Multihost Experimental Evolution of the Pathogen Ralstonia
solanacearum Unveils Genes Involved in Adaptation to Plants
Alice Guidot,1,2 Wei Jiang,z,1,2 Jean-Baptiste Ferdy,3 Christophe Thebaud,3 Patrick Barberis,1,2
Jer^ome Gouzy,1,2 and Stephane Genin*,1,2
1
INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
3
UPS-CNRS-ENFA, Laboratoire E volution et Diversite Biologique (EDB), UMR5174, Universite Paul Sabatier, Toulouse, France
z
Present address: State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, College of Life Science and
Technology, Guangxi University, Nanning, Guangxi, China
*Corresponding author: E-mail: [email protected].
Associate editor: Eduardo Rocha
2
Abstract
Ralstonia solanacearum, the causal agent of a lethal bacterial wilt plant disease, infects an unusually wide range of hosts.
These hosts can further be split into plants where R. solanacearum is known to cause disease (original hosts) and those
where this bacterium can grow asymptomatically (distant hosts). Moreover, this pathogen is able to adapt to many plants
as supported by field observations reporting emergence of strains with enlarged pathogenic properties. To investigate the
genetic bases of host adaptation, we conducted evolution experiments by serial passages of a single clone of the pathogen
on three original and two distant hosts over 300 bacterial generations and then analyzed the whole-genome of nine
evolved clones. Phenotypic analysis of the evolved clones showed that the pathogen can increase its fitness on both
original and distant hosts although the magnitude of fitness increase was greater on distant hosts. Only few genomic
modifications were detected in evolved clones compared with the ancestor but parallel evolutionary changes in two genes
were observed in independent evolved populations. Independent mutations in the regulatory gene efpR were selected for
in three populations evolved on beans, a distant host. Reverse genetic approaches confirmed that these mutations were
associated with fitness gain on bean plants. This work provides a first step toward understanding the within-host
evolutionary dynamics of R. solanacearum during infection and identifying bacterial genes subjected to in planta selection. The discovery of EfpR as a determinant conditioning host adaptation of the pathogen illustrates how experimental
evolution coupled with whole-genome sequencing is a potent tool to identify novel molecular players involved in central
life-history traits.
Key words: in vivo experimental evolution, bacterial plant pathogen, Ralstonia solanacearum, tomato, fitness, virulence
evolution, full genome sequence, parallel evolution, reverse genetic.
Public, veterinary, and plant health are increasingly challenged
by rapidly evolving pathogens. Virulence and the ability of a
pathogen to cause an epidemic both rely on its ability to
adapt to its hosts. Unraveling the mechanisms that drive
pathogen adaptive evolution is therefore critical for dealing
with newly emerging infectious diseases (Griffin et al. 2004).
Although many theoretical studies have examined the role of
ecological and evolutionary factors on the evolution of parasite virulence, current understanding of the genetic basis of
pathogen adaptation, which is critical to disease emergence
through novel traits within the same or different hosts,
remains rudimentary (Griffin et al. 2004; Toft and
Andersson 2010; Gandon et al. 2013; Marvig et al. 2013;
Barroso-Batista et al. 2014). Pathogens adapt to their hosts
by the successive fixation of beneficial mutations, the dynamics of which may be influenced by several factors, including
the host characteristics (Gandon et al. 2013). This might be
especially important when pathogens colonize novel hosts,
where the distribution of the fitness effects of beneficial mutations likely depends upon differences in selection pressures
between novel and original hosts. When colonizing distant
hosts, pathogen populations are initially less adapted to the
hosts but typically they show quickly large fitness gains which
then tend to decelerate over time (Elena and Lenski 2003).
Current theory suggests that such effect might reflect two
types of processes predicted by Fisher’s geometrical model of
adaptation: 1) A greater proportion of mutations that are
beneficial in the novel host, and 2) selection which tends to
fix mutations with progressively smaller effects as the population gets closer to the new adaptive optimum (Orr 1998,
2003; MacLean et al. 2010; Tenaillon 2014).
Studies in experimental evolution provide a powerful
framework to investigate the evolutionary dynamics of pathogens by making it possible to test quantitative assumptions
and predictions from theoretical models of adaptive evolution on the fitness effects and the dynamics of new mutations
(Bataillon et al. 2012). Through replicated experiments in
ß The Author 2014. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: [email protected]
Mol. Biol. Evol. 31(11):2913–2928 doi:10.1093/molbev/msu229 Advance Access publication August 1, 2014
2913
Article
Introduction
Guidot et al. . doi:10.1093/molbev/msu229
controlled environments, it is possible to investigate how
bacterial phenotype and genotype evolve over hundreds
and even thousands of generations (Lenski et al. 2003;
Herring et al. 2006; Pelosi et al. 2006). The combination of
experimental evolution with whole-genome sequencing has
recently enabled the characterization of entire genomes of
evolved lineages and led to the identification of the specific
genetic changes that arose during the course of evolution,
providing explicit links between genotype and phenotype
(Brockhurst et al. 2011; Dettman et al. 2012; Kawecki et al.
2012). A major outcome of a classic long-term experimental
evolution with Escherichia coli was that large and rapid fitness
increases were seen in the early phase of adaptation even
though the genome seemed to accumulate changes in a regular manner for up to 20,000 generations (Barrick et al. 2009).
A new mutation that had a 10% advantage took approximately 250 generations to become fixed in the population
whereas a mutation with only a 0.1% advantage required
20,000 generations to reach that frequency. Such dynamics
were also observed in other studies (Brockhurst et al. 2011;
Dettman et al. 2012). An outstanding question arising
from these studies concerns the overall reproducibility of
evolutionary changes in independent experiments (Herring
et al. 2006; Barrick et al. 2009; Conrad et al. 2009; Stanek et al.
2009). Parallel evolution provides key evidence for adaptive
changes as it is unlikely that identical or similar changes
would become fixed in independent populations through
purely stochastic processes (Pelosi et al. 2006; Gandon
et al. 2013).
Most evolution experiments conducted to date under
laboratory conditions were performed in homogeneous
environments in controlled culture media that resulted in
constant selection pressure. Although the dynamics of adaptation has been studied in fluctuating environments through
in vitro experiments (Rainey and Travisano 1998), very few
evolution experiments have addressed this question using
microorganisms in changing and spatially structured environments. Pathogens are typically confronted to such heterogeneity, being able to reproduce in multicellular hosts with
different cell types and tissue specificities and to cope with
the multiple host responses encountered during biotic interactions. Some recent studies, mostly involving viruses (see,
e.g., Deardorff et al. 2011; Bedhomme et al. 2012), have
successfully used in vivo host serial passages experiments
(SPE) (Ebert 1998) of pathogens coupled to genome resequencing to identify the genes underlying adaptive
changes, highlighting the importance of such an approach
to investigate the mechanisms of evolution in host–pathogen
systems.
In this study, we performed in vivo host SPE of the bacterial
plant pathogen Ralstonia solanacearum coupled with wholegenome sequencing to investigate the genetic basis of adaptation in the pathogen on different host plants, including
both original and distant hosts. Ralstonia solanacearum is
the causal agent of bacterial wilt affecting an unusually
large host range of more than 250 plant species and is recognized as one of the most destructive bacterial plant diseases
(Hayward 1991; Mansfield et al. 2012). This soil borne
2914
MBE
pathogen enters plant roots, invades the xylem vessels, and
spreads rapidly to aerial parts of the plant through the vascular system. In original hosts, high population levels (up to
1010 colony-forming units per gram of fresh weight) are easily
reached within a few days and the vascular clogging induced
by this extensive colonization causes wilting symptoms
and eventual plant death (Peeters, Guidot et al. 2013).
Ralstonia solanacearum can also multiply asymptomatically
in multiple plant species, which are in this case designated as
distant hosts (Hayward 1991). The ability of this pathogen to
adapt to many host plants is also supported by field observations reporting emergence of strains with enlarged pathogenic properties and being able to colonize novel hosts
(Hayward 1991; Wicker et al. 2007, 2009). Ralstonia solanacearum strain GMI1000 is an attractive model to investigate
host adaptation as this strain infects several host species belonging to different plant families. Moreover, GMI1000 belongs to a phylogenetic group described as possessing a high
evolutionary potential as well as a high ability to recombine
(Wicker et al. 2012). In addition, the genome of GMI1000 is
completely sequenced (Salanoubat et al. 2002) and many
studies have identified the genetic factors involved in pathogenesis and host colonization (Genin 2010; Peeters, Carrère
et al. 2013). Essential to pathogenesis are the type III protein
secretion system and a complex virulence regulatory network
coordinating the expression of multiple additional virulence
factors (Genin and Denny 2012; Peeters, Carrère, et al. 2013;
Peeters, Guidot, et al. 2013).
