Resistant tomato restricts colonization and invasion by the pathogen Ralstonia 1 solanacearum at four organismal levels 2

We show the spatio-temporal dynamics of the tomato- Ralstonia solanacearum interaction, revealing an out-of-the-xylem spread. We set the foundations to study the 45 complex molecular mechanisms that control each restriction point. Abstract 49 Ralstonia solanacearum is a devastating bacterial vascular pathogen causing bacterial 50 wilt. In the field, resistance against this disease is quantitative and only available for 51 breeders in tomato and eggplant. To understand the basis of resistance in tomato, we 52 have investigated the spatio-temporal bacterial colonization dynamics using non- 53 invasive live monitoring techniques coupled to grafting of susceptible and resistant 54 varieties. We revealed four different restrictions to the bacterium in resistant tomato: 55 root colonization, vertical movement from roots to shoots, circular vascular bundle 56 invasion and radial apoplastic spread in the cortex. We also show that the radial 57 invasion of cortical extracellular spaces occurs mostly at late disease stages but is 58 observed throughout plant infection. This work shows that resistance is expressed both 59 in root and shoot tissues and highlights the importance of structural constraints to 60 bacterial spread as a resistance mechanism. It also shows that R. solanacearum is not 61 only a vascular pathogen but spreads “out of the xylem”, occupying the plant apoplast 62 niche. Our work will help elucidate the complex genetic determinants of resistance, 63 setting the foundations to decipher the molecular mechanisms that limit pathogen 64 colonization, which may provide new potential precision tools to fight bacterial wilt in the 65 field. 66

5 solanacearum-infected tomato plants was proposed to correlate with the extent of 125 bacterial invasion into the secondary xylem tissues (Nakaho, 1997a,b). This limitation of 126 pathogen movement from the protoxylem or the primary xylem to other xylem tissues 127 was found most conspicuous in Hawaii 7996 (Nakaho et al., 2004). Other studies 128 described that cell walls were thicker in parenchyma and vessel cells of infected xylem 129 tissues in the resistant LS-89 than in susceptible Ponderosa or mock-inoculated plants 130 (Nakaho et al., 2000). Accumulations of electron-dense materials in vessels and 131 parenchyma cells were also described as more apparent in LS-89, while Ponderosa 132 showed necrosis in all parenchyma cells adjacent to vessels with bacteria (Nakaho et 133 al., 2000). A recent report microscopically studied R. solanacearum distribution in roots 134 of Hawaii 7996 and the susceptible cultivar West Virginia 700 and found that 135 colonization of the root vascular cylinder was delayed and movement inside the 136 vasculature was spatially restricted in Hawaii 7996 (Caldwell et al., 2017). 137 Together, these studies underscore the existence of a complex set of events that 138 restrict bacterial colonization in space and time in resistant varieties. However, a 139 systematic investigation of R. solanacearum invasion patterns at a whole plant and 140 tissue-system level is lacking. 141 Here, we have applied luminescent and fluorescent bacteria for the characterization of 142 bacterial wilt resistance in tomato root, hypocotyl, and stem organs at the tissular level. 143 We have compared highly susceptible, moderately resistant, and highly resistant grafted 144 tomato plants using a standard soil-based seedling grafting method and an in vitro 145 grafting method. We propose an integrative model for bacterial wilt in resistant tomato 146 lines that highlights the importance of four different restriction levels that limit bacterial 147 colonization: 1) Invasion of the root 2) vertical movement upwards to the stem, 3) 148 circular passage from vessel to vessel and 4) xylem escape and radial spread into the 149 pith/cortex tissues. For in vitro grafting, seeds were sown onto sterile filter paper placed on MS-containing 174 plates. Eight days after germination (seven for Marmande to obtain equivalent stem 175 diameters), cotyledons were removed and the plants were cut at a perpendicular angle 176 1 to 2 cm below the cotyledons using sterile tools. For double grafted plants, two 2 to 3 177 cm-distant-cuts