Identification of Metabolic Pathways Expressed by Pichia anomala Kh6 in the Presence of the Pathogen Botrytis cinerea on Apple: New Possible Targets for Biocontrol Improvement

Yeast Pichia anomala strain Kh6 Kurtzman (Saccharomycetales: Endomycetaceae) exhibits biological control properties that provide an alternative to the chemical fungicides currently used by fruit or vegetable producers against main post-harvest pathogens, such as Botrytis cinerea (Helotiales: Sclerotiniaceae). Using an in situ model that takes into account interactions between organisms and a proteomic approach, we aimed to identify P. anomala metabolic pathways influenced by the presence of B. cinerea. A total of 105 and 60 P. anomala proteins were differentially represented in the exponential and stationary growth phases, respectively. In the exponential phase and in the presence of B. cinerea, the pentose phosphate pathway seems to be enhanced and would provide P. anomala with the needed nucleic acids and energy for the wound colonisation. In the stationary phase, P. anomala would use alcoholic fermentation both in the absence and presence of the pathogen. These results would suggest that the competitive colonisation of apple wounds could be implicated in the mode of action of P. anomala against B. cinerea.


Introduction
Post-harvest diseases cause significant economic losses for fruit and vegetable producers. From 20-25% of the production in industrialised countries and up to 35-50% in developing countries [1,2] are lost. The control of post-harvest diseases is mainly based on the use of chemical fungicides [3]. With the increasing consumer demand for chemical residue-free products and the emergence of resistant fungi, the interest of the producers in alternative solutions has increased. One of the most promising technologies is the use of antagonistic organisms. Some antagonistbased products are already commercially available and others are currently at varying stages of development [3,4]. Understanding the mode(s) of action of antagonists is one of the parameters of product development and is relevant for marketing purposes [5]. It allows for the identification of useful traits that could be upgraded by genetic tools and improves performance reliability through the development of formulations [6]. Competition for space and nutrients [7][8][9][10][11][12], secretion of lysis enzymes [6,13,14] and activation of plant defences [15] have already been reported to play a role in the antagonistic mode(s) of action. However, the cellular mechanisms that support these modes of action are not clear. Recently, cDNA-AFLP was used on an in situ model close to the one used in this study, to search for Candida oleophila strain O genes of biocontrol relevance against B. cinerea on apples [16] and identified primary metabolisms that could be involved in the colonisation of apple wounds.
Pichia anomala strain Kh6 (Saccharomycetales: Endomycetaceae) was identified as an antagonist of the apple pathogen Botrytis cinerea (Helotiales: Sclerotiniaceae) [17]. Microbiological, biochemical and molecular approaches [13,[18][19][20][21] revealed the complexity of P. anomala mode of action. Jijakli and co-workers [17] demonstrated that there was a quantitative relationship between pathogen spore concentration and the amount of the antagonist needed for disease control on apple, suggesting that the protective effect of the antagonistic yeast was closely related to wound colonisation [17,22]. When cultured on a medium supplemented with B. cinerea cell walls as the sole carbon source, P. anomala overexpressed genes that exhibited b-glucosidase, transmembrane transport, citrate synthase and external amino acid sensing and transport functions [20]. More particularly, in these conditions, a higher activity of bglucanases (PAEXG1 and PAEXG2) was detected [13]. When mutant strains of P. anomala strain K (with two genes coding for PAEXG1 and PAEXG2 disrupted separately or simultaneously) were applied on apple wounds, their biocontrol efficiency against B. cinerea decreased [21]. However, the study also highlighted that both high yeast concentrations and apple ripening stage appeared to compensate for the inactivation of the PAEXG genes [21]. Although the above-mentioned studies identified different modes of action of P. anomala against B. cinerea, however, cellular mechanisms implied in this biocontrol still remain to be identified.
Proteomic techniques couple bi-dimensional electrophoresis  with mass spectrometry to separate and identify hundreds of proteins present in a complex mixture and is powerful to observe changes in protein expression in organisms under various environmental conditions [23]. Regarding biocontrol, proteomics is a promising technique for identifying the mechanisms associated with the biocontrol process [24][25][26] and for optimising formulations and enhancing antagonist efficacy.