Specifically, we investigated the genetic bases of host
adaptation through experimental evolution of R. solanacearum by starting replicated experiments using a single
clone of the model strain GMI1000 inoculated to three different original hosts and two different distant hosts. We then
performed SPE in which the pathogen was transferred from
one plant individual to another of the same species in order
to maintain pathogen populations on the same host over 300
bacterial generations. This enabled us to determine whether
clones from the SPE-derived populations acquired enhanced
fitness on a given host compared with the ancestor. For each
plant species, five independent lineages of derived clones
were generated by conducting five SPE in parallel to test
whether the same or similar changes were selected in the
different populations. Nine of the obtained laboratoryevolved clones on the tomato original host and on the
bean distant host were characterized through wholegenome sequencing in order to identify all genomic
modifications that arose in the evolved clones. We further
characterized some of these mutations and demonstrated
their contribution to the enhanced fitness in the experimental host. We then examined whether our findings were
consistent with the predictions from theoretical models
of adaptive evolution. We specifically examined whether
larger fitness gains were detected in the clones evolved
on distant hosts relative to those evolved on original hosts.
We were also able to compare the frequencies of beneficial
mutations, that is, contributing to the enhanced fitness
in plants, between populations evolved in either original or
distant hosts.
MBE
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
Results
SPE of R. solanacearum on Different Plants Can Select
for Evolved Clones with Enhanced In Planta Fitness
Generation of Populations Evolved In Planta for More than
300 Bacterial Generations
Experimental evolution of the R. solanacearum strain
GMI1000 was conducted by SPE on five different plant species
belonging to four botanical families, as outlined in figure 1.
Tomato, eggplant, and pelargonium were selected as original
hosts of strain GMI1000, whereas bean and cabbage were
considered as distant hosts. SPE were performed by force
inoculating the bacteria in the plant stem. This method
was chosen to be able to monitor the pathogen load at
each passage. The experimental protocol was designed so
that bacteria were continuously maintained into the plant
host by direct reinoculation of fresh plant material. Bacteria
were recovered by allowing them to diffuse in water from the
xylem vessels of stem sections (transversal and longitudinal)
Ancestral
clone
(GMI1000)
collected above the inoculation point. Collection time was
determined by the appearance of the first wilting symptoms
or executed 15 days after infection in the case of distant hosts.
In order to detect potential parallel evolution, that is, the
independent fixation of similar mutations in response to similar selective pressures, we conducted five biological SPE replicates for each of the plant species tested. Each plant species
thus generated five populations of derived clones (named
“A,” “B,” “C,” “D,” and “E”) from the common GMI1000 ancestral clone corresponding to five independent SPE on a
specific plant line.
A total of 26 SPE was sufficient to reach at least 300 bacterial generations for every plant except for cabbage which
required 36 SPE (table 1). This is explained by the fact that
bacteria multiplied less efficiently in cabbage, as evidenced by
the reduced generation number estimated per SPE (table 1).
Hence, the time course of the experiment to reach at least 300
generations was longer for distant hosts (bean and cabbage)
varying from 336 to 492 days than for original hosts (tomato,
Experimental Evolution
(Serial Passage Experiment)
SPE1
SPE2
SPEn
≈ 300
Bacterial
generations
Genome
sequencing of
evolved clones
Measuring in planta
adaptation
(competition assays with
the ancestral clone)
Derived
clones
FIG. 1. Experimental evolution scheme of a Ralstonia solanacearum clone through SPE on host plants. SPE were conducted to reach approximately
300 bacterial generations. In planta adaptation was measured using competition assays with the ancestral clone. Genomic resequencing was performed
for evolved clones displaying enhanced fitness on plants compared with the ancestral clone.
2915
MBE
Guidot et al. . doi:10.1093/molbev/msu229
Table 1. Number of SPE Necessary to Reach 300–350 Bacterial Generations Starting with a Clone of the Ralstonia solanacearum Strain GMI1000.
Plant Species
Tomato var. Marmande
Eggplant var. Zebrina
Pelargonium var. Maverick
Ecarlate
Cabbage var. Bartolo
Bean var. Blanc Precoce
Botanical Family
Original/Distant
Host of GMI1000
Original
Original
Original
Estimated no. of
Generations at Each
SPE (mean SD)
13 4
15 2
14 2
No. SPE to Reach
300–350
Generations
26
26
26
Solanaceae
Solanaceae
Geraniaceae
Brassicaceae
Fabaceae
Distant
Distant
10 4
12 4
36
26
Time to Reach
300–350
Generations (days)
200
137
184
492
336
NOTE.—At each SPE, the number of bacterial generations was estimated by comparing the effective number of cells injected into the healthy plant at SPEn and the effective
number of cells recovered from the infected plant at SPEn+1. SD, standard deviation.
eggplant, and pelargonium) varying from 137 to 200 days
(table 1).
Characterization of Phenotypic Changes
During the course of the SPE, no difference in plant symptoms
was observed with the evolved bacterial populations compared with the behavior of the ancestral clone. Although
bacterial wilt symptoms repeatedly appeared 5–7 days
postinoculation on the three original hosts, only rare and
very discrete symptoms were detected in distant hosts at
the 15 days postinoculation threshold, whatever the
number of SPE performed.
Phenotypic conversion (PC), a specific shift from a mucoid
to a nonmucoid colony morphology (Brumbley et al. 1993),
occurred in populations A and D evolved on tomato. The PCtype mutants were initially detected after 14–15 SPE and
represented 82% of the clones in population A and 23% in
population D. In the following SPE, PC-type mutants were
always detected in these two populations at frequencies varying from 35% to 91%. PC-type mutants were also detected,
but at lower frequencies in pelargonium (population C isolated at SPE 13) and in cabbage (population B isolated at SPE
9) but never in eggplant or bean.
The in planta fitness of each SPE-derived population was
measured relative to a gentamicin-resistant variant of the
ancestral clone using competition experiments. We calculated a competitive index (CI) that was used as a fitness estimator (see Materials and Methods). For each of the five host
plant species, we determined the CI for a total of 25 clones
representing five clones randomly isolated from each five
independent populations generated by SPE during at least
300 generations on the investigated plant species.
The CI obtained for each of the 125 tested derived clones
on the five different plant species is illustrated in figure 2.
Most of the derived clones had a CI value significantly superior to the CI value with the ancestral GMI1000 clone and
none of them had a CI inferior to the CI with GMI1000 whatever the investigated plant species. On the three original
hosts, serial passage transmission resulted in variable fitness
gains. On one hand, for tomato and eggplant, 24 and 22 of the
25 tested clones have a CI value significantly superior to the CI
value with GMI1000, respectively. On these two original hosts,
SPE thus largely resulted in enhanced fitness. The mean CI for
2916
these adapted clones was 4.2 (SE = 0.3) and 3.5 (SE = 0.2),
respectively. On the other hand, only 6 of the 25 analyzed
clones obtained after 300 generations on the pelargonium
original host had a CI significantly superior to the CI with
GMI1000. Remarkably, five of the six adapted clones belonged
to population E, indicating host adaptation only for this population, the mean CI for these five clones being 2.9 (SE = 0.2).
The lower adaptive response on pelargonium could be due to
various selective and demographic factors and suggests that
more SPE are probably required to evolve clones better
adapted to this host. Concerning the clones obtained on
the two distant hosts, 22 of 25 bean-clones and all 25 cabbage-clones had a CI value significantly superior to the CI
value with GMI1000, the mean CI for these clones being 6.3
(SE = 0.3) and 5.3 (SE = 0.2), respectively (fig. 2).
We then tested the effect of evolved clones within population and the effect of populations on CI variation for each
plant species using a mixed model (see Materials and
Methods). No or very little CI variation was detected
among clones within populations evolved on distant hosts,
whereas strong CI variation was found among clones within
populations evolved on the two original hosts tomato and
eggplant (table 2). This suggests that on average, populations
evolved on original hosts are more genetically diverse (with
the exception of pelargonium) than populations evolved on
distant hosts. Strong population effect was detected on CI
variation for eggplant, pelargonium, and cabbage (table 2).
For pelargonium and cabbage, this effect was mainly due to
one population (population E) being highly differentiated
from the other four populations. This differentiation could
be due either to a more rapidly accumulation of adaptive
mutations or to the fixation of different adaptive mutations
than in the other populations.
When considering the mean CI across all 25 derived clones
per plant species, no difference could be detected between CI
for clones evolved on the two distant hosts bean and cabbage
(Wilcoxon test, P = 0.089; table 3). However, these mean
CIs for clones evolved on distant hosts were significantly
superior to the mean CIs obtained for clones evolved on
the three original hosts (table 3). Altogether these data indicated that for both original and distant hosts the estimated
number of 300 bacterial generations in planta was sufficient
to select for evolved clones with enhanced in planta fitness,
except for certain SPE conducted on pelargonium.
