Soil bacteria as sources of virulence signal providers promoting plant infection by Phytophthora pathogens

Phytophthora species are known as “plant destroyers” capable of initiating single zoospore infection in the presence of a quorum of chemical signals from the same or closely related species of oomycetes. Since the natural oomycete population is too low to reach a quorum necessary to initiate a disease epidemic, creation of the quorum is reliant on alternate sources. Here, we show that a soil bacterial isolate, Bacillus megaterium Sb5, promotes plant infection by Phytophthora species. In the presence of Sb5 exudates, colonization of rhododendron leaf discs by 12 Phytophthora species/isolates was significantly enhanced, single zoospores of P. nicotianae infected annual vinca and P. sojae race 25 successfully attacked a non-host plant, Nicotiana benthamiana as well as resistant soybean cultivars with RPS1a or RPS3a. Sb5 exudates, most notably the fractions larger than 3 kDa, promoted plant infection by improving zoospore swimming, germination and plant attachment. Sb5 exudates also stimulated infection hypha growth and upregulated effector gene expression. These results suggest that environmental bacteria are important sources of virulence signal providers that promote plant infection by Phytophthora species, advancing our understanding of biotic factors in the environmental component of the Phytophthora disease triangle and of communal infection of plant pathogens.

that release metabolites into the environment 25,26 . Some of metabolites play important roles in the life cycle of Phytophthora species. They enhance the growth and reproduction of Phytophthora species [27][28][29][30] . They also may act as plant hormone mimics stimulating the opening of plant stomata to allow pathogens to enter the apoplast and impair defense gene expression [31][32][33][34] . They may also act as transformers of plant products promoting infections by these pathogens. For example, bacteria can convert isoflavones produced by soybean plants, into a substance that is 100 times more attractive to zoospores of P. sojae 35 , consequently facilitating zoospore aggregation on or around plant surface, a prerequisite of infection process. Similarly, plant colonization by Phytophthora species is associated with abundant bacterial populations in the rhizosphere 36 . However, such beneficial impacts of environmental microorganisms on plant pathogens are largely overlooked in current research 24 . Instead, negative or detrimental impacts have been a focus aiming to identify a biocontrol agent for the diseases although effective products are rare [37][38][39] .
In contrast, the importance of a beneficial relationship between pathogens and the surrounding microorganisms has been better recognized in medical science 40,41 . For example, infection by Candida albicans in humans has been linked to co-colonized bacterial biofilms 42 . Infection of immunocompromised patients by Cryptococcus neoformans is attributed to virulence precursors provided by Klebsiella aerogenes bacterial microflora 43 . These studies have provided unparalleled insights into pathogen ecology, pathogenicity, and disease epidemiology, facilitating our understanding of how plant infections occur under natural conditions and ways to counteract.
Here, we explore the beneficial effects of soil bacterial exudates on Phytophthora species during plant infection. We focus on a soil isolate of Bacillus megaterium, Sb5, and demonstrate that environmental bacteria are sources of signals for Phytophthora virulence as well as fitness, and communal infection is a life strategy of these pathogens.

Soil bacteria enhance the colonization of rhododendron leaf discs by Phytophthora species.
In an initial assay with P. nicotianae isolate 1B11, soil water extract (SWE) and its five major bacterial component species (Sb1, 3 to 6) were compared with sterile distilled water (SDW) and surface rinsate of potato dextrose (PDA) agar medium used to grow bacteria (SDWms) for their impact on pathogen colonization. This was evaluated by submerging leaf discs of Rhododendron, a universal suscept, overnight in zoospore suspensions prepared with autoclaved bacterial cell suspensions of individual isolates in PDA plates, autoclaved SWE (ASWE), SDW or SDWms. Sb5 and ASWE significantly enhanced colonization of leaf discs by P. nicotianae when compared to SDW (Fig. 1). In contract, the other 4 bacterial isolates (Sb 1, 3, 4 and 6), and SDWms had no impact on the Phytophthora colonization.
