SEARCH

SEARCH BY CITATION

Keywords:

  • compensatory evolution;
  • epistasis;
  • experimental evolution;
  • fludioxonil;
  • fungicide resistance;
  • gene interaction;
  • mycelial growth rate;
  • parasexual analysis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We have studied compensatory evolution in a fludioxonil resistant mutant of the filamentous fungus Aspergillus nidulans. In an evolution experiment lasting for 27 weeks (about 3000 cell cycles) 35 parallel strains of this mutant evolved in three different environmental conditions. Our results show a severe cost of resistance (56%) in the absence of fludioxonil and in all conditions the mutant strain was able to restore fitness without loss of the resistance. In several cases, the evolved strain reached a higher fitness than the original sensitive ancestor. Fitness compensation occurred in one, two or three discrete steps. Genetic analysis of crosses between different evolved strains and between evolved and ancestral strains revealed interaction between compensatory mutations and provided information on the number of loci involved in fitness compensation. In addition, we discuss the opportunities for the experimental study of evolutionary processes provided by the filamentous fungus A. nidulans.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The use of antibiotics has resulted in persistent resistance in microbial populations (Schrag & Perrot, 1996; Bjorkman et al., 2000). By a single mutation, an organism can become resistant to an antibiotic it encounters. It is common that these resistance mutations come with a cost: in the absence of the antibiotic, the resistant type often has a lower fitness than its sensitive relatives (Cohan et al., 1994; Andersson & Levin, 1999; Gilliver et al., 1999; Cowen et al., 2000; Reynolds, 2000). Thus, loss of resistance might be expected when a resistant population is placed back in an antibiotic-free environment. However, instead of reverting to the sensitive wild type, a resistant genotype may gain fitness by accumulating compensatory mutations that decrease the costs while resistance remains unaffected (Cohan et al., 1994; Schrag et al., 1997; Andersson & Levin, 1999; Levin et al., 2000; Lipsitch, 2001). Clearly, the phenomenon of compensatory evolution in antibiotic-resistant strains is highly relevant for understanding the persistence of resistance in natural populations.

Several experimental studies have investigated compensatory evolution in antibiotic-resistant prokaryotes (see also Lenski, 1997). Bouma & Lenski (1988) showed that fitness costs associated with a plasmid encoded resistance to chloramphenicol in Escherichia coli decreased over 500 generations. Schrag & Perrot (1996) reported compensatory adaptation in streptomycin resistant E. coli populations. Cohan et al. (1994) discussed the amelioration of fitness costs in rifampin resistant Bacillus subtilis and have suggested that the potential for compensatory evolution depends on the magnitude of the cost of resistance. Moore et al. (2000b) indeed observed greater compensatory adaptation for deleterious mutations of large effect than for those of small effect. Levin et al. (2000) studied compensatory evolution in streptomycin resistant E. coli and concluded that rather than back-mutation to the high fitness sensitive type, mainly intermediate fitness compensatory mutations arise that do not affect the level of resistance. Reynolds (2000) reached similar conclusions in rifampin resistant E. coli.Bjorkman et al. (2000) found different compensatory mutations in streptomycin resistant Salmonella typhimurium depending on whether bacteria evolved in mice or laboratory medium. The beneficial effect of a compensatory mutation may depend on the presence of the deleterious resistance mutation. Schrag et al. (1997) observed negative fitness effects of compensatory alleles from a streptomycin resistant E. coli strain when transferred into a sensitive background.

We have studied antibiotic resistance and compensatory evolution in a multicellular eukaryote system, the filamentous fungus Aspergillus nidulans. In order to do this, we have developed an experimental evolution protocol for Aspergillus nidulans, studying evolution over 3000 cell cycles. The emerging antibiotic resistance is a problem in fungi like Aspergillus spp., some of which are human pathogens (Cowen et al., 2001; Moore et al., 2000a). Aspergillus nidulans shares the well-known advantages of unicellular micro-organisms in evolution experiments such as short generation time, easily controlled and replicated environments, large populations and small space requirements (Elena & Lenski, 2003). Furthermore, Aspergillus nidulans has the additional advantages of multicellularity and a eukaryotic sexual cycle that can facilitate genetic analysis (Pontecorvo et al., 1953; Pontecorvo & Kafer, 1958; Clutterbuck, 1974; Swart & Debets, 2004). Filamentous fungi grow on the surface of a substrate and form aerial asexual and sexual spores. The way the fungus expands during growth on solid medium is drawn schematically in Fig. 1. From a mononucleate spore, the fungus eventually forms mycelium consisting of a network of hyphae with the nuclei in the mycelium occupying spatially defined positions. Twenty-four hours after their formation hyphal cells can differentiate into foot cells, giving rise to conidiophores, each carrying up to 10 000 asexual spores. The production of sexual spores commences after 4 days (Adams et al., 1998).

image

Figure 1.  Schematical drawing of a growing Aspergillus nidulans colony. From a spore with one single nucleus, growth starts by a few cycles of nuclear division. When there are more than eight nuclei, hyphae start to develop (Fig. 1a–c). After cell stretching, septa are formed in the hyphae (Fig. 1d), eventually leading to the situation shown in Fig. 1e: the tip of the mycelium has gone one large apical cell containing 20–40 nuclei. During a cell cycle, all nuclei in this apical cell divide simultaneously. After the nuclear division, septa are formed. Half of the nuclei are then located in the subapical cells formed by the septa, the other half remain in the apical cell to enter the next cycle. Each subapical cell will contain three to four nuclei and will have a length of approximately 40 μm, or one nucleus per 13 μm mycelium. The nuclei in the subapical cells do not divide anymore; they are only metabolically active. Mitotically active nuclei are shaded (from Kaminskyj & Hamer, 1998 with permission).

