This protocol describes how to maintain the Escherichia coli Long-Term Evolution Experiment (LTEE) by performing its daily transfers and periodic freeze-downs and how to conduct competition assays to measure fitness improvements in evolved bacteria. These procedures can serve as a template for researchers starting their own microbial evolution experiments.
The Long-Term Evolution Experiment (LTEE) has followed twelve populations of Escherichia coli as they have adapted to a simple laboratory environment for more than 35 years and 77,000 bacterial generations. The setup and procedures used in the LTEE epitomize reliable and reproducible methods for studying microbial evolution. In this protocol, we first describe how the LTEE populations are transferred to fresh medium and cultured each day. Then, we describe how the LTEE populations are regularly checked for possible signs of contamination and archived to provide a permanent frozen "fossil record" for later study. Multiple safeguards included in these procedures are designed to prevent contamination, detect various problems when they occur, and recover from disruptions without appreciably setting back the progress of the experiment. One way that the overall tempo and character of evolutionary changes are monitored in the LTEE is by measuring the competitive fitness of populations and strains from the experiment. We describe how co-culture competition assays are conducted and provide both a spreadsheet and an R package (fitnessR) for calculating relative fitness from the results. Over the course of the LTEE, the behaviors of some populations have changed in interesting ways, and new technologies like whole-genome sequencing have provided additional avenues for investigating how the populations have evolved. We end by discussing how the original LTEE procedures have been updated to accommodate or take advantage of these changes. This protocol will be useful for researchers who use the LTEE as a model system for studying connections between evolution and genetics, molecular biology, systems biology, and ecology. More broadly, the LTEE provides a tried-and-true template for those who are beginning their own evolution experiments with new microbes, environments, and questions.
In February of 1988, Richard Lenski inoculated twelve flasks containing a defined glucose-limited growth medium with clonal cultures of Escherichia coli at the University of California, Irvine1. The following day, he transferred 1% of the culture from each flask to a set of new flasks containing fresh growth medium. This 1:100 dilution allowed the bacterial populations to expand 100-fold before exhausting the available glucose, corresponding to roughly 6â…” generations of cell divisions. This procedure was repeated the following day and has been every day since, with a few interruptions. These daily transfers have continued, even as the experiment was relocated, first to Michigan State University in 1992, and then to The University of Texas at Austin in 2022. All the while, new mutations have continuously generated genetic variation in these E. coli populations and natural selection has led to evolved cells outcompeting their ancestors.
Lenski designed this experiment, now known as the Long-Term Evolution Experiment (LTEE), to investigate the dynamics and repeatability of evolution. In order to answer these questions, he included several important features in the design of the experimental setup and its protocols2. One of these features was the careful choice of a model organism. The original twelve populations were all started from single colonies that shared an immediate common ancestor, Escherichia coli B strain REL606. This strain was chosen because it had already been commonly used in lab settings, reproduced completely asexually, and contained no plasmids or intact prophages3,4 — all of which make studying its evolution simpler. Another choice that simplified the experiment was to use a very low concentration of glucose in the growth medium to limit the density of cells in each flask after growth. Using a low cell density was intended to make it easier to analyze changes in population fitness by reducing the potential for the evolution of ecological interactions within populations (e.g., by cross-feeding)5.
REL606 is unable to use ʟ-arabinose as a carbon and energy source (Ara−) due to a point mutation in the araA gene. Prior to starting the LTEE, a spontaneous mutant with a restored araA sequence, designated REL607, was isolated from REL6066. REL607 is capable of growing on ʟ-arabinose (Ara+). REL606 was used to start six of the LTEE populations, and REL607 was used to start the other six. Arabinose is not present in the growth medium used during the LTEE, so REL607 behaves the same as REL606 under these conditions. However, when plated on tetrazolium arabinose (TA) agar, Ara− and Ara+ cells form red and white colonies, respectively. This method for discriminating between the two ancestral E. coli strains and their descendants is quite useful. It can be used to detect cross-contamination between LTEE populations. It also aids in measuring the fitness of an Ara− strain or population relative to an Ara+ one when they are competed against one another. Fitness is measured by setting up a co-culture of oppositely marked competitors and then monitoring how the frequencies of red and white colonies (obtained by spreading dilutions of the culture on TA plates) change between when the competitors are initially mixed and after one or more growth cycles under the same conditions as the LTEE. The representation of the more-fit cell type will increase during each growth cycle.
