Powdery mildew-induced changes in phyllosphere microbial community dynamics of cucumber

Abstract As an important habitat for microorganisms, the phyllosphere has a great impact on plant growth and health, and changes in phyllosphere microorganisms are closely related to the occurrence of leaf diseases. However, there remains a limited understanding regarding alterations to the microbial community in the phyllosphere resulting from pathogen infections. Here, we analyzed and compared the differences in phyllosphere microorganisms of powdery mildew cucumber from three disease severity levels (0% < L1 < 30%, 30% ≤ L2 < 50%, L3 ≥ 50%, the number represents the lesion coverage rate of powdery mildew on leaves). There were significant differences in α diversity and community structure of phyllosphere communities under different disease levels. Disease severity altered the community structure of phyllosphere microorganisms, Rosenbergiella, Rickettsia, and Cladosporium accounted for the largest proportion in the L1 disease grade, while Bacillus, Pantoea, Kocuria, and Podosphaera had the highest relative abundance in the L3 disease grade. The co-occurrence network analysis of the phyllosphere microbial community indicated that the phyllosphere bacterial community was most affected by the severity of disease. Our results suggested that with the development of cucumber powdery mildew, the symbiotic relationship between species was broken, and the entire bacterial community tended to compete.


Introduction
Microbes can benefit plants by decomposing compounds that cannot be absorbed (Cha parr o et al. 2012, Almario et al. 2017, Li et al. 2022a ), and can also stimulate plant defense responses (Berendsen et al. 2012 ).The phyllosphere is a complex microbial habitat (Vorholt 2012 ), as with the rhizosphere, host species and genotypes , complex en vironmental conditions , and various biological and abiotic factors can affect the phyllosphere microbial community (Singh et al. 2018, Beilsmith et al. 2019, Sc hlec hter et al. 2019, Li et al. 2021 ).Micr oor ganism-plant inter actions hav e a huge impact on plant health (Getzke et al. 2019 ).For example, Bacillus subtilis effectiv el y contr olled soybean root rot and impr ov ed r oot biomass, plant height, and c hlor ophyll content (Jia et al. 2021 ).Pseudomonas , Sphingomonas , and Bacillus species in the phyllospher e, alle viated the dama ge to plants caused by the pathogen (Tsa vkelo v a et al. 2007, Inner ebner et al. 2011 ).Members of the Methylobacterium , Microbacterium , and Stenotrophomonas could impr ov e the host plant's gr owth and nutritional status by producing indoleacetic acid and fixing nitrogen (Madhaiyan et al. 2015 , Schoenfelder andFraser 2019 ).When challenged by pathogens, plants increased the abundance of specific bacteria or fungi through biosynthesis, transport, and secretion processes (Canarini et al. 2019, Liu et al. 2020 ), and also attracted beneficial micr oor ganisms in response to disease stress (Carrion et al. 2019, Liu et al. 2021 ).
Cucumber is an important vegetable crop in the world, which is often affected by various diseases during growth and de v elopment.Powdery mildew, caused by Golovinom y -cescic horacearum or P odosphaera xanthii (Kuzuy a et al. 2006, P erez-Gar cia et al. 2009 ), is widely distributed and spreads rapidly, making it one of the most de v astating diseases of cucumber plants (Fukino et al. 2013 ).Curr entl y, c hemical a gents ar e widel y used to contr ol powdery milde w (Cerkauskas and Ferguson 2014 ).Ho w ever, the widespread and long-term use of c hemical a gents has made pathogenic bacteria resistant and caused environmental pollution (Fondevilla andRubiales 2012 , Rubiales et al. 2015 ).Biological control has the advanta ges of str ong persistence and eco-friendl y, whic h meets the r equirements of agricultural sustainable development (Tanaka et al. 2017, Rur et al. 2018 ).Exploring the c hanges in the phyllospher e micr obial comm unity thr oughout the de v elopment of cucumber powdery mildew is of great significance for the biological control of the disease.
Ther efor e, in this study, we investigated changes in the phyllospher e micr obial comm unities at differ ent gr ades of disease with the objectives of: (1) understanding the diversity and structural differences of phyllosphere microbial communities at different le v els of disease incidence, (2) investigating the correlation between the se v erity of cucumber powdery mildew and changes in the phyllosphere microbial communities, and (3) providing a theoretical basis for biological control studies of powdery mildew.

