| Bacterial wilt disease,caused by the plant pathogenic bacterium Ralstonia solanacearum,limits crop production for cash and subsistence growers worldwide.This pathogen presents in all the continents around the climate warm regions.The extensive genetic diversity of strains responsible for the bacterial wilt disease has in recent years led to the concept of a R.solanacearum species complex(RSSC).Genome sequencing of more than 100 representative strains from each phylogenetic groups has broadened our knowledge of the evolution and speciation of this bacterium and led to the identification of novel virulence gene functions.Thus,many studies about plant-R.solanacearum interactions have been carried out and have shed light on the genetics,molecular biology,and disease development.In other words,these studies involve in how plants prevent R.solanacearum infection via nonspecific and specific immune defenses and how bacteria evade from these defenses,sequester nutrients essential for their replication,and cause disease via pathogenicity traits.These qualitative achievements are fundamental to understand the pathogenesis of R.solanacearum,but they are insufficient to know whether plants are diseased or not.These fundamental questions eventually require the quantitative understanding of the processes responsible for the rise,dissemination and fall,and even evolution,of the infection population of R.solanacearum.Our work pioneered the quantitative study in plant bacterial pathogens from addressing a question about “How many R.solanacearum individuals enter from the root to establish the bacterial wilt disease on the host plant?” It allowed us to determine what factors control the infection dynamics of R.solanacearum within the host plant.1 Novel approach for bacterial wilt phenotypingTo enhance the capacity of current phenotyping on bacterial with disease,we monitored both disease index and weight loss dynamics of tomato plants within 9 days post inoculation with varying strains or concentrations of R.solanacearum.The weight matrix over time was transformed into the relative water loss or cumulative water loss data frame as parameter of plant status upon R.solanacearum infection.Then,we developed an approach to depict the phenotype of bacterial wilt according to this physiological water loss in the host plant.The results showed that there were significant differences in the water loss of plants in different trials.Interestingly,the cumulative water loss is more sensitive to the water loss and has a strong correlation with disease dynamics.It is very powerful to characterize the development of bacterial wilt.The detection of the water loss dynamics is a stable and effective phenotype on the host plant during the infection by R.solanacearum.Thus water loss dynamics may be a new physical indicator for plant under biological or abiotic stress studies.This approach would be useful to integrate into high-throughput phenotypic analysis platform for the plant science.2 Hypothesis of on infection dynamics model and definition of model parametersThe infection cycle of R.solanacearum in host plant can be divided in five subsequent steps: 1)root invasion,2)passage through the root cortex,3)proliferation within the xylem vessels,4)a phenotypic switch,inducing EPS production,and 5)plant wilting and pathogen transmission.This initial population entered the host plant can be regarded as the infection founding population.We modelled the complete infection cycle allowing for a holistic approach of bacterial pathogenesis in order to reach beyond detailed cellular and molecular studies of R.solanacearum.A combination of model prediction and experimental plant bioassays,mimicking the natural infection conditions,were used to analyse both the infection and the within-host population dynamics.Seven parameters were refined from the whole life cycle of R.solanacearum in host plant.These parameters are: 1)the plant-pathogen association constant(K),2)the maximum entry flow(Vfmax),3)the reduction of the maximum entry flow over time(z),4)delay time to reach the xylem(d),5)the growth rate in xylem(?),6)the quorum sensing threshold(Q),7)the steepness of plant wilting response to EPS concentration(?).We designed a mathematical model to investigate impacts of these parameters on the infection founding population(Ni).In our model simulations,five of these parameters lead to a reduction of the amount of founders when their value increased.These parameters are: 1)the plantpathogen association constant(K),2)the reduction of the maximum entry