Download figure to PowerPoint

The nuclei at the growing front of the colony go through the largest number of mitoses; sampling from the growing front makes the effective number of cell cycles in the context of experimental evolution remarkably high. Based on the time needed for a cell cycle as well as on the distribution of the fixed nuclei in the mycelium, it can be estimated that the fungus can go through about 19 cell cycles a day, see Methods (Bergen & Morris, 1983; Kaminskyj & Hamer, 1998). Here, the time needed for a cell cycle is defined as the time needed for a full replication of the genome. Each nucleus is omnipotent and potentially can give rise to homokaryotic sexual as well as asexual spores. Therefore, the haploid nucleus in a mycelium is the evolutionarily relevant unit, as is the haploid cell in a bacterial culture.

All nuclei in a mycelium are mitotically derived from a single ancestral nucleus, resulting in a genetically homogeneous colony. The only source of genetic variability in a fungal colony is mutation resulting in segments of mycelium with different genotypes. The effect of a beneficial mutation that arises in the mitotically active hyphal tips may therefore be seen at the growing front of the colony as a sector with accelerated growth (Fig. 2). The process of compensatory evolution is studied under three experimental regimes, reflecting different agricultural conditions (see Table 1).

image

Figure 2.  The fungicide resistant station strain (WG561) growing under optimal conditions (left, minimal medium with fungicide) and under suboptimal conditions (right, MM without fungicide) after 6 days of incubation. The colony formed when growing under optimal conditions has a perfectly circular shape; the surface is covered by asexual conidiospores. At the edge, the colony has a lighter color; here, the asexual spores have not yet fully developed, this takes around 24 h (Adams et al., 1998). Growing under suboptimal conditions (in this case medium without fungicide) spontaneous mutations can develop that give rise to clear sectors with a beneficial mutation. From the point in the mycelium where the mutation occurred, a sector with higher growth rate developers. The colony shows in this example has two such mutations.

Download figure to PowerPoint

Table 1.   Conditions and strains used in the evolution experiment. For genotypes of strains see Table 2. Fludioxonil is the fungicide used.
 ConditionStrains used (no. of replicate strains)Fludioxonil resistant (R) or sensitive (S)
AMM without fludioxonilWG 561 (8) and WG 615 (7)R
BMM with decreasing fludioxonil concentrations, at the start 0.2 ppm, after the fourth week starting to decrease with weekly 0.02 ppm, thus reaching zero at week 14 and remaining at zero thereafter up to the end of the experimentWG 561 (5) and WG 615 (5)R
CMM alternation presence and absence of fludioxonil in a four week cycle; without fludioxonil every first three weeks and with 0.2 ppm every fourth weekWG 561 (5) and WG 615 (5)R
WMM without fludioxonilWG 562 (5)S

Here, we demonstrate the presence of costs associated to fludioxonil resistance in A. nidulans. Second, we show that compensatory evolution occurs in a fludioxonil resistant strain over 3000 cell cycles in the following environmental conditions: in the absence of fungicide, with diminishing concentrations of fungicide and with alternating high and low fungicide concentrations. Third, we demonstrate that a limited number of one to three mutations are involved in the process of compensation without loss of the fungicide resistance. Fourth, genetic analysis demonstrates the presence of epistatic interaction between the resistance mutation and compensatory mutations. Finally, we discuss consequences and implications for the persistence of resistance.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Strains and genetic techniques

Standard techniques for genetic analysis of A. nidulans were used (Pontecorvo et al., 1953; Pontecorvo & Kafer, 1958). The A. nidulans strains used in this study were derived from the original Glasgow strains (Clutterbuck, 1974), see Table 2. WG562 is the ‘wild-type’ lab strain with a lysine marker to facilitate genetic analysis. WG561 is a spontaneous mutant of WG562 that is resistant to 0.2 ppm of the fungicide fludioxonil (Novartis). We crossed WG561 with WG145, which has a pyridoxin deficiency and white asexual spores, and determined that fludioxonil resistance is due to a single locus (fldA1). From this cross, we recovered WG615, which is fludioxonil resistant, pyrodoxin deficient and has white asexual spores. Compensatory mutations to alleviate the costs of fludioxonil resistance were named flc. We determined the chromosomal location of the resistance locus and compensatory loci by parasexual analysis essentially as described by (Kafer, 1958). The diploid nuclei are induced to haploidize with 1.7 ppm benomyl to give nuclei with new combinations of parental chromosomes in the near absence of intra-chromosomal recombination. The new combinations can be detected using selective media.

Table 2.   Strains used.
StrainGenetic markers per chromosome
IIIIIIIVVVIVIIVIIIUnlocalized
  1. fldA1, resistance to fludioxonil; flc, mutation compensating for costs associated with fludioxonil resistance; other markers as described by Clutterbuck (1974): acrA1 acriflavine resistance; actA1, actidione resistance; chaA1, chartreuse conidia; choA1, choline req.; lysB5, lysine req.; nicA2, nicotonic acid req.; nicB8, nicotonic acid req.; pabaB22, p-aminobezoic acid req.; phenA2, phenylalanine req.; pyroA4, pyrodoxine req.; pyrG89, pyrimidine req.; riboB2, riboflavin req.; sB3, sulphate non-utilizer; sC0, sulphate non-utilizer; wA2, white conidia; yA2, yellow conidia.