Pa thogen identifica tion
Conidia of powdery milde w wer e collected from cucumber (Jin You No. 1) leaves infected with powdery mildew and made into 2.5-5.0 × 10 5 conidia/mL of spore suspension.The spore suspension was inoculated on healthy cucumber seedlings by spray method for multiple isolations and purifications .T hen, the powdery mildew spores were placed on the glass slide with sterile forceps and observed using a 10 × 40x optical microscope.
DN A w as extr acted fr om cucumber powdery milde w pathogen and amplified with ITS1 (5 -TCCGTA GGTGAA CCTGCGG-3 ) and ITS4 (5 -TCCTCCGCTT A TTGA T A TGC-3 ) primers.PCR reactions were performed with the following pr ogr am: 3 min at 95 • C, follo w ed b y 35 c ycles of 30 s at 95 • C, 25 s at 56 • C, 1 min at 72 • C, and a final extension step of 5 min at 72 • C. The PCR products were detected on 1% a gar ose gel electr ophor esis and sent to Tsingke Biotechnology Co., Ltd.(Beijing, China) for Sanger sequencing.The gene sequence of powdery mildew strains was uploaded to the NCBI database and obtained accession No. OQ216740.Mega 7.0 softw are w as used for multiple sequence alignment.Construct a phylogenetic tree using the neighbor connection method and repeat testing with 1000 bootstr a ps.

Po wdery milde w pa thogen inocula tion and sample collection
Cucumber seeds w ere so wn in medium with a ratio of peat, vermiculite, and perlite of 3:1:1 and cultivated in a climate chamber with 60% r elativ e humidity, day/night temper atur e 25/15 • C, photoperiod 16/8 h, light/dark.After two euphylla times, they wer e tr ansplanted into the gr eenhouse and inoculated with a certain concentration of spore suspension on cucumber lea ves .Tw o w eeks later, based on the proportion of cucumber po wdery mildew lesions (0% < L1 < 30%, 30% ≤ L2 < 50%, L3 ≥ 50%) ( Fig. S1 ), leaf samples were collected from the same location and size using a five-point sampling method.The classification method of morbidity le v els used was in accordance with China's national standard (GB/T 17980.30-2000)(Luo et al. 2019 ).Eac h disease le v el has 5 replicates, and all the samples were transported to the laboratory at r efriger ated conditions.

DN A extr action and purifica tion
The method of microbial isolation in the phyllosphere was based on an earlier study (Redford and Fierer 2009, Bodenhausen et al. 2013, Xie et al. 2015 ).That is, after weighing the leaves, immersed them in a 100 mL sterile conical flask, and added a certain volume of 0.1 M PBS with 0.01% Tween 80. Shaken the flask at a speed of 250 rpm at 28 • C for 30 min, and then ultr asonic tr eatment for 10 min.The phyllospher e micr obiome wer e harv ested using a 0.22 μm filter.Then tr ansfer to −80 • C r efriger ator for stor a ge to subsequent DN A extraction.Accor ding to the manufacturer's plan, the MP FastDNA ® SPIN Kit for Soil (MP Biochemicals, Solon, OH, USA) was used to extract DNA fr om the leav es surface.Bacteria wer e amplified using the specific primers 799F (5 -AACMGGA TT AGA T ACCCKG-3 ) and 1193R (5 -ACGTC ATCCCC ACCTTCC-3 ).T he fungi were amplified using ITS1F (5 -CTTGGTCA TTT A GA GGAA GTAA-3 ) and ITS2R (5 -GCTGCGTTCTTCA TCGA TGC-3 ).Finally, the data was analyzed on Majorbio Cloud Platform's online platform.