To confirm the activity of Sb5, additional tests were conducted with 11 other Phytophthora species/isolates. Sb5 significantly promoted disc colonization by all the species and isolates assessed ( Table 1). ASWE was also included for the purpose of comparison. As expected, it also promoted disc colonization by all the species and isolates. Sb5 exudates promote plant infection by Phytophthora species. Sb5 was further assessed with the cell-free filtrate (CFF) for its ability to promote infection of whole leaves and intact plants by Phytophthora species using three phytopathosystems: annual vinca -P. nicotianae, soybean -P. sojae, and tobacco -P. sojae. In the annual vinca -P. nicotianae system, a single zoospore infection assay 11 was performed with detached leaves to determine whether Sb5 extracellular products may be used as quorum sensing signals by the pathogen to initiate infection. P. nicotianae caused extensive necrotic lesions on leaves 72 h after inoculation in the CFF treatment but did not in the negative control with SDW (Fig. 2a). Similarly, the percent sites blighted was significantly greater in the CFF treatment than negative controls (Fig. 2b). Like CFF, ASWE had the same level of infection-promoting Figure 1. Soil bacterial products enhance colonization of rhododendron leaf discs by Phytophthora nicotianae. Leaf discs were submerged in a suspension at 10 zoospores/ml in an autoclaved filtrate of bacterial cell suspension of individual isolates (Sb1, Sb3, Sb4, Sb5 or Sb6) prepared from potato dextrose agar (PDA), autoclaved soil water extract (ASWE), sterile distilled water (SDW) or rinsate of PDA plate surface with SDW (SDWms). The discs were examined for colonization 48 h after being plated on PARP-V8 agar. The graph depicts colonization rates (%) of the treatments. Each column is a mean of three repeating experiments. Bars depict standard error. Columns with the same letter are not different according to LSD test at P = 0.05.
In the soybean -P. sojae system, responses of five differential soybean plants to its counterpart P. sojae race 25 were examined. Susceptible Williams and L77-1863 plants developed significantly larger lesions and had greater infection rates in the CFF treatment than the SDW control (Fig. 3). Similar infection-promoting effect was observed in two of the three resistant differential plants. L88-8470 plants had significantly larger lesions and higher infection rates in the CFF treatment than the controls, so did L83-570 (Fig. 3). The only exception was L92-7857 plants carrying RPS3c resistance gene; no infection developed in either treatment (Fig. 3). These results indicate that CFF altered the interaction between race 25 of P. sojae and the differential plants except for L92-7857 with RPS3c.  In the Nicotiana benthamiana -P. sojae system, CFF was evaluated for its impact on the interaction between a nonhost plant and a Phytophthora species. Nearly 70% of detached leaves were infected when inoculated with zoospores suspended in CFF. Blight symptoms developed 3 days after inoculation (Fig. 4a left). In contrast, the control leaves developed no symptoms or slight yellowing (Fig. 4a right). P. sojae was isolated from surface-sterilized leaf tissue with the blight symptom as confirmed using ITS-SSCP analysis of resultant cultures ( Fig. S1) but did not from that with the yellowing symptom. These results indicate that P. sojae was the causal agent of the symptom and CFF triggered leaf infection on a nonhost plant by P. sojae. CFF also triggered infection of N. benthamiana seedlings by P. sojae. N. benthamiana seedlings did not develop any disease symptoms after the roots were submerged for a week in the suspension of either P. sojae at 3,200 zoospores/ml or Sb5 at 10 6 cells/ml alone. In contrast, those with roots submerged in a mixture suspension of P. sojae zoospores and Sb5 cells at the same concentrations developed severe damping-off and blight symptoms. The number of diseased plants decreased with decreasing bacterial concentration in the mixture; Sb5 cells at ≤ 1,000/ml did not result in a significant difference from the control (Fig. 4c). As from infected leaves, P. sojae, but not the bacterium, was isolated from surface-sterilized roots of diseased plants. This again confirmed that P. sojae but not Sb5 was the causal agent of the disease, and Sb5 was a stimulator of N. benthamiana infection by P. sojae. Sb5 exudates stimulate zoospore homing and morphogenesis. To understand how CFF facilitated plant infection by Phytophthora, zoospore behaviors associated with infection process were examined. Zoospores of all the Phytophthora species tested exhibited prolonged swimming in CFF, 14-20 h longer than those in SDW. In the case of P. nicotianae, 90% zoospores maintained swimming for at least 2 h in CFF. Among these, about 10% kept swimming for longer than 24 h. In contrast, the majority of zoospores in SDW encysted in less than 2 h, and only few of them made overnight. In the case of P. sojae, promotion of zoospore swimming by CFF was not as evident. Many zoospores encysted before the treatment, indicating that zoospore encystment of treatments was not resulted from osmotic shock.