AN116yA2wA2sC0pyroA4     
WG145 wA3 pyroA4     
WG492pyrG89 phenA2  sB3nicB8riboB2 
WG525pyrG89acrA1actA1pabaB22nicA2sB3choA1chaA1 
WG561  fldA1 lysB5    
WG562    lysB5    
WG615 wA3fldA1pyroA4     
WG619  fldA1 lysB5flcA4   
WG621 wA3fldA1pyroA4    flc5, flc6, flc7
WG622  fldA1 lysB5   flc8
WG626 wA3fldA1pyroA4 sB3nicB8riboB2 
WG628  fldA1 lysB5   flc9, flc10
WG629 wA3fldA1pyroA4    flc11
WG630 wA3fldA1pyroA4    flc12
WG647  fldA1 lysB5   flc19
WG648 wA3fldA1pyroA4    flc20, flc21, flc22

Culture media and growth conditions

We cultured the strains in Petri dishes with Minimal Medium (MM) or Complete Medium (CM). MM consists of 6.0 g NaNO3, 1.5 g KH2PO4, 0.5 g MgSO47·7H2O, 0.5 g KCl, 10 mg of FeSO4, ZnSO4, MnCl2 and CuSO4 and agar 15 g + 1000 mL H2O (pH 5.8). For CM, 2.0 g neopeptone, 1.0 g vitamin assay casamino acids, 1.0 g yeast extract and 0.3 g ribonucleic acids for yeast was added to MM. After sterilization of the media, sucrose was added to a final concentration of 25 mmol L−1 to both media. CM was supplemented with a vitamin solution after sterilization (2 mL solution/1000 mL medium) containing thiamin 100 mg L−1, riboflavin-Na 1.25 g L−1, para-aminobenzoicacid 100 mg L−1, nicotinamide 1.0 g l−1, pyridoxin-HCl 500 mg L−1, D-pantothenic acid 100 mg L−1 and biotin 2.0 mg L−1. MM always was supplemented with lysine to a final concentration of 2.0 mmol L−1 in the medium and with pyrodoxin to a final concentration of 0.1 mg L−1. In the assay used to determine reaction norms and in two of the conditions of the evolution experiment MM with the fungicide fludioxonil (4-(2,3-difluoro-1,3-bezodioxol-4-yl)pyrrol-3-carbonitril) was used. Fludioxonil (supplied by Novartis, UK) was dissolved in methanol to a concentration of 100 ppm. This stock solution was used to add fludioxonil to the medium to a concentration of 0.2 ppm; the methanol concentration in the medium never exceeded 0.8% and methanol evaporated rapidly since it was added to the medium when warm (>45 °C). Spore suspensions were made in saline (distilled water with NaCl 0.8 g L−1) supplemented with Tween80 (0.05%) by washing off all spores from the surface of a plate with 5 mL saline-tween. Inoculation of agar plates was done either with an inoculation needle or with 5 μL of spore suspension. Plates were incubated at 37 °C.

Mycelial growth rate

Relative mycelial growth rate (MGR) is used as fitness measure (Pringle & Taylor, 2002). After 6 days of growth, we determined the MGR by averaging the colony diameters as measured in two randomly chosen perpendicular directions in mm. The relative MGR is defined as the MGR divided by the colony diameter of the fungicide sensitive ancestor WG562, grown on minimal medium without fungicide.

MGR is found to be highly reproducible and relatively independent of inoculum size. Three suspensions with decreasing numbers of spores (100 000, 100 or 10 spores per inoculum of 5 μL) were plated in 5-fold in Petri dishes with Minimal Medium and incubated at 37 °C. After 6 days, we measured the colony diameter. For the inoculum sizes of 100 000 and 100 spores no difference in colony diameter was observed (t-test; t11 = 1.01, n.s.); we found a colony diameter of 69.1 mm (standard error of the mean – SEM = 0.31) and 69.5 mm (SEM = 0.50), respectively. Only in the case of the extremely small inoculum size of 10 spores the colony diameter was significantly lower with a diameter of 65.0 mm (SEM = 0.10 mm; t-test; t11 = 24, P < 0.001). We conclude that MGR measurements are highly reproducible and that inoculum size in our evolution experiment (transferring between 10 000 and 50 000 nuclei, see below) did not influence the MGR.

Experimental evolution procedure

To facilitate genetic analysis and sexual crosses between evolved strains, we used both resistant strains WG561 and WG615 for conditions A–C of the evolution experiment (see Tables 1 and 2). We measured the MGR of WG561 and WG615 on media with and without fungicide, showing that the genetic markers are neutral in terms of MGR (t-test; t8 = 0.367, n.s.). WG562 was used as control stain (condition W), it evolved under conditions without fungicide to detect any adaptation of this fungicide sensitive strain to our experimental conditions.

At the start of the experiment, we inoculated approximately 1000 spores at the center of a Petri dish. After 6 days of incubation at 37 °C, all plates were put at 4 °C for at least 24 h (but no longer than 60 h). The (mean) MGR of the colony was measured and the part of the colony with the highest MGR was identified by visual inspection (see Fig. 2 for an example). Part of the mycelium of this sector was taken off with a needle and transferred to the middle of a Petri dish with fresh medium; in this way between 10 000 and 50 000 nuclei were transferred. This cycle was repeated 27 times. Conidiospore samples of all replicates were stored weekly at −80 °C in a solution with 0.7% neo-peptone and 30% glycerol in water. Each week we measured the MGR of the reference strain (the nonevolved WG562) by starting five replicate cultures from the stock stored at −80 °C. The plates with the reference strain were incubated together (but randomized) with those with evolving strains.