MBE
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
Distant hosts
50
20
20
10
10
CI on Bean
50
5
2
1
5
2
1
0.5
0.5
0.2
0.2
50
50
20
20
CI on Cabbage
10
5
2
1
10
5
2
1
0.5
0.5
0.2
0.2
GMI1000
CI on Eggplant
CI on Tomato
Original hosts
CI on Pelargonium
50
a1 a2a3a4 a5 b1 b2b3b4 b5c1 c2 c3 c4 c5 d1d2d3d4 d5 e1e2e3e4 e5
Population
A
Population
B
Population
C
Population
D
Population
E
20
10
5
2
1
0.5
GMI1000
0.2
a1 a2a3a4 a5 b1 b2b3b4 b5c1 c2 c3 c4 c5 d1d2d3d4 d5 e1e2e3e4 e5
Population
A
Population
B
Population
C
Population
D
Population
E
FIG. 2. CI values of the derived and ancestral clones into the three original and two distant hosts used for SPE. The three original hosts of Ralstonia
solanacearum are tomato, eggplant, and pelargonium. The two distant hosts of R. solanacearum are bean and cabbage. For each plant species, the CI
was measured for 25 clones derived from the ancestral GMI1000 clone after SPE during approximately 300 bacterial generations into stem of the
investigated plant species. These 25 clones represent five clones randomly isolated from each five independent populations generated by SPE into plant
stems (populations A, B, C, D, and E).
Table 2. CI Variation among Populations and among Clones within Populations for Each Plant Species Used.
Plant Species
Original Host
Tomato
Model terms
Population
Clone (population)
F or LRT
0.89
24.5
P value
0.4881
0.0001
Eggplant
F or LRT
5.57
25.2
P value
0.0033
0.0001
Distant Host
Pelargonium
F or LRT
16.14
0
P value
0.0001
1
Bean
F or LRT
0.2
5
Cabbage
P value
0.9371
0.0253
F or LRT
21.44
0
P value
0.0001
1
NOTE.—The model random term “clone (population)” was tested with a likelihood ratio test (LRT) model with and without this effect. Gray shades indicate P values below 0.05.
2917
MBE
Guidot et al. . doi:10.1093/molbev/msu229
Table 3. Effect of the Plant Species on the Mean CI of the Derived Clones Obtained by SPE during 300 Generations on a Fixed Plant Species.
Original Host
Egplant
Pelargonium
Cabbage
Bean
Tomato
Mean CI = 4.00
SE = 0.26
6.80E-03
2.20E-16
1.93E-08
8.92E-04
Distant Host
Eggplant
Mean CI = 3.22
SE = 0.22
Pelargonium
Mean CI = 1.56
SE = 0.08
Cabbage
Mean CI = 5.26
SE = 0.24
3.24E-16
7.96E-16
3.94E-08
2.20E-16
2.20E-16
0.08949
Bean
Mean CI = 5.98
SE = 0.33
NOTE.—Mean CI values were compared using a Wilcoxon test and the table gives the obtained P value. A total of 25 derived clones per plant species were analyzed representing
five clones randomly collected from each five populations generated by independent SPE during 300 generations on a specific plant species. Significant differences between mean
CI are highlighted in gray. SE, standard error.
Enhanced In Planta Fitness Is Associated with
Improved In Vitro Growth for Some but Not
All Evolved Clones
As the evolved clones were more competitive than the wildtype strain after mixed infections in planta, we tested whether
this behavior was associated with improved in vitro growth in
minimal medium. For that purpose, we analyzed the in vitro
growth kinetics of the 25 tomato-clones and 25 bean-clones
characterized in mixed infections in planta as described
above. The in vitro growth curves of these 50 clones were
determined during growth in minimal medium supplemented with glucose by monitoring OD600 over 30 h
(fig. 3). Significant differences in growth rates and densities
in stationary phase were observed between the evolved
clones and the ancestral clone (fig. 3B and C). The results
shown in figure 3B and C revealed that the in vitro growth
rates and the densities in stationary phase of all but one
(b3tomato) of the evolved clones were either similar or superior
to their ancestor. Eleven and fourteen clones evolved on
tomato and bean, respectively, exhibited a growth rate significantly higher (varying between 0.24 and 0.33 h1) than the
ancestral clone (Student t-test, mean = 0.22, P < 0.05; fig. 3B).
Improved growth rates in minimal medium were noticeable
in several populations such as Ctomato, Cbean, Dbean, and Ebean
as all or four of the five tested clones acquired this property.
Five and fifteen clones evolved on tomato and bean, respectively, exhibited a density at the entry of stationary phase
significantly higher (OD600 varying between 1.82 and 2.22)
than the ancestral clone (Student t-test, mean = 1.61,
P < 0.05; fig. 3C). Increased cell densities in stationary phase
were observed for populations Btomato, Cbean, Dbean, and Ebean.
The only evolved clone showing a reduced in vitro growth
rate and a reduced density in stationary phase (fig. 3) compared with the ancestor was the b3tomato clone, which was
also the only one showing a reduced in planta fitness (fig. 2).
However, when comparing the in planta fitness value (mean
CI) with the in vitro growth rates or with cell densities when
reaching the stationary phase of the 25 tomato-clones and 25
bean-clones, no direct correlation was found (Kendall’s rank
correlation, P 4 0.05).
Genotypic Analysis of Experimentally Evolved Clones
To determine the nature of genetic events that occurred
during SPE of strain GMI1000 on plants, we sequenced the
2918
entire genome of nine evolved clones that showed enhanced
fitness on their experimental host: Three on the tomato original host (a1tomato, b2tomato, and e1tomato), and six on the bean
distant host (a2bean, c1bean, c3bean, c4bean, d2bean, and e2bean).
The ancestral GMI1000 clone was also resequenced. Sequence
data were mapped to the GMI1000 reference genome and
analyzed using a new protocol (see Materials and Methods).
All detected mutations were validated through polymerase
chain reaction (PCR) amplification and Sanger sequencing,
whereas large deletions were confirmed by comparative
genomic hybridization on the GMI1000 microarray.
Table 4 lists all the genetic modifications identified in the
sequenced clones compared with the GMI1000 reference
genome. The amount of genetic changes was very limited
as no more than three mutations per sequenced clones
were detected. Changes were either deletions (1–33 kb),
single base substitutions, or IS (insertion sequence) movements. These mutations modified coding sequences (CDS)
or intergenic regions; the latter was observed twice upstream
RSc1976 and RSc2508 genes. It is likely that such events
could affect the expression of the downstream gene(s). A
large deletion of 33 kb was detected in the a1tomato clone.
This deleted region is surrounded by two IS transposase
elements and contains 27 genes, including the pme gene required for a pectin degradation and other genes of undefined
function. IS movement was detected in three distinct clones;
two of them (a1tomato and e1tomato) which evolved in tomato
but originated from independent populations (A and E) were
affected by the same IS movement, an insertion of the ISRso9
120 nucleotides before the start codon of the RSc2508 gene.
In addition, SNPs (single nucleotide polymorphisms) were
detected both within gene sequences and intergenic regions.
Remarkably, most of the SNPs detected in CDS were nonsynonymous mutations (eight of nine) (table 4).
Three major classes of genes appeared to undergo genetic
changes in the evolved clones. A first class corresponds to
genes involved in various metabolic processes, a second one
to genes encoding for transcriptional regulatory proteins,
and a third to genes encoding for hypothetical proteins of
unknown function. Only one of all mutations altered a gene
formerly described as a virulence gene in R. solanacearum,
that is, the phcS regulatory sensor kinase-encoding gene.
Parallel evolution was observed for two genes: 1) The
RSc2508 gene, encoding a protein of unknown function,
whose expression could be affected by an IS insertion
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
MBE
FIG. 3. In vitro growth in minimal medium of the evolved clones isolated after 26 SPE on tomato and on bean. (A) In vitro growth curves. (B) In vitro
growth rates. (C) Final OD in stationary phase. For each plant species, in vitro growth was investigated for 25 evolved clones representing five clones
randomly isolated from each five independent populations generated by SPE into plant stems (populations A, B, C, D, and E). In vitro growth was
investigated during culture in minimal medium supplemented with glucose at a 20 mM final concentration by monitoring OD600 over 30 h. The growth
rates and the final OD of the evolved clones were compared with the growth rate and final OD of the ancestral GMI1000 clone using a Student t-test
(*P < 0.05, **P < 0.01, ***P < 0.001).
2919
2920
27 genes
RSp0128–0154
Del 33 kb
C 80 T
T 26 M
b2
Tomato-Evolved Clones
C 976 G
R 326 G
ISRso9
120
a1
ISRso9
120
e1
b
C 163 T
R 55 C
A 394 ()
a2
G 88a A
c1
C 278 A
P 93 Q
G 88a A
c3
C 278 A
P 93 Q
G 88a A
c4
C 278 A
P 93 Q
Bean-Evolved Clones
ISRso9
1,129
d2
C4T
L2L
Del 21 nt
155–175
e2
G 145 A
D 59 N
NOTE.—For each gene, the modification type is indicated. For SNPs, the original nucleotide is followed by the position of the SNP in the coding sequence of the gene and by the new nucleotide. The associated modification in the protein sequence is
indicated in bold. ISRso9 indicates an insertion of the transposable element ISRso9; the number underneath ISRso9 indicates the position of the IS insertion upstream the start codon of the gene or in the coding sequence of the gene.
a
Mutation upstream the start codon of the gene.
b
Single nucleotide deletion; Del, deletion; nt, nucleotides.
