Biocontrol potential of grapevine endophytes against grapevine trunk pathogens

Grapevine Trunk Diseases (GTDs) are a major challenge to the grape industry worldwide. GTDs are responsible for considerable loss of quality, production, and vineyard longevity. Seventy five percent of Chilean vineyards are estimated to be affected by GTDs. GTDs are complex diseases caused by several species of fungi, including Neofusicoccum parvum, Diplodia seriata, and Phaeomoniella chlamydospora. In this study, we report the isolation of 169 endophytic and 209 epiphytic fungi from grapevines grown under organic and conventional farming in Chile. Multiple isolates of Clonostachys rosea, Trichoderma sp., Purpureocillium lilacium, Epiccocum nigrum, Cladosporium sp., and Chaetomium sp. were evaluated for their potential of biocontrol activity against fungal trunk pathogens. Tests were carried out using two dual-culture-plate methods with multiple media types, including agar containing grapevine wood extract to simulate in planta nutrient conditions. Significant pathogen growth inhibition was observed by all isolates tested. C. rosea showed 98.2% inhibition of all pathogens in presence of grapevine wood extract. We observed 100% pathogen growth inhibition when autoclaved lignified grapevine shoots were pre-inoculated with either C. rosea strains or Trichoderma sp.. Overall these results show that C. rosea strains isolated from grapevines are promising biocontrol agents against GTDs.


Introduction 30
Grapevine trunk diseases (GTDs) are a major challenge to viticulture worldwide, because they 31 compromise the productivity and longevity of grapevines (Vitis vinifera L.) and increase production 32 costs (Munkvold et al. 1994  Sosnowski and Mundy, 2018). GTDs are managed mostly by practices that aim to prevent infections 49 (Gramaje et al. 2018; Mondello et al. 2018). Widely adopted preventive practices include late 50 pruning (Petzoldt, 1981;Munkvold et al 1994), double-pruning (Weberet al. 2007), and the 51 application of protectants on fresh pruning wounds . Pruning wounds can be 52 protected by benomyl and tebuconazole (Bester et al. 2007), inorganic compounds as boric acid 53 (Rolshausen and Gubler, 2005), or natural antifungal compounds as organic extracts (Mondello et al. 54 2018). Manual applications of these formulations as paints are effective, but costly and time-55 consuming, while spray applications are difficult due to the small surface and orientation of pruning 56 wounds (Bertsch et  in addition to plant-defense induction and antibiosis, they could also compete for space and nutrients 77 with GTD pathogens (Zabalgogeazcoa, 2008). 78 Here we report the isolation and identification of endophytic and epiphytic fungi from grapevines 79 grown in commercial vineyards in Chile. From this collection, we selected antagonist candidates and 80 evaluated them for growth inhibition activity against the main GTD fungal species found in Chile, in 81 co-culture, and in planta assays. We provide compelling evidence that endophytic and epiphytic 82 strains of C. rosea are strong antagonists of the main GTD species, which makes this species The isolation of endophytic fungi was performed following the methodology described in (Pancher et 96 al. 2012). Briefly, shoots (50 cm long) and roots were cut into 10-cm-long fragments. Fragments 97 were surface disinfected by rounds of 2 min serial immersions in 90% ethanol, then 2% sodium 98 hypochlorite solution, and, 70% ethanol, followed by double-rinsing in sterile distilled water under 99 laminar airflow. Absence of microbial growth on surface-sterilized shoots was confirmed by plating 100 the distilled water from the last wash step on potato dextrose agar (PDA; BD-Difco) in Petri dishes, 101 that were then incubated for 2 weeks at 25ºC. After disinfection, fragments were further cut into 2.5 102 mm pieces. Each section was placed on Petri dishes (90-mm diameter), placing the vascular bundle 103 towards the growing media, containing: i) PDA (39 g L -1 ; BD-Difco), ii) malt extract agar (MEA, 104 33.6 g L -1 ; BD-Difco), and iii) plain agar (AA, 20 g L -1 ; Difco), each one with antibiotics 105 (streptomycin, 0.05 g L -1 , and chloramphenicol, 0.05 g L -1 ). All Petri dishes were incubated at 25ºC 106 for 7 to 10 days under 12 h of light and 12 of darkness. Different colonies were tentatively identified 107 based in morphology ( Barnett and Hunter, 1955). Pure cultures were obtained from hyphal tip 108 transfer to PDA media and maintained at 5ºC. 109

