Proteomic Analysis of Banana Vascular Sap Provides Insight Into Resistance Mechanisms to Fusarium Oxysporum F. Sp. Cubense Tropical Race 4

Background: Fusarium oxysporum f. sp. cubense tropical race 4 (Foc TR4) is the causal agent of Fusarium wilt, and is the most destructive soil-borne and vascular invasive fungus of banana. The sap circulating in vascular cells transports proteins including those that might be involved in disease-resistance processes. However, there is no research to analyze changes in banana vascular sap protein response to TR4 to date. Results: To gain an integrated understanding of differential protein abundance in banana vascular sap during TR4 infection, we performed a comparative proteomic analysis of vascular sap of the resistant ‘Pahang’ and the susceptible ‘Brazilian’ bananas inoculated with TR4. We identied 129 differential expression proteins (DEPs) between resistant and susceptible tested combinations. Of these DEPs, hypersensitive-induced response protein 1 (HIR1) and E3 ubiquitin ligase (E3) decreased in abundance in Pahang with no change in Brazilian under TR4 infection; chalcone isomerase (CHI) and glycine-rich RNA-binding protein (GRP) increased in abundance in Pahang but no signicant changes in Brazilian under TR4 infection; carboxylesterase (CXE) and GDSL lipase (GLIP) were specically in higher abundance in Pahang response to TR4 compared to that of Brazilian. It suggested that these proteins played important roles in bananas against TR4. Conclusions: Our study identied 129 DEPs in vascular sap between resistant and susceptible tested combinations. Of which, HIR1, E3, CHI, GRP, CXE and GLIP played important roles in bananas response to TR4. To our knowledge, this is rst report to analyze changes in banana vascular sap proteins in response to TR4, which help us to explore the molecular mechanisms of banana defense to Fusarium wilt. and TCEP: Tris(2-carboxyethyl)phosphine; IAM: Iodoacetamide; TEAB: Tetraethylammonium bromide; RPLC: Reversed-phase liquid chromatography; LC-MS/MS: Liquid chromatography-tandem mass spectrometry; HCD: High energy collisional dissociation; False discovery rate; of Orthologous Groups; Gene of Genes and SAR: systemic resistance; PAMP-triggered CaMCML: Calcium-binding protein CML; ligase; CCR: Cinnamoyl-CoA alcohol dehydrogenase; Caffeoyl-CoA O-methyltransferase; CHS: Chalcone synthase; 3',5'-hydroxylase; DFR: Dihydroavonol-4-reductase; LACS: Long chain acyl-CoA synthetases; PLDα1: recognition motif; TMV: Tobacco mosaic virus.

Genetically modi cation of a susceptible commercial banana is a promising alternative for banana improvement [14][15][16]. However, defense mechanisms of banana against TR4 are not well understood.
TR4 is the most destructive soil-borne and vascular invasive fungus. It invades root vascular bundles and extends upward to the aerial parts. We previously investigated transcriptomics in the corm to identify the pathways involved in the resistance [17]. Proteomics, complementary to transcriptomics, can provide insights into complex biological processes in banana [18][19][20]. Large-scale studies of proteomics previously focused on dissecting interactions between bananas and Foc [21][22][23]. The proteins related to PR response, cell wall strengthening and antifungal compound synthesis were involved in banana defense to TR4 [22]. β-1,3-glucanase and chitinase were reported to function in banana against TR4 at the early defense stage [21]. The expression patterns of proteins related with cell cytoskeleton, natural killer cell mediated cytotoxicity and lipid signaling were different in banana during Foc1 and Foc4 infection, suggesting these proteins participated in mediating different resistance to Foc1 and Foc4 in banana cultivar 'Brazilian' [23]. These studies help us to understand the defense mechnism of banana against TR4.
Plants transport signal molecules as well as water and minerals over long distance via the vascular bundles [24]. The signal molecules are vital for plant adaption to abiotic and biotic stress [25,26]. The vascular sap proteomics have been applied to characterize the processes associated with plant defense to Fusarium wilt [27][28][29][30][31]. However, to our knowledge, there is no research to analyze changes in banana vascular sap proteins response to TR4 to date. In this study, we performed a comparative proteomics analysis of vascular sap in resistant and susceptible bananas inoculated with TR4. 129 DEPs were identi ed between resistant and susceptible tested combinations, among which HIRP1, E3, CHI, GRP, CXE and GLIP involved in banana defense against TR4. This study provides integrated insight into the resistant mechanism of banana against Fusarium wilt.