Galactose mutarotase-like protein
RSc3437
Transketolase
tktA
RSc2750
Transcription regulator protein
Histidine kinase two-component
regulator
phcS
RSc2736
RSc3062
Amidophosphoribosyltransferase
Hypothetical protein
Hypothetical protein
Description
Transcription regulator protein
Gene
purF
Name
efpR
RSc1976
RSc2508
RSc1622
ID
RSc1097
Table 4. List of Validated Genetic Modifications in the Ralstonia Solanacearum-Evolved Clones.
Guidot et al. . doi:10.1093/molbev/msu229
MBE
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
MBE
upstream the gene in a1tomato and e1tomato clones as described above and by an SNP within the CDS detected in
a2bean clone; and 2) the RSc1097 regulatory gene altered by
three different SNPs in clones originating from three independent SPE populations from bean (populations C, D, and E)
(table 4).
Remarkably, these competition assays revealed that the
ancestral GMI1000 strain could not be detected anymore in
the bacterial populations in several cases after the second
passage. This occurred, for example, after SPE2 for GRS706
(complementation with c1bean allele)/GMI1000 competition
assays in two of the five plants tested, and after SPE3 in all the
five plants tested (fig. 4), thus indicating that after approximately 36 bacterial generations in planta the strain carrying
this c1bean allele (GRS706) outcompeted its ancestral form.
Altogether these results demonstrate that loss of function of
the RSc1097 gene or occurrence of the mutations selected
during the experimental evolution in this regulatory gene led
to a better adaptation of the parental strain to the bean
host. The RSc1097 regulatory gene was hereafter named
efpR (for mutant with enhanced fitness in plants).
Evidence That Mutations in the RSc1097 Gene
Improve R. solanacearum Fitness In Planta
To assess the fitness effects of the mutations discovered in the
RSc1097 gene, we first generated a deletion mutant of this
gene in the GMI1000 ancestor. The three allelic forms of
the RSc1097 gene detected in the evolved clones c1bean,
d2bean and e2bean, as well as the ancestral (wild-type)
RSc1097 gene copy as control, were PCR-amplified and
then used to complement the RSc1097-deleted mutant
(see Materials and Methods) (table 5).
The fitness of the RSc1097 strain and the various complemented derivatives was monitored in the same conditions
that prevailed during the evolution experiment. All the strains
were used in mixed inoculum with the GMI1000 wild-type
strain and stem-inoculated for competition assays in bean.
The CI was determined after each passage on plant; figure 4
presents the results obtained during the first three SPE. After
SPE1, probably because of the detection limit, the mean CI
of the deleted mutant (GRS704) and of two of the three
complemented derivatives with evolved alleles (c1bean and
d2bean RSc1097 alleles) were not significantly different from
the mean CI with GRS705 (RSc1097 complemented by the
wild-type RSc1097 allele; table 5) (fig. 4). Only the GRS708
strain (RSc1097 complemented by the e2bean RSc1097 allele;
table 5) had a mean CI in bean of 2.97 significantly superior to
the mean CI with GRS705 (Wilcoxon test, P = 0.015) (fig. 4).
After SPE2, RSc1097 mutant as well as the mutant strains
carrying the complementing c1bean, d2bean, and e2bean alleles
displayed a significant gain of fitness compared with the wildtype complementing allele which remained close to one
(fig. 4). This result demonstrated that both the deletion
of the RSc1097 gene and its complementation with the
RSc1097 evolved alleles improve competitiveness in bean
relative to the ancestral GMI1000 strain. These results were
confirmed after SPE3 (fig. 4).
Table 5. List of Strains Constructed in This Study.
Strain
GRS540
GRS704
GRS705
GRS706
GRS707
GRS708
Genotype
GMI1000 derivative, GmR
GMI1000 derivative, efpR, SpR
GRS704:: efpR-GMI1000, SpR, GmR
GRS704:: efpR-c1bean, SpR, GmR
GRS704:: efpR-d2bean, SpR, GmR
GRS704:: efpR-e2bean, SpR, GmR
NOTE.—efpR, RSc1097 gene; SpR, GmR indicate resistances to spectinomycin and gentamicin, respectively. efpR-GMI1000, efpR-c1bean,
efpR-d2bean and efpR-e2bean are efpR alleles in the ancestral
GMI1000 clone and in the c1bean, d2bean and e2bean clones evolved
on bean during approximately 300 generations, respectively.
Mutations in efpR (RSc1097) Are Rapidly Fixed
in Independent Populations Adapting on Bean
As the mutations in the efpR gene appeared to be beneficial
on bean, we investigated the temporal dynamics of adaptation in the bacterial populations. The mutation frequency was
determined in each of the five independent populations
that evolved on bean using a PCR/sequencing strategy on
96 clones randomly isolated from each population. After 26
SPE, which corresponds to an estimated number of at least
300 bacterial generations, 100% of the clones from populations A and B carried the ancestral efpR allele whereas 100%,
100% and 74% of clones from populations C, D and E carried
the efpR c1bean, d2bean and e2bean evolved alleles, respectively
(table 6). This indicated that these three alleles are clearly
predominant in the population. This result was already suspected for population C, as all three sequenced clones (clones
c1bean, c3bean, and c4bean) contained exactly the same SNP
in the efpR gene (table 4).
A similar analysis was performed after 16 SPE (~180 generations), 18 SPE (~200 generations), 20 SPE (~230 generations), and 23 SPE (~260 generations) in order to determine
the temporal dynamics of allele frequency change over time.
In populations A and B, the absence of mutational events in
the efpR-coding sequence was confirmed at these different
tested steps. In populations C and D, although no clone
showed any mutation in the efpR gene after 18 SPE in bean,
100% of the clones in bean showed mutations in the efpR
gene after 20 SPE (table 6). In population E, the efpR e2bean
allele was detected in 2% of the clones after 23 SPE in bean,
but the frequency rose to 74% after 26 SPE in bean (table 6).
These results show that the mutations in efpR have appeared
after approximately 200–250 generations in bean and then
quickly reached fixation (within two SPE), in accordance
with the gain in fitness conferred by the resulting alleles as
previously described (fig. 4).
Occurrence of mutations in efpR in populations evolved in
plant species different from bean was also investigated using
a similar PCR/sequencing analysis on 96 clones randomly isolated from each population. This analysis was conducted
on three independent populations propagated on the
tomato original host and on three independent populations
2921
MBE
Guidot et al. . doi:10.1093/molbev/msu229
FIG. 4. CI values of the efpR mutant strains after 1, 2, and 3 passages into bean plants. (A) SPE scheme starting with a mixed inoculum of the efpR
mutant and the wild-type GMI1000 strain in the same proportion. (B) CI values obtained after 1, 2, and 3 passages into bean plants. The ratio above the
boxes indicates the proportion of plants in which the wild-type GMI1000 strain disappeared. The CI values were compared with the CI obtained with
the GRS705 control mutant using a Wilcoxon test (*P < 0.05, **P < 0.01).
Table 6. Chronological Appearance of the Mutations in the efpR Gene for Each of the Five Populations
Evolved on Bean by SPE.
SPE
16
18
20
23
26
Frequency of efpR Mutations in Bean Populations
Estimated No. of Generations
180
200
230
260
300
A
0
0
0
0
0
B
0
0
0
0
0
C
0
0
100
100
100
D
0
0
100
100
100
E
0
0
0
2
74
NOTE.—Mutations in the efpR gene have been investigated for a total of 96 clones randomly collected in each of the five “bean
populations” after 16, 18, 20, 23, and 26 SPE. The table gives the frequency of the mutations (%) in the efpR gene for each population.
2922
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
MBE
propagated on the cabbage distant host after 26 SPE.
However, only one nonsynonymous change (G290A) in the
efpR gene sequence was detected at a very low frequency
(0.01) in only one of the six investigated populations, population E propagated on the tomato original host.
Discussion
potential of adaptation in distant, asymptomatic hosts
raises important questions about the evolutionary dynamics
of the bacterial wilt disease and the control policies to limit it.
Disease management practices based on cultural rotations
of “tolerant” hosts such as weeds (Lopez and Biosca 2005)
could paradoxically result in pathogen reservoirs with
enhanced adaptive potential.
SPE on Different Hosts Consistently Lead to
Changes in R. solanacearum Mean Fitness
Beneficial Mutations Accumulated during Adaptation
We used experimental evolution combined with wholegenome sequencing to identify the molecular determinants
that drive adaptation in the R. solanacearum pathogen when
grown on different host plant species, including both original
and distant hosts. A vast majority (88–100%) of the evolved
clones had acquired at generation 300 a significant fitness
advantage over the ancestral clone after serial passages on
two of the three original hosts (tomato and eggplant) and on
the two distant hosts (bean and cabbage).
