifR strain compared to wildtype (Fig. 4). Interestingly, the CI for rifB showed a slight increase immediately after being irradiated while both CI’s for rifA and rifB decreased significantly. Even more surprisingly, there was a 266-fold drop in the CI calculated for rifB just following UV exposure and the CI calculated for 24 hours after UV exposure. The same drop in rifA and rifC was only 22-fold and 35-fold, respectively. This suggests that the rifR mutation seen in rifB may not have an impact on the ability for the resistant bacteria to initially survive UV exposure, but significantly inhibits the long-term survival and ability to recover from the DNA-damaging event.

Figure 2
Figure 2: Graphs showing bacterial growth for the wildtype strain (blue) and rifampicin-resistant strain (orange) in both the a) first trial and b) second trial. The first time point shows CFUs/plate without UV exposure; second time point shows CFUs/plate immediately following UV exposure; and the third time point shows CFUs/plate after 24 hours of growth following UV exposure.
Figure 3
Figure 3: Ratio of rifampicin-resistant bacteria (orange) to wildtype E. coli (blue) in the co-culture, normalized to the relative population size of each. A) Results from first trial and B) results from second trial. Each set of three bars corresponds to the specific rifR strain in co-culture (rifA, rifB, or rifC). The first bar in each set represents the bacterial ratio without UV exposure; second bar in each set represents the bacterial ratio immediately following UV exposure; and the third bar in each set represents the bacterial ratio after 24 hours of growth following UV exposure.
Figure 4
Figure 4: Comparison of the Competitive Indexes (CI) of each rifampicin-resistant strain from the second trial. Horizontal black line with “x” represents the starting population. Green line denotes CI for the rifR strain immediately after UV exposure. Red line denotes CI for the rifR strain after 24 hours of growth following UV exposure. Vertical dashed line represents the fold decrease in the two CI’s of each rifR strain. CI graphed on a log10 scale.


The two experiments explore the interactions between micro-evolutionary forces, such as natural selection and mutation. We investigated how traits conferring either higher or lower relative fitness affected the evolutionary progression of two different organisms.

In our first fitness experiment, we assessed the rapid change of allelic frequency over the course of the semester using Drosophila simulans as a model. By adding a single red-eyed fly to a population of ancestrally white-eyed flies, we were able to map the phenotypic progression of the advantageous trait as well as determine the effects of natural selection on the surrounding genome. We predicted an increase in frequency of the red-eye allele over multiple generations which was relatively consistent with the results of our experiment, indicating a higher relative fitness of flies with red eyes. If random genetic drift is a significant determinant of directional evolution due to the small population size, we would expect to observe an increased proportion of white-eyed flies. If the relative proportion of allele frequencies remained unchanged, it would suggest that neither of them confer a fitness advantage under our experimental conditions.

Since sexually reproducing higher organisms undergo homologous recombination, natural selection that drives an increase in frequency of a beneficial mutation will inadvertently drag along neutral variants4, e.g. “high” and “low,” that reside on neighboring genes, represented by the “Near” and “Far” loci. The greater the distance between two loci found on a single molecule of DNA, the greater the opportunity for the loci to recombine. Similarly, the smaller the distance between two loci, the lower the chance of disassociation via recombination and the higher likelihood of them being observed together within a genome. These “hitchhikers” are said to be chromosomally linked and become associated with the spread of the variant being selected for. Following the selective sweep, i.e. fixation, of the red-eye allele, hitchhiking genes systematically reduce genetic variation in the surrounding regions despite their neutral fitness effect. This macro-evolutionary pattern is observed by the distinct presence of the “high” variant at all “Near” loci in red-eyed males. This pattern of linked genes demonstrates why some traits are commonly seen together in organisms.

On the other hand, alleles at genes further away from the original mutation will most likely fail to hitchhike due to a break in association via recombination, therefore maintaining higher levels of genetic variation4. This pattern is demonstrated by both red-eyed and white-eyed flies exhibiting both variants at the “Far” locus. However, it is important to note that two red-eyed males isolated by other groups contained the “low” variant at the “Near” locus. While natural selection is simultaneously driving adaptation and conserving genetic states through the spread of the beneficial mutation and hitchhiker genes, respectively, the disruptive forces associated with recombination attempt to restore the genetic variation back to the chromosome. The dynamic between these opposing forces mediates the progression of molecular evolution, similar to a system of checks and balances.

In our second fitness experiment, rifampicin-resistant E. coli were investigated for fitness effects relative to wildtype. By performing competition assays, we were able to address the differences observed in growth after being exposed to UV irradiation. We hypothesized that the rifR strains would have lower numbers of CFUs per plate, which was observed in our results. While each rifR strain had massively lower numbers than wildtype, rifB stood out to us. We predicted that rifA and rifC had the same mutation on the β subunit of their RNAP since their results were very similar, while rifB had a different mutation. After DNA sequencing of the rpoB gene that encodes the β subunit, we found that rifA and rifC had a S531F missense mutation and rifB had a D516G missense mutation. These two rpoB codons are frequently associated with rifampicin-resistance across many different species of bacteria5.