During the evolution experiment over twenty-seven weekly transfers the growing front of the cultures goes through at least 3000 cell cycles (successive mitotic divisions). The number of cell cycles is estimated in two ways. The first estimate is based on the time the fungus needs for one cell cycle. A cell cycle takes 90 min growing at 37 °C (Rosenberger & Kessel, 1967; Bergen & Morris, 1983). Therefore, in 1 week the fungus can complete about 112 cell cycles. The second way to estimate the number of cell cycles is by considering the position of the nuclei in the mycelium combined with the growth characteristics of the fungus, shown in Fig. 1. After six days of incubation at 37 °C, an A. nidulans colony (WG562) inoculated with a needle has grown for 30 mm from the point of inoculation (Wolkow & Hamer, 1996; Suelmann et al., 1997). Since there is on average one nucleus per 13 μm mycelium (Kaminskyj & Hamer, 1998), a hypothetical hypha extending radially from the colony center along a straight line would contain around 2300 nuclei after one week of incubation at 37 °C. How many mitoses are required to obtain 2300 nuclei? Formation of novel nuclei occurs at the growing front of the colony in the apical cells; nuclei ‘behind’ the growing front are only metabolically active. After every mitotic cycle in the apical cells, half of the newly formed nuclei remain in the apical cells and half are directed to the subapical cells. The number of nuclei in the apical cell varies and can reach up to 50 nuclei per apical cell (Wolkow & Hamer, 1996; Kaminskyj & Hamer, 1998); 20 seems to be a reasonable conservative estimate just before mitosis. Assuming 20 nuclei in the apical cell, each mitotic cycle will add 20 nuclei to the total present in the growing hypha. To reach the total number of 2300 nuclei, 115 cycles are required. This second estimate for the number of cell cycles is consistent with the first estimate based on the time needed for one cell cycle.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We isolated 30 spontaneous fludioxonil resistant mutants from the sensitive wildtype (WG562) on minimal medium with the fungicide concentrations ranging from 0.03 to 0.8 ppm. All mutants found suffered a considerable cost of resistance in terms of MGR: on average the MGR on standard fungicide-free medium was reduced by 56% (SEM = 0.76%) relative to the MGR of the sensitive ancestral strain grown under the same conditions.

For further analysis we selected mutant strain WG561, which was isolated from medium containing 0.2 ppm of fludioxonil. A sexual cross of the resistant mutant (WG561) with a sensitive strain (WG145) showed a 1 : 1 segregation of the resistance phenotype (Chi square test; n = 67; χ2 = 0.134; P > 0.05), indicating of a single resistance mutation. Other genetic markers used in the genetic analysis showed no effect on MGR. The resistance mutation was termed fldA1. Diploid analysis showed that fldA1 is recessive and located in linkage group III (data are in Table 3). Figure 3 shows the relative MGR of WG562 and WG561 on medium with different fungicide concentrations. Growth of the resistant mutant WG561 is highest in the presence of fludioxonil. At high concentrations the MGR is higher than at low concentrations and even higher than the MGR of the fungicide sensitive strain (WG562) on fungicide free medium, showing that WG561 is well adapted to conditions with fungicide. Under fungicide-free conditions strain WG561 suffers a fitness cost of approximately 50%.

Table 3.   Linkage data of parasexual analysis of diploids. Diploids were formed of the strain with an unlocalized allele and a tester strain. Segregation was induced with benomyl and haploid segregants were analyzed, see methods. Table 1 shows the strains and their markers.
Strain (with unlocalized allele)Linkage group tester strain Linked markers*Number of segregants analysedRecombinant frequency (number of recombinants)Linkage group
  1. Only the linked marker of the tester strain is shown. The other markers of the tester strains were unlinked to the allele of interest and segregated randomly.

WG651 (fldA1)WG525fldA1-actA11331.5% (2)III
WG651 (fldA1)AN116fldA1-sC02240% (0)III
WG619 (flcA4)WG626flcA4-sB31200.8% (1)VI
image

Figure 3.  Reaction norms of the sensitive (WG562) and resistant strain (WG561) used in the evolution experiment. Mycelial growth rate (MGR) is expressed relative to the MGR of the sensitive strain growing on minimal medium without fungicide. The reaction norms show that the presence of fungicide enhances growth of the resistant strain and inhabits the growth of the sensitive strain.

Download figure to PowerPoint

Evolution experiment over 3000 cell cycles

Strains carrying the fldA1 mutation were used (see Tables 1 and 2) to set up an evolution experiment over 27 weeks (corresponding to some 3000 cell cycles), weekly selecting for the part of the colony with the highest MGR in three different environments: fungicide free, decreasing fungicide concentrations over time and alternating absence and presence of fungicide (conditions A–C, respectively, see Table 1). In total, we used 35 replicates. In addition, strain WG562 carrying the wildtype sensitive allele (fldA+) was used as control on medium without fungicide (condition W, Table 1). The results of these experiments are shown in Fig. 4. For all conditions (A–C and W), evolved strains after 27 weeks had a higher mean MGR than at the start (t-tests; P < 0.01). The mean MGR at week 27 is not different for the conditions A–C (one way anova, F2,26 = 1.42; n.s.).

image

Figure 4.  Results of the evolution experiment (27 weeks, i.e. >3000 mitotic cells cycles). A–C, and W refer to the experimental conditions used (see Table 2). The data points in the graphs represent the weekly mean MGR of all evolving strains per condition, error bars represent the standard error of the mean (SEM). MGR was made relative to the MGR of the (nonevolved) fungicide sensitive ancestor (WG562) growing without fungicide. The resistant mutant when growing in the presence of high fungicide concentrations has a higher MGR than the sensitive ancestor when growing in the absence of fungicide (see Fig. 3). In conditions B and partially C this results in the relative MGR to be greater than 1.