More zoospores attached to the root surface of Arabidopsis in the presence of CFF (Fig. 5). Approximately nine cysts of P. nicotianae and three cysts of P. sojae were observed per microscopic field on root surface after 40-min incubation in CFF. Significant fewer cysts attached in the control (Fig. 5b). Number of attached cysts was also different between species. Fewer P. sojae zoospores attached to the roots compared to P. nicotianae within the same treatment, which may be associated with early encystment of P. sojae.
Cyst germination was also improved by CFF. Germination of P. nicotianae and P. sojae after 24 h in SDW was 33% and zero, while in CFF it was 55% and 75%, respectively. The process after germination also benefited from CFF. Cyst morphogenesis underwent upon further incubation in CFF, developing robust finger-like projections, but not in SDW where short and thin projections formed (Fig. S2). Formation of these structures was associated with successful plant infection by zoospores. In the annual vinca -P. nicotianae system, plant infection was significantly enhanced when zoospore inoculum was prepared with CFF but not with SDW (Fig. 6).
Sb5 CFF regulates expression of effector genes of P. sojae. To further understand how CFF may promote the interactions of P. sojae race 25 with its differential host plants and a nonhost plant, N. benthaminana, RNA from plants 48 h after inoculation in the CFF treatment and controls was examined for expression of P. sojae RXLR effector genes, Avr1a, Avr1b, Avr3a and Avr3c. All plants of the CFF treatment except for those of L92-7857 had transcript level elevation of one or more effector genes when compared to the controls (Fig. 7). In the L77-1863, all four effector genes were upregulated, with Avr1a and Avr3c elevated about 20 fold over the control. In the Williams, except for Avr1a, all other three genes were upregulated at low levels. Similarly, in the L83-570, except for Avr1b, other three genes were upregulated with Avr1a and Avr3c that elevated more than 5 fold. In the L88-8470 and nonhost tobacco plants, only one effector gene each was upregulated. Moreover, the elevated gene in L88-8470 was not Avr1a, an effector gene of P. sojae corresponding to RPS1a, the R gene of the plant, but was Avr3a that elevated two fold over the control, indicating CFF-triggered RPS1a recognition by Avr3a. The elevated gene in the tobacco was Avr1a that elevated near 100 fold over the control. This result indicates that upregulation of Avr1a allowed the pathogen to recognize R genes of this nonhost which facilitated tobacco plant infection. In contrast, none of effector genes elevated in inoculated L92-7857 that did not produce any symptoms with or without CFF, indicating that CFF did not affect the interactions between L92-7857 with RPS3c and race 25. Therefore, CFF-promoted effector gene expression appeared to be dependent on plant resistance genotypes.
Identification of Sb5 and active molecular size range of CFF. Sb5 was only one tested as Gram positive among five dominant isolates from soil water extracts. This was confirmed with fatty acid methyl ester analysis (FAME) which showed that Sb5 had a similarity index of 89% to Gram positive Bacillus megaterium while the other four isolates Sb1, 3, 4 and 6 all were Gram negative bacteria: Klebsiella pneumoniae, Klebsiella ornithinolytica, Serratia liquefaciens, and Burkholderia ceribiensis, respectively. Identity of Sb5 was further confirmed by sequenc-ing16S rDNA of the bacterium. The obtained partial sequence (accession number: FJ858254) was 99% identical to that of a Bacillus megaterium strain (accession number: NC014103.1) at a query coverage of 100% and E value of 0 in GenBank. Consequently, Sb5 was identified as Bacillus megaterium.  The size range of functional components of Sb5 CFF was determined by fractionation through Centriprep columns at molecular weight cutoff of 3, 10 or 30 kDa. Four fractions were obtained and examined for activities in promoting P. nicotianae zoospore behaviors in vitro and plant infection in the annual vinca -P. nicotianae system. All of the fractions prolonged zoospore swimming as CFF. They also promoted zoospore germination. However, only fractions of > 3 kDa stimulated zoospore morphogenesis or formation of finger-like projections and vesicles (Fig. 6 top). Likewise, these fractions with larger molecular weight enhanced single zoospore infection of annual vinca, resulting in significantly higher infection rates compared with the smaller fraction and control (Fig. 6  bottom). These results indicate that substances in small and large fractions contributed differently in the different steps of the plant infection process.