were performed. Rootstocks and scions were transferred to fresh plates 178 without filter paper and matched with the corresponding reciprocal tissues without any 179 stabilizing device. Plates were kept standing upright in the growth chamber. After 10 180 days, successfully healed plants were either pin-inoculated with the luminescent strain 181 and monitored over time or transferred to soil-containing pots and grown as described 182 for pathogenicity assays after acclimation for 48h in transparent boxes (Altuna 2594005, 183 Stewart Garden) with vented lids opened after 24h. 184 For standard grafting, plants grown with stems 1.5-2 mm in diameter (9 days after 185 sowing) were grafted 2 cm below the cotyledons using a 70º angle cut and 1.6 or 2 mm 186 diameter grafting clips (Bato Plastics B.V). Grafted plants were kept into misted 187 acclimation boxes in growth chambers and acclimated to light (24h darkness, 24h at 188 10% light, 24h at 50% light) and then to ambient humidity (opening the vents 4 days 189 after grafting and partly opening the lid for 48h before removing it). 190 191

Plant inoculation and pathogenicity assays 192
For in vitro assays, 10 day old plantlets or plantlets 10 days after grafting were pin-193 inoculated 1 cm below the root collar using a sterile 0.3x13mm-sized needle (30Gx½, 194 BD Microlance, Becton Dickinson) submerged in a 10 6 CFU·ml -1 (OD 600 =0.001) R. 195 solanacearum suspension. Plates were kept in growth chamber (25ºC day, 22ºC night) 196 and wilting symptoms recorded and bacterial invasion visualized as detailed below. 197 For soil drenching inoculations, plants were grown until they reached between the 7 and 198 9 true leaf stage (4 to 5 weeks after sowing, and 5 to 6 weeks for grafted plants). Touch Imaging System, Bio-Rad) as previously described (Cruz et al., 2014) using a 5-210 minute exposure time with the 3x3 sensitivity. Images were processed using the Image 211 Lab software (Bio-Rad). Inoculated soil-grown plants were uprooted, roots were 212 surface-sterilized in water with ~5 to 10% bleach for at least 1 minute followed by a 213 wash in water. Plants inoculated with the luminescent strain were sliced from apex to 214 roots using a sterile razor blade. One mm-thick transverse sections and the two halves 215 of 1 to 2 cm-length radial slices were placed flat on a square plate with a misted lid and 216 visualized using live imaging system as detailed before. For each location, a 0.5 cm 217 8 section was excised and incubated for at least 30 minutes into a sterile 2 ml tube with 218 200 µl of sterile distilled water. Luminescence was measured on a luminometer (FB 12, 219 Berthold Detection Systems). Relative Light Units per second (RLU·s -1 ) were related to 220 Colony Forming Units per gram of tissue (CFU·g -1 ) after dilution plating of samples and 221 CFU counting 24h later. 222 Plants inoculated with the fluorescent strain were dissected as before and 223 photographed using binocular microscopy with a UV fluorescent lamp (BP330-385 224 BA420 filter) and DP71 camera system-equipped SZX16 Stereo microscope (Olympus). 225 Quantification of mean fluorescence in the green, blue and red channels was achieved 226 using the ImageJ software. 1A). While all tomato roots were colonized 3 days post inoculation (dpi) (Fig. 1B left  252 panel), shoot colonization was clearly delayed and reduced in H7996 compared to 253 Marmande as indicated by the percentage of plants in which bacterial colonization was 254 detected (Fig. 1B right panel). A representative photograph of the assay at 4 dpi, when 255 the susceptible plants start to wilt, is presented in Figure 1C. This image shows that, Marmande shoots, while Marmande roots did not prevent colonization of the H7996 269 scions ( Fig. 2A). Interestingly, the presence of a resistant root system was sufficient to 270 cause a reduction in shoot colonization, as stem luminescence was comparable in 271 grafted plants with or without a resistant lower stem (Fig. 2B). hypocotyl regions of the other variety (Fig. S6A,B). The double-grafted plants were 336 grown on soil to 7-9 true leaf stage and infected with the luminescent R. solanacearum 337 strain ( Fig. S6C-G). As expected, plants that contained the roots and basal hypocotyl 338 from Marmande wilted similarly to plants with Marmande rootstocks (Fig. S2, S6D,E). 