Download figure to PowerPoint

The sensitive controls of strain WG562 (condition W) increase in average MGR from 1.00 (SEM = 0.0027) to 1.03 (SEM = 0.0055). This increase is statistically significant (t-test; t8 = 5.91, P < 0.001). In condition A (see Table 1), the average MGR of the 15 evolving strains increased from 0.50 (SEM = 0.0034) to 0.82 (SEM = 0.052). In condition B (see Table 1), the resistant mutant showed a higher MGR during the first four weeks when growing in the presence of the fungicide (1.08; SEM = 0.0014) than the populations of condition A that grew in the absence of fungicide (MGR = 0.50; SEM = 0.0034); the mean MGR diminished with decreasing fungicide concentration during weeks 5–14 and recovered during the complete absence of the fungicide (which was from week 14 onwards) to a MGR of 0.97 (SEM = 0.058). In condition C (see Table 1), the presence of fungicide leads to an increased MGR as shown by the peaks in the curve. By week 27, the evolving strains were adapted to both the absence and presence of fungicide in the environment: on fungicide-free medium the mean MGR increased from an initial value of 0.50 (SEM = 0.0034) to 0.90 (SEM = 0.092) on week 27; on fungicide-containing medium the MGR was 1.02 (SEM = 0.049) on week four when the strains first encountered fungicide and 1.01 (SEM = 0.049) on week 24 (this MGR difference is non-significant; t-test; t16 = 0.18, n.s.). MGR was made relative to the MGR of the (nonevolved) fungicide sensitive ancestor (WG562) growing without fungicide (this MGR was set to 1). As can be seen from Fig. 3, the resistant mutant growing in the presence of high fungicide concentrations has a higher MGR than the sensitive ancestor growing in the absence of fungicide. In conditions B and C this results in the relative MGR to be greater than 1.

Out of the 35 evolved fungicide resistant strains of conditions A–C, seven had a higher MGR than the original sensitive ancestor (WG562); the highest MGR was 1.24 (MGR relative to WG562; SEM = 0.005).

To set up conditions A–C, both strains WG561 and WG615 were used, both carrying fldA1 but with different genetic markers (see Tables 1 and 2). After 27 weeks, there was no difference in response (MGR and rate of adaptation) for evolved strains derived from strains WG561 and WG615 (t-tests, MGR after 27 weeks: t31 = 1.21, n.s.; rate of adaptation: t31 = 0.84, n.s.).

Figure 4 also shows the change in MGR in the experimental populations. The rate of adaptation over the course of the experiment is given by the slopes of the fitness trajectories. The mean slope for all strains per condition was computed over the weeks without fungicide in the environment. For A the slope was computed over the first 13 weeks, for B over weeks 15–27 and for C over weeks 1–17 omitting the weeks that fungicide was present in the medium. The slope in condition B is higher than the slope in A and C, however, statistically this is a borderline case (one way anova; F2,29 = 3.23; P = 0.054). The rate of adaptation (given by the slope) in conditions A–C is higher than in condition W (multiple t-tests; P < 0.001), showing that adaptation in the fungicide resistant strains was indeed compensation for fitness costs and not only adaptation to our laboratory conditions.

Changes in level of resistance

The results from condition C with the alternating presence and absence of fungicide shows that the resistance to fungicide is maintained, while the fitness in the absence of fungicide increases. In addition to the measurements of weekly linear growth, we tested all strains of conditions A–C after 27 weeks of evolution for their resistance to different concentrations of the fungicide. All strains maintained their resistance to the fungicide, showing that the compensatory changes did not involve back mutation of the resistance allele to the wild type, but mutations at other loci.

Stepwise fitness improvement

The appearance of clearly recognizable sectors displaying enhanced growth rate suggests that fitness improves in discrete steps. We investigated several strains from each condition in more detail, remeasuring MGR in six-fold on fungicide-free medium for five out of the 27 time points in the stored samples of two selected strains per condition. The measurements were done in a single assay to eliminate possible minor variation between weekly assays. Two reconstructions per condition are shown in Fig. 5. Results show that MGR improvement occurred in one to three major steps, suggesting the occurrence of one to three major compensatory mutations during the experiment. The differences in MGR before and after a step were highly significant in all cases (t-test; n = 12, P < 0.001). For one evolving strain of condition A (WG621), we remeasured the MGR of all 27 subcultures in one assay, see Fig. 6 left panel. For this fitness trajectory we found a clear stepwise fitness improvement. A step function with five steps gives the best fit to the data, see Table 4 (partial f-test; Elena et al., 1996). The evolved strains studied in these assays all have a different MGR after the 27 weeks of evolution, suggesting they followed different evolutionary trajectories. In the strains studied here, the first step in fitness improvement was always larger than subsequent steps.

image

Figure 5.  Adaptive trajectories of several independently evolving strains. A–C, and W refer to the experimental conditions used (see Table 2). Here, two evolving strains per condition are shown (derived from either WG561 or WG615, see Table 1). We measured the MGR at five times points in 3-fold in a single assay, error bars representing the SEM are smaller than the symbols. MGR was made relative to the MGR of the (nonevolved) fungicide sensitive ancestor (WG562) growing without fungicide. On the basis of MGR, steps in adaptation were identified, the difference in MGR between the steps is highly significant (t-tests, n = 6, P < 0.001).