Discussion
This study with Bacillus megaterium, a widespread soil bacterium [44][45][46] , demonstrates that environmental bacteria are sources of virulence signal providers for Phytophthora pathogens in plant infection. It advances our understanding of the roles of biotic environmental factor in the interaction between plant and pathogen, providing important new insights into disease development, a collective result of the interaction of susceptible host, virulent pathogen, and conducive environment 47 . We demonstrate that Phytophthora species can exploit environmental bacteria to improve their pathogenesis and fitness like fungal pathogens 24,40 . In the presence of bacterial metabolites, they can not only increase growth and proliferation [28][29][30]46 , but also improve zoospore homing and colonization as indicated in this study to enhance their likelihood of successful plant infection. They can also initiate or improve an infection in plants with bacteria. P. nicotianae initiated its infection at single zoospore level, and P. sojae broke up Nonhost defense barriers to infect tobacco plants as shown in this study. Strikingly, Phytophthora species have the ability to recruit these ubiquitous microorganisms for their benefits because they can produce the bacterial universal quorum sensing signal 23 . Given such beneficial relationships with bacteria or ability for communal infection, it becomes clearer why Phytophthora species remain the most destructive plant pathogens with relatively low populations in nature. More research on relationships between Phytophthora and the beneficial phytobiome as well as the chemistry of functional exudates is warranted to facilitate innovation of plant disease control technologies.
This also advanced our understanding of the natural functions of B. megaterium. Densities of B. megaterium in soil of cropping systems range between 2-6 log cfu/g 44 which is within the range that promoted plant infection by Phytophthora pathogens as demonstrated in this study. On the other hand, B. megaterium have been shown to suppress plant infection by P. capsici 39 . This inconsistency may be attributed to different modes of action of the bacterium created with different procedures. In the biocontrol experiment bacterial, rinsed bacterial cells at greater than 10 8 /ml were used prior to plant to be challenged with P. capsici 39 . This procedure may lead to occupation or blockage of infection sites prior to arrival of the challenging pathogen due to outcompeting of a high density of bacterial cells or other indirect antagonisms 44,48,49 . In contrast, in this study, exudates or metabolites of the bacterium at much lower concentrations (≤ 10 6 cells/ml) were used simultaneously with test Phytophthora to challenge plants. This procedure allows both microorganisms compete at the same time at population densities close to that in nature where the bacterial metabolites may act as mediators of intercellular signaling of virulence and cell differentiation to these pathogens. Metabolites of many bacteria possess a dual nature at different concentrations in the natural ecosystems; they function as antibiotics at high concentration on target microorganisms but as mediators of intercellular signaling of virulence and cell differentiation to these microorganisms at their natural densities 50 . Some bacilli have been shown to produce antibiotics that are antagonistic to Pythium 48 , a close relative of Phytophthora. It is interesting to see how these antibiotics behavior at subinhibitory concentrations. However, because we did not test a concentration greater than10 6 cells/ml, whether exudates of B. megaterium may have a dual nature at different concentrations remains to be elucidated. Other factors such as diversity of B. megaterium may also resolve the conflicting results between the biocontrol experiments and our study. B. megaterium has been reported as a weak plant pathogen 51,52 while in other studies it was shown to be a source of potential plant growth regulators 53 . Genetic differences were found between the isolates used in the P. capsici control study 39 and that used in our study. The former had FAME similarity indices of 52-55% to the type B. megaterium 39 , while the latter is 89% similar to the type. Additional investigation with multiple isolates would help to determine whether these differences may result in functional variations.