flow over time(z),3)the growth rate in xylem(?),4)the quorum sensing threshold(Q),and 5)the steepness of plant wilting response to EPS concentration(?).Overall,the number of founders is determined by a trade-off between bacterial entry flow into the roots(controlled by K,Vfmax and z)and the wilting kinetics(controlled by ?,Q and ?).For K and z the decrease of the number of founders is due to a reduction of an effective entry flow of cells.On the other hand increasing parameters ?,Q and ?,leads to a delayed transmission phase and thus reducing the amount of bacteria entering from subsequent re-infections.However,the maximum number of successful founders can dramatically increase the final founder flow value(Vfmax).3 Modelization on infection dynamics 3.1 Host plant can be re-infected by R.solanacearumWild-type R.solanacearum strain GMI1000 and its gentamicin-resistant GRS540-derived strain were used for sequential inoculation experiments on tomato plants.Tomato plants were first inoculated with GMI1000 strain,then re-inoculated with GRS540 strain after 1,8,23,and 48 hours post inoculation(hpi)with the first wild type GMI1000.At 5 dpi of GMI1000,all inoculated plant were subjected to detect the ratio of GRS540 infection.Our data showed that when the secondary infection occurred at 23 hpi,> 60% of the plants were still infected by GRS540,whereas at 48 hpi,only few plants was infected by GRS540 strain.Therefore,re-infection may continue to occur within 48 hours of inoculation with R.solanacearum,and the reduction of the maximum entry flow over time(z)was determined as 2 ± 0.1% h-1.3.2 Quantification of model parametersThe median infection dose(MID)of R.solanacearum was quantified to evaluate the infectivity of this pathogen in tomato by Reed and Muench method.The MID corresponds to the bacterial inoculums that trigger disease for,at least 50%,of the inoculated plants.The MID gradually declined until 10 dpi to an asymptotic value of 5.9?106 bacterial cells.To ensure that the MID is a feature of the soil-root interface,we performed a similar assay but using stem injection to bypass the roots.Here,the MIDstem was only 4.1×103 cells at 5 dpi.During the colonization,the bacterial load in infected plants was quantified at different times following root infection in order to evaluate the bacterial growth rate in the plant stem.Fitting data points between days 2 and 5 yielded an exponential growth rate of 0.196 h-1 ± 0.039 with low correlation coefficients,which may result from the intrinsic variability of the timing of the first infection associated with soil inoculations procedures.However,we found a growth rate at 0.233 ± 0.012 h-1 that is in the range of the in vitro growth rate in minimal medium: 0.253 h-1 ± 0.034 with high correlation coefficients.The Matthews correlation coefficient(MCC)was used to find the best predicting threshold value of bacterial load inducing the wilting symptom.A maximum MCC value of 0.896 was obtained for a bacterial threshold of 6.107 ± 0.97.107 cfu?g-1FW.This value is in strong agreement of published quorum sensing bacterial load.We further exploited a large data set of more than 500 individual infected plants to extrapolate the wilting time response to the EPS production in tomato plants.For this purpose,all wilting kinetics of diseased individuals were synchronized to the first day of wilting symptoms appearance.The steepness,?,of the fitted wilting time response curve was 0.612 ± 0.012.Then,we determined the in planta EPS concentration from the known flux of EPS production,VEPS,of R.solanacearum and the measured bacterial load and growth rate.This EPS concentration over time was used to assign the scaling factor,? = 2.31,to obtain the wilting dose response curve to EPS concentration.Finally,we determined the reduction of the proliferation,?-,due to the collapse of the host tissues at late infection.From measured data of advanced wilting symptoms and bacterial load,we fitted a value of 0.25 h-1 for the reduction of the proliferation,3.3 Model simulation and experimental validation of infection founder sizeThe above experimentally defined parameters,namely K = 5.9?106,z = 0.02,? = 0.23,q = 6.107,a = 2.31,b = 0.612 and ?