Arabidopsis thaliana responds to colonisation of Piriformospora indica by secretion of symbiosis-specific proteins

Plants interact with a wide variety of fungi in a mutualistic, parasitic or neutral way. The associations formed depend on the exchange of nutrients and signalling molecules between the partners. This includes a diverse set of protein classes involved in defence, nutrient uptake or establishing a symbiotic relationship. Here, we have analysed the secretomes of the mutualistic, root-endophytic fungus Piriformospora indica and Arabidopsis thaliana when cultivated alone or in a co-culture. More than one hundred proteins were identified as differentially secreted, including proteins associated with growth, development, abiotic and biotic stress response and mucilage. While some of the proteins have been associated before to be involved in plant-microbial interaction, other proteins are newly described in this context. One plant protein found in the co-culture is PLAT1 (Polycystin, Lipoxygenase, Alpha-toxin and Triacylglycerol lipase). PLAT1 has not been associated with plant-fungal-interaction and is known to play a role in abiotic stress responses. In colonised roots PLAT1 shows an altered gene expression in a stage specific manner and plat1 knock-out plants are colonised stronger. It co-localises with Brassicaceae-specific endoplasmic reticulum bodies (ER-bodies) which are involved in the formation of the defence compound scopolin. We observed degraded ER-bodies in infected Arabidopsis roots and a change in the scopolin level in response to the presence of the fungus.


Introduction
The roots of more than 80% of all land plants form mycorrhizal interaction [1]. The interactions can shift from mutualism to parasitism depending on environmental conditions and the communication between the organisms [2]. The communication between the symbionts is driven by molecules released into the rhizosphere by both partners including diverse protein classes [3][4][5][6]. Based on genome analyses, it has been estimated that the proportion of released fungal proteins represent about 3% to 10% of the corresponding total proteome [6]. Furthermore, the PLOS ONE | https://doi.org/10.1371/journal.pone.0209658 December 27, 2018 1 / 28 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 PYK10 level are overcolonised by P. indica, although PYK10 expression is unaffected by P. indica in wild type (WT) seedlings [43,44]. PYK10 hydrolyses multiple indole glucosinolates and scopoletin to scopolin [44,45]. Scopoletin, and its glycone scopolin, accumulates in roots and helps plants to fine tune their defence responses by scavenging of reactive oxygen species [46,47]. Scopoletin acts as antifungal component against Alternaria alternata in tobacco and is strongly suppressed in the beneficial symbiosis of tomato with R. irregularis [48,49]. A key enzyme for scopoletin biosynthesis is the feruloyl-CoA 6'-hydroxylase 1 (F6'H1), and the accumulation of this secondary metabolite is severely reduced in the corresponding knock-out line [46]. Interestingly, biosynthesis of scopoletin can also be stimulated by ectopic expression of the Arabidopsis PLAT1 and -2 (Polycystin, Lipoxygenase, Alpha-toxin and Triacylglycerol lipases) proteins in tobacco [50]. It has been hypothesized that the PLAT domain provides a docking platform for proteins to regulate their catalytic activities [51][52][53]. AtPLAT1 and -2 proteins are functionally and physically associated with ER-bodies [51]. This suggests an interaction with PYK10, although the exact function of the PLAT domain is still enigmatic.
In this study, we identified proteins which are found in the growth medium of A. thaliana, P. indica or P. indica co-cultivated with A. thaliana. Proteins which are only found in the co-culture or disappear from it, are considered symbiosis-specific and are mainly involved in growth and defence-related processes. Many enzymes belonging to the latter category are implicated in the production or release of secondary metabolites. The role of one of these proteins, PLAT1, was investigated in more detail and its postulated function in the formation of the defence compound scopolin was related to the degradation of ER-bodies and root colonisation.