Download figure to PowerPoint

image

Figure 6.  (left panel) Adaptive trajectory of strain WG621 (which was derived from WG615 and evolved in condition A). Stored samples of all weeks were retrieved from −80 °C and the MGR was measured in 3-fold in a single assay. MGR was made relative to the MGR of the (nonevolved) fungicide sensitive ancestor (WG562) growing without fungicide. Error bar represent the SEM. On the basis of MGR and a partial f-test (Elena et al., 1996), four steps in adaptation could be identified, see Table 4. (right panel) Reaction norms on MM with increasing fungicide concentrations of progeny from a cross between evolved strain WG621 after 27-week transfers (depicted in the left panel) and the non-evolved resistant ancestor (WG561). For comparison, part of the reaction norm of a fungicide sensitive strain WG562 (+++) is also shown. The wild-type allele is represented by +, the resistance allele by R and the compensatory mutations by C1 and C2. The mean MGR (±SE) of the four classes in the progeny (n = 40) is shown, indicating that two main loci are involved in the fitness compensation.

Download figure to PowerPoint

Table 4.   Analysis of variance (partial F-test; Elena et al., 1996) comparing the fit of several models with different numbers of steps to the fitness trajectory of WG621 (see Fig. 6).
ModelSSSSRFPd.f.Groupings
  1. Groupings show the assignment of the week points to intervals between steps. The time points at the MGR jumps were fixed based on the MGR observed and shown in Fig. 6 left panel (as was done by Elena et al., 1996). Several five step models were tested (data not shown), the model presented in this table gave the best fit. The P value indicates the significance of addition of the additional step to the model. SS, sum of squares; SSR, residual sum of squares; d.f., remaining degrees of freedom.

One step1.28E + 05   260–27
Two steps1.29E + 058.60826752.4<0.001250–1, 2–27
Three steps1.29E + 052.33443368.5<0.001240–1, 2–7, 8–27
Four steps1.29E + 050.54408279.0<0.001230–1, 2–7, 8–17, 18–27
Five steps1.29E + 050.4597226.530.018220–1, 2–7, 8–17, 18–21, 22–27
Six steps1.29E + 050.4776460.00n.s.210–1, 2–7, 8–17, 18–21, 22–25, 26–27

Sexual crosses to assess the number of mutational steps in fitness improvement

The number of phenotypic classes in the progeny of a sexual cross between an evolved strain and the ancestor can reveal the number of major compensatory mutations involved in the fitness trajectory. The phenotype is best given by the reaction norm on different concentrations of fungicide (Fig. 3).

First, we crossed an evolved strain of condition A, WG619 (which was derived from WG561) with the fungicide sensitive strain WG145. Based on MGR measurements showing one distinct increase of MGR (see Fig. 5), WG619 is expected to carry one compensatory mutation. Analyzing 40 progeny, we found four phenotypic classes when determining the reaction norm of the progeny on medium with fungicide (by growing the progeny on a series of plates with MM with increasing fungicide concentrations, see Fig. 3 and Table 5). One recombinant has the same phenotype as WG561 (the nonevolved fungicide resistant). The other recombinant is a ‘novel’ type and has a higher MGR than the ancestor WG619, see Table 5, suggesting that the compensatory mutation also has a beneficial effect in the absence of the resistance allele. Parasexual analysis (Kafer, 1958) using strain WG626 showed that the compensatory allele is located on chromosome VI (recombinant frequency of 0.8% to sB3, 120 segregants analysed).

Table 5.   Sexual cross between a strain of condition A after 27 weeks of adaptation (WG619; derived from WG651) and a fungicide sensitive strain (WG145) showed four classes in the progeny when measuring their MGR on MM without fungicide (no significant deviation from a 1 : 1 : 1 : 1 segregation; Chi-square test; χ2 = 7.6; d.f. = 3; P > 0.05). The MGR was made relative to the standard fungicide sensitive reference (WG562). SEM, standard error of the mean.
A)
 Sexual cross
  WG619+fldA1+lysB5flcA4
  WG145wA3+pyroA4++
 Progeny genotype*No. of progenyMGR (SEM)Phenotype description
  1. *Relevant markers are shown; fldA1, resistance allele; flcA4, compensatory mutation.

B)
 fldA1flcA491.18 (0.005)Evolved fungicide resistant
 ++41.00 (0.005)Fungicide sensitive ancestor
 fldA1+160.78 (0.005)Non-evolved fungicide resistant
 +flcA4111.12 (0.005)‘‘New’’ fungicide sensitive type

Second, we crossed another strain of condition A that had undergone 3000 cell cylces, (WG621, which was derived from WG615) with WG562, a fungicide sensitive strain. Based on MGR combined with a partial f-test (see Fig. 6 and Table 4), four steps in fitness improvement were identified in the evolutionary trajectory of WG621, see Fig. 6 left panel. By analysing the reaction norm on medium with increasing fungicide concentrations of 40 progeny of the cross, there were many phenotypic classes. Therefore, we decided to do a cross with the nonevolved resistant parent (WG561), eliminating one segregating locus. This cross, again analysing 40 progeny, gave four distinct phenotypic classes, indicating the presence of two segregating major compensatory mutations. (The ratio of the classes did not deviate from 1 : 1 : 1 : 1; χ2-test; P = 0.70.) The phenotype was given by the mean MGR (±the standard error of the mean) of strains of each of the classes grown on media with increasing fludioxonil concentrations (see Fig. 6 right panel). The result of this cross provides evidence of two major compensatory mutations in the evolved strain, rather than four as suggested by the MGR measurements (see Fig. 6) combined with the partial f-test. However, there could be a third and fourth compensatory mutation present since the variation in MGR between the progeny within the classes is relatively high (4.5 times higher than in the other MGR assays); possibly additional segregating loci have an effect on all four phenotypic classes. One of the recombinant classes has a higher fitness than both parental types; this represents the progeny with the second compensatory mutation only. The fact that the second compensatory mutation has a greater effect in the absence of the first compensatory mutation is a clear indication of interaction between both compensatory mutations.