Isolation of epiphytic fungi 110
For each plant, 1.5 g of soil in direct contact with roots was carefully collected. In a laminar flow 111 bench, 13.5 ml of sterile distilled water was added, before vigorous agitation for 20 min in a 112 horizontal position. After 5 min of decantation, serial dilutions of the supernatant were made. 10 -3 113 This is a provisional file, not the final typeset article and 10 -4 dilutions were used to inoculate PDA, MEA, and AA. To all media streptomycin, 0.05 g L -1 114 and chloramphenicol, 0.05 g L -1 were added. Plates were incubated for 7 to 14 days at 25ºC. 115

4
Taxonomic characterization of the fungal isolates 116 DNA extraction from cultivable isolated fungi (n=387 isolates) was performed as described in 117 Morales-Cruz et al. (2015), with the following modifications. Mycelia from 7 to 21 days old fungal 118 cultures were frozen with 3 mm metal beads in tubes at -80ºC. Tubes were shaken vigorously with a 119 vortex for 5 minutes at maximum speed. Disrupted mycelia were resuspended in 200 µL of nuclease-120 free sterile-distilled water and then homogenized in a vortex for 15 s. Mycelia was incubated at 121 100ºC for 10 min, followed by a centrifugation step at 14500 rpm for 2 min. An aliquot of 10 µL of 122 the supernatant was used for the PCR runs.

6.
Test of fungal antagonism 142 Initial assessment of antagonistic properties was conducted against D. seriata as pathogen. Further 143 evaluations on selected antagonists were carried out using D. seriata, N. parvum, and P. 144 chlamydospora. Agar discs from a 7-day old actively growing colony were used. Co-culture assays 145 were performed placing a 5 mm agar disc on one side of the Petri dish with PDA (39 g L -1 ; Difco) or 146 PA (200 g L -1 grapevine propagation material, 20 g L -1 agar) and on the opposite side a 5 mm agar 147 disc containing the antagonist strain. Plates were incubated at 25ºC for 7-28 days in darkness 148 (Badalyan et al. 2002) using a randomized complete block design. Registered bioproducts MAMULL 149 and TIFI were used as antagonistic controls. Pathogen growth area was evaluated at 7, 14, 21, and, 28 150 days post-co-culture (Schindelin et al. 2012). Inhibition percentage was calculated using the pathogen 151 growth area when was cultured alone (C) or in interaction with the antagonist (T) according to the 152 formula I = ((C-T)/C) * 100 (Thampi and Bhai, 2017). 153 An in planta assay was also performed. Annual shoots were used for the experimental set-up to verify 154 the antagonistic potential shown in plate co-culture. Several preliminary evaluations were carried out 155 in order to test variability caused by autoclave sterilization of pruning material, humid-chamber moist maintenance, type of inoculum and time needed for the pathogen to grow through the wood piece. 157 Even though tissue was death, the overall shoot matrix structure was conserved after autoclave 158 sterilization (data not shown). Internode portions of dormant cuttings were cut in 4.5 cm length 159 pieces and then used fresh or autoclaved for 25 min at 121 ºC. Agar mycelium plugs were evaluated 160 as inoculum. In 2 days, pruning material in contact with the pathogen and/or antagonist plugs were 161 covered in the mycelium. As the inoculum was too high, a spore suspension solution was used to 162 inoculate the wood pieces. Mycelium/spore mix suspension of the pathogens D. seriata and N. cutting, 20 g L -1 agar) with sterile distilled water. In the case of the antagonists Clonostachys rosea 165 (isolates CoS3/4.24, CoR2.15 and R31.6) a spore suspension adjusted to 1 x 107 conidia mL -1 was 166 used as recommended. Antagonist inoculation was carried out adding 40 uL of antagonist fresh spore 167 suspension until it reached the woody stem cut end by capillarity. Tebuconazole (60 mL/100L fields 168 recommended doses; SOLCHEM, Chile) or sterile distilled water was applied in the same manner as 169 controls. This experiment was carried out 5 times. Woody stem cuts were incubated in individual 170 humid chambers for 24 hours. Then, 10 uL of fresh pathogen mycelia/spore mix suspension was 171 inoculated on the same side where the antagonist was inoculated previously and immediately placed 172 in a horizontal position, preventing suspension diffusion. Incubation was carried out in humid 173 chambers for 3-7 days. Afterward, the surface of the woody stem was disinfected by rubbing with 174 70% ethanol. With a hot sterile scalp, the bark and 0.5 cm of the woody stem ends were removed. 175 Small pieces located at 1 and 2.5 cm from the inoculation point were collected and cultured in 176 individual PDA plates at 25ºC for 7 days. To evaluate the pathogen mycelia and spore suspension 177 viability, 10 uL of the solution was inoculated in one side of the wooden piece as described above 178 and immediately processed to obtain 3 mm pieces at 1 and 2.5 cm from the pathogen inoculation 179 point. Every piece was cultured in PDA at 25 ºC for 7 days. The presence of the pathogen on PDA 180 was evaluated under a light microscope. 181