Another critical feature of the LTEE is that samples of the evolving populations are periodically archived. When mixed with a cryoprotectant such as glycerol, E. coli cells can be frozen and later revived7. As part of the LTEE protocol, every 75th day (which equates to roughly 500 generations), a portion of each population that was not transferred to a new flask is mixed with glycerol, split up between multiple vials, and stored in a freezer. This frozen "fossil record" enabled researchers to perform the first studies of the LTEE, in which they revived the evolved E. coli populations from various timepoints and competed them versus the ancestral strains to track how rapidly fitness was increasing1. Fitness evolution has been remeasured periodically as more "strata" of the frozen "fossil record" have been preserved. The overall conclusion from these measurements is that fitness continues to improve in the LTEE to this day, even after so many generations of evolution in the same environment8,9,10.
What has allowed the LTEE to continue for so long? Many of the same features that enabled its original questions to be asked and answered have also served as safety measures and fail-safes against inevitable disruptions due to bad luck, human error, and world events. Every day, when the cultures are transferred to fresh growth medium, the researcher performing the transfers alternates between Ara− and Ara+ populations. Then, when the populations are frozen, they can be plated on selective and indicator agar to check if any "neighboring" populations have been accidentally cross-contaminated or mixed up (e.g., white colonies are in a population that should only form red colonies) or contaminated with foreign microbes (e.g., unexpected colony morphologies or cell densities). In the event that a population has been compromised, its progenitor can be revived from the freezer and carried forward in its place. The Ara markers and the frozen archive thus serve dual purposes as both experimental resources and safety measures.
Because its history is so well preserved and easily accessible, LTEE samples have been studied using technologies that did not exist when the experiment began. For example, whole-genome sequencing has been used to examine the dynamics of mutations in the LTEE populations11,12,13,14,15, and transcriptomics and ribosomal profiling have been used to examine changes in gene expression16,17. Genetic tools have been used to reconstruct strains that differ by single mutations or combinations of several evolved mutations to understand their effects on fitness and various phenotypes18,19,20,21. Samples from the frozen "fossil record" are easily replenished so that parts of or entire copies of the experiment's history can be shipped to other laboratories. LTEE samples now exist on all continents except Antarctica, and they are being studied by researchers who are younger than the experiment itself. The robust methods of the LTEE and evolved E. coli samples and strains from its historical record have also served as starting points for evolution experiments examining other questions and environments22,23,24,25,26,27,28,29.
Figure 1: Overview of LTEE procedures. Please click here to view a larger version of this figure.
Here, we demonstrate three core protocols used in the E. coli Long-Term Evolution Experiment (Figure 1). We describe: (1) how to perform the daily transfers, (2) how to archive population samples and clonal isolates, and (3) how to perform and analyze co-culture competition assays to measure fitness differences. Our hope is that these protocols foster the continued use of LTEE resources and inform the design of new microbial evolution experiments.
1. Daily transfers of LTEE populations
NOTE: The twelve LTEE populations are transferred daily by inoculating fresh medium with 1% of the cultures from the prior day's flasks. The steps in this process are summarized in Figure 1. The six Ara− populations started from strain REL606 are designated A−1 to A−6, and the six Ara+ populations started from strain REL607 are designated A+1 to A+6. Strict adherence to aseptic technique and to a schedule and order for transferring the populations minimizes the risk of contamination and other disruptions.
2. Archiving the LTEE populations
NOTE: Samples of the LTEE populations are frozen every 75 transfers. The populations grow ~6 â…” generations each day following the 100-fold transfer dilution, so this period corresponds to ~500 generations. During archiving, the LTEE populations are also plated on different types of agar media to check for contamination. Optionally, representative clones can be picked from these plates and archived at this time. These steps are summarized in Figure 1.
3. Competitive fitness assays
NOTE: In the LTEE, reproductive fitness is quantified in terms of the relative number of doublings that different bacteria achieve over one or more 24 h culture cycles under the same conditions as the daily transfers. Specifically, the relative fitness of one competitor to another is the ratio of their realized doubling rates when they compete head-to-head in a co-culture. Each competitor in a pair may be a full population or a clonal isolate that was previously archived as part of the frozen "fossil record" of the LTEE. Alternatively, one or both competitors may be a clone that has been genetically modified to add or remove specific mutations to test their effects. The two competitors must have opposite Ara+/Ara− states because this genetic marker is used to differentiate them during this assay. The overall workflow for a competition assay is shown in Figure 2. The duration of the co-culturing phase can be extended from one to three (or more) days to improve the precision of fitness estimates when testing for differences between competitors that are nearly evenly matched. See the Disussion for other critical considerations and possible modifications of this protocol.