Bacterial fitness was defined in this study by a CI value
measured for each experimentally evolved clone using an in
planta competition assay with the ancestral clone. This fitness
value is therefore an indicator of the bacterial multiplication
rate in the host plant, which is not correlated to disease
symptom production as individuals with significant enhanced
fitness did not induce wilting symptoms faster. A previous
study also reported that bacterial wilt symptoms are not
always correlated with the pathogen multiplication rate
(Angot et al. 2006). Because the stem-inoculation procedure
bypasses the natural infection route of R. solanacearum
through the plant roots, fitness as defined in our study applies
to the pathogen fitness during xylem colonization. Parts of
the R. solanacearum lifecycle such as survival in soil or root
infection were not considered in our experimental design
in order to maintain a uniform and continuous selective environment within each replicated experiment. As the fitness
value ensues from the bacterial multiplication rate in
planta, we investigated whether the competitive advantage
of evolved clones resulted from improved basal metabolic
capacities. This is indeed the case for several of these clones
as their in vitro growth rates and densities at stationary
phase in minimal medium were significantly higher than
the wild-type strain. This is in agreement with the general
view that minimal medium conditions partly mimic the
apoplastic or vascular plant environment (Genin 2010) and
suggests that the growth advantage of evolved clones during
SPE was dependent upon the host plant environment.
The R. solanacearum SPE showed that host species identity
influenced the process of fitness gain of the selected clones
during 300 generations of experimental evolution. Clones
evolved on distant hosts (bean and cabbage) displayed the
highest fitness gain with a CI around 5.5 in competition assays
with the ancestral clone. These fitness gains were significantly
higher to those measured for clones evolved on original hosts
(tomato, eggplant, and pelargonium). These findings are consistent with the theoretical prediction that larger fitness gains
are more likely to be observed in clones colonizing distant
hosts than in clones colonizing original hosts. This high
No more than one to three different genomic alterations were
detected per sequenced evolved clone compared with the
common ancestral clone. This is in full agreement with
the average number of SNP substitutions found in other
experimental evolution studies with bacterial and yeast species (Dettman et al. 2012). Very limited genetic changes were
also observed in a study of Pseudomonas aeruginosa adaptation to the airways of cystic fibrosis patients (Smith et al.
2006). Recently, an evolution experiment aiming to convert
the R. solanacearum strain GMI1000 into a legume-nodulating symbiont concluded after the sequencing of evolved
clones that 21–38 SNPs were generated for an estimated
maximum of 25 generations in the plant environment
(Marchetti et al. 2010). However, this unusually high mutation
rate probably does not result from a specific property of the R.
solanacearum genome nor from the plant environment but
rather from introducing a 550-kb symbiotic plasmid in the
GMI1000 strain (Marchetti et al. 2010). It should be noted
that the 33-kb deletion detected in the a1tomato clone was
perfectly similar to a deletion event also observed by
Marchetti et al. (2010) in three different clones in their evolution experiment on Mimosa pudica. As this region is flanked
by IS sequences, it might correspond to an unstable genomic
island.
Two lines of arguments suggest that the mutations detected in the R. solanacearum-evolved clones were strongly
selected to fixation. First, all but one SNP identified in CDS led
to nonsynonymous amino acid change, a pattern which likely
reflects the role of positive selection rather than random drift
under which a greater proportion of synonymous mutations
would be expected. Such biases favoring nonsynonymous
SNPs were also observed in evolution experiments of E. coli
in various selective environments (Shendure et al. 2005;
Barrick et al. 2009; Conrad et al. 2009; Wang et al. 2010).
Second, if the changes that we observed had been driven
by stochastic processes, we would not expect to see mutations in the same genes in clones originating from independently evolving populations. Here, two of the eight genes
were altered in more than two evolved clones, and importantly these changes corresponded to different mutation
types. Such parallel changes provide strong evidence for the
action of natural selection at these genes.
Classical examples of genes subjected to selection in plant–
pathogen interactions involve pathogenicity determinants,
with selection favoring mutants with the ability to avoid
the pathogen recognition in resistant plants (Kearney et al.
1988; Joosten et al. 1994). Evolution experiments of the pathogenic bacteria P. syringae pv. phaseolicola and Xanthomonas
2923
MBE
Guidot et al. . doi:10.1093/molbev/msu229
citri subsp. citri on resistant plants also revealed that the fixed
mutations were biased toward type III secretion system effectors that condition the compatibility of the plant/bacterium
interaction (Lovell et al. 2011; Trivedi and Wang 2013).
Among the genes targeted by mutations in our in planta
experimental evolution system, only one was previously
known to be involved in virulence, that is, the phcS sensor
kinase-encoding gene (Clough et al. 1997). PhcS composes
with PhcR a two-component regulatory system that responds
to threshold concentrations of the atypical quorum sensing
signal 3-hydroxy palmitic acid methyl ester by elevating the
level of functional PhcA, a master regulator of virulence in
R. solanacearum. PhcA directly or indirectly controls several
other regulatory genes and expression of functions, such
as bacterial motility, production of plant cell wall-degrading
enzymes, exopolysaccharides, and the type III secretion
system, some of which being essential for pathogenesis
(Genin and Denny 2012). It is not known whether the
nonsynonymous mutation in phcS leads to up- or downregulation of PhcA but it may be relevant that the pleiotropic
nature of this regulator controlling positively and negatively a
wide number of target genes could impact many phenotypic
traits of R. solanacearum in planta resulting in enhanced
fitness. More generally, transcription regulatory genes were
predominant targets in our in planta selection process as
three of the eight genes altered by SNPs belong to this class
whereas only 7% of the GMI1000 genes are predicted to
encode transcription regulators (Salanoubat et al. 2002).
This result supports the view that regulatory networks are
critical for bacterial adaptation across both physiological
and evolutionary timescales (Philippe et al. 2007).
In this study, the efpR regulatory gene was altered by three
different SNPs in clones originating from three independent
SPE populations from the bean distant host. In order to test
the hypothesis that changes in the efpR gene were associated
with the adaptive process, the mutations selected through
experimental evolution were reintroduced into the ancestral
clone carrying a deletion of efpR. This allowed us to demonstrate that the resulting strains harboring efpR-evolved alleles had acquired a fitness advantage compared with the
wild-type during infection of bean plants. Indeed, after only
three SPE, the efpR alleles allowed the recipient strains to
outcompete the wild-type ancestor strain in bean plants.
Because the deletion of the efpR gene was shown to confer
a competitive advantage over the wild-type strain in planta,
it is likely that the efpR SNPs identified in bean-evolved lineages led to a gene loss of function. Our data also show that
these mutations in efpR were rapidly fixed in the populations,
indicating that they should provide a significant fitness gain
inside the host plant. An additional SNP in efpR was detected
at a very low frequency in only one of the three investigated
populations evolved on the tomato original host but no mutations in efpR could be detected in the three investigated
populations evolved on the cabbage distant host. Further
studies will aim to determine whether the adaptive efpR
mutations are bean-specific or confer a fitness advantage
in other plants; on the other hand, the adaptive contribution
2924
of the additional efpR allele identified in the Etomato population remains to be investigated.
The efpR gene encodes a transcription regulator protein
of 113 amino acids, which shares similarity with XRE
(xenobiotic response element) family helix-turn-helix
proteins. Remarkably, the efpR gene is found in all the
R. solanacearum strains sequenced to date and it encodes
a conserved protein as there are only 8.8% variable amino
acids among sequenced R. solanacearum genomes representative of the diversity in this species complex (Salanoubat
et al. 2002; Remenant et al. 2010, 2011, 2012; Xu et al. 2011)
(supplementary fig. S1, Supplementary Material online).
The wide distribution and the low sequence polymorphism
of the efpR gene in the R. solanacearum species complex
suggest that this regulatory gene plays an important role in
some fundamental aspects of the life-cycle of this microorganism (Lefeuvre et al. 2013). For example, this EfpR protein
might be important for infection through the roots or for
survival in the soil, two steps of the R. solanacearum life-cycle
which were not integrated in our experimental evolution
scheme. The three efpR SNPs selected by in planta experimental evolution only targeted conserved nucleotides
(supplementary fig. S1, Supplementary Material online).
One SNP in population E occurred in the helix-turn-helix
domain (leading to a D49N change), whereas another SNP
in population C led to a P93Q change (supplementary fig. S1,
Supplementary Material online). The SNP in population D
led to a synonymous change in the second codon of the
gene, thus generating a new codon which is poorly used
based on the R. solanacearum codon usage (Nakamura
et al. 2000; Plotkin and Kudla 2011). We hypothesize that
this change could affect translation efficiency, protein turnover, or maturation processing as the influence of the second
codon on gene functionality was already established
(Tang et al. 2010).
Conclusions
This work provides a first step toward understanding the
within-host evolutionary dynamics of R. solanacearum
during infection and identifying bacterial genes subjected to
in planta selection. In vivo evolution experiments showed
that the pathogen can increase its fitness on both original
and distant hosts although the magnitude of fitness
increase was greater on distant hosts, which is consistent
with the predictions from theoretical models of adaptive
evolution. Although this approach could not address the
R. solanacearum complete lifecycle, it led to the identification
of a list of candidate genes that may contribute to adaptation
to host plants. The finding that alteration of the efpR gene
confers a fitness gain on the bean distant host demonstrates
that such experimental evolution approach coupled with
whole-genome sequencing is a potent tool to identify
novel molecular players involved in central life-history traits.