7
Test of antagonist mechanism 182 To characterize the mechanism of antagonism, the same experimental setup of co-culture was carried 183 out on water agar (AA, 20 g L -1 ; Difco) with a microscope sterile slide covered by a thin layer of the 184 same agar in its surface. Using a light microscope (MOTIC BA410), the sample was screened for 185 loops of the antagonist hyphae around N. parvum and D. seriata, indicating mycoparasitism. This 186 experiment was carried out 3 times. To determine antibiosis as the type of antagonist mechanism 187 used, isolated fungi E. nigrum R39.1, C. rosea CoS3/4.4, and Cladosporium sp. B38d.2 were 188 cultured in PDA plates (39 g L -1 ; Difco) over cellophane paper for 7 days. Cellophane paper with the 189 fungal colony was then removed from the plate and a mycelial plug of D. seriata or N. parvum was 190 placed in the centre. Plates were incubated for 7 days at 25ºC and pathogen growth was evaluated. 191 This experiment was carried out three times. that has not been managed for disease protection for over 150 years. A total of 222 and 166 201 morphologically distinct filamentous fungi and yeasts were isolated from the commercial vineyards 202 and the Codpa Valley, respectively. Fungi were isolated and characterized taxonomically using ITS1 203 and ITS4 sequences. All fungal sequences were at least 98% identical to the best BLASTn hit in the 204 UNITE database. We could assign taxonomy to a total of 300 isolates. The ITS sequence was 205 discriminant at the species level for 227 isolates. The remaining were assigned to the corresponding 206 genus or family. A total of 58 genera were represented, 37 and 38 among epiphytes and endophytes, 207 respectively. As expected, below ground samples (rhizosphere and roots) were more diverse (56 208 genera) than sprouts and woody stems (5 genera) (Figure 1). 209  To assess the antagonistic ability of the ten selected isolates, we co-cultured each one of them with D. 224 seriata and N. parvum, two of the main fungi causing GTDs in Chile. Co-cultures were carried out on 225 two different types of growth media: the commonly used potato dextrose agar (PDA) and a substrate 226 made of agar and ground woody grapevine tissue (aka, grapevine plant agar (PA)) that simulates in 227 planta nutrient composition (Massonnet et al. 2017). Isolates displayed a wide range of growth rates, 228 which often differed between PDA and PA (Figure 2). Interestingly, most endophytes, including all 229 C. rosea isolates, grew faster on PA than PDA. Different growth rates reflected the patterns of 230 inhibition of D. seriata and N. parvum (Figures 3 and 4). The Trichoderma Altair isolate grew faster 231 than the rest on PDA and reached its maximum inhibitory effect on both pathogens as early as day 7 232 in PDA. Growth