-= 0.25,were used as input in the mathematical model in order to predict the infection bottleneck size for R.solanacearum strain GMI1000 leading to successful tomato infection.First,we estimated the maximum effective entry flow of founders,Vfmax,from fitting the model prediction to experimental wilting onset.We found Vfmax to be in the range of 5 to 22 cells?h-1.The model predicts that the actual number of founders at 7 dpi should be in the range of 90 to 476 for an inoculum at 5×107 cell?m L-1.We then experimentally determined the size of the infection bottleneck,i.e.the number of founders(Ni)in tomato using a method based on the co-infection of GMI1000 mixed with minute fractions(from 1% to less than 0.1%)of GRS540.Importantly,both strains were shown to be of equal fitness in planta.The probability of an infected plant to also contain the marked strain is directly correlated with the proportion of the marked strain in the original inoculum(p < 1%)and the size(Ni)of the infection bottleneck.Sets of 32 plants were inoculated with given GMI1000/GRS540 ratio(p).The resulting proportion of plants infected by GRS540 was then used to generate a single estimation of Ni through a probabilistic analysis.After extensive,60 independent inoculations of sets of 32 plants,the median of Ni was estimated to be 458 cells.We observed a high variability of Ni with a two times standard deviation of 2734 cells.This variability could reflect the true biological variability of Ni,however it may also arise from the stochastic sampling of the marked strain inherent to our experimental setup.Thus,in order to evaluate the contribution of a random sampling of the marked strain on Ni variation,we simulated a random sampling based on the same parameters as the experimental setup(number of plants,concentration of marked strain in inoculum).For a true Ni value of 458 the stochasticity analysis gave a median Ni of 454 cells with a two times standard deviation of 507 cells.Hence,we evaluated that the stochasticity of individual sampling due to our experimental setup could represent up to 18%(507/2734)of the variability observed in the experimental Ni estimation.4 Infection bottleneck analysis of R.solanacearumThe bottleneck from disease triangle,i.e.the nature of host bottleneck,pathogen strategies and environmental changes continues to be a central topic in studies of host-pathogen interactions.Host defenses that counteract infection can be thought of as bottlenecks,as can some intrinsic features of the host environment,i.e.physical and immune barriers.Pathogens circumvent pathogenic determinants like secretion systems effectors for kill host defenses and bacterial competitions.To understand mechanisms restricting infection population size,we used a wounded-root inoculation procedure to estimate the contribution of plant natural root barriers to the infection bottleneck size.At 4 dpi after inoculation,all wounded-root plants had symptoms,indicating an accelerated infection and colonization.We used the same methodology of mixed inoculations to determine the impact of wounding on the infection bottleneck size,i.e.dramatically increasing to a median of 24215 cells.We then treated tomato plants with a hrp mutants,eps mutants and tss F mutants to evaluate impacts of bacterial virulence factors on the infection bottleneck size.At 10 dpi,the infection bottleneck size of the hrp defective R.solanacearum strain was estimated significantly lower than NWT at a median of 6 cells(and mean of 7 ± 4 cells)following soil drenching inoculation.The founder amount of the hrp mutant increased to a mean of 62 ± 40 cells and a median of 53 cells following the wounded-root inoculation.The mean of Nhrp_wounding was about 9 times higher than Nhrp but 482 times lower than NWT_wounding.The infection bottleneck was significantly restricted 22 and 19 times when tomato plants were exposed with the eps mutant and tss F mutant compared to the wild type of R.solanacearum.Interestingly,these bacterial factors had different contributions to the proliferations of this soil pathogen in planta following the stem injection.The hrp and tss F mutants decreased their growth rates within host colonization,but the growth rate of eps mutant is indistinguishable from the wild type strain.We performed same experiments of wilt type strain at 30 °C to evaluate the impact of high temperature on infection bottleneck size of R.solanacearum in tomato plants.High temperature increased the in planta growth rate of R.solanacearum but restricted 4 times of the infection bottleneck size compared to in common lab conditions.Furthermore,we tested whether this research scheme can be adapted to other model pathosystems,e.g.the in vitro pathosystem of Medicago-R.solanacearum.The median of the infection bottleneck size was experimental estimated around 289 cells that was no significantly different from the value of the model simulation.5.Cumulative