Figure 2: Competition assay flowchart. The full procedure for a one-day competition assay is shown. The three-day procedure continues with the alternative pathway on Day 1 and Day 2 until plating on Day 3 in the same way that is pictured for Day 1 of the one-day competition. Please click here to view a larger version of this figure.
Appearance and turbidity of LTEE cultures
Due to the low glucose concentration in DM25, the turbidity of fully grown LTEE populations is only barely visible in eleven of the twelve flasks. When examining the LTEE cultures by eye for normal growth and signs of contamination (step 1.6), each flask containing an LTEE population should be compared side-by-side to the blank (Figure 3A). The exception is population A−3, which evolved to use citrate as an additional carbon and energy source and therefore reaches a higher cell density32. The turbidity of DM25 cultures of the REL606 and REL607 ancestor strains is similar to that of a typical evolved population (Figure 3B). LTEE strains and populations grow to a higher density in DM1000 owing to the higher concentration of glucose, and a much higher density in LB (Figure 3B). The density of DM25 cultures of the A−3 LTEE population is intermediate between the densities of cultures of REL606 in DM25 and DM1000 (Figure 3C).
Figure 3: Appearance of LTEE cultures. (A) Flasks containing the twelve LTEE populations after 24 h of growth in DM25 on the day when the experiment reached 76,253 ⅓ generations are pictured alongside the blank. (B) Flasks containing cultures of the REL606 and REL607 ancestors grown for 24 h in DM25, DM1000, and LB are pictured alongside media blanks. (C) Zoomed in pictures of the same flasks side-by-side showing how the turbidity of the A−3 population flask in DM25 compares to the REL606 ancestor in DM25 and DM1000. Please click here to view a larger version of this figure.
Spectrophotometer readings of the optical densities at 600 nm (OD600) of cultures grown in DM25 (step 1.7) match these visual observations for both the LTEE populations (Figure 4A) and their ancestors (Figure 4B). These readings can be used to quantitatively compare and document growth when contamination or a mistake is suspected. For measurements of the LTEE populations between 76,000 and 76,500 generations, we found that the OD600 of A−3, the population that evolved to grow on citrate, was 0.223 on average (0.218-0.227, 95% confidence interval). The OD600 of the other eleven populations was 0.0252 on average (0.0239-0.0265, 95% confidence interval). There was slight, but significant variation in OD600 readings among the eleven normal populations (F10,88 = 5.1035, p = 7.5×10−6). The LTEE populations reach stationary phase after roughly 5-6 hours of incubation. If they are transferred in the morning, growth will be visible by mid- to late afternoon of the same day. Many species of microbes are able to grow aerobically on citrate. Therefore, increased turbidity in populations other than A−3 is likely a sign of outside contamination.
Figure 4: Turbidity of LTEE cultures. (A) Optical density at 600 nm (OD600) of the twelve LTEE populations after the 24 h growth cycle on three different days between 76,000 and 76,500 generations of the experiment. The OD600 values of three 1-mL aliquots on each of the three different days are plotted as points. The mean OD600 value of three different aliquots of the blank from the same day was subtracted from these values. Filled bars show averages. Error bars are 95% confidence limits. (B) OD600 of cultures of the REL606 and REL607 ancestors in DM25, DM1000, and LB. The OD600 values of three 1-mL aliquots on each of three different days of two separate cultures for each condition and strain are plotted as points. The mean OD600 value of three different aliquots of the blank from the same day was subtracted from these values. Filled bars show averages and error bars are 95% confidence limits. Grey shaded areas between the panels show how the OD600 axis is rescaled between the DM25 panel and the DM1000 and LB panels. Please click here to view a larger version of this figure.
Growth and morphology of LTEE colonies
When checking the populations for contamination by plating them on different media (step 2.15), the REL606 and REL607 ancestors and all evolved populations form white colonies with translucent and somewhat irregular edges on minimal glucose (MG) agar plates (Figure 5A). The composition of MG agar is the same as that of the DM25 used in the daily LTEE transfers, except with a higher concentration of glucose, so the evolved LTEE populations often form larger colonies on MG than the ancestors. Owing to its higher cell density in DM25, the A−3 population will have several-fold more colonies if the same volume is plated for it as for the other populations, and this may limit the size of the colonies. The most common types of contaminating microbes form stark white, opaque, and perfectly circular colonies on MG.