Secretome analysis of P. indica and A. thaliana reveal distinct changes in the co-culture
To identify symbiosis-specific proteins that are secreted during the early phase of the interaction between P. indica and A. thaliana, we co-cultivated both organisms in liquid plant nutrition media (PNM) for three days and compared them with cultures of the plant and fungus grown separately (S1 Fig). The spent media were filtered and proteins were isolated by solid phase extraction followed by LC-MS/MS (liquid chromatography-tandem mass spectrometry) analysis. The resulting MS/MS spectra were searched against the UniProt database (http:// www.uniprot.org/) [54] where a total of 590 different A. thaliana and 164 P. indica proteins were identified in three independent biological replicates (S1 Table). In the single cultures of Arabidopsis 199, 316 and 117 proteins were observed within three replicates, while the number of identified proteins in the single fungal cultures were 39, 76  For further investigations, we took all identified plant proteins and classified them (S2 Table) according to their GO (Gene ontology) terms using AmiGO (http://www.geneontology.org, version 1.8) [55][56][57]. Due to the absence of a functional annotation for a high number of fungal proteins, no GO terms have been captured for P. indica. The GO terms annotated to plant proteins were sorted within the categories "molecular functions", "biological processes" and "cellular components" (Fig 1A-1C). For each category, the 20 most abundant terms are shown in Fig 1A-1C. The functional categorisation revealed that secreted plant proteins in both cultures belong mainly Secreted proteins in P. indica-Arabidopsis symbiosis to identical GO terms, however, the number of protein entries varies for each term. As shown in Fig 1A, the most frequent term (10%) related to proteins found in the single culture of Arabidopsis is "binding of different compounds" (GO:0005488), whereas the most abundant term related to proteins from the co-culture (12%) is "catalytic activity" (GO:0003824). As another example, about 5% of the proteins in the co-culture play a role in hydrolysis (GO:00016787), while proteins associated with this GO term were not found in the single culture. Among the terms of the category "biological processes" (Fig 1B), slight differences between the single and co-culture were observed. For instance, proteins with the GO term "metabolic process" (GO:0008152) were found twice as often in the co-culture. Furthermore, the terms "defence", "external stimuli", "response to biotic stimulus" and "catabolic process" were found more often for co-cultures. The category "cellular component" (Fig 1C) included notably more proteins associated with the GO terms "extracellular region" and "cell periphery" in the co-culture. Contrary, GO terms associated with "cellular processes" or "organelles" were more frequently found in the single culture.
A. thaliana responds to the presence of P. indica by an altered secretion of proteins associated with defence and growth. To better understand changes in the composition of secreted proteins, we looked for specific differences between the co-culture and single culture (S2 Table). The analysis considered only proteins found in two or more root exudates and these proteins were classified based on their function. We observed 102 proteins to be differentially secreted (Fig 2). The most enriched proteins in the single culture are associated with the functions "primary metabolism", "abiotic stress response", "growth and development" and "biotic stress response" (~20% each). In the co-cultures, 55% of the proteins play a role in defence against biotic stresses, 20% have been described in the context of growth and development, 11% belong to the primary metabolism and 7% proteins are related to abiotic stress responses. Furthermore, 11% proteins are classified as miscellaneous. Interestingly, 7% of the differentially secreted proteins are associated with mucilage.
Among proteins, which were involved in development and growth and which were predominantly detected in the co-culture, are the ribonuclease T2 (At1G14220), the lipid recognition protein At5G23840 and the pectin methylesterase inhibitor At1G23205. Examples related to biotic stress responses include the pathogenesis-related protein 5 (At1G75040), which was strongly induced by the fungus. In Arabidopsis, it has been implicated in systemic acquired resistance [58]. Furthermore, the two functionally distinct germin like proteins (GLPs), GLP2 (At1G02335) and GLP4 (At1G18970) appear to be secreted into the co-culture upon interaction with the fungus. Additionally, PLAT1 (At4G39730) was found in the medium of the co-culture.
Symbiosis reprograms the P. indica secretome. A total of 164 P. indica proteins were found in all experiments (S1 Table): 36 of them were differentially enriched between the treatments (S2 Table). Among these proteins, two were found predominantly in the single fungal culture (S2 Table): PIIN_00867, an uncharacterised protein, -and PIIN_06517 related to glyoxal oxidase. 34 proteins were present only in the co-culture. Generally, many of the fungal proteins identified in this study are uncharacterised or possess homology to degrading enzymes. For instance, PIIN_01733, which was found in the co-cultures, is a putative betamannanase with homology to hydrolases from two saprophytic fungi [78]. Additionally, we analysed fungal proteins for subcellular localization using available bioinformatics tools: 21 contained an N-terminal secretory signal peptide (S2 Table). One of them is a indole-3-acetaldehyde dehydrogenase (PIIN_04899) which has been suggested to be involved in the formation of auxin [79]. P. indica-root colonisation influences PLAT1 and leads to changes in scopolin production. As a case example to study the secretome, we further analysed the role of the Polycystin, Lipoxygenase, Alpha-toxin and Triacylglycerol lipases 1 (PLAT1). This protein was found in the medium of the co-culture and has not been reported in context with P. indica before. We checked the expression level of PLAT1 in colonised WT roots and observed a significant lower level of plat1 mRNA at three and seven days of co-cultivation. After two weeks the level of plat1 mRNA was comparable to uncolonised plants ( Fig 3A). In shoots no regulation was observed ( Fig 3B). A weak but not significant downregulation was observed for the homologous plat2 mRNA, whereas for the putative pseudogene plat3 no regulation was observed. Interestingly, a ten-time higher expression in leafs compared to roots was observed for plat3. Nevertheless, for all tested time points and plant parts the expression level of plat3 was extremely low (S2B and S2C Fig). Since P. indica had the strongest effect on PLAT1 expression, we took a closer look at the colonisation of plat1 mutant seedlings by the fungus. After one week, we found a three times higher amount of fungal RNA in the plat1 mutant compared to the WT seedlings, whereas after two weeks no differences in colonisation between plat1 mutants and WT plants was observed (Fig 3C). This stronger colonisation did not have an impact on the total fresh weight of the plat1 mutants ( Fig 3D).