Finally, we performed test crosses to verify the reconstruction we made. A recombinant suspected of only having the second compensatory mutation without the first was crossed with the nonevolved fungicide resistant (WG561). Here we found two classes among 40 progeny (data not shown), thus confirming the presence of just one major mutation. We also performed a cross between the nonevolved resistant (WG561) and the evolved strain (WG621) after 6 weeks. As can be seen in Fig. 6 left panel, by this stage only the first compensatory mutation is present; analysis of 34 progeny confirmed this (data not shown).

Sexual crosses between evolved strains

A sexual cross of two strains from the same condition was used to elucidate whether compensatory mutations that have arisen independently in replicate strains lead to even higher fitness when combined. For each condition (A–C) two strains were selected on the basis of their increased fitness compared to the ancestor. For condition A, WG619 and WG621 were crossed (see Figs 5 and 6), for condition B and C the strains crossed are shown in Fig. 5 (for condition B WG628 and WG629; for condition C WG630 and WG647). From each cross, 40 random progeny were isolated; the MGR of all progeny and the parents of the cross was measured in six-fold on fungicide free MM, results are shown in Fig. 7. We compared the MGR of the progeny with the MGR of the highest parent. In cross A no progeny had a higher MGR than the highest parent (t-test; t10 = 1.00, n.s.). In cross B, 7 progeny had a higher MGR than the highest parent (t-tests; d.f. = 10, P < 0.05). In cross C, 1 progeny had a higher MGR than the highest parent (t-test; t10 = 6.00, P < 0.001). It was not possible to detect segregating classes among the progeny.

image

Figure 7.  Of conditions A–C used in the evolution experiment (see Table 2), two evolved strains were crossed and the MGR of 40 progeny was measured together with the parents of the cross. Graphs show MGR ranges with the number of progeny. Arrows indicate the MGR class of the parents, *indicates the MGR class of the nonevolved resistant (WG561 and WG615).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Experiments were performed to study the development of fludioxonil resistance and subsequent compensatory evolution, which eliminates resistance associated fitness costs in A. nidulans. The prediction that resistance will carry a cost in the original fungicide-free environment was borne out since all resistant mutants suffered a fitness reduction. A cost of resistance under drug-free conditions is commonly observed (Cohan et al., 1994; Schrag & Perrot, 1996; Levin et al., 2000; Reynolds, 2000). The average MGR of 30 resistant mutants was reduced by 56% compared to the sensitive ancestor when growing on fungicide free medium. One resistant mutant (WG561) was selected. This strain and WG615, which is isogenic to WG561 for the resistance mutation and has equal MGR, were used in the evolution experiment (Table 1). Relative MGR was used as sole fitness measure in our experimental procedure; however, MGR is highly correlated with other relevant fitness measures (Pringle & Taylor, 2002).

The evolution experiment tested the prediction that a resistant strain will improve its fitness when cultured under conditions where resistance is not functional and causing negative pleiotropic effects on fitness. In all conditions tested, the mean MGR measured on medium without fungicide had increased over the 3000 cell cycles. Some of the evolved strains reached a higher fitness than the original fungicide sensitive strain (WG562) or its descendants that evolved for the same period of time (condition W), suggesting that there are higher peaks in the fitness landscape than the one occupied by this ancestor, which did not evolve to such a higher peak in the course of our experiment (condition W), probably not being able to pass through an adaptive valley.

We studied the process of compensatory adaptation under three different experimental conditions: one constant condition (condition A, never fungicide) and two varying conditions (condition B with diminishing presence of fungicide and condition C with alternating absence and presence of fungicide). In conditions B and C the evolving strains tended to show a more rapid adaptation (the slope of their fitness trajectories is steeper, see Fig. 4) and a higher final fitness than the evolving strains in condition A. What may have happened in terms of adaptive fitness topography is that in our experiment the continuing change of environment has prevented strains from getting ‘stalled’ on a relatively low fitness peak. In condition B where fungicide concentrations were high at first but declined in the course of the experiment, compensatory mutations only were evident when the fungicide concentration approached zero (this was at week 14, see Fig. 4 and Table 1). Possibly, compensatory mutations were only seen as escaping sectors (and thus selected) when the cost of resistance had become really high. In condition C, the alternating presence of the fungicide appears not to have hampered adaptation in the fungicide-free environment. Interestingly, these strains evolving into generalists by having to adapt to both the presence and absence of fungicide did as well as the specialists growing in condition A were the environment was always fungicide-free (Elena & Sanjuan, 2003).

It could be argued that the fitness gain we observed in our experimental evolution lines is not (only) a compensation of the cost of resistance, but possibly also of one or more deleterious mutations that might have hitchhiked to fixation along with the resistance mutation during the selection procedure used to obtain resistant strains. However, we consider this possibility as unlikely, because in crosses (both sexual and parasexual) involving the resistant strain we never found any indication of negative fitness effects segregating in the progeny, other than the fitness cost associated with the resistance mutation.