inhibition only occurred upon physical contact between colonies of Trichoderma sp. 233 and the pathogens. The faster growth on PA of the endophytes Clonostachys, Chaetomium, 234 Epicoccum, and Cladosporium was associated with greater pathogen inhibition rates on this substrate 235 compared to PDA, especially for the Clonostachys isolates. In PA, C. rosea overgrew the pathogen 236 colony at least 7 days earlier than in PDA. All C. rosea strains inhibited over 98% pathogen growth 237 in PA at day 21 (Figure 4) slow and limited growth of Neofusicoccum parvum was also visible in the halo produced by 247 Purpureocillium. Cladosporium sp. B38d.2 showed an interesting difference in antagonist activity 248 against N. parvum in PA, reaching its higher inhibition rate (Figure 4). When cultured with this 249 pathogen, Cladosporium strongly sporulated, covering the entire plate, and stopped N. parvum early 250 growth. 251   When C. rosea epiphytic strain CoS3/4.24 was co-cultured with D. seriata or N. parvum, pathogen 291 growth terminated before direct contact with C. rosea in correspondence of the halo surrounding the 292 antagonist. In this case, the inhibitory activity of C. rosea may depend on a secreted antibiotic 293 compound. This was also observed when Cladosporium sp. B38d.2 was used as antagonist. To test 294 the inhibitory activity of the C. rosea secretome, we inoculated C. rosea on a sterilized cellophane 295 membrane overlaid on PDA and incubated for seven days. The cellophane membrane was shown to 296 be permeable to metabolites secreted by fungi (Dennis and Webster, 1971;Chambers, 1993; 297 Sharmini et al. 2004;Rodriguez et al. 2011). After removing the cellophane membrane together with 298 the C. rosea mycelia, we inoculated the plates with pathogens and measured their growth in 299 comparison with normal PDA. Pathogen growth was significantly reduced on plates previously 300 This is a provisional file, not the final typeset article incubated with C. rosea, likely due to the secreted metabolites that permeated through the cellophane 301 membrane (Figure 7). The inhibition caused by the secreted metabolites of C. rosea CoS3/4.24 led to 302 a 47.2% and 50.1% reduction in growth of D. seriata and N. parvum, respectively. In the case of 303 Cladosporium sp., 34.26% and 42.46% inhibition was observed against N. parvum and D. seriata, 304 respectively. Changes in the pathogen colony morphology were also observed, especially when in 305 contact with C. rosea CoS3/4.24 isolate secondary metabolites. N. parvum colony turned into several 306 flat independent colonies with undulate margins, while D. seriata grew as one colony with irregular 307 shape. 308 309 FIGURE 7. Pathogen growth over secondary metabolites produced by antagonists C. rosea CoS3/4.24, Cladosporium sp.