genes on the pathogenicity of R.solanacearumWe perform a cumulative gene disruption analysis based on the rip H gene family to initially understand how the multigene contributions to disease dynamics.The Rip H family consists of Rip H1,2 and 3 named as HLK1,2 and 3 before.In general,rip H family genes showed the functional redundancy in virulence to tomato plant.The triple rip H mutant GRS522 of R.solanacearum dramatically attenuated on tomato plants compared to the wild type strain.Interestingly,infection dynamics study of GRS522 revealed that three rip H collectively contribute to reducing infection rather than colonization in tomato.We performed the bacterial enumeration in planta to determine the internal growth curve of the triple mutant GRS522.It show a reduction or delay of bacterial infection in planta,but the bacterial distribution of bacterial load showed no statistically different from the wild type.We generated the several multiple effector mutants based on the genome background of GRS522.These mutants are MEM1(GRS522-Rip A2),MEME3(MEM1-Rip R),MEM33(MEM3-Rip AA)and MEM38(MEM33-Rip D).All the mutant have strongly delayed the symptom onset in tomato plant compared the wild type and dramatically reduced the pathogenicity.Survival analysis showed there no significant difference between each multiple mutants.None of these mutants were able to define the median survival time.MEME1 triggered more disease plants than other mutants but was unable to statistically distinguish from its original strain GRS522.6 Effect of pesticides on disease dynamic and model validationCommonly pesticides on bacterial wilt control,Streptomycin sulfate(SS),Thiodiazole copper,(TC)Pseudomonas fluorescens(PF)and Bacillus subtilis(BS),were used to understand the influence of the regulation means on the infection dynamic model of bacterial wilt.The results showed that bacterial wilt severity was reduced by these four treatments on tomato plants.Among them,TC was shown as the most effective pesticide on bacterial wilt control,while SS had strongest inhibitory effect on the motility of R.solanacearum.The colonization of R.solanacearum in soil was significantly reduced by the treatments of SS,PF and BS but TC had no effect on its colonization in soil.We took further to understand the mechanism of TC regulation on bacterial wilt disease,i.e.impacts of TC on growth rate and infection founders size of R.solanacearum.Our data indicated there are no significant difference between growth rates of R.solanacearum in TC treated(0.1876 ± 0.026 h-1)and mock plants(0.194 ± 0.042 h-1).Infection founding population size was significantly reduced from 476 cells to 227 cells when tomato plants were in the treatment with TC.This was in the agreement with model simulation for TC treatment,i.e.a value of 208 cells which was no difference from the measured value.This work indicated that our model is suitable for predicting the effect of pesticide regulation on the occurrence of bacterial wilt.In this thesis,we have addressed and answered some important questions concerning bacterial wilt in natural conditions.The pathogen infects plants through roots,colonises within the xylem preventing the movement of water into the upper portion of the plant tissue,and eventually kills the plant.To accomplish the entire life cycle,what complexity in the nature,how many bacteria first get inside plant,and how these penetrated pathogens replicate and spread in their hosts are lacking in current molecular/qualitative studies.We used a combination of model prediction and experimental plant bioassays,and aimed to propose a holistic view of bacterial pathogenesis as well as the physical water loss caused by pathogen infections,to reach beyond the many detailed molecular studies that evaluate smaller parts of the interaction.Finally,the effects of the three types of secretion-system effector genes and their regulatory effects on the dynamic model of R.solanacearum infection were evaluated,and the feasibility of the model was further verified.The pathogenic mechanism of R.solanacearum and plant responses to this infection were evaluated.This work provides a more systematic and comprehensive understanding of the host and pathogen interaction mechanism.Meanwhile,influence of pesticides on the model was characterized and evaluated.It clarified that this generalized model was suitable for predicting the influence of pesticide regulation on bacterial wilt,which will be important implications for bacterial wilt management and precise pesticide use. |
PDF Full Text Request