The S531 residue is located in the deep cleft of the rifampicin binding pocket and forms a hydrogen bond with the napthalene ring on rifampicin7. In a study investigating an S331L mutation, a major difference in the crystal structure was observed in the mutant β subunit. The electron density was weak and scattered which consequently left half of the rifampicin naphthalene ring exposed to solvent, reducing ~40% of the contact area between RNAP and the rifampicin molecule. This alteration dramatically reduces the van der Waals interactions resulting in a large loss of binding free energy. Collisions between rifampicin and the Leu side chain of the S531L mutant may also help in inhibiting the function of the antibiotic6. If this collision is seen with leucine, we predict that a similar collision involving the phenylalanine benzene ring in the S531F mutant could also prevent rifampicin binding.

In E. coli, the D516 residue in RNAP forms part of the sidewall of the rifampicin binding pocket and assists in positioning the rifampicin molecule to form two hydrogen bonds in the rifampicin-resistance determining region (RRDR) cluster I6.  In a study investigating a D516V mutation, there was no major structural change in RRDR cluster I in the mutant RNAP. However, the altered electrostatic distribution of the rifampicin-binding pocket may make it less favorable for binding the relatively apolar rifampicin molecule6. Even though valine is extremely hydrophobic, it is possible that a glycine in the D516 position could also alter the electrostatic charge enough to disrupt binding, conferring resistance.

These two missense mutations in the active site of RNAP could change the structure of the region just enough to prevent rifampicin binding, but not enough to inhibit normal function. However, when a lesion caused by UV damage enters this part of the active site, we predict that the damaged site gets stuck and prevents its removal by Mfd. Another possibility could be that these mutations prevent a conformational shift of RNAP caused by a different protein in the nucleotide excision repair pathway. However, future investigations of these distinct mutations are needed to confirm the structural basis of their rifampicin resistance. More biochemical studies need to be done on the rifampicin molecule.

A slight change in the chemical structure of rifampicin could suppress these functional mutations in RNAP and allow for the antibiotic to bind even in their presence. Rifampicin is a very important antibiotic used in M. tuberculosis treatment3,9, however, resistance is becoming increasingly frequent. If the structure of rifampicin could be improved even slightly to compensate for the steric hinderance caused by common resistance mutations, it could open the door for a whole new treatment option and prevent devastating resistance mutations from becoming fixed in entire strains of M. tuberculosis.

The results from both relative fitness studies enhance our understanding of the interplay among micro-evolutionary forces and its impact on macro-evolutionary patterns. In the first fitness study, we were able to focus our attention on how the DNA itself can promote molecular evolution by conferring fitness advantages under environmentally neutral conditions. Using D. simulans, we observed how the introduction of a single beneficial allele could propagate over multiple generations and eventually become fixed within a population. In the second fitness study, we took the naturally selective process a step further by applying an additional source of environmental stress. Using antibiotic-resistant E. coli, we were able to focus our attention on how differing environments can encourage evolution by directionally selecting for alleles that do not incur a relative fitness cost. These experimental approaches can be used to further our understanding of the evolutionary relationship between natural selection and mutation.

Literature Cited

  1. Wade, M. Evolutionary Genetics. The Stanford Encyclopedia of Philosophy. (2008).
  2. Michod, R. Darwinian Dynamics: Evolutionary Transitions in Fitness and Individuality. (2000).
  3. Koch, A., Mizrahi, V., & Warner, D.F. The impact of drug resistance on Mycobacterium tuberculosis physiology: what can we learn from rifampicin?. Emerging Microbes and Infections. 3:e17. (2014).
  4. Heil, C.S., Hunter, M., Noor, J., Miglia, K., Manzano-Winkler, B., McDermott, S., & Noor, M. Witnessing Phenotypic and Molecular Evolution in the Fruit Fly. Evolution.  5(4): 629–634. (2012).
  5. Goldstein, B. Resistance to rifampicin: a review. The Journal of Antibiotics. 67: 625–630. (2014).
  6. Molodtsov, V., Scharf, N., Stefan, M., Garcia, G., & Murakami, K. Structural basis for rifamycin resistance of bacterial RNA polymerase by the three most clinically important RpoB mutations found in Mycobacterium tuberculosis. Mol Microbiol. 103(6):1034–1045. (2017).
  7. Zaw, M., Emran, N., & Lin, Z. Mutations inside rifampicin-resistance determining region of rpoB gene associated with rifampicin-resistance in Mycobacterium tuberculosis. Journal of Infection and Public Health. 605-610. (2018).
  8. Selby C. Mfd Protein and Transcription-Repair Coupling in Escherichia coli. Photochemistry and Photobiology. 93(1): 280–295. (2017).
  9. Dookie, S., Sturm, W., & Moodley, P. Mechanisms of first-line antimicrobial resistance in multi-drug and extensively drug resistant strains of Mycobacterium tuberculosis in KwaZulu-Natal, South Africa. BMC Infectious Diseases. 16:609. (2014).


Delaney would like to thank Dr. Jonathan Eggenschwiler, Dr. Andrea Sweigart, Margot Popecki, and Shaugnessy McCann for guiding and providing expert advice on the research project. She would also like to thank Brittany Borzillo for helping with experiments as well as Dr. Lindsey Harding for the refinement of this piece. Finally, she would like to thank her professor and mentor, Dr. Vincent Starai, for his generosity and invaluable insights.

Citation style: Nature