This study reveals diverse roles of exudates of B. megaterium in plant infection by Phytophthora species such as triggering quorum sensing-mediated infection and promoting zoospore morphogenesis. The latter as indicated by formation of infection hyphae or appressorium-like structures that are used by fungi and oomycetes to exert pressure to overcome host-cell turgor and/or to secrete enzymes to degrade the plant cell wall and penetrate the cuticle 54 . Appressorium formation in Phytophthora species has been related with plant colonization in the early stage of infection 55 . In this study, we found an association between production of such structures and successful plant infection by P. nicotianae, suggesting the pathogen colonization promoted plant infection. However, since colonization does not necessarily result in infection, the enhanced infection may be due to a persistent effect of the bacterial exudates, stimulating formation of haustoria that function at the biotrophic stage of infection for effector delivery in the plant cells and pathogen virulence promotion 56 . This is supported by upregulation of RXLR effector genes in the infected host and nonhost plants of P. sojae after CFF treatment. Upregulation of these genes often occurs with formation of haustoria during oomycete entry into plant cells to suppress plant defense signaling 57 . In the P. sojae-infected host and nonhost plants promoted by CFF, there was upregulation of one or more of these effector genes in inoculated soybean and tobacco plants in which disease development was significantly enhanced in the presence of CFF. L88-8470 was infected with assistance of CFF despite the fact that Avr3a but not Avr1a corresponding to RPS1a was upregulated, and the transcript level of Avr3a was not very high. Similarly, CFF treated L83 produced HR with much lower upregulation of Avr3a, than Avr1a and Avr3c, despite correspondence of Avr3a with RPS3a of the plant. In the infected N. benthamiana with CFF, Avr1a was Scientific RepoRts | 6:33239 | DOI: 10.1038/srep33239 upregulated substantially which may give rise to abolishment of the nonhost resistance. Yet, plant resistance genotype is likely a determinant in the assisted interaction between Phytophthora and resistant plants by bacteria exudates. For example, in inoculated L92-7857, RPS3c may provide an adequate defense to P. sojae which was not affected by Sb5; none of the effector genes of the pathogen were upregulated. This suggests that pathogen network mediated modification of effector gene expression contributes to susceptibility of plants which is dependent on genetic background of the plant.

Materials and Methods
Soil bacteria isolation, growth conditions, and identities. Bacteria were isolated from soil water extract (SWE). SWE is a solution that is widely used for induction of sporangia and zoospore production due to its containing functional bacterial metabolites and maintaining the activity after autoclaving 8,18,[27][28][29] . SWE was prepared by mixing 10 g of sandy loam soil from the rhizosphere surface of camellia plants in eastern Virginia with 1 liter of dH 2 O on a magnetic stirrer overnight, followed by filtering through Whatman No. 4 and 1 filter paper. To isolate bacteria, aliquots of 200 μ l of SWE were spread on ten 90-mm PDA (potato dextrose agar) plates and incubated at 25 °C for 24 to 48 h. Emerging colonies were grouped by morphology, and representatives of predominant groups were selected, grown on PDA at 25 °C. The Gram nature of these isolates was determined by KOH assay 58 . To prepare fresh cultures, a colony of Gram negative or positive isolates was first grown overnight in a broth of potato dextrose or J (5 g tryptone, 15 g yeast extract, 3 g K 2 HPO 4 , 2 g glucose, pH 7.4) 59 , and then streaked onto the same agar medium plates for 16-24 h at 25 °C. The identities of the isolates were determined by fatty acid methyl ester analysis (FAME) at the Microbial ID laboratory (Newark, DE, USA) and listed in the Table S1.
Bacterial cell suspension and filtrate preparation. Bacterial cell suspensions were prepared by mixing 10 loops of fresh culture into100 ml sterile distilled water (SDW). This normally gave a cell suspension concentration of 10 7-8 /ml as determined with a Neubauer hemocytometer after a serial dilution. A suspension of 10 7 cells/ml was used as a stock for further applications. Cell suspension filtrates were prepared in slightly different ways depending on experiment. For experiments of leaf disc colonization by Phytophthora zoospores, the filtrate was prepared by first agitating the stock suspension overnight, followed by filtering through Whatman 1 and 4 filters and finally autoclaving. For other experiments, cell free filtrate (CFF) was prepared by a centrifugation of the overnight-agitated stock suspension at 5,883 g for 5 min, followed by a filtration of the supernatant through a 0.22-μ m filter. For both filtrate preparations, a 1: 10 dilution was made with SDW to get a 10% working concentration.

Phytophthora species, growth conditions and zoospore suspension preparation. Thirteen
Phytophthora species/isolates were used in this study. Apart from those listed in Table 1, P. sojae isolate 28G4 (P7081, race 25) and P. nicotianae 1B11 were used extensively. All isolates were cultured in 10% clarified V8 broth or on 5% V8 agar at 23-25 °C. Zoospore suspensions were prepared from cultures in broth medium or 2 to 6 day-old agar culture plugs with methods described previously 18 . A zoospore suspension of 10,000/ml SDW was used as a stock for further dilution into the various solutions including autoclaved SWE (ASWE), filtrates of suspensions or CFF of selected bacterial isolates, fractionated CFF, SDW as well as rinsate of agar medium surface with 10 ml of SDW (SDWms).
