Dissection of bacterial Wilt on Medicago truncatula revealed two type III secretion system effectors acting on root infection process and disease development.

Ralstonia solanacearum is the causal agent of the devastating bacterial wilt disease, which colonizes susceptible Medicago truncatula via the intact root tip. Infection involves four steps: appearance of root tip symptoms, root tip cortical cell invasion, vessel colonization, and foliar wilting. We examined this pathosystem by in vitro inoculation of intact roots of susceptible or resistant M. truncatula with the pathogenic strain GMI1000. The infection process was type III secretion system dependent and required two type III effectors, Gala7 and AvrA, which were shown to be involved at different stages of infection. Both effectors were involved in development of root tip symptoms, and Gala7 was the main determinant for bacterial invasion of cortical cells. Vessel invasion depended on the host genetic background and was never observed in the resistant line. The invasion of the root tip vasculature in the susceptible line caused foliar wilting. The avrA mutant showed reduced aggressiveness in all steps of the infection process, suggesting a global role in R. solanacearum pathogenicity. The roles of these two effectors in subsequent stages were studied using an assay that bypassed the penetration step; with this assay, the avrA mutant showed no effect compared with the GMI1000 strain, indicating that AvrA is important in early stages of infection. However, later disease symptoms were reduced in the gala7 mutant, indicating a key role in later stages of infection.

Very few phytopathogenic bacteria colonize intact plant roots, even if the local soil provides a favorable environment for bacterial growth. Penetration seems to be the limiting step, as most bacteria need natural openings or wounds to invade their hosts and there are few opportunities in roots (Loria et al., 2003). Ralstonia solanacearum, a gram-negative soil-borne pathogen responsible for bacterial wilt, naturally invades plants through axils of secondary roots. R. solanacearum has to attach to the root, find nutrients, multiply, migrate into plant tissues, and penetrate the xylem in order to cause disease (Vasse, 1995). It has a wide and still-expanding host range that includes several hundred susceptible species in at least 50 different plant families, which include economically important hosts such as potato (Solanum tuberosum), tomato (Solanum lycopersicum), peanut (Arachis hypogaea), banana (Musa spp.), pepper (Capsicum annuum), eggplant (Solanum melongena), and tobacco (Nicotiana tabacum; Hayward, 1991). The high incidence of plant mortality and the lack of effective control methods make R. solanacearum one of the world's most destructive plant pathogens (Prior et al., 1998). This bacterium is a model organism extensively studied by genetic, molecular, and biochemical techniques in order to develop strategies for the control of bacterial wilt.
R. solanacearum virulence responds to several environmental stimuli and involves the production of multiple virulence factors (Schell, 2000;Genin and Boucher, 2002). For example, Exopolysaccharide I (EPS I), a large nitrogen-rich acidic exopolysaccharide (Orgambide et al., 1991), is thought to be an important virulence factor. Produced in large amounts by the bacteria, EPS I enhances the speed and extent of stem colonization (Saile et al., 1997) and is presumed to cause wilting by restricting water flow through xylem vessels (Garg et al., 2000). In addition to EPS I, R. solanacearum secretes plant cell wall-degrading enzymes through the type II secretion system (T2SS). Pectinolytic enzymes fragment pectin into oligomers, which act as substrates for the bacterial growth (Tans-Kersten et al., 1998); the breakdown of pectin enhances virulence by facilitating movements through pectin-rich regions such as vascular bundles (Gonzalez and Allen, 2003). Cellulytic enzymes also facilitate invasion of roots and/or penetration of xylem vessels by degrading cellulosic glucans in the cell wall (Liu et al., 2005).