These results could be confirmed by microscopy (Fig 4): after seven days we observed a stronger colonisation in the plat1 mutant ( Fig 4A) compared to the WT (Fig 4B), where almost no fungus could be detected. Since AtPLAT1, which was identified in the co-culture, promotes scopoletin biosynthesis [50], and an upregulation of scopoletin in colonised A. thaliana roots has been described before [12], we analysed the connection between PLAT1, scopoletin and P. indica colonisation.
In a first step we examined the influence of scopolin and scopoletin on P. indica in vitro and found that both chemicals inhibit the growth of P. indica with a half maximal inhibitory concentrations (IC 50 ) of 0.34 mM and 0.45 mM, respectively ( Fig 5).
Scopolin is much more abundant than scopoletin and can easily be measured in high quantities, in Arabidopsis roots [80]. Therefore, we studied the influence of P. indica on scopolin production in Arabidopsis roots in vivo. After two weeks, the mock treated WT roots showed a four-time upregulation of scopolin compared to the other time points, whereas in colonised roots the scopolin level was not altered. Instead, the same level of scopolin was observed for all time points and treatments in the plat1 mutant (Fig 6).
To examine whether scopolin alone is responsible for the altered colonisation in the plat1 mutant, we investigated the performance of the fungus in the f6'h1 Arabidopsis mutant line where scopolin and scopoletin production is severely reduced (Fig 7A). When comparing the fresh weight between WT and f6'h1 plants, we observed, similar to plat1 (Fig 3D), that the mutant lines are not influenced by the presence of the fungi. Furthermore, the mutant lines were smaller than WT plants ( Fig 7B). Since PYK10 can be involved in scopolin formation, we checked its expression: pyk10 is not influenced by the fungus neither in the WT plants nor in the plat1 or the f6'h1 mutant background ( Fig 7C). Other genes, which are also associated with scopolin formation, are not influenced by the fungus in f6'h1 or WT plants as well (S3 Fig).
We further analysed the colonisation of Arabidopsis roots by microscopy. Seven days after co-cultivation most of the hyphae associated with WT roots were detected on the root surface between the epidermal and subepidermal root cells and barely detectable within root cells, whereas in f6'h1 mutant roots, hyphae could be detected around and within root cells. This effect was even stronger 14 and 19 days after co-cultivation (Fig 8).
P. indica-root colonisation causes degradation of ER-bodies. The Arabidopsis lipases PLAT1 and -2 are functionally associated and co-localise with ER-bodies in vivo [51]. Thus, we analysed the fate of ER-bodies during the establishment of symbiosis (Fig 9). The size of ER-bodies in uncolonised roots is not altered during the development of the plant but decreases during the interaction with the fungus (Fig 9A). After three days, the ER-bodies in the WT have the normal rod-shaped structure ( Fig 9B) but are slightly smaller in the colonised roots ( Fig 9C). After seven days, the ER-bodies in the fungal treated roots have a significant reduced size and, First, we determined the scopolin levels in root of WT and plat1, and the peak areas of HPLC chromatograms were normalised by an internal standard. All experiments were repeated four to five times with at least ten plants each. Error bars represent standard deviation. Different letters represent significant differences between treatments at p<0.05 (Two-Way ANOVA, followed by a Bonferroni correction). https://doi.org/10.1371/journal.pone.0209658.g006 Secreted proteins in P. indica-Arabidopsis symbiosis moreover, a diffuse signal was observed ( Fig 9D). This effect diminishes after two weeks and the size and structure of the ER-bodies was again comparable to the mock treated samples (Fig 9E).
We further determined the size of ER-bodies in the liquid culture at the time point of the secretome study. No difference in structure was observed between the mock treated and colonised roots (S4 Fig). But when comparing the liquid culture with the solid culture, a slight sizereduction was observed for the ER-bodies in untreated roots at 3 dpi (S5 Fig).