On minimal arabinose (MA) agar, the REL607 ancestor and the Ara+ populations typically all form slightly translucent white colonies. This typical growth pattern has persisted for the Ara+ populations through 76,000 generations, except A+6, which has evolved a defect in growth on arabinose and no longer forms colonies on MA (Figure 5B). There is no selection to maintain growth on arabinose during the LTEE transfers in DM25, so other Ara+ populations may also eventually stop forming colonies on MA agar plates as the experiment continues. With the exception of A−3, the Ara− populations do not form colonies on MA agar, though close examination may reveal microcolonies owing to trace nutrients in the agar. The A−3 population forms numerous small colonies on MA, as these cells can grow on the citrate that is also present in this medium. Contaminant colonies on MA are rare.
Figure 5: Plating LTEE populations to detect contamination. Dilutions of the REL606 and REL607 ancestors and the twelve LTEE populations on the day when the experiment reached 76,026 â…” generations were plated on (A) MG, (B) MA, and (C) TA agar plates and photographed after 24 h and 48 h. The same dilutions were made for all cultures, but half as much volume was plated for the ancestors as is described in the protocol for the LTEE populations, to account somewhat for their higher cell densities. Please click here to view a larger version of this figure.
On tetrazolium arabinose (TA) agar, the REL606 ancestor and all Ara− populations are expected to form red colonies, while the REL607 ancestor and all Ara+ populations should generally form colonies that are white (which can include light pink or peach shades) (Figure 5C). The LTEE ancestors form robust colonies, which are easily identifiable as Ara− and Ara+ on TA agar within 16-24 hours. Originally, this difference could be used to detect cross-contamination between Ara− and Ara+ populations. However, TA agar has a more complex nutrient composition than the chemically defined DM25 medium used in the daily transfers, and there has not been an evolutionary pressure for E. coli in the LTEE to maintain an ability to robustly grow under these conditions. Consequently, some evolved LTEE populations now exhibit poor growth on TA plates, taking 48 hours to form colonies or not reliably growing at all. The colors and morphologies of colonies formed on TA by the evolved LTEE populations have also changed relative to the ancestors and diverged from one another. The presence of a few aberrant colonies is not always an indication of contamination. Spontaneous mutations can occur that switch the Ara marker state of LTEE strains, especially from Ara+ to Ara− due to the higher likelihood of loss-of-function mutations affecting arabinose utilization versus reversion mutations that restore araA activity. Mutations switching Ara marker states are more common in populations that have evolved hypermutation (A−1, A−2, A−3, A−4, A+3, and A+6)13. On TA agar, contaminating microbes of other species often (but not always) form small, perfectly circular colonies with red centers ringed by distinct white boundaries that are unlike those formed by any LTEE strains or populations.
Co-culture competition results
Competitions between all Ara− and Ara+ pairs of the two LTEE ancestors (REL606 and REL607, respectively) and the A−5 and A+5 population samples archived at 20,000 generations (REL8597 and REL8604, respectively) show how colonies with different Ara marker states can be differentiated and counted on TA agar (steps 3.4.8 and 3.6.6) (Figure 6). Colonies were counted for six replicate flasks for each pair of competitors before and after one-day and three-day assays that began with revival in DM1000 (Table 1). The total numbers of colonies observed for the same dilution and volume plated vary with which competitors were mixed because cultures of evolved LTEE populations reach lower cell densities than cultures of the ancestor strains in DM25. This difference is a consequence of the evolution of increased cell size, which occurred in all LTEE populations during the first few thousand generations of the experiment8, 33.
Figure 6: Competition assays plated on TA agar plates. Examples of TA agar plates from competition assays. REL606 and REL607 are the Ara− and Ara+ ancestors of the LTEE, respectively. REL8597 and REL8604 are the 20,000 generation A−5 and A+5 populations, respectively, from the frozen "fossil record" of the LTEE. TA plates corresponding to one replicate assay between each pair of strains are shown for Day 0, Day 1, and Day 3 of the competition. Plates were photographed after 24 h of growth at 37°C. Cells of the REL606 and REL8597 competitors are Ara− and form red colonies. Cells of the REL607 and REL8604 competitors are Ara+ and form white colonies. Please click here to view a larger version of this figure.