Like many other pathogenic bacteria, R. solanacearum possesses a type III secretion system (T3SS), encoded by the hypersensitive response and pathogenicity (hrp) genes. In contrast to T2SS, which secretes degrading enzymes, the T3SS allows the injection of effector proteins directly into the plant cell through the hrp pili. The T3SS is essential for pathogenicity, since R. solanacearum hrp mutants cannot induce disease in many susceptible plants such as tobacco, tomato, petunia (Petunia hybrida), and Medicago (Vasse et al., 2000;Kanda et al., 2003;Zolobowska and Van Gijsegem, 2006;Vailleau et al., 2007). All effectors are presumed to facilitate virulence, but on some host cultivars they can be recognized by a cognate resistance protein, thereby triggering host defenses (Stavrinides et al., 2008). Thus, when the cognate resistance protein is present in the host plant, the effectors act as avirulence factors. This recognition response is associated with effector-triggered immunity, which results in a fast and intense plant defense response known as the hypersensitive response (HR). The HR is characterized by localized programmed cell death that helps to limit the infection to the area of initial contact (Jones and Dangl, 2006). In R. solanacearum strain GMI1000, 74 T3SS effectors (T3Es) have been identified ). To date, three of them have been reported to act as avirulence factors: AvrA is responsible for eliciting an HR in tobacco (Carney and Denny, 1990;); PopP1 acts as a host-specificity determinant toward tobacco ) and petunia (Lavie et al., 2002); and PopP2 interacts with the Arabidopsis (Arabidopsis thaliana) RESISTANT TO RALSTONIA SOLANACEARUM1-R resistance protein (Deslandes et al., 2003). Some T3Es are virulence factors: they promote pathogenic functions of the bacteria and disrupt defense responses of the host immune system. Several mechanisms by which T3Es suppress plant defense have been described, including the inhibition of plant signaling components and the alteration of host transcription or posttranscriptional processes. Some T3Es are proteases that can degrade defense components through the host ubiquitination machinery and the 26S proteasome (Block et al., 2008). To date, only a few effectors from R. solanacearum have been shown to have pathogenicity functions. Gala T3Es are the best characterized; they have an F-box and a Leu-rich repeat domain and interact with ASK (for Arabidopsis SKP1-like) proteins in Arabidopsis (Angot et al., 2006). F-box and ASK proteins are part of ubiquitin E3 ligase complexes that attach ubiquitin to specific proteins, which are subsequently degraded by the 26S proteasome.
This work aims to study the sequence of events related to the T3SS-dependent infection process of R. solanacearum on intact plant roots. Many previous studies of plant-R. solanacearum interaction involved bacteria delivered directly into the vascular system through artificially wounded roots, in order to increase penetration potential (Deslandes et al., 1998;Vailleau et al., 2007). However, these inoculation procedures bypassed normal infection routes and potential resistance mechanisms. This study focuses on the early steps of the R. solanacearum infection process on intact roots. In previous work, the R. solanacearum strain GMI1000 was shown to infect the model legume plant Medicago truncatula  in an in vitro wounded root system, and susceptible (Jemalong A17) and resistant (F83005.5) lines were identified. In this work, the in vitro inoculation procedure was adapted for intact M. truncatula roots, thereby enabling us to study the sequence of events related to the T3SSdependent infection process. Screening of loss-offunction T3E mutants in the GMI1000 strain on intact roots of the susceptible M. truncatula line led to the identification of two virulence effectors, Gala7 and AvrA. We investigated the role of these effectors in various stages of the infection process.

RESULTS
The Interaction between R. solanacearum and M. truncatula Leads to T3SS-Dependent Root Symptoms In order to study the infection process of R. solanacearum in M. truncatula, we used an in vitro inoculation procedure of intact roots adapted from Vailleau et al. (2007). It enabled us to observe root symptoms resulting from the interaction between R. solanacearum GMI1000 strain and either susceptible (A17) or resistant (F83005.5) lines of M. truncatula. The first root symptoms observed, in both susceptible and resistant lines, were an arrest of root hair elongation 18 h post inoculation (hpi) and a substantial reduction in root growth (Student's t test, P = 2e-16), with no growth occurring after 24 hpi.