P. indica alters the secretome of colonised plants
The availability of the complete genome sequence and a system of hydroponic cultivation makes P. indica an interesting model endophyte for studying the molecular basis of beneficial symbiotic interactions. In this study, P. indica and A. thaliana were cultivated alone and together in hydroponic media and the secreted proteins were compared after three days of cocultivation. This approach allowed us to identify symbiosis-relevant proteins secreted during the early stage of interaction. At the time point of this study, the fungus and the plant have not established a mutualistic symbiosis and the fate of the interaction is still undetermined [12,38,81]. Our results demonstrate that 102 plant proteins were differentially enriched the presence of the fungus (S2 Table and Fig 2). The most severe alterations in the level of plant proteins affect biotic and abiotic stress responses, the primary metabolism and mucilage synthesis. and f6'h1 (B) Growth promotion of WT and f6'h1 by P. indica after 7 and 14 days of co-cultivation. At 14 dpi a slightly higher growth-stimulating effect was visible for f6'h1 compared to the WT. (C) Expression of PYK10 in the roots of WT and f6'h1 mutants 14 days after co-cultivation with or without P. indica. All experiments were repeated three to four times with at least ten plants each. Error bars represent standard deviation. Different letters represent significant differences between treatments at p<0.05 (Two-Way ANOVA, followed by a Bonferroni correction).
https://doi.org/10.1371/journal.pone.0209658.g007 In the single culture, proteins belonging to the primary metabolism (24%), abiotic stress (23%), growth and development (19%) and biotic stress (19%) were present with a similar frequency. On the contrary, in the co-culture proteins involved in biotic stress responses (55%) as well as growth and development (20%) were more abundant. The high proportion of proteins associated with these GO terms is not surprising, as P. indica influences many aspects of plant growth and defence mechanisms [5,34,37,81,82].
Abiotic stress response was one of the most dominant categories in the single culture but not the co-culture. Interestingly, we identified several proteins associated with cadmium stress in the single culture. It is unclear whether these proteins play a direct role in cadmium stress responses or whether they have additional function, but intriguingly, for tobacco and sunflower it was reported that P. indica alleviate cadmium specific stress [83].
While abiotic stress, such as osmotic stress, can be expected in a liquid cultivation system, we were surprised to find almost no proteins which play a role in this category in the co-culture. As exception, the abiotic stress-related cysteine proteinase RD21A, the Aspartic Protease in Guard Cell 1 and PLAT1 were identified in the co-culture [73]. PLAT1 is induced by ABA and plays a role in abiotic stress response, but our and previous findings indicate an additional role of PLAT1 in plant-microbial interaction [50]. ABA plays a vital role in the interaction of plants with P. indica. It accumulates stage-specific in colonised roots, which is required for successful colonisation [81,84]. Thus, an altered level of ABA induced by the fungus may be involved in or responsible for the observed differences in PLAT1 accumulation.
Besides PLAT1, we identified 34 differentially regulated proteins with a role in biotic stress. Some of the proteins have already been described in the context of P. indica colonisation before. For example, the polygalacturonase inhibitor 1, which restricts programmed cell death of the host [85], was found in the co-culture. A homologue of this protein was recently identified as a target of P. indica in Chinese cabbage [86]. Another protein from the co-culture, the pectin methylesterase inhibitor At1G23205 is induced by volatiles from a plant growth-promoting Bacillus subtilis strain [87].
In a recent proteome study, Brassica napus was co-cultivated with P. indica. Similar to our study, proteins associated with biotic stress as well as growth were identified. Another investigation showed that a GLP was downregulated in rape by the fungus after 28 days of co-cultivation [88]. In our short-time assay, homologues of this protein, GLP2 and GLP4, were found in the medium of the co-culture. While GLP4 has an important part in controlling Blumeria graminis infection in wheat and barley [89,90], GLP2 plays a role in root development, possibly by controlling the distribution of carbohydrates between different root parts [91]. These findings are not surprising since the roots are in contact with different fungal cell wall components and the fungus might manipulate the carbohydrate metabolism in roots to stimulate plant growth [92]. Furthermore, the beginning of interaction is characterised by a stressful biotrophic-like phase [81]. At the early time point studied here, the fungus establishes the colonisation, which is accompanied by the balanced regulation of defence compounds. Interestingly, we found that the salicylic acid level is reduced while jasmonate, indolic glucosinolates and flavonoids are increased [12, 37]. Apparently, after 3 days of co-culture, jasmonate-dependent defence strategies dominate in the co-culture which further support the concept that a beneficial interaction phase has not yet been reached.