Sta tistical anal ysis
To assess phyllosphere microorganism diversity and richness, alpha diversity indexes, including Shannon index, Simpson's index, Chao and ACE index, were analyzed by Mothur (V1.30.2).Principal coordinate analysis (PCoA) and Venn diagram analysis using the v egan pac ka ge in R softwar e (V.3.2.5) (Lozupone et al. 2006 ).Calculation of species numbers in multiple groups or samples and differences in species composition among samples or subgroups at different taxonomic levels using R software (V3.3.1).The Kruskal-Wallis test was used to investigate the abundance and composition of phyllosphere microbial communities at different taxonomic le v els.Linear discriminant anal ysis Effect Size (LEfSe) was used to identify dominant species in phyllosphere microbial communities of cucumber powdery mildew based on linear discriminant analysis (LDA threshold above 4).
To e v aluate the effect of disease se v erity on the phyllospher e micr obial comm unity, the co-occurr ence network of OTUs with the top 300 r elativ e abundance was constructed.The "psych" pac ka ge in the R language was used for pairwise comparison and Spearman correlation coefficient analysis .T he correlation with the Spearman correlation coefficient ( R > 0.6 or < -0.6) and statistical significance ( P < 0.05) was selected for network construction.The inter activ e platform Gephi 0.9.2 was used to visualize the network.The node color r epr esents the corr esponding phylum for each OTU, the node size is proportional to the number of connections, and the edge r epr esents the correlation among the nodes in the microbiome network.

Pa thogen identifica tion
As shown in Fig. 1 , the conidia of powdery mildew pathogen were oval with fibrous structure, and the number of fibrous bodies in eac h conidia v aries fr om 1 to 10.The conidiophor es consisted of spores in tandem.Molecular biological identification confirmed that the ITS sequence of powdery mildew pathogen was highly consistent with that of Podosphaera xanthii ( Fig. S2 ).Ther efor e, we determined that the pathogen causing cucumber powdery mildew was Podosphaera xanthii in the present study.

Effect of disease severity on the diversity of microbial community in the phyllosphere
A total of 3600945 bacterial and 3857627 fungal raw sequences were obtained from the 15 leaf samples (Table 1 ).These original sequences were classified as 3104 bacterial and 1454 fungal OTUs with 97% similarity, r espectiv el y ( Table S1 ).The r ar efaction curv es tend to be le v el off, meaning all samples r eac h sequencing depths ( Fig. S3 ).
The α diversity including Shannon, Simpson, Chao, and ACE indices.For bacterial communities α di versity, the di versity and richness of L3 incidence levels are significantly higher than other  incidence le v els (Fig. 2 A).The fungal α diversity sho w ed a decreasing trend with increasing disease grade, with no significant change in richness (Fig. 2 B).The analysis of PCoA displayed that the phyllospher e bacterial comm unity structur e differ ed significantl y at different incidence levels, while the fungal community structure differed less ( P < 0.05) (Fig. 3 ).

Effect of different disease severities on the composition and structure of the phyllosphere microorganisms
Venn dia gr ams wer e used to r epr esent unique and shar ed OTU numbers at different incidence levels (Fig. 4 ).The highest number of unique OTUs was found in bacterial L3 (1149) and fungal L1 (358).The quantity of shared OTUs was 13.92% in bacteria and 16.38% in fungi.
At the genus le v els (Fig. 5 C, D), the Rosenbergiella r elativ e abundance in the bacterial community accounted for the largest proportion in L1 and nearly none in L3.Relative abundance of Pseudomonas sho w ed a first incr easing and then decr easing tr end from L1 to L3. Candidatus_Portiera and Rickettsia relative abundance gr aduall y declined and Bacillus gr aduall y incr eased fr om L1 to L3.For the fungal community, P odosphaera w as the most relative abundant in L3 (83.19%).Wallemia and Cladosporium were the most r elativ e abundant in L1 (23.65% and 9.78%) in the fungal community.