Results
Vascular saps of Pahang (resistant) and Brazilian (susceptible) inoculated with TR4 or mock at 14 dpi were collected for comparative proteomic analysis. A total of 261,038 spectra were acquired through iTRAQ quantitative proteomics analysis, among which 31,450 spectra were matched to 6,503 peptides and 1036 proteins.
The vascular proteome of banana All identi ed proteins (1,036) were conducted functional analysis, of which 938, 923 and 779 proteins were annotated with COG, GO and KEGG databases, respectively.
In terms of COG (Fig. 2), 938 proteins were assigned into 23 functional categories. Of which 'posttranslational modi cation, protein turnover, chaperones', 'energy production and conversion' and 'carbohydrate transport and metabolism' were the top 3 largest categories. Related to plant defense were 'lipid transport and metabolism', 'cell wall/membrane/envelope biogenesis', 'secondary metabolites biosynthesis, transport and catabolism', 'signal transduction mechanisms' and 'defense mechanism' categories.
In terms of GO (Fig. 3), 923 proteins were assigned into 43 GO terms, and divided into 3 groups, including biological process, cellular component and molecular function. 'Response to stimulus', 'signaling', 'detoxi cation', 'immune system process' and 'antioxidant activity' were usually regarded to relate to disease resistance among molecular functional groups.

Differential expression proteins analysis
A total of 129 unique DEPs were identi ed in the 4 pairwise comparisons between mock and inoculated in resistant and susceptible genotypes ( Fig. 5a and 5b), for which we performed expression and function annotation analysis (Supplementary Table 2

Discussion
TR4 is a vascular-invading fungus, and it colonizes in the vascular system of banana and completes its life cycle [32]. Vascular sap contains macromolecules, such as proteins, involved in disease-resistance processes [27,28,31]. To gain an integrated understanding of the changes of banana vascular sap proteins during TR4 infection, we performed a comparative proteomic analysis of vascular sap in resistant diploid 'Pahang' and susceptible triploid 'Brazilian' inoculated with TR4 at 14 dpi. The amount of fungal biomass and the degree of necrosis in Pahang tissues were signi cantly less than their levels in Brazilian at 14 dpi [17,33].

Signal transduction
Signal transduction pathways are responsible for induction of plant defense against pathogen [34,35], such as mitogen-activated protein kinase (MAPK) cascades and plant hormone signals [36]. Once plant perceives the invading pathogen, the activation of MAPKs is one of the earliest signaling events [37]. In the present study, we found 10 proteins related with MAPK signaling pathway-plant (Supplementary Table 1), such as nucleoside diphosphate kinases known to be an inducer of MPK3/6 expression through phosphorylation leading to hypersensitive response (HR) cell death in plant response to pathogen attack [38].
Salicylic acid (SA) and jasmonic acid (JA) are essential hormone signals of plant immunity [39]. SA and JA antagonize each other [40]. It is generally considered that SA enhances resistance to biotrophs, while JA is effectively against necrotrophs and insects [41,42]. Fusarium oxysporum was classi ed as hemibiotrophs [43]. SA metabolism activation and signal transduction, and JA induced defense responses improved banana resistance to TR4 [44,45]. In this study, we identi ed 2 proteins associated with SA and JA signaling (Supplementary table 1), including pathogenesis-related protein 1 (PR1), a marker for systemic acquired resistance (SAR) from SA signaling pathway [46]; and coronatineinsensitive protein homolog (COI1), a key regulator for JA-dependent induced systemic resistance (ISR) [41,47]. Further research is needed to determine whether SA-dependent SAR and JA-dependent ISR are simultaneously activated in banana.

Environmental adaptation
In their natural habitats, plants are threatened by various abiotic and biotic stress. Over the evolutionary course during plant-pathogen interaction, plants have developed multi-layered innate immune system to defend against pathogen. The preliminary layer of immune is pathogen-associated molecular pattern (PAMP) perceived by pathogen recognition receptors (PRRs), and induces a series of physiological changes leading to PAMP-triggered immunity (PTI) [48]. These physiological changes include reactive oxygen species (ROS) bursts and calcium (Ca 2+ ) concentrations changes [49,50]. Ca 2+ acts as an important second messenger whose concentration is sensed by Ca 2+ -binding proteins, such as calciumdependent protein kinase (CDPK) and calcium-binding protein CML7 (CaMCML) that we detected in the banana vascular sap (Supplementary Table 1). It initiates downstream signaling processes [51], such as hypersensitive response and cell wall reinforcement.