Viability of the epidermis and root hairs was monitored by fluorescein diacetate (FDA), which stains living cells (Fig. 1, A, D, and G). The epidermis of the mock inoculation control was viable, as evidenced by green fluorescence. Also, root hair initiation and elongation occurred (Fig. 1, A-C). After inoculation with GMI1000, a loss of fluorescence was first observed at 24 hpi and was complete at 72 hpi, indicating a loss of viability of the root tip epidermis (the root tip was considered the first 2-3 mm of the root; Fig. 1G). In addition, no root hair initiation or elongation was detected, and root tip browning and swelling were observed (Fig. 1, H and I). Root growth arrest, root tip swelling, browning, and epidermis loss of viability will be subsequently referred to "root symptoms." Despite the similarity in the root symptoms between resistant and susceptible lines, foliar symptoms were only observed in the susceptible line (A17). The chlo-rosis and wilting of leaves and cotyledons led to the death of an average of 50% of A17 plants when the experiment was terminated 28 d post inoculation (dpi).
Infection with R. solanacearum strains carrying mutations in the regulatory hrpB gene (GMI1525) or in the structural T3SS gene hrcS (GMI1402) resulted in the absence of root growth arrest at 24 hpi, an intact root tip epidermis at 72 hpi, no swelling and browning at 7 dpi ( Fig. 1, D-F), and no wilting of the shoots up to 28 dpi. Root and foliar symptoms, therefore, were dependent on a functional T3SS. Bacterial mutants defective in EPS or T2SS biosynthesis were altered in their ability to induce foliar symptoms but induced root symptoms similar to those caused by GMI1000 (data not shown). Therefore, root symptoms were specifically dependent on T3SS.
The Infection of M. truncatula by R. solanacearum Strain GMI1000 Involves Two Distinct Root Colonization Steps The bacterial distribution and the internal structure of the root tip were observed at 3 and 7 dpi on the susceptible line using semithin resin sections (Fig. 2). After mock inoculation, the root tip was typically organized ( Fig. 2A; Groot et al., 2004). At 3 dpi with GMI1000, bacteria adhered to the epidermis surface ( Fig. 2C) and started to colonize the inner root tip, where they were detected in cortical cells and extracellular spaces (Fig. 2D). The root tip (the first 2-3 mm of the root) was the only penetration site detected, with no preferential point of entry in that area. Colonization through axils of lateral roots was never observed. Concomitantly, the root tip structure was altered: the epidermal layer exhibited severe cell damage (as illustrated by the FDA test), enlarged cortical cells were highly vacuolated, supernumerary vessels were differentiated, and the disorganized meristematic zone showed unconventional vacuolated cells ( Fig. 2). At 7 dpi, the cortex was heavily colonized, showing enlarged cells containing numerous bacteria. R. solanacearum started to penetrate and proliferate in the vessels of the susceptible line (Fig. 2, E and F) but not in the resistant line.
In order to quantify the differential bacteria colonization, infections were conducted using a R. solanacearum strain transformed with a constitutive GUS construct (GMI1559; Etchebar et al., 1998). The GUS activity was revealed in planta, and the observations were performed with a stereomicroscope. At 14 dpi, bacteria had crossed the epidermis and colonized the cortical root tip cells of 45% of the F83005.5 plants and 85% of the A17 plants. Bacteria were never observed in the vessels of F83005.5, whereas 55% of the A17 plants had their vessels colonized at 14 dpi.
Thus, two stages in the infection process were clearly identified. The first one, common to the susceptible and the resistant lines, was the bacterial colonization of the root tip cortical cells. Interestingly, a restriction of bacterial propagation was found in the resistant line F83005.5, indicating the possible onset of resistance. The second stage, specific to the susceptible line, was the colonization of the vascular bundles, leading to plant death.
The Resistant F83005.5 Line Showed a Higher Cell Wall Autofluorescence and Earlier Endodermis Lignification Than the Susceptible Line A17 after Inoculation with R. solanacearum GMI1000 To identify potential cell wall modifications in response to infection, cytological studies were performed on transverse root tip sections of the susceptible and resistant lines after pathogen infection. Tissues were stained with FDA, which produces bright green fluorescence in living cells. A and D, In the control conditions, the epidermis was intact and showed root hair differentiation and elongation. G, With GMI1000 inoculation, the epidermis showed drastic loss of viability and no root hair differentiation could be observed. Bars = 100 mm. B, E, and H, Root tip transverse sections at 3 dpi (1-2 mm from the root tip). Following GMI1000 inoculation, the root tip was enlarged compared with control conditions. Bars = 70 mm. C, F, and I, Root tip at 7 dpi. Compared with control conditions, roots inoculated with GMI1000 strain showed epidermal browning and root tip deformation. Bars = 100 mm.