Another observation was that proteins of the primary metabolism were found predominantly in the single culture and not the co-culture (24% and 7%, respectively). Since secreted time point and treatment four roots were used and the size of 20-100 ER-bodies per root was measured. (B) Fluorescence microscopy of ER-bodies in A. thaliana root without P. indica at 7 dpi and with P. indica at 3 dpi (C), 7 dpi (D) and 14 dpi (E).
https://doi.org/10.1371/journal.pone.0209658.g009 proteins were analysed, this observation is difficult to explain. Based on the results discussed above, one would expect that the fungus represses the primary metabolism to shift metabolites towards defence during the early time point investigated in this study. Under this assumption the amount of both cellular and exudated proteins involved in the plant primary metabolism could be reduced when P. indica is present. The strong reduction of these proteins in the exudate preparations might be due to their easier accessibility for fungal proteases. The high number of putative degrading enzymes in the P. indica secretome supports this hypothesis. However, whether plant proteins involved in the primary metabolism are specifically degraded, is unclear. An interpretation of these results requires more information about the targets of the fungal proteases, which is still poor.
We also found proteins associated with mucilage in the media of the different cultures. Recent evidence points to mucilage function in seed germination, root tip growth and the formation of arbuscular mycorrhiza [93][94][95]. The importance of mucilage for arbuscular mycorrhiza has been shown for several plant symbioses including the interaction of maize roots with Gigaspora gigantea in which secreted plant components induce hyphal branching [96]. The function of mucilage during the interaction of Arabidopsis with microbes, such as P. indica, has not been studied before even though mucilage was investigated in Arabidopsis seeds [73]. Proteins associated with mucilage were identified in the single culture, in the co-culture or in both treatments (S2 Table). For example, the subtilisin-like protease 1.7 (SBT1.7) was present in both cultures (S1 Table), while its homologue SBT1.8 was only detected in the single culture. These quite specific effects suggest that the mucilage-related proteins have specific functions in the symbiosis and are differentially regulated by the fungus. Barely anything is known about the role of mucilage in controlling hyphae entry into root cells, the regulation of plant genes and proteins involved in their synthesis, although it is obvious that they might be an important barrier for fungal invasion, besides other functions which can be envisioned in symbiotic interactions cf. [93][94][95][96].

Secretome from P. indica
Even though the genome of P. indica is sequenced, the function of many proteins is still unknown. In our study, most of the differentially regulated proteins were found in the medium of the co-culture. For one of them (PIIN_04899, a putative indole-3-acetaldehyde dehydrogenase), Hilbert and colleagues (2012) showed an induction of this gene by arabinose, a cell wall component of A. thaliana [97,98]. For another protein (PIIN_04899), a homologous protein from the Pseudomonas syringae strain DC3000 was described. This protein produces indole acetic acid (IAA) via indole-3-acetaldehyde dehydrogenase (ALDA) that catalyses the formation of IAA from indole-3-acetaldehyde. Disruption of aldA and a close homolog, aldB, leads to reduced IAA production and virulence of DC3000 on A. thaliana. McClerklin and colleagues (2017) further showed that the IAA produced by this pathway in DC3000 suppresses salicylic acid-mediated defences in Arabidopsis [99]. P. indica can produce IAA in vitro and its growth promoting effect in Arabidopsis can be imitated by this hormone [79]. In the interaction of P. indica with barley, IAA is required to establish a symbiotic interaction, but not for the growth promotion effect [97]. Hence, the identified fungal protein might trigger the IAA synthesis and, in turn, confer beneficial traits to Arabidopsis.