Biosynthesis of secondary metabolites
Plant secondary metabolites contribute to all aspects in plant and pathogen interactions [52]. In the biosynthesis of secondary metabolites, phenylpropanoid and avonoid biosynthesis have been proved to encompass a wide range of constitute and inducible immunity through lignin and phytoalexin synthesis [53]. In our experiment, we found 21 proteins involved in the phenylpropanoid biosynthesis (Supplementary Table 1). We detected synthetic enzymes of lignin leading to strengthen cell walls [17], including phenylalanine ammonia-lyase (PAL), 4-coumarate-CoA ligase (C4L), 2 cinnamoyl-CoA reductases (CCR), cinnamyl alcohol dehydrogenase (CAD) and 3 peroxidases (POD) (Supplementary Table 1). In addition, caffeoyl-CoA O-methyltransferase (CCoAOMT) associated with lignin production resulting in quantitative resistance to multiple pathogens [54], was found. 13 proteins were assigned to avonoid biosynthesis, such as chalcone synthase (CHS) as the gatekeeper of avonoid biosynthesis which can help plant to produce more avonoids, iso avonoid-type phytoalexins [55], three P450 enzyme avonoid 3',5'-hydroxylase and dihydro avonol-4-reductase (DFR) as precursors for the production of catechins and pro-anthocyanidins involved in plant resistance [56].

Lipid metabolism
Lipids and fatty acids involved in lipid metabolism are considered as signal transduction mediators of plant disease resistance [57,58]. We found two long chain acyl-CoA synthetases (LACS) involved in fatty acids metabolism known to act in the synthesis of cutin to confer plant resistance to fungal pathogen [59,60]. Additionally phospholipases D α1 (PLDα1) involved in lipid metabolism which promote phosphatidic acid and ROS [61] were detected in vascular sap (Supplementary Table 1 Table 2). It suggests that TR4 did not induce highly dramatic changes in the overall vascular sap proteome. This result was similar to the proteomic analysis of phloem sap in melon defense against viral infection [62]. Nevertheless, these DEPs present in vascular sap might play important roles in banana response to TR4.
Among those, we detected 4 genes of interest. First, the hypersensitive-induced response protein 1 (HIR1) which decreased in abundance in Pahang with no change in Brazilian under TR4 infection HIR1 is known to act as regulators of plant immunity by triggering hypersensitive cell death [63,64]. It would suggest that Pahang decreases HIR1 expression to suppress the cell death as a resistance mechanism to TR4 due to Foc as hemibiotroph or necrotroph [43,65].
Ubiquitin involved in the ubiquitination system are key for plant immunity [66]. Ubiquitination is mediated by three step enzymatic cascades, including activating (E1), conjugating (E2) and ligating (E3) enzymes [67]. E3 ubiquitin ligase RING1 gene, CaRING1, played a positive role in pepper (Capsicum annuum) response to microbial pathogens [68]; whereas a homologous triplet of U-box type E3 ubiquitin ligases acted as negative regulators of PTI in Arabidopsis [69]. In the study, E3 with a RING zinc-nger domain decreased in abundance only in Pahang response to TR4. However, further studies are needed to prove whether this protein played a negative role in banana response to TR4.
CHI is an important enzyme of avonoid pathway involved in the production of phytoalexin [70]. We found a chalcone-avonone isomerase (CHI) that increased in abundance in Pahang under TR4 infection, as well as in Pahang control compared with Brazilian control.
Finally, GRP containing an RNA recognition motif (RRM) domain was increased in abundance in Pahang response to TR4, but no signi cant changes in other pairwise comparisons. GRPs act as regulators in diverse cellular processes, including response to stress in plants [71,72]. Over expressing TaRZ1, a wheat (Triticum aestivum) zinc nger-containing GRP, in Arabidopsis thaliana increased resistance against necrotrophic bacteria Pseudomonas syringae [73]. Although functional validation would be needed to con rm the actions, our results unraveled a list of promising candidate genes to explore.