Colonization of rhododendron leaf discs by multiple Phytophthora species. Leaf discs of rhododendron, a plant susceptible to numerous Phytophthora species 8 , were used to investigate the effects of five selected dominant soil bacterial isolates on Phytophthora zoospore colonization. Leaves were surface sterilized with 75% ethanol to limit effects of other microorganisms and cut into 4-mm diameter discs. 15-20 of the discs were submerged in 100-ml plastic bottles with a 60 ml zoospore suspension at 10/ml in a bacterial suspension filtrate, ASWE, SDW or SDWms. The discs were kept in the solutions at 23 °C overnight then plated on PARP-V8 agar, a Phytophthora selective medium 62 for an additional 48 h before determination of pathogen colonization. Leaf discs colonized by Phytophthora were counted. Each experiment included three replicates and was repeated at least once.
Phytopathosystems and infection assays with P. nicotianae and P. sojae. Three phytopathosytems, annual vinca -P. nicotianae, soybean -P. sojae, and tobacco -P. sojae were used. In the annual vinca -P. nicotianae system, a single zoospore infection assay was conducted with detached leaves as described in a previous study 11 . Briefly, zoospores were suspended in CFF, SDW or ASWE to be tested to obtain an inoculum concentration of 100/ml. To make single zoospore inoculation, 10-μ l drops of the inoculum were placed on 10 different locations on the surface of a leaf and incubated at 23 °C. Infected sites were counted 72 h post inoculation. The experiment included 18 leaves or 180 inoculated sites and was repeated twice.
In the soybean -P. sojae system, the hypocotyl inoculation method was performed as described previously 60 . Specifically, P. sojae 28G4 inocula were prepared with a 2-day-old mycelial mat grown in V8 broth which was washed twice with SDW and drained to remove the nutrients from the media. At inoculation, 50 mg of mycelia was mixed with 1 ml of CFF or SDW and macerated with a syringe. Each hypocotyl was injected with 20 μ l of the slurry and kept in a moist chamber at 23-25 °C for 2-4 days. Inoculated plants were measured for infection based on presence of a dark lesion and lesion sizes at the inoculation site after 3 days. Nine replicates were included and assay was repeated twice.
In the tobacco -P. sojae system, both detached leaves and plant seedlings were used. To inoculate detached leaves, a 20-μ l drop of a zoospore suspension at 12,500/ml SDW or CFF was placed on 4 sites of the abaxial side of a leaf. Inoculated leaves were rated for disease incidence based on occurrence of yellowing or blight symptoms after 72 h at 25 °C. Each assay included 15 replicate leaves and was repeated once. To inoculate seedlings, roots first were rinsed off Metro Mix with tap water and SDW, respectively, and were then placed in a 30-ml beaker with10 ml zoospore suspension at 3,200/ml SDW, a live Sb5 cell suspension at 10 6 /ml or a mixed suspension of zoospore and bacteria at the same individual concentrations. To determine a concentration effect of bacterial cells, a 10-fold dilution series of Sb5 from 10 6 /ml was used. Plants were kept in the dark overnight after inoculation then incubated for an additional 6 days under the plant growth conditions before disease assessment. Plants showing dieback or blight symptoms were recorded as diseased. Each assay included 3-7 plant replicates depending on experiment and was repeated at least once.
To determine the causal agent of the inoculated tobacco with Sb5 and/or P. sojae, symptomatic plant tissues were surface sterilized with 10% bleach to eliminate attached microorganisms on the surface and prevent false positive. They were then plated on PARP-V8 and J-agar at 23 or 25 °C respectively, in the dark for 3 days to allow growth of the causal agent from inside of plant tissues. Mycelial or bacterial growth from the tissue was recorded. Growing mycelia were used to prepare DNA to determine the identity with ITS1-SSCP analysis for Phytophthora species 20 .
Microscopy of effects of Sb5 exudates on zoospore behaviors. Zoospore homing behaviors including motility, plant tissue attachment and cyst germination in response to CFF were examined under an Olympus IX71 inverted microscope at magnification of 40-160× . Experiments investigating zoospore motility and cyst germination were conducted in vitro. Approximately 500 zoospores were suspended in 100 μ l CFF or SDW in a depression slide and incubated at 23 °C for 2, 4, 6, and 16 h. To assess mobility of zoospores, quantity of swimming zoospores in each of three microscopic fields was estimated at each time point and compared to that at zero point and duration between treated and untreated zoospores. Similarly, to determine germination in response to CFF, germinated cysts in both treatments were rated in three fields 16 h after incubation. Each observation included two replication slides.