Role of PLAT1, ER-bodies and scopolin in the colonisation of P. indica
We found PLAT1 in the supernatant of the co-culture. PLAT1 functions as positive regulator of abiotic stress tolerance and promotes plant growth [51]. Furthermore, PLAT1 expression is regulated by the ABA-dependent transcription factors 1-4 [51] which prompted us to have a closer look at the function of the protein in the interaction. In the beginning of the colonisation the expression of this gene is reduced in colonised roots; in confirmation with this results plat1 mutants are stronger colonised by the fungus after seven days. After two weeks, no difference to the wild-type can be detected. PLAT domains are found in different membrane-or lipid-associated proteins [52,[100][101][102][103] and exhibit similarity to the C2 domain involved in protein-protein and protein-membrane interactions [52,104]. Therefore, it has been hypothesised that the PLAT domain provides a docking platform for proteins to regulate their catalytic activities and that the proteins per se do not have enzymatic activities [51][52][53]. Hyun and colleagues (2014) further observed a co-localisation of PLAT1 with ER-bodies [51]. Since PYK10, which is found in ER-bodies, is involved in the establishment of symbiosis between P. indica and A. thaliana [43], one could speculate that besides PYK10, also PLAT1 might play an important role, by regulating defence processes during early phases of the symbiotic interaction. This is supported by the stronger colonisation of plat1 mutants seven days after colonisation and by the altered level of scopolin in plat1 roots. Furthermore, the scopolin mutant f6'h1 plants are stronger colonised by the fungus. This indicates that F6'H1 or its product scopolin restricts the entry of P. indica hyphae into roots cells. In agreement with these results, scopolin inhibits growth of P. indica and other fungi in vitro [39]. Downregulation of the plat1 mRNA in the infected roots seem to be associated with the observed degradation of the ER-bodies, whereas it is not clear whether the expression of PLAT1 in the nucleus or the ER-bodies in the cytoplasm, or both are the primary target of the fungus in Arabidopsis roots, and whether both effects are directly connected to each other. The appearance of the PLAT1 protein in the exudate fraction might be caused by the disruption of cells, which occurs when the ER-bodies become gradually smaller due to the activation of defence processes against P. indica. Ultimately, smaller ER-bodies in colonised roots might restrict the liberation of defence compounds such as scopolin from its precursor scopoletin, allowing a more efficient colonisation of roots by the fungus.
Apparently, the early phase of interaction between A. thaliana and P. indica includes an arms-race in which P. indica tries to restrict the plant defence by reducing the amount and function of ER-bodies. At later time points, when a beneficial interaction starts and the fungus is recognized as a friend, the defence compounds PLAT1 and scopolin appear to be less important. These processes can be triggered by altered phytohormone levels. For example, Camehl and colleagues (2010) showed that ethylene can inhibit the fungus and is required for balancing the interaction between P. indica and A. thaliana [104]. Other defence-related phytohormones have been reported to be activated during different phases of the symbiotic interaction. Although they may explain alterations in the expression of genes involved in scopolin/scopoletin biosynthesis after exposure of the roots to P. indica, the role of PLAT1 and the coumarins in this scenario require more investigation.

Conclusion
In this study, we analysed the secretion of proteins during the early interaction between A. thaliana and P. indica. The identified fungal proteins demonstrate that root colonisation results in an almost complete alteration in the fungal exudation profile. Unfortunately, barely anything is known about the function of the secreted fungal proteins, making it difficult to draw meaningful conclusions.
Several of the identified Arabidopsis proteins are differentially regulated. They function in defence and might be required to restrict entry of the hyphae into the host cells. This includes remorin 3, germin like protein 4, osmotin like protein 34 or pathogenesis-related protein 5. For other defence-related proteins, a clear function has not yet been described. Our analysis demonstrated that multiple peroxidases and proteins with a role in growth processes, such as patellin 1 and 2 as wells lipid transfer protein 8, were influenced by the fungus. Furthermore, proteins related to mucilage formation, such as RD21A, RD21B or aspartic protease 1, respond to the fungus. The quite unique regulation of the individual mucilage-related proteins in response to the fungus suggests that they play an important and so far, little investigated role, and that symbiosis-specific signals manipulate mucilage formation.
The establishment of symbiosis requires a reprograming of fungal and plant proteins. Benefits for the host only become visible after successful root colonisation and the recognition of the microbe as mutualistic symbiont. Several identified proteins in the single and co-culture suggest that this phase has not yet been reached three days after infection, and that our analyses was performed at a time point when the arms-race between the two symbionts is ongoing.
ER-bodies are involved in defence and the degradation of ER-bodies in response to fungal colonisation might play a role in balancing defence and growth responses. The downregulation of plat1 by the fungus seems to cause a downregulation of the defence compound scopolin and thus allowing a better colonisation. However, other enzymes and proteins with not welldefined functions are present in the ER-bodies and lower PLAT1 and scopolin levels after root colonisation may only be one aspect in the arms-race that involves ER-body functions.
Cultivation of A. thaliana with P. indica for secretome study. Seeds were planted on MS medium (Murashige and Skoog) [107] in Petri dishes closed with micropore band. After vernalisation at 4˚C for two days the plates were kept under an 8 h light (60 μmol photons � m -2� s -1 )/ 16 h dark period with a constant temperature of 21˚C for 12 days. For the experiment ten petri dishes were filled with 20 ml liquid PNM without sugar for each treatment. For the P. indica single culture and the co-culture, a fungal plaque with 1 cm diameter was added to the plates and cultivated at room temperature for three days, whereas for the single A. thaliana culture and the mock control (medium without fungus or plant) a KM plaque with the same diameter was added. After three days, four plants were added to the petri dishes of the A. thaliana single culture and the coculture. To the P. indica single culture and the mock control, a plaque of MS media was added instead. The mock treatment was performed to check for contaminations. After three days, the supernatant was collected and used for the solid phase extraction.