The purpose of this work was to compare the proteome of P. anomala grown on apple fruit in the absence or presence of B. cinerea and at different growth phases in order to identify P. anomala influenced metabolic pathways while taking into account the antagonist/pathogen/host tripartite interaction as in natural infection conditions. Cellular processes potentially involved in the protective effect of P. anomala are also proposed.

Organisms and model
All the experiments were carried out in situ on Malus x domestica cv. Golden Delicious purchased from a local supplier (Van Dyck Freres SA, Namur, Belgium). To study the yeast proteome within the triple apple/antagonist/pathogen interaction, an in situ model was recently developed [27] that allows for nutrient exchanges between apples and microorganisms and for the extraction of proteins compatible with a 2-DE study, both quantitatively and qualitatively. Forty and ten apples were used in the exponential (7 hrs) and stationary (24 hrs) growth phases, respectively. Four different samples corresponding to P. anomala incubated alone for 7 hrs (Kh6-1) or for 24 hrs (Kh6-2) and P. anomala co-inoculated with B. cinerea for 7 hrs (Kh6B-1) or for 24 hrs (Kh6B-2) were produced in five biological replicates.
Protein extraction, two-dimensional electrophoresis, image analysis and protein identification by mass spectrometry The hot SDS/acetone protein extraction protocol and 24 cm 2-DE procedure based on Delaplace et al. (2006) [28] were adapted and are detailed by Kwasiborski et al. [27]. The protein content was determined using the Bradford method. The obtained 2-DE gels were scanned using the Typhoon Variable Mode Imager 9400 (GE Healtcare) at a resolution of 100 mm and analysed using 2-DE image analysis software Decyder v7 (GE Healthcare). The proteome of P. anomala inoculated alone on apple wounds was compared to the proteome of P. anomala in the presence of B. cinerea in the exponential and stationary growth phases: Kh6-1 vs. Kh6B-1 after 7hrs and Kh6-2 vs. Kh6B-2 after 24hrs.
Statistical procedures for the selection of spots of interest were carried out using XLstat addinsoft version 7.5.3 (Microsoft Corporation, Seattle, USA). The normal distribution of our data was verified using the Shapiro-Wilcoxon test and proteins with abnormal distributions were normalised using the Box-Cox procedure. Protein abundances were considered as statistically different between our conditions using a one way analysis of variance (p,0.05). Then differentially expressed proteins across all biological replicates with an absolute ratio of at least 1.5-fold were selected.
The mass spectrometry identifications were carried out by the proteomic platform based in the Centre de Recherche Public G. Lippmann (Belvaux, Luxembourg). Spots of interest were excised using the Ettan Spot Picker from the Ettan Spot Handling Workstation (GE Healthcare). After washing and desalting in 50 mM ammonium bicarbonate/50% v/v methanol, followed by 75% v/v ACN, spots were then digested with 10ng/mL Trypsin Gold (MSgrade, Promega, Madison, WI, USA) in 20 mM ammonium bicarbonate. The peptides were extracted using a 50% ACN containing 0.1% TFA and analysed using the Applied Biosystems 4800 Proteomics Analyser (Applied Biosystems, Foster City, CA, USA). Calibration was carried out with the peptide mass calibration kit for 4700 (Applied Biosystems). Proteins were identified from their peptide mass fingerprintings and searching the NCBI protein sequence database using the MASCOT software (Matrix Science, http://www.matrixscience.com, London, UK). The search parameters allowed for carboxyamidomethylation of cysteine as fixed modification and oxidation of methionine, oxidation, dioxydation and kynurenin of tryptophan as variable modifications. After manual checking of the identifications, proteins with at least two matching peptides and a total ion score above 60 were considered as significantly identified. Figure 1 presents the growth time-course of P. anomala Kh6 in the presence or absence of B. cinerea on the developed apple model. P. anomala Kh6 grew in the same way in the absence and presence of B. cinerea and showed the usual phases, i.e. latent (0 to 2 hrs), exponential (2 to 10 hrs) and stationary (from 10 hrs onward).

Results
One hundred and five protein spots in the exponential growth phase and 60 spots in the stationary growth phase displayed significant differences depending on the absence or presence of B. cinerea. Out of these, 61% and 38% were successfully identified by mass spectrometry in the exponential and stationary phases, respectively ( Table 1). The remaining proteins were either not found in the databases or quantities in the excised spots were too low to allow for identification by mass spectrometry.