To study epistatic interactions, the trajectory of several evolved strains was analysed to investigate the process of compensatory evolution in detail. We used two methods to estimate the number of steps in fitness compensation. First, we looked for discrete and abrupt increases in MGR (Figs 5 and 6 left panel.). Next, we made a sexual analysis by crossing an evolved strain with the non-evolved resistant (see Table 5).

In evolved strain WG619 a sexual cross showed the presence of one compensating locus. Recombinants with the compensatory allele in the sensitive background showed that the compensatory allele has a general fitness enhancing effect, although this effect is greatest in the resistant background. We crossed evolved strain WG621 with the resistant ancestor WG561 and found two segregating compensatory mutations (see Fig. 6 right panel.). Here, we could study the interaction between compensatory mutations in the resistant background. Within the progeny there was a recombinant class having the second compensatory mutation without the first. The fitness of this recombinant was higher than all other classes from this cross. Apparently, the second compensatory mutation has a greater effect on fitness, but in this particular case some of its beneficial effect is removed by the first compensatory mutation. Our data show that in all strains studied in detail the first step in fitness improvement was larger than subsequent steps. This is compatible with the idea that mutations with large benefits are more likely in individuals that are distant from fitness optima and less likely in those close to optima (Fisher, 1930; Burch & Chao, 1999).

In only one (the cross in condition B) of the three crosses between evolved strains (see Fig. 7) we found clear evidence of recombinant progeny with a higher fitness than both parental classes, indicative of some additive effects on fitness between different compensatory mutations and/or synergistic epistasis. However, most progeny has a lower MGR than the parents, probably caused by a breaking-up of favorable combinations of compensatory mutations (antagonistic epistasis). We conclude that epistatic fitness interactions between compensatory mutations are common. Although our observations are relatively few, and genetically not well characterized, they are in accordance with the findings by Sanjuan et al. (2005) on a simple RNA genome, where antagonistic epistasis among beneficial mutations was the rule. Also Poon & Chao (2005) found epistatic interactions between deleterious mutations and their compensating beneficial mutations.

The MGR is an easy to assess and highly reproducible fitness measure for filamentous fungi (Pringle & Taylor, 2002). Other possible fitness measures would be total (a) sexual spore production or the total production of biomass from a fixed quantity of carbon source. These fitness measures are more difficult to assess and give a larger variation between replicate measurements than the MGR. De Visser et al. (1997) found that MGR and these other two fitness measurements are highly correlated when studying fitness in A. niger, a related filamentous fungus. By always including the same reference strain, MGRs of different experiments can be compared (Bruggeman et al., 2003). Under more natural conditions, MGR clearly will not be the only trait under selection and is likely to be an inadequate measure of lifetime reproductive success. However, we stress that in our experiment we choose to define adaptive success exclusively in terms of MGR and that therefore MGR represented total fitness in our experiment.

In evolving asexual populations in liquid media and without recombination, clonal interference (competitive elimination of mildly beneficial mutations by a superior beneficial mutation) is an important factor that affects the rate of adaptation (Gerrish & Lenski, 1998). In an expanding fungal colony, clonal interference may play a role in several ways. In contrast to global resource competition, which prevails in liquid cultures, physical interference and local competition are likely in a growing fungus. Beneficial mutations can become fixed in a sector, which is unlikely to affect the rest of the colony as a whole. Another beneficial mutation could start independently of the first mutation somewhere else in the mycelium. Our selection method implies maximal clonal interference; we select only one fit sector eliminating other potential sectors. In a separate study we have compared the selection method used here (transferring mycelial tissue from the fastest growing sector) with transferring a sample from all spores produced. The latter method allowed even faster compensatory evolution (Schoustra et al., 2005).

The main aim of our study has been to elucidate fundamental aspects of adaptation. However, this study has practical implications as well. Based on our experiments, the prediction can be made that the use of fludioxonil in agriculture can lead to resistant strains that by subsequent accumulation of compensatory mutations can achieve an equal or even higher mycelial growth rate than the original fungicide sensitive types. The fungicide used, fludioxonil, is widely used in agriculture to control emerging fungal infections on plants, grape vines in particular (Cabras & Angioni, 2000; Rosslenbroich & Stuebler, 2000; Verdisson et al., 2001). It is thought to act by interfering with sugar transport and sugar phosphorylation and by disordering membrane functions (Hilber et al., 1995; Ziogas & Kalamarakis, 2001). Our findings under the different environmental conditions that mimic possible agricultural conditions suggest that prolonged presence of fungicide followed by gradual degradation of the fludioxonil (condition B) or re-application of fludioxonil in a cycle (condition C), do not retard the emergence of resistant types with a high fitness in both environments.

In addition to addressing questions regarding the development of resistance and the subsequent compensatory evolution, we have shown that the filamentous fungus A. nidulans is a useful organism for the experimental study of evolutionary processes. The spatially structured growth results in isogenic sectors with a beneficial mutation. Sexual crosses are a valuable tool for genetic analysis of evolved strains. Future work with this fungal experimental system will involve comparisons between evolution in haploids and diploids, and between sexual and asexual reproduction, for which this system had promising possibilities.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We wish to thank Arjan de Visser for assistance with the statistical analyses and suggestions and comments. We also thank Véronique Perrot and Daniel Rozen for helpful discussions and/or comments. The Netherlands Organization for Scientific Research (NWO-ALW) has funded this project.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References