Autofluorescence of the inner root tip and callose, suberin, and lignin deposits were observed 3 and 7 dpi in both A17 and F83005.5 lines after GMI1000 (Fig. 3) or mock (data not shown) inoculation. Autofluorescence of the cell wall of the cortical cells was observed only after inoculation (Fig. 3). Sections with (Fig. 3A) or without (Fig. 3B) cell wall autofluorescence were observed in both resistant and susceptible roots; however, at 7 dpi, roots of the resistant line showed cell wall autofluorescence at a significantly higher proportion than was observed in the susceptible line (frequency test, P = 0.027).
Callose and suberin accumulations were observed equally in susceptible and resistant lines (data not shown). Lignin deposits were observed in the endodermis of the two lines only after inoculation with the GMI1000 strain (Fig. 3). Sections with (Fig. 3C) or without (Fig. 3D) endodermis lignifications were observed in the inoculated roots, but at 3 dpi, F83005.5 roots showed lignin deposits on the endodermis, at a significantly higher frequency (frequency test, P = 0.011) than did A17 roots. Interestingly, at 7 dpi, A17 and F83005.5 root tip endodermis were lignified to a similar extent (frequency test, P = 0.366), indicating that rapid lignification may be important in resistance.

Root Symptom Development in the Susceptible A17 Line Depends on Two T3Es
As seen previously, root symptoms were specifically dependent on the presence of the T3SS. To identify bacterial molecules involved in the development of these root symptoms in the susceptible line, we used FDA staining to screen a collection of 67 loss-of-function T3E mutants in the GMI1000 background (Supplemental Fig. S1; . Because the analysis of root tip epidermis viability revealed intermediate phenotypes, we established a scale, ranking from a score of 0 (intact root tip epidermis after mock or T3SS-defective control inoculation) to 4 (complete loss of viability of the root tip epidermis as observed after inoculation with GMI1000). The screening revealed two mutants that were partially impaired in their capacity to induce epidermal cell death, gala7 and avrA. Inoculation of A17 with gala7 and avrA mutants led to median scores of 3 and 3.5, respectively (Fig. 4). Root growth arrest at 24 hpi and root tip swelling were reduced after infection with the two mutant strains compared with GMI1000 but were more pronounced than after mock or T3SS-defective control inoculation.
In order to uncover a potential additive effect of the two effectors on the root symptom, an avrA and gala7 double mutant strain was generated. The FDA test revealed a median score of 2, showing an additive effect on epidermal cell death (Wilcoxon rank sum test, P = 2.10 24 ; Fig. 4). Root growth measurements showed high variability, and no additive effect could be demonstrated.
Among the effector mutants showing a minor effect (Supplemental Fig. S1), two of them were selected: GRS118 (GMI1000 DRSp0846-RSp0847) and GRS230 (GMI1000 DRSp0842; Cunnac et al., 2004). Two triple mutants were constructed with avrA and gala7. They were both assayed with the FDA test, and no significant difference could be observed with the double mutant. Thus, in our conditions and from among this group of effectors, avrA and gala7 seem to be the main determinants of root symptom development. However, it is likely that other effectors have lesser individual effects. After observing the involvement of AvrA and Gala7 in root symptom development after R. solanacearum inoculation, we characterized their roles in root bacterial colonization and the development of foliar symptoms. At 14 dpi, the gala7 mutant was never detected in the root tip cortex and therefore did not colonize the vascular system (Fig. 5), nor did it induce wilting of the A17 plant (Fig. 6). This indicated that Gala7 was necessary for the bacterial penetration into the root of the susceptible M. truncatula line.