Future investigation of evolved R. solanacearum clones
should broaden our vision of the functions and genes
contributing to the pathogen’s fitness inside its host(s)
far beyond the current knowledge of pathogenicity
determinants.
MBE
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
Materials and Methods
Bacterial Strains, Plant Material, and Growth
Conditions
The evolution experiment was investigated with a clone from
the R. solanacearum model strain GMI1000 (Salanoubat et al.
2002). The list of GMI1000 derivatives used in this work
is given in table 5. The strain GRS540 that is a gentamicinresistant derivative of GMI1000 was generated by stable
integration of plasmid pAGS1 (a pRCG-GWY-based plasmid
developed as a R. solanacearum chromosomal insertion
system [Monteiro et al. 2012]) after linearization by ScaI
and natural transformation as previously described
(Gonzalez et al. 2011). Ralstonia solanacearum strains were
grown at 28 C in complete BG medium (Plener et al. 2010).
When needed, antibiotics were added to the medium at
the following final concentrations (mg l1): gentamicin, 10;
spectinomycin, 40.
The plants used in this study were three original host plant
species of strain GMI1000: Tomato (Solanum lycopersicum
var. Super Marmande), eggplant (S. melongena var. Zebrina)
and pelargonium (Geranium sanguineum var. Maverick
Ecarlate), and two distant plant species: Cabbage (Brassica
oleracea var. Bartolo) and bean (Phaseolus vulgaris var.
Blanc Precoce) (table 1). Four- to five-week-old plants
(for tomato, eggplant, pelargonium, and cabbage) and 15to 20-day-old bean plants were used for the inoculations.
Plant experiments were conducted in a growth chamber
under the following conditions: 75% humidity, 12 h light
28 C, and 12 h darkness 27 C.
Evolution Experiment
A single colony from strain GMI1000 was grown overnight
in liquid complete BG medium at 28 C under agitation at
180 rpm and 10 ml of a suspension at 106 CFU/ml was injected
into the stem of five plants for each of the five selected species
using a microsyringe. SPE on each plant were then conducted
as follows. At each SPE, the bacterial populations that developed into the plants were recovered as soon as the first
wilting symptoms appeared (minimum 5 days postinoculation) for original hosts or maximum 15 days postinoculation
for distant hosts. In order to recover bacterial populations
from plants, the whole stem (with petioles) was cut at 1 cm
above the inoculation point, truncated into 1 cm segments,
and then all segments were incubated in 3 ml of sterile water
at room temperature for 30 min so that bacteria naturally
diffuse from stems and petioles to sterile water. Ten microliters of the recovered bacterial suspension diluted to approximately 106 CFU/ml was directly injected into the stem of
a healthy plant of the same species. Serial dilutions of the
recovered bacterial suspension were plated onto BG medium
and incubated at 28 C in order to determine the effective
output concentration. At each SPE, the number of bacterial
generations was estimated by comparing the effective
number of cells injected into the healthy plant at SPEn and
the effective number of cells recovered from the infected
plant at SPEn+1. However, this total number of generations
is underestimated as the bacterial cells were not recovered
from the whole plant. The population size is therefore likely
larger and underestimated in our calculation. Finally, 1 ml of
the recovered bacterial suspension was also stored in glycerol
at 80 C. The inoculated plants were incubated in a growth
chamber under the previously described conditions.
Bacterial Competition Assays In Planta
The method used to estimate in planta fitness of
R. solanacearum clones derives from a previously described
method (Macho et al. 2010) based on competition assays
between a tested clone and strain GRS540, a gentamicinmarked variant of GMI1000. Competition assays were
performed using identical conditions as for SPE. Stems were
inoculated by injection of 10 ml of a 106 CFU/ml bacterial
suspension, containing equal CFU of the derived clone and
the GRS540 clone. Serial dilutions of the inoculum were
plated onto BG medium with and without gentamicin
using an easySpiral machine (Interscience, France) at 5 and
15 dpi for original and distant plants, respectively, in order
to determine the bacterial ratio between the derived and
GRS540 clones recovered from the host tissues.
The CI was defined as the derived/ancestral-clone ratio
within the output sample divided by the ratio in the corresponding inoculum (Baumler et al. 1997). As expected, when
the ancestral GMI1000 clone was coinoculated with its
genetically marked variant (GRS540) in the same proportion
into the stem of the five different plant species, the obtained
CI was not significantly different from one (Wilcoxon test,
at least ten replicates, P 4 0.05). The 25 derived clones
evolved on a same plant species were analyzed at the same
time on a same plant lot. Three to five experiments were
conducted to determine the CI and three plants per investigated clone were used in each experiment.
A similar protocol was used to determine the in planta
fitness of the five different GMI1000 mutants constructed in
this work (GRS704–GRS708). The five mutants were analyzed
at the same time on a same plant lot. Five replicates were
performed (five plants per mutant) in each experiment.
Monitoring In Vitro Growth Rates
The growth rates of evolved clones were measured in cultures
growing in MP minimal medium supplemented with glucose
at a 20 mM final concentration (Plener et al. 2010). Overnight
cultures grown in this MP-glucose medium at 28 C with
180 rpm shaking were used to inoculate three replicates of
200 ml of fresh MP-glucose medium with an initial OD600 at
0.05. Bacterial growth was performed in 96-well plates
and monitored using the FLUOstar Omega microplates
reader (BMG Labtech). Measures of OD600 was performed
every 5 min.
In vitro growth rates were determined using a model in
the R statistical software as follows: Optical density (OD) data,
which we used to infer in vitro kinetics, were first log-transformed and smoothed using a moving average technique,
each average being computed over 20 consecutive measurements. We then identified the period of 20 measurements
2925
MBE
Guidot et al. . doi:10.1093/molbev/msu229
corresponding to the maximum increase in OD, and estimated the maximum rate of increase as the slope of a
linear regression performed on these 20 measurements.
Genome Resequencing and Detection of Mutations
All genomic sequences were produced using the Illumina
technology. Paired-end libraries were prepared following the
protocol recommended by Illumina Inc. (http://www.
illumina.com). More than 12 million paired-end reads
(2 100 bp) were generated leading to an approximately
100 total coverage of the reference genome. The VAAL
software has been used to detect SNP and small InDels
(Nusbaum et al. 2009). The IGV (Integrative Genomics
Viewer) software was used to visually validate all detected
polymorphisms. In order to identify all potential false positive
polymorphisms, the genome of the reference GMI1000 strain
was resequenced in this work. The obtained sequence was
compared with the GMI1000 genomic sequence produced in
2002 (Salanoubat et al. 2002) and only 32 sequence differences were detected and reported in supplementary table S1,
Supplementary Material online.
In order to sensitively and specifically identify all modification types, an original bioinformatics protocol has been designed. In a first step, kmer libraries (counts of occurrences of
words of length k, with k = 15 nt, minimum occurrences = 10)
were built for all derived clones. In addition, the list of kmers
present in the ancestral genome was computed both using
the Illumina reads generated in this study and using the set of
kmers found in the GMI1000 reference genome. Then, for
each derived clones we identified read pair containing
kmers that were frequents in the clone data set but absents
(i.e., very rare, count < 10) in the ancestral genome. Read pairs
showing an original kmer composition based on the comparison counts from derived clones and the reference data were
extracted and assembled using the accurate assembler programs Cap3 and Mira. Finally, longer and reliable consensus
sequences obtained were aligned on the reference genome in
order to analyze the genome modifications.
All detected polymorphisms were checked by PCR amplification and sequencing with the Sanger technology using the
primers reported in supplementary table S2, Supplementary
Material online. Large deletions were confirmed by comparative genomic hybridization on the GMI1000 microarray
using the protocol described in Guidot et al. (2007).
Frequency and Chronology of Appearance of
Mutations
The frequency of the detected mutations in populations of
derived clones was determined by PCR amplification and
Sanger sequencing of the mutated region from cell suspensions of 96 clones randomly isolated in each population. To
determine the chronology of appearance of the mutations,
the nucleotide sequence of the mutated regions was determined in populations of derived clones obtained after 18, 20,
23, and 26 SPE.
2926
Engineering of an efpR Mutant and Complementing
efpR Alleles
The strain GRS704 carrying a deletion of efpR (RSc1097 gene)
was created using the pWJ2 plasmid as a marker-exchange
deletion vector. pWJ2 was constructed as follows: Two border
fragments of the efpR gene were PCR-amplified from genomic
DNA of GMI1000 using the primers RSc1097_L-Band-L and
RSc1097_L-Band-R (downstream fragment) and RSc1097_RBand-L and RSc1097_R-Band-R (upstream fragment) that are
listed in supplementary table S2, Supplementary Material
online. Both fragments were cloned into the pGEM-T
vector (Promega) and the interposon (Prentki and Krisch
1984) was cloned into the single EcoRI restriction site between
the downstream and upstream fragments. The resulting plasmid pWJ2 was linearized by NdeI and introduced into strain
GMI1000 strain by natural transformation (Gonzalez et al.