Experiments for zoospore attachment were conducted with detached clean roots of 5-week old Arabidopsis plants. Individual roots were submerged in a 50 ml zoospore suspension of P. nicotianae or P. sojae at 10,000/ml SDW or CFF for 40 min at 23 °C and then rinsed with water to remove unattached zoospores. Plant tissue was examined for cyst attachment after stained with trypan blue solution 63 . Two randomly selected roots were observed. Attached cysts were counted in three fields of the microscope.
Analyses of gene expression of RXLR effectors in response to Sb5 exudates. Experiments of expression of RXLR effector genes of P. sojae in the inoculated plants were conducted under the MIQE guidelines 64 . Samples were taken from soybean 2-cm sections of inoculated sites from five hypocotyls or three roots of inoculated tobacco seedlings, ground in liquid nitrogen and stored at − 80 °C after plants were inoculated for 48 h with P. sojae in the presence or absence of CFF or Sb5 as described above. RNA was extracted from the frozen samples with the RNeasy Mini Kit and RNase-Free DNase Set (Qiagen, Valencia, CA). cDNA was obtained by reverse transcription of 5 μ g of RNA with the High Capacity cDNA Archive Kit (Life Technologies Carlsbad, CA). 2.5 μ l of the cDNA of each sample or the PCR negative control dH 2 O was amplified in a 25 μ l PCR reaction with a pair of primers listed in Table S10 and the Power SYBR ® Green PCR Master Mix in the ABI 7500 Real Time PCR System (Life Technologies Carlsbad, CA). Each sample was run for at least three times at the 9700 mode with 50 °C for 2 min at first stage, 95 °C for 10 min at second stage followed by 40 cycles of 95 °C for15 sec and 60 °C for 1 min.
Gene expression was based on Ct values (threshold cycle), standard errors of Ct (Ct std err) and average Ct (AvgCt) of genes in each sample from the PCRs as determined with the Relative Quantification Study of the software SDS v.1.3.1 at the default setting (Life Technologies Carlsbad, CA) (Table S7). To get relative expression levels of a target gene, AvgCt of the target gene in a Sb5 treated or untreated sample was normalized against AvgCt of the mean of AvgCt of the three reference genes (Actin A, β-tubulin and ubiquitin) (Table S10) in the same sample, or dAvgCt (=AvgCt target − AvgCt refs ). To obtain expression levels of the target genes in treated samples, dAvgCt of treated samples were calibrated with dAvgCt of untreated to get ddAvgCt (=AvgCt Sb5 − AvgCt SDW ) followed by quantification with 2 −ddAvgCT . The maximum and minimum ddAvgCt was calculated based on the standard errors of dCt (Table S7).

Identification of Sb5 and fractionation of CFF.
In addition to (FAME), the identity of Sb5 was determined by sequencing the amplicons of 16S rDNA of the isolate with primers 968F and 1410R 65 . Both strands of the sequences were processed with BioEdit V7.2.5 (http://www.mbio.ncsu.edu/BioEdit/bioedit.html) 66 and Scientific RepoRts | 6:33239 | DOI: 10.1038/srep33239 BLASTed against 16S rDNA sequences of microbes at http://blast.ncbi.nlm.nih.gov to match to a species level. The final sequence was then deposited into GenBank.
Sb5 CFF was fractionated with centrifugal filter columns to estimate the sizes of active components. CFF was first loaded into Centriprep YM-3 (MWCO 3 kDa) and centrifuged as instructed by the manufacture (Millipore. com). The filtrate was designated as < 3 kDa fraction. The remaining CFF in the column was subsequently loaded to YM-10 (MWCO 10 kDa) and centrifuged. The filtrate collected was designated as 3-10 kDa fraction while the remaining in the second column was processed with the third column, Centriprep YM-30. The resulting filtrate was designated as 10-30 kDa fraction and the solution that did not go through was designated as > 30 kDa fraction. These four fractions were then tested together with the controls CFF and SDW for bioactivities in the microscopy and infection assays described above.
Statistical analyses. Proc ANOVA in SAS with least significance difference (LSD) and t-Test at equal and unequal variances in Excel were used for evaluating statistical significance and comparison of means between treatments. P-value < 0.05 was considered significant. Standard errors were calculated based on the standard deviation of the replicates.