Analysis on the dominant species of the phyllosphere microbial community under different degrees of disease
As shown in Fig. 6 , there were significant differences between dominant genera of phyllosphere microbial communities at differ ent disease le v els.In the bacterial community, the relative abundance of Rosenbergiella and Rickettsia in L1 (26.5% and 11.04%) was higher than in L2 (21.28% and 7.29%) and L3 (0.07% and 3.98%).In contr ast, the r elativ e abundance of Bacillus , Curtobacterium , Pantoea , Kocuria, and Virgibacillus was the highest in L3.In the fungal comm unity, the r elativ e abundance of Podosphaera was highest in L3 (83.19%) and lo w est in L1 (43.07%),Wallemia had the highest r elativ e abundance in L2 (25.75%),Cladosporium was the highest in L1 and the lo w est in L3.
LEfSe analysis revealed differences in the phyllosphere microbial communities at different levels of incidence from phylum to genus (Fig. 7 ).Bacterial community analysis found no dominant species enrichment in L2 compared with other r anks.Pr oteobacteria, Rhizobiales (from order to gen us), Rick ettsiales (from order to genus), Gamma pr oteobacteria (fr om order to genus), and Mor ganellaceae wer e significantl y enric hed in L1.In the L3 grade, Firmicutes and Actinobacteriota (all two from phylum to genus), P antoea, Micr ococcaceae (fr om famil y to genus) and Propionibacteriales (from order to genus) were enriched.The fungal communities Dothideomycetes (from class to genus), Pleosporales, Sordariom ycetes, and Basidiom ycota (fr om phylum to famil y) wer e

Analysis of microbial co-occurrence networks in the phyllosphere
By analyzing the co-occurrence network, the complexity of the relationship between the species of the phyllosphere microbial community at different disease levels was revealed (Fig. 8 , Table 2 ).It was found that most of the nodes of the co-occurrence network at different incidence levels belonged to 8 bacterial phyla and 2 fungal phyla.The number of bacterial community edges was inv ersel y pr oportional to the degr ee of disease, and the number of fungal edges first decreased and then increased.In the bacterial netw ork, there w as no significant change in the percentages of positiv e corr elation edges fr om L1 to L2 le v el, but the percenta ges of positive correlation edges obviously decreased at L3 level.While in the fugal co-occurrence network, it showed a trend of first increasing and then decreasing from L1 to L3 level.In addition, the av er a ge degr ee and the number of edges of L1 in both bacterial and fungal co-occurrence networks were higher than those of L2 and L3.