Proteome changes during the exponential phase Table 2 shows the P. anomala proteins up-or down-regulated by the absence or presence of the pathogen during the exponential phase. The abundance of forty four proteins was decreased, while the abundance of 63 was increased in the presence of B. cinerea. These proteins are implied in a broad range of metabolic pathways: genome expression, carbohydrate, energy and cellular metabolisms and cell process (Table 1).
Proteins down-represented in the presence of B.
cinerea. Out of the 13 proteins implied in genome expression, only guanylate kinase (spot 944) has a function in nucleotide synthesis. The other 12 proteins are implied in protein synthesis pathways. Methionine synthase (spots 79 and 80), homocitrate synthase (spot 352), acetohydroxy-acid isomeroreductase (spot 539), and NADP-specific glutamate dehydrogenase (spot 1048) have specific functions in amino acid metabolism. The translation elongation factor 2 (spot 131), heat shock protein SSB1 (spots 199, 207 and 213), the mitochondrial isoform of GTPase elongation factor Tu (spot 469) and 40S ribosomal protein S0 (spots 696 and 1460) are involved in the translation process.
Six proteins are involved in the carbohydrate and energy metabolisms. 2,3-bisphosphoglycerate-independent phosphoglycerate mutase (spot 333) and enolase 1 (spots 441 and 452) are two glycolytic enzymes. The b-subunit of succinyl-CoA synthetase (spot 506) belongs to the citric acid cycle and the b-subunit of the F1F0-ATPase complex (spot 385) is a constituent of the respiratory chain.

Metabolism
Metabolic pathway Number of proteins over-represented in  Proteome changes during the stationary phase Table 3 shows proteins influenced by the absence or presence of B. cinerea during the stationary phase. Twenty proteins were downrepresented, while 40 were over-represented in the presence of the pathogen.
Proteins down-represented in the presence of B. cinerea. Seven proteins are implied in the carbohydrate and energy metabolisms. Glyceraldehyde-3-phosphate dehydrogenase (spot 1177) and fructose-1,6-biphosphate aldolase (spots 1350 and 1377) are glycolytic enzymes. Pyruvate decarboxylase (spots 241, 247 and 322) is implied in alcoholic fermentation. Mitochondrial F-ATPase b (spot 1135) is a constituent of the respiratory chain.

Discussion
The objective of the present study was to identify metabolic pathways of P. anomala influenced by the presence of B. cinerea, under conditions close to natural infection conditions. This allowed us to correlate proteome expression with the biocontrol process and propose molecular links between both datasets. Exponential and stationary phase proteomic profiles differed, suggesting different physiological states of the yeast. The proteome of P. anomala inoculated alone on apple wounds was compared to the proteome of P. anomala co-inoculated with B. cinerea, first in the exponential growth phase and second in the stationary growth phase. The large number of proteins influenced by the presence of B. cinerea in the exponential compared to the stationary phase seemed to indicate that the early response of P. anomala to the introduction of the pathogen involves large metabolic modifications.
The pentose phosphate pathway: a possible early answer to the introduction of B. cinerea In the early stages after inoculation, i.e. in the exponential growth phase, our results may indicate that P. anomala modified its energetic metabolism in order to respond to its needs.
In the absence of B. cinerea, the over-representation of two isoforms of enolase 1 (spots 441 and 452) and 2,3-bisphosphoglycerate-independent phosphoglycerate mutase (spot 333) suggests that the glycolysis pathway is activated during the growth and wound colonization. Indeed, glyceraldehyde-3phosphate dehydrogenase and other glycolytic enzymes were the most abundant proteins isolated from Saccharomyces cerevisiae grown on a glucose medium in the exponential phase [29]. Apples are composed of 12.6% carbohydrates, mainly in the form of glucosederived molecules such as fructose (a glucose isomer), sorbitol and pectin (a heteropolysaccharide of galacturonic acid, arabinose, galactose, rhamnose and xylose [30]) that could enter the glycolytic pathway to provide P. anomala with the energy required for its growth. Likewise; the higher proportion of proteins involved in the protein synthesis process in the absence of B. cinerea should be the consequence of the high rate of protein synthesis that occurs in the exponential phase. Indeed, the highest incorporation rate of 35S-amino acids in a culture of S. cerevisiae on YPD medium was observed during the exponential phase [31].