At 14 dpi, the avrA mutant strain was found in the root tip cortex of 18% of the plants and in the vascular system of 4% of the plants, in comparison with GMI1000, which was found in the root tip cortex of 85% of inoculated plants and in the vascular system of 55% (Fig. 5). The plant wilting rate was also 5 times lower following inoculation with avrA compared with GMI1000 ( Fig. 6). At 28 dpi, unwilted plants did not show any sign of infection. Therefore, the avrA mutant was not delayed in the induction of wilting symptoms but showed a reduced aggressiveness in comparison with GMI1000 strain. All wilted plants inoculated with the avrA mutant showed root growth arrest similar to that found after GMI1000 inoculation. Therefore, AvrA appears to be not essential for the root penetration of the susceptible M. truncatula line, but it does play a positive role in this process.
In conclusion, the effectors Gala7 and AvrA are involved in the appearance of root symptoms and also play roles in root colonization and disease development. Gala7 was necessary for bacterial penetration of  root cells, whereas AvrA was partially required for root colonization.

AvrA and Gala7 Are Differentially Involved in Bacterial Propagation in Vessels and Wilting Symptom Development
After showing the involvement of the two effectors in root penetration, it was interesting to assay their role in the subsequent steps of vascular propagation and wilting development. Therefore, we used an inoculation procedure that bypassed root penetration using root tips that were cut just before the bacterial inoculation . Using this procedure, 100% death of susceptible plants was observed at 14 dpi with either GMI1000 or the avrA mutant. Only 40% of plants died after inoculation with the gala7 mutant, and a similar death rate was observed with the double mutant strain, which showed 30% of wilted plants (Fig. 7). This suggested that AvrA is not necessary for bacterial propagation in the vessels and wilting development, whereas Gala7 is involved in those steps.

DISCUSSION
In order to study the root infection process of the devastating pathogen R. solanacearum, we developed an in vitro inoculation procedure using intact M. truncatula roots. The root tip was identified as the key site for T3SS-dependent root symptom development and root penetration of R. solanacearum. The pathosystem consisted of susceptible and resistant M. truncatula lines infected with the R. solanacearum strain GMI1000. Examination of this pathosystem allowed the dissection of sequential stages of pathogen infection and the identification of two effectors important in this process.

Root Symptom Development in Response to Infection Is Genetically Controlled by the Bacteria
The first step in the infection (Fig. 8) was the development of root symptoms, which included root growth arrest, root tip swelling and browning, and epidermal cell death. These root symptoms were not dependent on the plant genetic background, as they were common to both susceptible and resistant lines. They were also observed on other M. truncatula lines infected with R. solanacearum, four resistant (TN7.22,TN3.17,TN6.17,and DZA315.16) and three susceptible (TN1.21, TN9.1, and TN6.2) lines. They appeared to be the first step in bacterial infection, but they were not always associated with the wilting of the plant. The progression of early steps of the infection processes is not always associated with resistance, as described in other systems (Dita et al., 2007). Indeed, specific recognition leading to defense mechanisms may initiate after penetration of the pathogen into the host tissues (Freytag et al., 1994).
R. solanacearum infection processes in intact roots have been described in tomato (Vasse, 1995) and petunia (Zolobowska and Van Gijsegem, 2006). No tomato root symptoms were mentioned. However, in petunia, Zolobowska and Van Gijsegem (2006) ob- Figure 5. Mutations in two bacterial effectors alter the ability of the pathogen to colonize root tissues. Bacterial repartition in the root tip of the susceptible A17 line at 14 dpi. Roots were inoculated with a GMI1000 derivative containing a construct that gave constitutive GUS expression (GMI1559). The GUS reporter fusion from strain GMI1559 was recombined into the T3E mutant strains to allow detection of the resulting strain, and inoculated plants were assayed for the location of the infecting bacteria. White bars indicate no bacterial penetration in the root tip, gray bars indicate detection in cortex, and black bars indicate detection in vessels. No penetration in the root tip could be observed with the gala7 mutant. Bacteria were found in the cortex and the vessels of A17 inoculated with GMI1000 and avrA mutant but to a lesser extent with the bacterial mutant. Values are means of three biological repeats (10 individual plants per repeat). Figure 6. Mutations in two bacterial effectors alter the ability of the pathogen to cause foliar symptoms when inoculated on intact roots. Percentage of wilted A17 plants after GMI1000 (squares), avrA mutant (circles), gala7 mutant (diamonds), or double mutant strain (triangles) inoculation on intact root tips. GMI1000 inoculation led to 50% wilted plants at 28 dpi. This rate was 5 times lower with the avrA mutant. The gala7 mutant and double mutant strain inoculations led to no wilting symptoms. Values are means of three biological repeats (10 individual plants per repeat), and error bars represent the confidence interval at 95%. served swelling of the root tip, reduced elongation of main and lateral roots, inhibition of lateral root development, and formation of root lateral structures. Similarly, we found M. truncatula root growth arrest, which could result from the disruption of the root meristem structure that was observed. At the same time, internal cell maturation may still occur, as illustrated by the enlarged vacuolated cells and supernumerous mature vessels normally encountered farther from the root tip (Fig. 2B). However, root alterations may not be a specific virulence mechanism but may just be a consequence of infection.