Pahang speci c reaction during TR4 infection
We previously observed with gene expression on the same genotypes that Pahang exhibits constitutive defense responses before TR4 infection [17]. Proteomic results indicated that the higher number of DEPs (84) is between Pahang and Brazilian without infection. We also identi ed 26 DEPs by comparing TR4 inoculated Pahang with that of Brazilian, but could nd only one protein with increased abundance in both infected bananas. These results show our susceptible and resistant banana genotypes have de nitively a different response to TR4 infection. Among the 26 DEPs, 7 proteins have an unknown function but two proteins were highly associated with resistance to pathogen in plant, annotated as carboxylesterase and GDSL esterase/lipase. [74] while constitutive expression of PepEST, a fungus-inducible carboxylesterase in pepper (Capsicum annuum) increased resistance against the hemibiotrophic anthracnose fungus (Colletotrichum gloeosporioides) [75]. In the present study, a CXE protein (Ma06_t34160) was speci cally in higher abundance in TR4 inoculated Pahang compared to Brazilian.
Overexpressing GLIP1 in Arabidopsis improved resistance against hemibiotrophic and necrotrophic pathogens [77,78]. In the study, a GLIP protein (Ma11_t05100) showed an increased abundance in TR4 inoculated Pahang compared with that of Brazilian. This is consistent with our previous transcriptomic study in which this GLIP gene was also activated by TR4 attack [17].

Conclusions
To gain an integrated understanding of the changes of banana vascular sap proteins during TR4 infection, we performed a comparative proteomic analysis of vascular sap in resistant diploid 'Pahang' and susceptible triploid 'Brazilian' inoculated with TR4 at 14 dpi. A total of 1036 proteins were detected in vascular sap, some of these proteins are involved in 'biosynthesis of secondary metabolites', 'environmental adaptation', 'lipid metabolism' and 'signal transduction', which are commonly considered as disease-resistance pathways. Since the vascular sap contained defense-related proteins the constitutive presence of certain proteins could contribute toward resistance. 19 proteins were signi cantly more abundant after TR4 inoculation, and 26 proteins were only upregulated in the resistant genotype. Of these DEPs, hypersensitive-induced response protein 1 (HIR1) and E3 ubiquitin ligase (E3) decreased in abundance in Pahang with no change in Brazilian under TR4 infection; chalcone isomerase (CHI) and glycine-rich RNA-binding protein (GRP) increased in abundance in Pahang but no signi cant changes in Brazilian under TR4 infection; carboxylesterase (CXE) and GDSL lipase (GLIP) were speci cally in higher abundance in Pahang response to TR4 compared to that of Brazilian. It suggested that these proteins played important roles in bananas against TR4. To our knowledge, this is rst report to analyze changes in banana vascular sap proteins response to TR4 to date, which provided insight into resistant mechanisms of banana defense against Fusarium wilt. In the next steps, the function of these proteins will be further validated.

Plant inoculation and vascular sap collection
We selected Musa acuminata 'Pahang' (AA, ITC0609), with o cial SMTA-2015 from the International Musa Germplasm Transit Centre (ITC), and Musa Cavendish 'Brazilian' (AAA, commercial cultivar in China). Pahang is resistant to TR4, while Brazilian is susceptible to TR4 [33,79,80]. The banana inoculation was performed similarly [33] with minor modi cations. The roots of banana with 6-8 leaves were cut to 5 cm and immersed into TR4 conidia suspension for 30 min at a concentration of 10 6 conidia/mL. The plants were soaked into sterile water as mocks. All plants were transplanted to pots lled with sterile vermiculite, and placed in an arti cial climate chamber at 30℃, 80% humidity, and 8 h light/16 h dark (Fig. 1a, b). At 14 days post inoculation (dpi), the pseudostems were transected at 0.5 cm above the corms with a sterile blade (Fig. 1c). After removed the exudate from the cut cells, the vascular sap exuded spontaneously from the remaining pseudostems was collected with a pipette (Fig. 1d-1f). Vascular sap isolated from at least 30 plants was pooled into one independent biological replicate, and three independent biological replicates were conducted. The vascular sap was frozen with liquid nitrogen until protein extraction.
Added equal volume of Tris-saturated phenol, and vortexed at 4℃ for 10 min. Centrifuged with 12,000 g at 4℃ for 20 min to take the phenol phase, and added an equal volume of BPP solution, and vortexed for 10 min at 4℃. Centrifuged with 12,000 g at 4℃ for 20 min to collect the phenol phase. The proteins were precipitated overnight at -20℃ from the phenol phase with pre-cooled ammonium acetate methanol in a ratio of 1:5. Centrifuged with 12,000 g at 4°C for 20 min the next day and discarded the supernatant. Washed the pellet twice with 90% pre-cooled acetone. Used 8 M urea and 1% sodium dodecyl sulfate (SDS) with a protease inhibitor to dissolve the pellet. Centrifuged with 12,000 g at 4℃ for 20 min to collect the protein supernatant.