The proteomic map of P. anomala in the presence of B. cinerea during the exponential phase showed that proteins involved in the carbohydrate and energy metabolisms, genome expression and cellular metabolism were highly represented, suggesting a high metabolic activity. This high activity may be related to the capacity of P. anomala to exhibit similar growth time-courses in the absence or presence of B. cinerea (Figure 1) [27]. In the presence of the pathogen, proteins involved in the PPP are over-represented and may allow P. anomala to efficiently use apple nutrients and support growth. Arabinose, an apple constituent, can be converted into xylose by L-xylulose reductase (spot 724, Table 2 and Fig.  2), an enzyme of the pentose and glucuronate interconversion pathway [32] that is over-represented in the presence of B. cinerea. PPP has been demonstrated as the main route for xylose catabolism in yeast [33]. The newly produced amount of xylose from arabinose or xylose derived from apple could then enter the PPP (Fig. 2) under the xylulose-5P form and be converted into fructose-6P and glyceraldehyde-3P, two glycolysis intermediates, via the action of transketolase (spots 120 and 121) and transaldolase (spot 631, Fig. 2), two enzymes over-represented in the presence of B. cinerea. The over-representation of five glycolysis proteins, particularly triosephosphate isomerase (spot 814) and fructose-1,6-biphosphate aldolase (spot 686) suggests the entry of fructose-6P and glyceraldehyde-3P into glycolysis and the production of high amount of pyruvate. Then pyruvate may be metabolised to supply substrates for oxidative phosphorylation inside mitochondria as suggested by the overrepresentation of seven proteins involved in that pathway (Table 2 and Fig. 2).
Moreover, fructose, sorbitol, galactose and galacturonic acid, other apple constituents could be converted into glucose-6P. The over-representation of 6-phosphogluconate dehydrogenase (spots 351 and 390), the first enzyme of the PPP which catalyses the irreversible decarboxylation of 6-phosphogluconate into ribulose-5-phosphate, suggests that the newly-formed glucose-6P could enter the PPP to form ribulose-5P together with synthesis of NADPH,H+ (Fig. 2), a powerful reducing-oxidative agent involved in protection against the toxicity of reactive oxygen species and in glutathione regeneration [34,35]. The newly synthesised NADPH,H+ may probably be used during oxidative phosphorylation in order to protect the yeast against reactive oxygen species (Fig. 2). Indeed, PPP plays an important role in the oxidative stress response in Saccharomyces cerevisiae [36].
These results are linked to the over-representation of 17 proteins involved in nucleotide metabolism, transcription, amino acid metabolism and translation. Via the PPP, newly-synthesised ribulose-5P may be converted into ribose-5P, the precursor of nucleic acids (Fig. 2), in order to respond to its high metabolic activity in the presence of B. cinerea. Moreover, the capacity of P. anomala to exhibit similar growth time-courses in the absence or presence of B. cinerea (Fig. 1) and the over-representation of proteins involved in cell division, cell division control protein Cdc48 (spot 53) implicated in protein processing in the endoplasmic reticulum, ATP-dependent molecular chaperone HSC82 (spot 98), a constitutively expressed protein involved in the cellular cycle [37], and Nucleosome assembly protein (spot 341), a protein involved in mitosis [38], suggest that P. anomala could mainly use ribose-5-P for its multiplication and growth.
Thus, in the presence of B. cinerea, our results suggest that the PPP may supply the yeast with an efficient consumption of apple nutrients and consequently an adaptation of its metabolism to meet its immediate needs. Then, we could hypothesize that P. anomala may be an efficient coloniser of the wound and a nutrient competitor for B. cinerea. Actually, competition for space of specific infection sites is admitted as a mode of action of various biocontrol agents [7][8][9][10][11][12]. Our results may link the PPP with the protective effect of P. anomala against B. cinerea on apple. However, additional experiments have to be done in order to confirm this hypothesis. For example, the gene coding for 6-phosphogluconate dehydrogenase could be disrupted in order to inhibit the PPP and observe its influence on the protection level.