4.
Effect of fungal antagonists on the growth of GTD fungi in one-year old grapevine woody 314 shoots 315 As both growth and inhibition rates of GTD pathogens were significantly different in media 316 containing grapevine annual shoot extract (plant agar, PA), we extended the testing of antagonism by 317 using one-year-old lignified shoots (aka canes) as a substrate for co-cultures. We tested both sterile 318 (autoclaved) and non-sterile canes. After 7 days, C. rosea, N. parvum, and D. seriata colonized 319 completely the internal tissue of 4.5 cm-long autoclaved canes. The antagonists C. rosea strains were 320 recovered in all pathogen co-inoculated samples after 7 days (Figure 8). No pathogen growth was 321 observed at 0.5 cm from the pathogen inoculation point when treated with the antagonists. 322 Interestingly, under the same conditions, Tebuconazole, a commercial synthetic fungicide, did not 323 reduce D. seriata nor N. parvum growth.

328
We also performed the co-culture experiments on canes that were not subjected to autoclaving. 329 Pathogens colonized the entire cane in 7 days in absence of any antagonist. In less than 0.1% and 330 10% of the co-culture assays, N. parvum and D. seriata were recovered from plant tissue previously 331 inoculated with C. rosea isolates, respectively. In the case of CoS3/4.24 isolate, N. parvum and D. 332 seriata growth inhibition was observed in 80% and 100% of the assays, respectively. In summary, 333 the antagonistic potential of the C. rosea isolates shown in agar plate was confirmed in grapevine 334 propagation material. 335

Discussion 336
We isolated fungi from grapevines to find potential biocontrol agents against GTDs. As they share 337 the same host with pathogens, these fungi may provide longer-lasting protection of grapevine tissues 338 than biocontrol agents identified on other plant species (Zabalgogeazcoa, 2008;Latz et al. 2018). 339 Three hundred eighty-seven different fungi and yeast were isolated and identified from multiple 340 grapevine tissues and pest management systems. The observed diversity was limited to culturable 341 This is a provisional file, not the final typeset article fungi, since no cultivation-independent identification tools were applied. Taxa were determined 342 solely based on the ITS sequence. Further validation using other informative sites, such as nu-SU-343 0817-59 and nu-SU-1196-39 (Borneman and Hartin, 2000) or TEF-1a (Ichi-Ishi and Inoue, 2005),  344 would provide additional resolution for some of the isolates we were not able to characterize at the 345 species level. As expected, rhizospheric soil showed to hold more fungal diversity than roots, and 346 sprouts showed less cultivable diversity than any other sample. This was in agreement with previous 347 studies using amplicon sequencing (Tan et al. 2017). 348 As the focus of this work was to find microorganisms able to colonize the grapevine persistently, we 349 conducted and CoR2.15 were endophytic, while CoS3/4.24 was isolated from the rhizosphere. Although we did 386 not find the same pattern when autoclaved tissue was used, the different behavior of endophytic and 387 epiphytic isolates supports the overall strategy to search for potential biocontrol agents among the 388 natural inhabitants of grapevines. 389 Generally recognized control mechanisms for fungal biocontrol agents are (1) competition for 390 nutrients and space, (2) induced resistance in the plant, both consisting in an indirect interaction with 391 the pathogen, (3) inhibition through antibiosis, and, (4) mycoparasitism (Latz et al. 2018;Köhl, 392 2019). The formation of short loops of the antagonist's hyphae around hyphae from another fungal 393 species also called hyphal coiling (Assante et al. 2004; Barnett and Lilly, 1962;Gao et al. 2005). The 394 coiling establishes an intimate contact with the parasitized hypha, penetrating the hypha and 395 delivering antibiotic compounds and cell-wall degrading enzymes (Barnett and Lilly, 1962). This 396 type of mycoparasitism has been commonly found in the genus Trichoderma (Howell, 2003;Benítez 397 et al. 2004) and reported in C. rosea (Barnett and Lilly, 1962;Morandi, 2001). The Trichoderma sp. 398 Altair isolate produced hyphal coils and also the C. rosea strains we tested. In all cases, we found a 399 strong correlation between coiling and antagonism suggesting that mycoparasitism plays an 400 important role in the interaction with the pathogens. In the case of C. rosea CoS3/4.24, a yellowish 401 halo around the antagonist colony was present. Antibiosis was previously described for this species 402 (Iqbal et al. 2017), but not all strains of the species show antibiotic production (Moraga- Suazo,403 2016). Further studies should be performed with the C. rosea isolates as this might have important 404 applications in agro-industrial areas (Karlsson et al. 2015). Direct interaction with the pathogen mode 405 of action, as mycoparasitism and antibiosis, are highly desirable mechanisms for further production 406 of commercial biocontrol agents, as they expose lower risks of human, plant and, environmental 407 toxicity (Köhl, 2019). 408

Acknowledgments 409
We would like to thank the VSPT group for let us collect samples from their vineyards and 410 supporting this project. We would also like to thank Patricio Muñoz for providing samples of old 411 grapevines from the Codpa valley. 412

Conflict of Interest 413
No conflict of interest declared. 414