Protein Digestion and iTRAQ Labeling
Protein concentrations were detected using BCA Protein Assay Kit (Pierce, Thermo, USA). Protein digestion was carried out according to the standard procedure. Brie y, took 100 μg protein of each sample. Added the nal concentration of 10 mM Tris(2-carboxyethyl) phosphine (TCEP) and incubated at 37°C for 60 min. Added the nal concentration of 40 mM iodoacetamide (IAM) to react at room temperature in the dark for 40 min. Added pre-cooled acetone (acetone: sample volume ratio = 6:1) to each tube, and incubated at -20°C for 4 h. Centrifuged at 10,000g for 20 min, and discarded acetone and took the precipitate. Resuspended the precipitated protein with 100 µL 100 mM tetraethylammonium bromide (TEAB) buffer. Added trypsin solution (1:50) to each tube, and incubated at 37°C overnight. The resulting peptide mixture was labeled using the 8-plex iTRAQ reagent (Applied Biosystems, 4390812) according to the manufacturer's instructions [81].
Functional annotation of peptides MS/MS spectra were searched using Proteome Discoverer (Thermo Scienti c, Version 2.2) against Musa acuminata database from UniProtKB-SwissProt [82] (http://www.uniprot.org/proteomes/UP000012960) and the decoy database as the following parameters. The highest score for a given peptide mass (best match to that predicted in the database) was used to identify parent proteins. The parameters for protein searching were based on below criteria: tryptic digestion with up to two missed cleavages, carbamidomethylating of cysteines as xed modi cation, and oxidation of methionine and protein Nterminal acetylation were considered as variable modi cations. False discovery rate (FDR) of peptide identi cation was set as FDR≤0.01. In order to support protein identi cation, a minimum of one unique peptide identi cation was used.
For 26 genes (Fig. 5e), we added the RefSeq (NCBI) annotation locus codes and the V2 annotation names (column "C" and "D", respectively) retired at the Banana Genome Hub [83]. Moreover, we added a more informative or alternative functional annotation which was available we added it in the "E" column. Finally, we checked if one or more paralogs or similar genes were present in Musa acuminata genome V2 [84] (Column "F" and "G", respectively) (Supplementary Table 2).
iTRAQ quantitative proteomics analysis The basic information analysis process for iTRAQ quantitative proteomics using the free online platform of Majorbio Cloud Platform (www.majorbio.com). First, the raw mass spectra generated by the mass spectrometer were subjected to the peak identi cation. Secondly, the UniProtKB-SwissProt reference proteomic database of banana (http://www.uniprot.org/proteomes/UP000012960) was established to identify peptides and proteins. All identi ed proteins were functional annotated using Cluster of Orthologous Groups of proteins (COG, http://eggnogdb.embl.de/#/app/home), Gene Ontology (GO, http://www.geneontology.org/) and Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/) with E-value≤1×10 -5 and identity≥0.98. The DEPs were identi ed with fold change>1.2 (upregulation), fold change<0.83 (downregulation) and P value<0.05 [23], and analyzed including DEPs Venn and expression pattern analysis.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
The authors con rm that the work described has not been published before. All authors read and approved the nal manuscript.

Availability of data and materials
Musa acuminata 'Pahang' (AA, ITC0609) is imported with o cial SMTA-2015 from the International Musa Germplasm Transit Centre (ITC). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository [85] with the dataset identi er PXD018261.

Competing interests
Program (CRP) on Roots, Tubers and Bananas (RTB) provide project research funds and analysis, and interpretation of data and in writing the manuscript.
Authors' contributions LZ and SJZ: conceptualization. LZ: performed the experiments, analyzed the data and wrote the paper. LL, SL, TB, SX, HF, KY, and PH: analyzed the data. MR, AC and SC: provided feedback on data analyses and reviewed the manuscript. YW and WT proof writing. SJZ: conceived and funding acquisition, designed the experiments and proof writing. All authors have read and approved the manuscript.