339 We observed and quantified bacterial movement along the xylem in the two 340 combinations of grafted plants using luminescence (Fig. 4). Marmande rootstocks were 12 heavily colonized by R. solanacearum, and bacterial density decreased as soon as the 342 pathogen crossed the first grafting junction and encountered H7996 tissue. When R. 343 solanacearum moved upwards into susceptible tissue for the second time, it multiplied 344 again to high levels ( Fig 4A,B top panel and graph). The complementary result was 345 observed in the reciprocal grafting: colonization was hampered in H7996 rootstocks, 346 especially at 10 dpi (Fig S6F,G), reached its peak on Marmande hypocotyls and 347 decreased when R. solanacearum crossed the second grafting junction and faced again 348 Marmande tissues below the graft (Fig. 6B and Fig. S8 left panel). In contrast, the 395 section at the graft junction level showed that H7996 tissues immediately blocked the 396 spread of the bacterium circularly through the xylem ring and radially to the pith and 397 cortical tissues (Fig. 6B). These restrictions became more apparent at higher sections, 398 consisting exclusively of resistant tissue (Fig. 6B and Fig. S8  wilting. Figure 6C shows these H7996 shoot sections confronted with a high bacterial 403 inoculum introduced from the susceptible rootstock, compared to Marmande shoot 404 sections. Noticeably, radial bacterial movement from the highly colonized xylem bundles 405 became strongly restricted in H7996 shoots, even in these extreme cases where the 406 xylem tissue was highly colonized (Fig. 6C right panel). This restriction could also be 407 observed when the fluorescent R. solanacearum strain was directly pin-inoculated into 408 the shoots (Fig. S9). 409 Finally, we performed a time-course invasion assay in which we quantified the amounts 410 of bacteria that were moving outside the vascular ring over time (Fig. 7). We observed 411 that R. solanacearum was escaping from the vascular ring as early as 5 dpi and heavily 412 colonized the pith and cortical tissues by 9 dpi (Fig. 7A top panels and 7B). Moreover, 413 the amount of bacteria located outside the vascular tissues was directly correlated with 414 the extent of vascular ring colonization (Fig. 7B). This contrasted with the ability of 415 H7996 shoots to impede pathogen escape from the vascular ring ( Fig. 7A top panels, 416 shows how quantitative resistance impacts these parameters (Fig. 8). Systematic  We discuss below each of these four important levels that can turn the scales towards 432 host resistance or successful plant colonization. 433

Restriction of root colonization 435
We analyzed the R. solanacearum interaction with tomato using two main variables: limitation of bacterial movement between xylem tissues (Nakaho et al., 2000(Nakaho et al., , 2004. 523 The detailed description of the process we present here will be crucial to decipher the 524 genetic determinants and the composition of these vascular coatings, which remain 525

unknown. 526
Circular restriction in the stem is a very efficient confinement strategy, since it is still 527 acting when high loads of bacteria are forced into the stem through root-inoculations 528 using H7996 scions grafted onto Marmande rootstocks (Fig. 6). However, there seems 529 to be an upper limit of bacterial inoculum beyond which this restriction is no longer 530 effective (see Plant number 4 in Fig. 6C lower panel). This is in agreement with previous 531 reports showing that delivering a high R. solanacearum inoculum (10 9 CFU ml -1 ) directly 532 in tomato stems overcomes resistance (Nakaho, 1997b). This idea relates to the 533 concept of a density threshold in the interaction between tomato and R. solanacearum. 534 Earlier observations established the onset of bacterial wilt symptoms at a density in the 535 stem between 10 6 and 10 8 CFU g -1 of fresh tissue (Grimault and Prior, 1994;Nakaho, 536 1997a; Huang and Allen, 2000; Nakaho et al., 2004). We have characterized this 537 threshold systematically assessing bacterial densities throughout the plant in large 538 populations of grafted tomatoes with varying resistance. We conclude that, both in 539 resistant and in susceptible varieties, symptom appearance invariably takes place when 540 bacterial populations in the hypocotyl exceed a threshold of 10 7 CFU per gram of tissue 541 ( Fig. 5 and Fig. S7). Plating dilutions of homogenized tissues is labor intensive, but we 542 show that light emission from tissues inoculated with a luminescent strain is a useful 543 measure of bacterial counts (correlation coefficient 0.9). Since bacterial density and 544 distribution is predictive of the degree of disease resistance, we have started using 545 luminescent strains to screen potato germplasm for resistance to bacterial wilt as a way  (Figs. 6 & 7). 