The alcoholic fermentation: the energy metabolism in the later stages In the later stages after inoculation, i.e. in the stationary growth phase, the proteome of P. anomala in the absence or the presence of B. cinerea seems to present no differences in the carbohydrate and energy metabolism. Proteins implied in glycolysis (spots 363, 534, 765, 814, 1177, 1344, 1350 and 1377) and alcoholic fermentation (pyruvate decarboxylase, spots 241, 247, 261, 264 and 322) are expressed. Alcoholic fermentation was further identified as a metabolic pathway providing energy in the post-diauxic phase that precedes the stationary phase. On the contrary, when yeast was grown on a rich medium for a long time, energy was provided by oxidative respiration in the stationary phase [39]. According to our growth time-course (Figure 1), 24h after inoculation, P. anomala presented stagnation in its population suggesting that yeast entered the stationary phase [27]. Then P. anomala should use the oxidative respiration and not the alcoholic fermentation. However, another study demonstrates that P. anomala, when inoculated on a glucoserich medium, used alcoholic fermentation as a way to provide energy in response to oxygen limitation [40]. In our model, inoculated apples were enclosed in plastic boxes in order to maintain a high relative humidity. Moreover, wounded apple sites represent media that are rich in glucose-derived molecules. During the first growth steps, the two microorganisms may possibly consume a large part of the available oxygen, or oxygen proportion may have been modified by microorganism carbon dioxide release or apple ethylene release. In these conditions, oxygen limitation may induce fermentative metabolism in P. anomala [40].
These last results could be a new way to explore in biocontrol. Indeed; in enclosed conditions, as in our experiment, yeast could use alcoholic fermentation and may produce volatile compounds like ethanol or ethyl acetate recently described as possible antimicrobial and antifungal compounds [41]. Then we could hypothesize that this volatile compound production could then contribute to the B. cinerea biocontrol by P. anomala. Finally, monitoring oxygen availability in storage rooms could improve the efficiency of biocontrol by P. anomala. However, we cannot rule out the implication of competition for space and nutrients in the inhibitory effect against the pathogen in the later stages. In the stationary phase, yeast was totally established on the wound site, leaving no space or nutrients for B. cinerea growth. Thus, P. anomala could maintain its inhibitory effect on the pathogen without necessarily inducing any specific metabolic pathway involved in the biocontrol modes of action, as in the exponential phase. However, additional experiments have to be done in order to confirm this hypothesis.

Observation of a higher protein synthesis induced by the introduction of the pathogen in the later stages
In the later stages after the co-inoculation, proteins implied in genome expression seem to be over-represented: glutamine synthetase (spot 471), which catalyses the fixation of a second amine group on glutamate to synthetise glutamine (ter Schure et al., 2000), orotate phosphoribosyltransferase (spot 850), which catalyses orotidine monophosphate formation during pyrimidine metabolism [42], NADP-dependent glutamate dehydrogenase (spot 343), involved in the conversion of aketoglutarate into glutamate [34,43], acetohydroxy-acid isomeroreductase (spot 530), involved in valine, leucine and isoleucine synthesis [44,45], 40S ribosomal protein S0 (spot 662), which is required for the assembly and/or stability of the 40S ribosomal subunit [46] and eukaryotic translation initiation factor 5A (spot 973), which plays a role in the formation of the first peptide bond [47]. This would suggest a higher protein synthesis rate than in the absence of the pathogen. Introducing the pathogen induced metabolic modifications such as the implementation of new metabolic pathways during the exponential phase. Here, the over-representation of proteins implied in genome expression could be the first signs of a new metabolic pathway or alternatively, in the presence of the pathogen, yeast may have maintained a high level of alcoholic fermentation activity implying high corresponding enzyme synthesis levels.
Finally, this study aimed to identify P. anomala metabolic pathways that are differentially expressed after the introduction of B. cinerea on apple. In the early stages of the co-inoculation P. anomala seems to set up the pentose phosphate pathway in order to supply higher amounts of energy and nucleic acid and support its high metabolic activity. This result suggests that PPP supports the efficient apple wound colonization. In the later stages, P. anomala seems to use alcoholic fermentation which is associated with an increased protein synthesis potential.
Nevertheless, this study confirms the complexity of the interaction between B. cinerea and P. anomala. Different mechanisms are influenced by growth stages and experimental conditions. However, these results suggested new targets in the study of the yeast mode of action against the pathogen in apple. Different pathways could be investigated in order to improve our knowledge of the biocontrol agent P. anomala.