Co-cultivation of A. thaliana with P. indica for expression and growth promotion studies. For the gene expression study, A. thaliana and P. indica were cultivated as described before [105]. A plaque with one cm diameter of P. indica and KM, respectively, was transferred to petri dishes with PNM covered by a sterilised nylon membrane. After seven days, four twelve-day old seedlings were transferred to the fungal plate and the control plate, respectively. For all experiments, WT and mutant plants were grown in parallel. Roots and leaves were harvested and snap frozen in liquid nitrogen prior further analysis.

Analysis of secreted proteins
Solid phase extraction. Supernatant of liquid spent culture medium was filtered (0.22 μm PES (polyethersulfone) Corning bottle top filter Sigma-Aldrich, Munich, Germany) and treated with a protease inhibitor cocktail (1 tablet cOmplete Ultra, Mini, EDTA (ethylenediaminetetraacetic acid)-free, Easypack (Roche, Basel, Switzerland) per 100 ml medium). Trifluoroacetic acid (TFA) was added to gain a final concentration of 0.1% TFA. Chromabond C4 SPE (solid phase extraction) cartridges (Macherey-Nagel, Düren, Germany) were reconstituted with 3 ml acetonitrile (ACN) and 3 ml 0.1% TFA. Each 20 ml spent medium of 10 biological replicates of plant, fungus, co-culture or mock samples were aspirated (vacuum-driven) through the SPE cartridges. Subsequently, the cartridges were washed with 3 ml 5% MeOH (methanol), 0.1% TFA, and the enriched proteins were eluted from the C4 resin with 3 ml 0.1% TFA in ACN/ A stainless-steel emitter in the Nanospray Flex Ion Source (Thermo Fisher Scientific) was used to generate positively charged ions at a spray voltage of 2.2 kV. The quadrupole/orbitrap instrument was operated in Full MS / data-dependent MS2 (Top15) mode. Precursor ions were monitored at m/z 300-1500 at a resolution of 120k FWHM using a maximum injection time (ITmax) of 100 ms and an AGC (automatic gain control) target of 1e6. HCD fragmentation at 30% normalised collision energy (NCE) generated MS2 ions, which were scanned at 15k FWHM using a maximum ITmax of 100 ms and an AGC target of 2e5. Dynamic exclusion of precursor ions was set to 30 s. The LC-MS/MS instrument was controlled by Chromeleon 7.2, QExactive HF Tune 2.8 and Xcalibur 4.0 software (Thermo Fisher Scientific).
Protein database search. Tandem mass spectra were searched against the UniProt database of A. thaliana and P. indica (2017/02/05) using Proteome Discoverer (PD) 2.1 (Thermo Fisher Scientific) and the algorithms of Mascot 2.4 (Matrix Science, London, UK), Sequest HT (integral version of PD 2.1) and MS Amanda 1.0. Two missed cleavages were allowed for trypsin digestion. The precursor mass tolerance was set to 10 ppm and the fragment mass tolerance was set to 0.02 Da. Modifications were defined as dynamic methionine oxidation and static cystheine carbamidomethylation. At least 2 peptides per protein and a strict false discovery rate (FDR) < 1% were required for positive protein hits. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [109] partner repository with the dataset identifier PXD009563.
For the GO-Term analysis, all identified proteins were searched against the GO consortium database using AmiGO (http://www.geneontology.org, version 1.8) [55][56][57]. The frequency (in %) of a specific GO-Term was determined individually for each treatment and biological replicate. The mean frequency for the different treatments was calculated and the 20 most abundant terms were used for further analysis.

Gene expression studies
The co-culture of P. indica and A. thaliana was performed as described before [105]. Roots and leaves were collected for RNA extraction and cDNA (complementary Deoxyribonucleic acid) synthesis followed by RT-qPCR (reverse transcriptase quantitative polymerase chain reaction) as described in [81]. All experiments were repeated in three to four biological and three technical replicates. Primers are given in S3 Table. The efficiency of expression was manually determined by LinRegPCR 2014 [110] followed by the ΔΔ Ct (threshold cyclce) method [111]. Expression levels of genes were normalised by glyceraldehyde 3-phosphate dehydrogenase (GAPDH), whereas the relative normalised expression depends on the respective control as stated in the figures.