Most colonies on a typical competition TA plate will be well-separated or overlap in ways for which it is easy to count how many initially circular colonies of different types grew together (Figure 7A). However, some situations may arise in which it is not obvious how to count an atypical colony or growth that is a mixture of the two colors. First, when a white Ara+ colony and a red Ara− colony overlap, the Ara+ colony tends to overgrow and envelope the Ara− colony. In this situation, one should count a small red patch or translucent "gap" in the larger Ara+ colony as an Ara− colony (Figure 7B). Second, spontaneous Ara+ mutants will occasionally arise in Ara− colonies. These mutants typically appear as white sectors (papillae) spreading more quickly out of the interior of a red colony because they grow more quickly once they gain access to arabinose as an additional nutrient (Figure 7C). These white-sectored colonies are counted as one Ara− colony and no Ara+ colonies. This situation becomes more common if plates are incubated for 48 h or longer. Third, sometimes translucent pinkish colonies are observed (Figure 7D). These are formed by the Ara- competitor. Finally, a small number of circular colonies with interiors that are a slightly different shade of red sometimes grow on TA plates when they are contaminated by a few outside microbial cells during preparation of the agar or when spreading culture dilutions on their surfaces (Figure 7E). These contaminant colonies should not be counted. If contamination of a competition culture is suspected because there are many atypical colonies on any of its TA plates, that replicate should be excluded.
Figure 7: Edge cases encountered when counting Ara− and Ara+ colonies on TA agar. In each panel, some Ara− and Ara+ colonies that should be counted are marked with solid red and black arrows, respectively. Colonies that should not be counted are indicated with dashed arrows corresponding to the type that they appear to be. All photos were taken after 24 h of incubation except in panel C. (A) Examples of normal Ara− and Ara+ colonies. (B) Examples of Ara+ colonies overgrowing nearby Ara− colonies, including one that is only barely visible as a transparent gap in the outside of the white colony. Count each of these cases as two colonies, one of each type. (C) Examples of Ara− colonies giving rise to Ara+ mutant sectors. Count each case as only a single Ara− colony. The white sector (papilla) that arises is due to an Ara+ mutant arising within the colony. The same field of colonies is shown following 24 h, 48 h, and 72 h of growth. (D) Example of a translucent pink colony. Count it as Ara−. (E) Examples of colonies formed by outside contamination by a microbe that is not E. coli. These are red but smaller and perfectly circular with a distinct white boundary. Please click here to view a larger version of this figure.
Analyzing the colony counts from these competitions using the Excel spreadsheet (Supplemental File 1) or by running the fitnessR package functions in R on colony counts entered into the CSV template (Supplemental File 2) shows that the two ancestors are indistinguishable in terms of their fitness within the precision of the assay, that both the 20,000-generation A−5 and A+5 populations are significantly more fit than the ancestors, and that neither evolved population is significantly more fit than the other (Welch's t-tests, p > 0.05) (Figure 8). The precision of the relative fitness estimate improves in the three-day competitions versus the one-day competitions for one of the closely matched pairs (REL606 vs. REL607). The precision of these measurements could be increased further by conducting longer competitions with more growth cycles, if so desired. However, the results from multi-day competitions are not informative once one competitor becomes so abundant relative to the other after the additional days of competition that the ratio of the two strains cannot be accurately determined because there are very few to no colonies of the less-fit type to count. This is the case for the three-day competitions of the ancestors against the evolved 20,000 generation populations (REL606 vs. REL8604 and REL607 vs. REL8597) (Figure 6 and Table 1).
Table 1: Colony counts from competitive fitness assays. One-day and three-day competition assays with six replicates were performed for all pairwise combinations of two Ara− and the two Ara+ competitors. REL606 and REL607 are the Ara− and Ara+ ancestors of the LTEE, respectively. REL8597 and REL8604 are the 20,000-generation A−5 and A+5 populations, respectively, from the frozen "fossil record" of the LTEE. Please click here to download this Table.