2011). Recombinant clones were selected after growth on
BG medium supplemented with spectinomicin and the deletion of the efpR gene was checked by PCR using the omega1
or omega2 primers (Prentki and Krisch 1984) and the
RSc1097_L-Band-L primer.
Complementation of strain GRS704 with the efpRGMI1000, efpR-c1bean, efpR-d2bean, or efpR-e2bean alleles was
performed through chromosomal insertion using the pWJ3,
pWJ4, pWJ5, or pWJ6 plasmids. These four plasmids were
constructed as follows: A DNA sequence comprising the
efpR open-reading frame plus 800-bp upstream and 187-bp
downstream was PCR-amplified from genomic DNA of
GMI1000, c1bean, d2bean, and e2bean clones using the primers
RSc1097_C_L and RSc1097_C_R (supplementary table S2,
Supplementary Material online). These four allelic versions
of efpR were then each cloned into the unique XbaI–KpnI
site of pNP267 (a pRCG-GWY-based plasmid [49]) to generate the pWJ3, pWJ4, pWJ5, and pWJ6 plasmids, respectively.
The four plasmids thus generated were linearized by ScaI and
introduced into GRS704 strain by natural transformation
(Gonzalez et al. 2011). Ralstonia solanacearum transformants
were selected after growth on BG medium supplemented
with gentamicin and the double crossing-over was confirmed
by PCR using the primers oNP611, oNP612 and oNP613 (supplementary table S2, Supplementary Material online) and
Sanger sequencing. The resulting efpR complemented strains
were called GRS705, GRS706, GRS707, and GRS708 (table 5).
Statistical Analysis
Differences between mean CI values were tested using a
Wilcoxon test. Differences between in vitro growth rates
and between in vitro final OD were tested using a Student
t-test. Correlation between in vitro growth rates and CI values
was tested using a Kendall correlation test. These three tests
were performed in the R statistical software.
CI variation among populations and among clones within
populations was studied for each plant species by using the
following mixed model:
CIpc ¼ CI þ populationp þ clonec populationp þ "pc :
Multihost Experimental Evolution . doi:10.1093/molbev/msu229
MBE
In this model, “CI” corresponds to the competitive index,
“m” is the overall mean, “population” measures the effect
of the five experimental populations, “clone” corresponds
to the effect of the fixed clone within each population, and
“"” is the residual term. “Population” was treated as a fixed
effect, whereas “clone” was treated as a random effect. Model
fitting was conducted using the PROC MIXED procedure in
SAS 9.3 (SAS Institute Inc., Cary, NC).
Deardorff ER, Fitzpatrick KA, Jerzak GVS, Shi P-Y, Kramer LD, Ebel GD.
2011. West Nile virus experimental evolution in vivo and the tradeoff hypothesis. PLoS Pathog. 7:e1002335.
Dettman JR, Rodrigue N, Melnyk AH, Wong A, Bailey SF, Kassen R. 2012.
Evolutionary insight from whole-genome sequencing of experimentally evolved microbes. Mol Ecol. 21:2058–2077.
Ebert. 1998. Experimental evolution of parasites. Science. 282:1432–1435.
Elena SF, Lenski RE. 2003. Evolution experiments with microorganisms:
the dynamics and genetic bases of adaptation. Nat Rev Genet. 4:
457–469.
Gandon S, Hochberg ME, Holt RD, Day T. 2013. What limits the evolutionary emergence of pathogens? Philos Trans R Soc B Lond Biol Sci.
368:20120086.
Genin S. 2010. Molecular traits controlling host range and adaptation
to plants in Ralstonia solanacearum. New Phytol. 187:920–928.
Genin S, Denny TP. 2012. Pathogenomics of the Ralstonia solanacearum
species complex. Annu Rev Phytopathol. 50:67–89.
Gonzalez A, Plener L, Restrepo S, Boucher C, Genin S. 2011. Detection
and functional characterization of a large genomic deletion resulting
in decreased pathogenicity in Ralstonia solanacearum race 3 biovar
2 strains. Environ Microbiol. 13:3172–3185.
Griffin AS, West SA, Buckling A. 2004. Cooperation and competition
in pathogenic bacteria. Nature 430:1024–1027.
Guidot A, Prior P, Schoenfeld J, Carrère S, Genin S, Boucher C. 2007.
Genomic structure and phylogeny of the plant pathogen Ralstonia
solanacearum inferred from gene distribution analysis. J Bacteriol.
189:377–387.
Hayward AC. 1991. Biology and epidemiology of bacterial wilt caused
by Pseudomonas solanacearum. Annu Rev Phytopathol. 29:65–87.
Herring CD, Raghunathan A, Honisch C, Patel T, Applebee MK, Joyce
AR, Albert TJ, Blattner FR, van den Boom D, Cantor CR, et al. 2006.
Comparative genome sequencing of Escherichia coli allows observation of bacterial evolution on a laboratory timescale. Nat Genet. 38:
1406–1412.
Joosten MH, Cozijnsen TJ, De Wit PJ. 1994. Host resistance to a fungal
tomato pathogen lost by a single base-pair change in an avirulence
gene. Nature 367:384–386.
Kawecki TJ, Lenski RE, Ebert D, Hollis B, Olivieri I, Whitlock MC. 2012.
Experimental evolution. Trends Ecol Evol. 27:547–560.
Kearney B, Ronald PC, Dahlbeck D, Staskawicz BJ. 1988. Molecular basis
for evasion of plant host defence in bacterial spot disease of pepper.
Nature 332:541–543.
Lefeuvre P, Cellier G, Remenant B, Chiroleu F, Prior P. 2013. Constraints
on genome dynamics revealed from gene distribution among the
Ralstonia solanacearum species. PLoS One 8:e63155.
Lenski RE, Winkworth CL, Riley MA. 2003. Rates of DNA sequence
evolution in experimental populations of Escherichia coli during
20,000 generations. J Mol Evol. 56:498–508.
Lopez M, Biosca E. 2005. Potato bacterial wilt management: new prospects for an old problem. In: Allen C, Prior P, Hayward AC, editors.
Bacterial wilt disease and the Ralstonia solanacearum species
complex. St Paul (MN): APS Press. p. 205–224. Available from:
about:newtab.
Lovell HC, Jackson RW, Mansfield JW, Godfrey SAC, Hancock JT, Desikan
R, Arnold DL. 2011. In planta conditions induce genomic changes
in Pseudomonas syringae pv. phaseolicola. Mol Plant Pathol. 12:
167–176.
Macho AP, Guidot A, Barberis P, Beuzon CR, Genin S. 2010. A competitive index assay identifies several Ralstonia solanacearum type III
effector mutant strains with reduced fitness in host plants.
Mol Plant Microbe Interact. 23:1197–1205.
MacLean RC, Perron GG, Gardner A. 2010. Diminishing returns from
beneficial mutations and pervasive epistasis shape the fitness landscape for rifampicin resistance in Pseudomonas aeruginosa. Genetics
186:1345–1354.
Mansfield J, Genin S, Magori S, Citovsky V, Sriariyanum M, Ronald P,
Dow M, Verdier V, Beer SV, Machado MA, et al. 2012. Top 10 plant
pathogenic bacteria in molecular plant pathology. Mol Plant Pathol.
13:614–629.
Supplementary Material
Supplementary tables S1 and S2 and figure S1 are available at
Molecular Biology and Evolution online (http://www.mbe.
oxfordjournals.org/).
Acknowledgments
The authors thank C. Boucher, N. Peeters, and the members
of the SG lab for support and advices during the course of this
work. They also thank F. Roux for critical reading of the manuscript and S. Alves-Carvalho for help with bioinformatic
analyses. They are grateful to Jean Luc Pariente, Serge Bosc,
and Claudette Icher for plant preparation. They also thank
the “Plateforme Genomique du Genopole Toulouse MidiPyrenees” for whole-genome DNA sequencing. This work
was supported by the French Laboratory of Excellence project
“TULIP” (ANR-10-LABX-41; ANR-11-IDEX-0002-02) and the
ANR grant D07 GMGE 003.
References
Angot A, Peeters N, Lechner E, Vailleau F, Baud C, Gentzbittel L, Sartorel
E, Genschik P, Boucher C, Genin S. 2006. Ralstonia solanacearum
requires F-box-like domain-containing type III effectors to promote
disease on several host plants. Proc Natl Acad Sci U S A. 103:
14620–14625.
Barrick JE, Yu DS, Yoon SH, Jeong H, Oh TK, Schneider D, Lenski RE, Kim
JF. 2009. Genome evolution and adaptation in a long-term experiment with Escherichia coli. Nature 461:1243–1247.
Bataillon T, Joyce P, Sniegowski P. 2012. As it happens: current directions
in experimental evolution. Biol Lett. 20120945.