Microscopy
Three, seven and 14 days after infection of A. thaliana with GFP-labelled P. indica (gift from Prof. P. Schäfer, University Warwick) root colonisation was imaged using an LSM (laser scanning microscope) 880 (Zeiss Microscopy GmbH, Jena, Germany), with the 488 nm laser line of an argon multiline laser. Images were taken with a 40x objective (Plan-Apochromat 40x/ 0.8). Pictures for quantification of ER-bodies were taken with an Axiolmager M2 microscope (Zeiss). Digital images were processed using ZEN (ZEISS Efficient Navigation) software.

Extraction of scopolin/ scopoletin from plant material for HPLC analysis
The procedure is based on Döll [112]. For 100 mg frozen plant material, 400 μl methanol and 2mM trans-cinnamic acid (Sigma-Aldrich) per sample as internal standard was added and homogenized for 60 seconds at full speed by an electric homogenizer (RZR 2102 Control (Heidolph, Schwabach, Germany). The samples were transferred to a centrifuge tube and centrifuged at 15.000 rpm and 4˚C for 10 min. The supernatant was transferred to a new tube and the pellet resolved in 400 μl methanol. After a second centrifugation step, the supernatants of identical samples were mixed and centrifuged again for 10 min. Finally, the samples were filtered through a 20 μm filter into a glass vial.

HPLC measurement
For the measurement a JASCO HPLC 900 system (JASCO Germany GmbH, Gross-Umstadt, Germany) was used. It was equipped with a DG-2080-53 degaser, a LG-980-02 solvent mixer, a PU-980 vacuum pump, an AS-950-10 autosampler and a MD-910 multiwave length detector. We used C18 column (Kromasil C18 HPLC column 5μm particle size, pore size 100 Å, length 150 mm, inner diameter 4.6 mm, K08670356, Sigma-Aldrich) at 20˚C with a flowrate of 0.6 ml/min. The device was coupled via a JASCO LC-Net II/ADC interface with the computer controlled by ChromPass/Galaxie 1.10.0.5590 from JASCO. Purging was performed for 1 min at a flow rate of 5 ml/min with 0.1% (v/v) formic acid in acetonitrile. This was followed by degassing with 0.1% (v/v) formic acid in water for 1 min at the same flowrate. The column was washed with 0.1% (v/v) formic acid in acetonitrile for ten minutes. For equilibration 2% (v/v) acetonitrile + 0.1% (v/v) formic acid in 98% (v/v) water + 0.1% formic acid (v/v) was used until constant pressure was reached. 10 μl of sample were injected and eluted over 27 min. In the beginning the acetonitrile content in the mobile phase was 2% (v/v) increasing to 100% over 15 min kept at 100% for 5 min, decreased to 2% over 1 min and kept for 6 min at 2%. As an authentic standard 10 μl of 0.1 mM scopolin and 10 μl of 0.625 mM scopoletin (Phytolab, Vestenbergsgreuth, Germany) were used. The amount of scopolin and scopoletin in the sample was determined by the peak area normalised with the peak for trans-cinnamic acid (Sigma-Aldrich).

Growth inhibition experiment of P. indica by scopoletin and scopolin
The procedure was performed as described by [113]. Scopolin and scopoletin was purchased from Sigma-Aldrich and diluted in methanol. Scopolin, scopoletin and methanol (for the control) were added to six-well plates. After evaporation of the solvent, the wells were filled with 1.5 ml liquid KM medium. 4.5x10 5 fungal spores were added per well. After 11 days of growth, P. indica was harvested and dried (60˚C, overnight) to determine the dry weight for calculation of the inhibition of fungal growth in %. Bars represent standard deviation. The experiment was repeated in three biological replicates.

Statistics and data deposition
All statistic was performed by Sigma Plot 13.0.083 (Systat Software, Erkrath, Germany). Details are given in the figure legends.
The MS proteomics data in this paper have been deposited via ProteomeXchange with identifier PXD009563.
Supporting information S1 Table. List of proteins identified in the secretomes from single and co-cultures. Selected proteins were identified by at least two unique peptides and a false discovery rate of <1%. The frequency of occurrence is given as Low  Table. List of proteins enriched across all three secretomes. Identified proteins were considered differentially enriched, when they were found in at least two replicates in one treatment and not in the other treatment. Furthermore, proteins that were found with a high frequency in one treatment and with a low frequency in the other treatment, were considered differentially secreted as well. Information about the function was obtained using the BLAST algorithm for homology searches against the UniProt database [54,114]. The location was predicted using TargetP, SignalP and Predotar algorithms. Table. List of primers used in this study. (XLSX) S1 Fig. Colonisation of A. thaliana by P. indica in liquid media at 3