Figure 8: Relative fitness measured using competition assays. Results of one- and three-day competition assays between LTEE ancestors and the 20,000 generation A−5 and A+5 LTEE populations. The diagram on the left shows the four pairwise competitions as color-coded double-headed arrows. Each combination of the two Ara− (red labels) and the two Ara+ (black labels) competitors was tested with six-fold replication. Colony counts from Table 1 were analyzed in R using the fitnessR package31, and the results were plotted using the ggplot2 package (version 3.4.0)34. Fitness is displayed as the competitor the arrow in the label is going toward relative to the competitor the arrow is coming from (e.g., REL8604 relative to REL606). Relative fitness values estimated from the colony counts for each competition assay replicate (points), mean relative fitness values for the pair of competitors (bars filled with the same color-coding as the diagram), and 95% confidence intervals (error bars) are shown. Relative fitness values could not be determined (N.D.) for the three-day competitions between the ancestors and the evolved populations because there were zero or very few colonies of the ancestors on the Day 3 plates (see Table 1). Please click here to view a larger version of this figure.
Supplemental File 1. Excel spreadsheet file for calculating relative fitness. Please click here to download this File.
Supplemental File 2. Comma-separated values input file template for calculating relative fitness in R using the fitnessR package. Please click here to download this File.
Long-term resilience of the LTEE and its methods
The E. coli Long-Term Evolution Experiment (LTEE) is now in its fourth decade. For a microbial evolution experiment of any duration, it is critical to maintain a reproducible environment, avoid contamination, archive samples, and accurately measure fitness. The LTEE demonstrates several time-tested strategies for achieving these objectives, including the use of well-shaken flasks that create a homogenous environment and a chemically defined growth medium that supports a low cell density. Moreover, the LTEE employs ancestor strains that differ in a genetic marker that gives a phenotype (colony color) that is both easily screened and selectively neutral in the evolution environment. This experimental design feature provides a means of identifying internal and external contamination and facilitates measuring fitness. However, not all of the procedures and safeguards that have been used by the LTEE since 1988 have proven equally robust. Some methods that were reliable when the LTEE began have become less effective as the E. coli populations have evolved. Fortunately, these problematic methods can now be augmented or replaced using technologies developed since the experiment's inception.
Detecting contamination
Detection of contamination is critical to the LTEE. Contamination can be of two sorts: between LTEE populations (cross-contamination) and with microbes from the environment (outside contamination). For the most part, careful use of aseptic techniques and close attention during media preparation and the daily transfers prevent both types of contamination, but they do happen. Early in the experiment, plating on TA agar could be used to detect instances of cross-contamination because transfers have always alternated between Ara− and Ara+ populations. The fingerprint of sensitivity and resistance of these E. coli to certain bacteriophages was also intended to be a design feature that could differentiate the LTEE populations from commonly used E. coli laboratory strains that might contaminate them4. However, these genetic markers have become unreliable as the experiment has progressed (e.g., some populations no longer form colonies on TA agar)10,35. Fortunately, the populations have genetically diverged as they have experienced separate evolutionary histories during the experiment, which has created new genetic markers that can now be used to detect cross-contamination. For instance, each population has evolved a unique combination of mutations in the pykF and nadR genes14,36,37. We sometimes PCR amplify and Sanger sequence these two genes to test whether colonies with unusual morphologies or colors are due to cross-contamination. As the costs of whole-genome and whole-population sequencing continue to drop, routine sequencing of the LTEE populations may soon be possible, thereby presenting new opportunities to monitor them for signs of contamination.
Measuring competitive fitness
Another case in which the LTEE has outgrown its original methods is that the fitness of the evolved E. coli has increased in the experimental environment to such a degree that one can no longer directly measure the fitness of the populations of today relative to their ancestors using the protocol described here. The evolved populations outcompete the ancestors to such an extent that few to no ancestor colonies remain to count after a one-day competition. One approach for dealing with this large fitness difference is to use unequal starting ratios of the strains, weighting the initial volumes that are mixed toward the less-fit competitor (e.g., 90 µL ancestor and 10 µL evolved competitor). A second approach is to identify an evolved Ara− clone that has a higher fitness than the LTEE ancestor, isolate a spontaneous Ara+ revertant mutant of it by selection on MA agar, and then verify that the revertant strain has the same fitness as its parent using a competition assay6,38. This new Ara−/Ara+ pair can then be used as a set of common competitor strains in lieu of REL606/REL607. Ideally, the evolved Ara− clone chosen as a common competitor (and its Ara+ revertant) will have intermediate fitness relative to all strains of interest in an experiment. Over the first 50,000 generations of the LTEE, these two approaches (using unequal starting ratios or a common competitor) did not produce meaningfully different fitness measurements versus the typical approach39.