552 These metabolically active cells are in close contact with the xylem vessels through the 553 pits and are thought to be pivotal for the induction of plant defense against xylematic 554 pathogens, although very little is known about the mechanisms regulating this response. 555 Earlier works detected widespread R. solanacearum colonization of stem parenchyma 556 cells in susceptible tomato varieties at late stages of infection, when plants showed 557 extensive wilting (Nakaho, 1997a;Nakaho et al., 2000). These cells appeared filled with 558 bacteria and displayed necrosis symptoms and signs of degeneration. On the contrary, 559 in resistant tomato varieties, necrotic parenchyma cells containing bacteria were 560 observed occasionally (Nakaho et al., 2000). Our data confirm these observations and 561 additionally show that parenchyma cell invasion starts at earlier times (5 dpi) in 562 susceptible plants and spreads massively through the pith at late time points (8-9 dpi, 563 Fig. 7A). In contrast, colonization remains limited to xylem vessels in resistant tomato 564 (Fig. 6). 565 As for the previously described bacterial movements, radial restriction out of the xylem 566 in resistant varieties can be partially overridden by grafting to susceptible rootstocks that 567 enable high bacterial densities to access resistant tissues, as can be seen in some of 568 the images in Fig. 6C. This is in agreement with a previous report showing that when 569 high bacterial inocula were used (10 9 ), R. solanacearum could also be detected in the 570 parenchyma cells of resistant tomato (Nakaho, 1997b). Thus, restriction of radial 571 bacterial movement is no longer effective when bacterial densities surpass a certain 572

threshold. 573
Structural changes in cell walls and pit membranes in response to R. solanacearum 574 infection are more conspicuous in resistant tomato (Nakaho et al., 2000). Thus, bacteria 575 may be prevented to escape the xylem in resistant tomato by a combination of inducible 576 structural defense mechanisms that may appear later and/or with less intensity in 577 susceptible lines, rendering them ineffective to restrict colonization. Very interestingly, 578 slightly decreased invasion can also be observed in the susceptible hypocotyls of the 579 Marmande-H7996 grafting combination (Fig. 7). This finding could be explained by a   solanacearum. Shoot sections were obtained at 6 dpi and photographed in a live imager. In (A), photographs represent each bisected fragment and its top and bottom slices exposed. Sections were obtained at the base of the hypocotyl, the distal 28 hypocotyl (right below the cotyledons), and the internodes 1, 2 and 3. In the Image Lab software (Bio-Rad) the following 'High'/'Low'/'Gamma' values were used for low and high exposure settings, respectively: 10000/60/1 and 1300/60/2. In (B), sections were obtained above and below the graft junction. The arrowheads and dotted lines indicate the position of the graft junction.  Only one self-grafted H7996 plant wilted, hence the lack of boxplot. Values between 0 and 4 lie below the threshold for luminescence detection (see Supplementary Figure   S3) and are here considered as zeros. From left to right, sections correspond to: taproot, basal hypocotyl, distal hypocotyl, internodes 1, 2 and 3. The dashed red line highlights the location of the grafting union. Letters above each boxplot indicate significant statistical difference by Fisher's LSD (α=0.05). Within each boxplot, the whiskers extend from the hinges to the largest (upper whisker) or smallest (lower whisker) value no further than 1.5 * IQR from the hinge (where IQR is the inter-quartile range, or distance between the first and third quartiles). Dots beyond the end of the whiskers are outliers. The band inside each box indicates the median.