Root tip epidermal cell death, caused by effectors secreted by the bacteria, could be beneficial to R. solanacearum. It could lead to the release of specific attractant compounds, as chemotaxis has been described to play a role in fitness and virulence (Yao and Allen, 2006). The release of nutrients may also contribute to the ability of the bacteria to grow and therefore participate in the quorum-sensing mechanism (Von Bodman et al., 2003). Root tip epidermal cell death could also be directly linked to the invasion of the bacteria. The cell death could be induced before bacterial penetration, or bacteria could invade still intact cells as described for the bacterium Streptomyces scabies in potato (Loria et al., 2003).
The specific requirement of a functional T3SS for root symptom development led us to test a panel of T3E mutants for their ability to cause epidermal cell death. This screen resulted in the identification of two effectors: Gala7 and AvrA. These two effectors were involved in epidermal cell death and had an additive effect. Other effectors may be involved to a lesser extent, as the root tip epidermis was still partially altered after infection with the double mutant strain. The two effectors were also involved in root growth arrest, but no significant additive effect could be measured due to the variability of root growth. These effectors were both demonstrated to be translocated into plant cells by the T3SS (Angot et al., 2006;, so they may modulate host pathways, for example, acting on the host proteasome pathway in the case of Gala7 (Angot et al., 2006 The second step in the infection process (Fig. 8) was the bacterial invasion of root tip cortical cells. The root  Step 1, Root symptom development. Root growth arrest, root tip swelling, and epidermal cell death were common to resistant F83005.5 and susceptible A17 lines. A cumulative effect of AvrA and Gala7 on epidermal cell death was seen.
Step 2, Bacterial cortex colonization. Bacterial invasion of the cortex was observed in the susceptible and resistant lines, but to a lesser extent in the resistant line. The gala7 mutant was never detected in the cortex of the A17 line, and the avrA mutant presence was strongly reduced compared with GMI1000.
Step 3, Bacterial vessel colonization. Plant genetic background was predominant, as the GMI1000 strain was never detected in the vessels of the resistant F83005.5 line. Invasion of the avrA mutant into the A17 line was strongly reduced compared with GMI1000.
Step 4, Wilting symptom development on wounded roots. Inoculation on A17 wounded roots revealed no difference in wilting symptom development after inoculation with the avrA mutant or GMI1000. Inoculation with the gala7 mutant revealed a strong reduction of wilting symptoms. Symbols are as follows: X, root symptoms; C, cortex; V, vessels; gray oval, bacteria.
tip was the only zone where we found that the bacteria could penetrate into M. truncatula through intact root tissue and not through induced or natural wounds, like axils of secondary roots on tomato (Vasse, 1995). To date, Streptomyces species and Erwinia chrysanthemi were the only pathogenic bacteria described as being able to penetrate intact root tissues (Nieves-Brun, 1985;Loria et al., 2003).
Cortical cell invasions occurred in both M. truncatula lines but were reduced in the resistant line. The cell wall is the primary interface for pathogen interactions and therefore is an important site of potential defense mechanisms, such as cell wall deposition and strengthening (Cano-Delgado et al., 2003). Autofluorescent phenolic compounds are mainly implicated in cell wall strengthening; an autofluorescence assay, therefore, allows a general view of potential defense mechanisms (Carver et al., 1998;Torregrosa et al., 2004). Stronger autofluorescence was detected in F83005.5 cortical cell walls and could be linked with the reduced bacterial penetration in the root.