These modifications to the competition protocol make certain simplifying assumptions that may not always be true. One is that fitness measurements are transitive. That is, if we compete two populations each versus a common competitor strain separately, then we can infer the relative fitness of the two populations to one another. This relationship has been found to be true for the LTEE40, for the most part, but it is not for other experiments41. One reason for this discrepancy can be the evolution of negative frequency-dependent fitness effects. This situation occurs when strains isolated from two different diverged lineages from population A−2 of the LTEE are competed against one another19,42. Each has an advantage when rare, due to cross-feeding, which stabilizes their co-existence. Sequencing data showing long-term co-existence of lineages with different sets of mutations suggests that similar interactions may also have arisen in other LTEE populations14,43, though it is not clear whether they are strong enough to noticeably alter fitness estimates. Finally, the evolution of aerobic growth on citrate in population A−3 of the LTEE32 means that the fitness of these cells now incorporates the use of a "private" resource when they are competed against cells that cannot use citrate, which complicates interpreting these results. Despite these exceptions, the use of a low glucose concentration and well-shaken environment has undoubtedly simplified making fitness comparisons between LTEE strains and populations.
At later generations, some of the LTEE populations no longer form colonies on TA agar, which makes performing competition experiments using even modified protocols difficult or impossible10. Alternative methods that do not require colony growth can potentially be used to determine the relative representation of two competitors, such as FREQ-seq which uses next-generation sequencing to count the proportion of reads containing two alternative alleles in an amplicon44. This method or a similar one could potentially be used with the Ara alleles or with newly evolved mutations, such as those in pykF and nadR, versus the ancestral sequence. Performing genetic modifications that introduce other types of neutral markers can also be used to measure relative fitness. For example, fluorescent protein genes have been inserted into the chromosomes of cells in LTEE offshoot experiments so that competitors can be counted using flow cytometry45. Another approach, which opens up the possibility of mixing together more than two strains in the same competition flask, is to insert barcodes that can be PCR amplified and sequenced into the genomes of different competitors. This approach has been used for lineage tracing in evolution experiments46. Both flow cytometry and barcode sequencing can accurately measure much more extreme ratios of two strains versus colony counting (because they can query > 10,000 cells/genomes versus the < 500 that can be counted on an agar plate), so using these methods also promises to increase the dynamic range in terms of fitness differences that can be measured relative to a common competitor.
Alternative designs for long-term microbial evolution experiments
For all its virtues, the LTEE is not perfect. Certain aspects of its design make it labor intensive and susceptible to human error. For instance, each day a researcher must come into lab and pipette between Erlenmeyer flasks to continue the experiment. Competition experiments can also pose daunting logistical hurdles, given that requirements for sterile glassware, media, incubator space, and colony counting rapidly escalate when even a small number of competitors are being tested with modest replication. We are often asked why we don't take advantage of laboratory automation systems, such as pipetting robots that operate on 96-well microplates, or continuous culture systems, such as chemostats or turbidostats. The answer is simple: the LTEE is, in a sense, a prisoner of its own long history. We dare not deviate from 10 mLÂ cultures shaking at a specific speed in 50 mL Erlenmeyer flasks because this would risk fundamentally changing the experiment. Subtle aspects of the environment to which these populations have been adapting for decades (e.g., the amount of aeration), would be altered in microplates or continuous culture systems. The population bottleneck at each transfer might also be different (smaller in microplates, for example), changing the evolutionary dynamics. In short, deviating from the methods described here would make the LTEE a different experiment, or at the very least risk introducing a discontinuity that would disrupt evolutionary trajectories.
Researchers designing new evolution experiments should consider these other ways of propagating microbial populations, while being aware of their potential benefits and drawbacks. Using pipetting robots to transfer populations in microwell plates is logistically simpler in some ways and can prove quite powerful due to the high numbers of replicate populations that can be propagated in this way47,48,49. However, automated transfers in most current setups do not take place under completely sterile conditions, which increases the likelihood of outside contamination. To prevent contamination, the growth medium is often supplemented with antibiotics, which become a feature of the environment that affects evolution. Transfers in microwell plates are also more prone to cross-contamination events. Finally, the environment of microwell plates — particularly if they are not shaken — tends to select for wall growth, aggregation, and other phenomena that can complicate evolution by creating multiple niches in one well. Using rich media or high concentrations of nutrients to keep population sizes large in small wells is likely to exacerbate these complexities. If such interactions arise, they can make measuring and interpreting fitness much more difficult.