Barroso-Batista J, Sousa A, Lourenço M, Bergman ML, Sobral D,
Demengeot J, Xavier KB, Gordo I. 2014. The first steps of adaptation
of Escherichia coli to the gut are dominated by soft sweeps. PLoS
Genet. 10:e1004182.
Baumler AJ, Tsolis RM, Valentine PJ, Ficht TA, Heffron F. 1997.
Synergistic effect of mutations in invA and lpfC on the ability of
Salmonella typhimurium to cause murine typhoid. Infect Immun. 65:
2254–2259.
Bedhomme S, Lafforgue G, Elena SF. 2012. Multihost experimental evolution of a plant RNA virus reveals local adaptation and host specific
mutations. Mol Biol Evol. 29:1481–1492.
Brockhurst MA, Colegrave N, Rozen DE. 2011. Next-generation sequencing as a tool to study microbial evolution. Mol Ecol. 20:972–980.
Brumbley SM, Carney BF, Denny TP. 1993. Phenotype conversion in
Pseudomonas solanacearum due to spontaneous inactivation of
PhcA, a putative LysR transcriptional regulator. J Bacteriol. 175:
5477–5487.
Clough SJ, Lee KE, Schell MA, Denny TP. 1997. A two-component
system in Ralstonia (Pseudomonas) solanacearum modulates
production of PhcA-regulated virulence factors in response to
3-hydroxypalmitic acid methyl ester. J Bacteriol. 179:3639–3648.
Conrad TM, Joyce AR, Applebee MK, Barrett CL, Xie B, Gao Y, Palsson
BØ. 2009. Whole-genome resequencing of Escherichia coli K-12
MG1655 undergoing short-term laboratory evolution in lactate
minimal media reveals flexible selection of adaptive mutations.
Genome Biol. 10:R118.
2927
Guidot et al. . doi:10.1093/molbev/msu229
Marchetti M, Capela D, Glew M, Cruveiller S, Chane-Woon-Ming B,
Gris C, Timmers T, Poinsot V, Gilbert LB, Heeb P, et al. 2010.
Experimental evolution of a plant pathogen into a legume symbiont. PLoS Biol. 8:e1000280.
Marvig RL, Johansen HK, Molin S, Jelsbak L. 2013. Genome analysis of a
transmissible lineage of pseudomonas aeruginosa reveals pathoadaptive mutations and distinct evolutionary paths of hypermutators.
PLoS Genet. 9:e1003741.
Monteiro F, Sole M, van Dijk I, Valls M. 2012. A chromosomal insertion
toolbox for promoter probing, mutant complementation, and pathogenicity studies in Ralstonia solanacearum. Mol Plant Microbe
Interact. 25:557–568.
Nakamura Y, Gojobori T, Ikemura T. 2000. Codon usage tabulated from
international DNA sequence databases: status for the year 2000.
Nucleic Acids Res. 28:292.
Nusbaum C, Ohsumi TK, Gomez J, Aquadro J, Victor TC, Warren RM,
Hung DT, Birren BW, Lander ES, Jaffe DB. 2009. Sensitive, specific
polymorphism discovery in bacteria using massively parallel sequencing. Nat Methods. 6:67–69.
Orr HA. 1998. The population genetics of adaptation: the distribution of
factors fixed during adaptive evolution. Evolution 52:935.
Orr HA. 2003. The distribution of fitness effects among beneficial mutations. Genetics 163:1519–1526.
Peeters N, Carrère S, Anisimova M, Plener L, Cazale A-C, Genin S. 2013.
Repertoire, unified nomenclature and evolution of the Type III effector gene set in the Ralstonia solanacearum species complex. BMC
Genomics 14:859.
Peeters N, Guidot A, Vailleau F, Valls M. 2013. Ralstonia solanacearum, a
widespread bacterial plant pathogen in the post-genomic era. Mol
Plant Pathol. 14:651–662.
Pelosi L, Kuhn L, Guetta D, Garin J, Geiselmann J, Lenski RE, Schneider D.
2006. Parallel changes in global protein profiles during long-term
experimental evolution in Escherichia coli. Genetics 173:1851–1869.
Philippe N, Crozat E, Lenski RE, Schneider D. 2007. Evolution of global
regulatory networks during a long-term experiment with Escherichia
coli. BioEssays 29:846–860.
Plener L, Manfredi P, Valls M, Genin S. 2010. PrhG, a transcriptional
regulator responding to growth conditions, is involved in the control of the type III secretion system regulon in Ralstonia solanacearum. J Bacteriol. 192:1011–1019.
Plotkin JB, Kudla G. 2011. Synonymous but not the same: the causes and
consequences of codon bias. Nat Rev Genet. 12:32–42.
Prentki P, Krisch HM. 1984. In vitro insertional mutagenesis with a
selectable DNA fragment. Gene 29:303–313.
Rainey PB, Travisano M. 1998. Adaptive radiation in a heterogeneous
environment. Nature 394:69–72.
Remenant B, Babujee L, Lajus A, Medigue C, Prior P, Allen C. 2012.
Sequencing of K60, type strain of the major plant pathogen
Ralstonia solanacearum. J Bacteriol. 194:2742–2743.
Remenant B, Coupat-Goutaland B, Guidot A, Cellier G, Wicker E, Allen
C, Fegan M, Pruvost O, Elbaz M, Calteau A, et al. 2010. Genomes of
three tomato pathogens within the Ralstonia solanacearum species
2928
MBE
complex reveal significant evolutionary divergence. BMC Genomics
11:379.
Remenant B, de Cambiaire J-C, Cellier G, Jacobs JM, Mangenot S, Barbe
V, Lajus A, Vallenet D, Medigue C, Fegan M, et al. 2011. Ralstonia
syzygii, the Blood Disease Bacterium and some Asian R. solanacearum strains form a single genomic species despite divergent
lifestyles. PLoS One 6:e24356.
Salanoubat M, Genin S, Artiguenave F, Gouzy J, Mangenot S, Arlat M,
Billault A, Brottier P, Camus JC, Cattolico L, et al. 2002. Genome
sequence of the plant pathogen Ralstonia solanacearum. Nature
415:497–502.
Shendure J, Porreca GJ, Reppas NB, Lin X, McCutcheon JP, Rosenbaum
AM, Wang MD, Zhang K, Mitra RD, Church GM. 2005. Accurate
multiplex polony sequencing of an evolved bacterial genome.
Science 309:1728–1732.
Smith EE, Buckley DG, Wu Z, Saenphimmachak C, Hoffman LR,
D’Argenio DA, Miller SI, Ramsey BW, Speert DP, Moskowitz SM,
Burns JL, Kaul R, Olson MV. 2006. Genetic adaptation by
Pseudomonas aeruginosa to the airways of cystic fibrosis patients.
Proc Natl Acad Sci U S A. 103:8487–8492.
Stanek MT, Cooper TF, Lenski RE. 2009. Identification and dynamics of a
beneficial mutation in a long-term evolution experiment with
Escherichia coli. BMC Evol Biol. 9:302.
Tang S-L, Chang BCH, Halgamuge SK. 2010. Gene functionality’s influence on the second codon: a large-scale survey of second codon
composition in three domains. Genomics 96:92–101.
Tenaillon O. 2014. The utility of Fisher’s geometric model in evolutionary
genetics. Annu Rev Ecol Evol Syst. 45:179–201.
Toft C, Andersson SGE. 2010. Evolutionary microbial genomics: insights
into bacterial host adaptation. Nat Rev Genet. 11:465–475.
Trivedi P, Wang N. 2013. Host immune responses accelerate pathogen
evolution. ISME J. 8:727–731.
Wang L, Spira B, Zhou Z, Feng L, Maharjan RP, Li X, Li F, McKenzie C,
Reeves PR, Ferenci T, et al. 2010. Divergence involving global regulatory gene mutations in an Escherichia coli population evolving
under phosphate limitation. Genome Biol Evol. 2:478–487.
Wicker E, Grassart L, Coranson-Beaudu R, Mian D, Guilbaud C, Fegan M,
Prior P. 2007. Ralstonia solanacearum strains from Martinique
(French West Indies) exhibiting a new pathogenic potential. Appl
Environ Microbiol. 73:6790–6801.
Wicker E, Grassart L, Coranson-Beaudu R, Mian D, Prior P. 2009.
Epidemiological evidence for the emergence of a new pathogenic
variant of Ralstonia solanacearum in Martinique (French West
Indies). Plant Pathol. 58:853–861.
Wicker E, Lefeuvre P, de Cambiaire J-C, Lemaire C, Poussier S, Prior P.
2012. Contrasting recombination patterns and demographic histories of the plant pathogen Ralstonia solanacearum inferred from
MLSA. ISME J. 6:961–974.
Xu J, Zheng H, Liu L, Pan ZC, Prior P, Tang B, Xu JS, Zhang H, Tian Q,
Zhang LQ, et al. 2011. Complete genome sequence of the plant
pathogen Ralstonia solanacearum strain Po82. J Bacteriol. 193:
4261–4262.