At this stage, the bacterial effectors were the major determinants of infection, as the gala7 mutant was never detected in cortical cells of line A17 and avrA mutant presence was strongly reduced. Our inoculation procedure highlights a new role for Gala7 as a determinant of bacterial penetration.

M. truncatula Defenses Can Prevent Root Tip Vessel Invasion, Providing a Key Resistance Mechanism against R. solanacearum
The third step in the infection process (Fig. 8) was bacterial propagation in the vascular tissue. The avrA mutant was detected in vessels of A17 plants, but at a much lower rate than GMI1000, showing bacterial determination at that stage. Plant genetic background was also important at that stage, as GMI1000 was never detected in the vascular tissue of F83005.5 plants. As seen previously, we detected autofluorescent cell wall cortical cells, suggesting the presence of phenolic compounds. Therefore, we assayed for the presence of lignin, a phenylpropanoid known to strengthen the cell wall and inhibit further pathogen ingress (Hijwegen, 1963;Cano-Delgado et al., 2003). Earlier lignification of the F83005.5 root tip endodermis was observed and could constitute a barrier preventing bacterial propagation in the vessels.
This third stage was tightly linked to the next step (wilting development; Fig. 8). No wilting was observed when bacteria did not reach the vessels. This suggests that foliar symptom development depends on the bacterium's ability to penetrate the root tip vasculature in M. truncatula, which was not described before in other systems. Indeed, bacteria could be detected in the vascular system of resistant tomato plants, but at a low level that does not induce plant wilting (Grimault et al., 1995).
The avrA mutant strain was able to reach the same stages as GMI1000, but to a lesser extent at each step; therefore, it had a reduced pathogenicity on M. truncatula. Although the AvrA gene is widely distributed in R. solanacearum, disruption of avrA in GMI1000 strain did not lead to a reduced pathogenicity on susceptible hosts such as tomato and Arabidopsis . This report shows that avrA is required during the infection process as an important determinant of disease development in M. truncatula. It is often hypothesized that avirulence genes may have been conserved through evolution because the encoded protein may promote disease in another pathosystem (Stavrinides et al., 2008). The demonstration of a role for avrA in disease development on M. truncatula supports this hypothesis.

AvrA and Gala7, Different Effects in Steps of the Infection Process
The gala7 mutant never colonized intact roots and therefore never induced foliar wilting, whereas the avrA mutant showed reduced aggressiveness. Therefore, we wondered if those two effectors were also involved in the virulence of R. solanacearum after xylem invasion and wilting symptom development. Invasion of the bacteria was made possible by cutting the root tip at the time of inoculation. In this condition, the avrA mutation did not affect disease development. In contrast, the gala7 mutation led to a reduction of wilting symptoms. These results suggested that the two effectors have different roles in R. solanacearum virulence, with AvrA only involved in the penetration steps.
The effects of the two T3Es have been studied on both root symptom development and disease development of leaves of the susceptible line, and two different mechanisms were revealed. The involvement of the two T3Es differed between early and later stages in infection: they had additive effects on root tip epidermal cell death, whereas disease development in vascular tissues involved gala7 but not avrA. Root penetration likely requires a large number of molecules, consistent with the fact that the development of root symptoms involves a nonspecific mechanism, observed on both resistant and susceptible lines and, at least partially, on petunia (Zolobowska and Van Gijsegem, 2006). Bacterial colonization and disease development seemed to require host-specific signaling pathways and therefore likely require highly specific bacterial effectors.

Bacterial Strains and Culture Conditions
The wild-type strain GMI1000 of Ralstonia solanacearum and hrp mutant derivatives (GMI1525, a hrpB regulatory mutant, and GMI1402, a hrcS T3SS structural mutant) were described previously . Mutant strains carrying disruptions of the global virulence regulator phcA (GMI1610; Genin et al., 2005) and the T2SS (gspD gspN deletion mutant strain GRS465) were also used. GMI1559 is a GMI1000 derivative carrying a constitutively expressed GUS (uidA) fusion, and GMI1560 is a hrcV mutant also carrying this