Continuous culture systems for microbial evolution include chemostats, in which fresh medium is constantly pumped in and culture is pumped out, and turbidostats, in which cultures are periodically diluted through automated sensing and pumping to maintain cells in a state of constant growth. These systems are very useful when one wants to model microbial physiology and evolution because they avoid having microbes transition between growth and starvation by keeping them in an environment that always has nutrients50. One can even add sensors that make real-time measurements of optical density, O2 consumption, pH, and other aspects of a culture's environment and growth. However, current continuous culture systems either require expensive equipment purchases or specialized expertise to build custom setups51,52,53,54. Also, wall growth, in which cells escape dilution by adhering to the culture chamber, bedevils evolutionary dynamics in continuous culture systems unless they are periodically sterilized. Due to these constraints, most chemostat and turbidostat evolution experiments to date have been of limited duration and/or involved relatively few independently evolving populations compared to serial transfer evolution experiments.
Conclusion
The methods we demonstrate here for the LTEE are critical for studying its unique historical record and continuing the open-ended evolution of these E. coli populations. They also provide a starting point for others who are considering new evolution experiments that may take advantage of laboratory automation or add back various elements of the complexity found in natural environments that were purposefully omitted from the LTEE. Since 1988, experimental evolution has flourished as a field. During this time, researchers in laboratories across the globe have demonstrated the immense flexibility of this approach for studying evolution, innovating by introducing creative experimental designs and monitoring the results using new technologies. The methods of the LTEE do not represent an endpoint, but we hope they will continue to inspire and provide a foundation for the field far into the future.
We thank Richard Lenski and the many researchers who have studied and contributed to maintaining the Long-Term Evolution Experiment with E. coli, including especially Neerja Hajela. The LTEE is currently supported by the National Science Foundation (DEB-1951307).
Name | Company | Catalog Number | Comments |
2,3,5-Triphenyltetrazolium chloride (TTC) | Sigma-Aldrich | T8877 | |
20 mL Glass Beaker | Sigma-Aldrich | CLS100020 | |
50 mL Erlenmeyer Flasks | Sigma-Aldrich | CLS498050 | |
Agar | Sigma-Aldrich | A1296 | |
Ammonium Sulfate | Sigma-Aldrich | AX1385 | |
Antifoam | Sigma-Aldrich | A5757 | |
Arabinose | Sigma-Aldrich | A3256 | |
Freezer Box (2") | VWR | 82007-142 | |
Freezer Box (3") | VWR | 82007-144 | |
Freezer Box Cell Divider (49-place) | VWR | 82007-150 | |
Freezer Box Cell Divider (81-place) | VWR | 82007-154 | |
Freezer Vials (1/2-Dram) | VWR | Â 66009-816Â | |
Freezer Vials (2-Dram) | VWR | Â 66010-560Â | |
Glucose | Sigma-Aldrich | G8270 | |
Glycerol | Fisher Scientific | G33 | |
Magnesium Sulfate | Sigma-Aldrich | M7506 | |
Metal Tray | Winco | SPJP-202 | |
Petri Dish | Fisher Scientific | FB0875712 | |
Potassium Phosphate Dibasic Trihydrate | Sigma-Aldrich | P5504 | |
Potassium Phosphate Monobasic | Sigma-Aldrich | P5379 | |
Sodium Chloride | Sigma-Aldrich | M7506 | |
Sodium Citrate Tribasic Dihydrate | Sigma-Aldrich | C7254 | |
Test Tube Cap (18mm) | VWR | 10200-142 | |
Test Tube Rack (18mm, steel) | Adamas-Beta | N/A | Test Tube Racks Stainless Steel Grid Arrangement 72 Holes (17-19 mm) |
Test Tubes (18 x 150 mm) | VWR | 47729-583 | |
Thiamine, Hydrochloride | Millipore | 5871 | |
Tryptone | Gibco | 211705 | |
Yeast Extract | Gibco | 212750 |
This article has been published
Video Coming Soon
ABOUT JoVE
Copyright © 2024 MyJoVE Corporation. All rights reserved