Discussion
Micr obial comm unities hav e long been consider ed an important component of the plant ecosystem and contribute significantly to plant growth and disease resistance (Cordovez et al. 2019 ).An increasing number of studies have shown that when plants are infested by pathogens, plant microbial communities change and recruit beneficial microbes from the environment to cope with the effects of disease stress (Trivedi et al. 2012, de Assis Costa et al. 2018, Masenya et al. 2021 ).This phenomenon in which plants activ el y seek to cooperate with microorganisms to enhance stress resistance is known as the "call for help" strategy (Bakker et al. 2018 ).At pr esent, ther e ar e man y r eports about the effects of pathogen on rhizosphere microbial communities, while the phyllospher e micr obiome has been less studied (Yin et al. 2021 ).
The study found that the bacterial diversity and richness index at L3 le v el was significantly higher than that at L1 level.We speculated that with the increase of disease index, plants would recruit more beneficial bacteria to cope with disease stress.This is consistent with the r esearc h on cucumber angular leafspot (Luo et al. 2019 ).Conv ersel y, fungal α div ersity index decr eased gr aduall y with the increase of disease se v erity.PCoA analysis also sho w ed that there were significant differences in microbial comm unity structur e in the phyllospher e at differ ent disease Figur e 5. T he r elativ e abundance of bacteria (A and C) and fungi (B and D) at phylum and gener a le v el under differ ent incidence le v els.se v erity.When powdery milde w occurr ed, Podospaera xanthii dominated in the phyllosphere microbiome, inhibiting the growth and r epr oduction of other fungi.This was the result of symbiosis and competition among microbial species.More and more powdery mildew pathogen was colonizing the lea ves , seriously disrupting the original balance between phyllosphere microbiota, and plants exhibited more severe disease symptoms.On the other hand the Podospaera xanthii belongs to specialized parasitic fungi, which compete for host nutrition and inhibit host defense responses (Zhang et al. 2018 ).Ther efor e, with the incr ease of leaf powdery milde w cov er a ge, cucumber leav es wither, become brittle, shrink, and lose photosynthetic function.Plants can maintain the stability of microbial communities by selecting and enriching certain micr oor ganisms, whic h hav e ac hie v ed the effect of r educing diseases (Santhanam et al. 2015 ).At the phylum le v el, the bacterial community is mainly composed of Proteobacteria, Actinobacteria, Firmicutes, and Bacteroidetes in our research, and the fungal community is composed of Ascomycota and Basidiomycota, whic h ar e also the major components of the phyllosphere communities of most plants .T he relative abundance of Actinobacteria and Firmicutes were highest at L3 grade.Actinobacteria and Firmicutes , which ha ve the functions of producing high-grade secondary metabolites and participating in catabolic and transformation processes (Kim et al. 2011, Palaniyandi et al. 2013 ), are able to suppress fungal disease (Kim et al. 2007, Mendes et al. 2011 ).At the genus le v el, the r elativ e abundance of Rosenbergiella and Rickettsia in the bacterial community was significantly negatively correlated with the disease grade .T here is limited research on Rosenbergiella and Rickettsia , and their functions are still unclear.In contrast, Bacillus , Pantoea , and K ocuria w er e positiv el y corr elated with disease se v erity.Pantoea and Bacillus as biocontr ol a gents pr evented plant diseases by antagonizing pathogens (Pusey 2002 ) and inducing plant defense responses (Johnson and Stockwell 1998, Wang et al. 2017, Fira et al. 2018 ).In addition, Pantoea has a unique biodegr adation ca pability and can degr ade herbicides and other toxic compounds (Giddens et al. 2002, Smits et al. 2010 ).The relative abundance of Wallemia and Cladosporium gradually decreases with the severity of powdery mildew, as they are pathogen of many plants (Abdelfattah et al. 2015 ).
Inter actions between micr oor ganisms ar e influenced by v arious biotic and abiotic factors (Faust and Raes 2012 ).Network analysis is a common method to study the correlation between different microbiomes (Li et al. 2017, Xu et al. 2020, Yuan et al. 2021 ).
Ther efor e, the co-occurr ence netw ork w as used to further explore the relationship among microbial populations and link microbial interactions with disease stress .T he phyla with the higher r elativ e abundance in the microbial netw orks w er e Pr oteobacteria, Actinobacteria, Basidiomycota, and Ascomycota.Proteobacteria and Ascomycota are the main microbial phyla, which play an important function in the co-occurrence network.They can r a pidl y decompose and absorb organic compounds and are more competitive than other fungi (Cui et al. 2019, Qiu et al. 2020 ).The network's topological features can indicate the complexity of microbial networks (Li et al. 2022b ).Compared with L2 and L3, the L1 bacterial network showed a higher number of edge connections ( 4515), an av er a ge degr ee (30.201), and a lo w er av er a ge path length (3.263).These topological features sho w ed that L1 has a more stable and complex ecological network, consistent with pr e vious findings (Zhang et al. 2018 ).The high negative correlation index in co-occurrence networks is associated with ecological imbalances and competition between micr oor ganisms (Gauthier et al. 2019 ), while the high positive correlation index indicates that there were less competitive interactions and more mutual benefits (Zhang et al. 2022 ).The number of edge connections, the proportion of positive correlations, and the av er a ge degree of fungal network showed first a downward and then incr eased tr end, while the bacterial netw ork sho w ed a do wnw ar d tr end fr om L2 to L3.This indicated that the se v erity of the disease increased and the fungal network tended to be stable, while the bacterial network tended to be unstable.Together, these studies suggest that disease se v erity has a greater impact on bacterial communities.
In conclusion, in this study, it was found that disease se v erity was an important factor in the changes of the phyllosphere micr obial comm unity.Ther e wer e significant differ ences in the structure of the phyllosphere microbial community under different disease grades.Bacillus , Pantoea , Kocuria , and Podosphaera were significantl y enric hed in the phyllospher e with the most se v er e disease.With the increase of the disease grade, the stability of the micr obial comm unity was br oken, and the bacterial comm unity tended to compete with each other rather than mutualism.These results provide a theoretical basis for exploring the changes of the phyllosphere microbial community of cucumber, screening and using beneficial micr oor ganisms to inhibit the occurrence of cucumber powdery mildew .T o further validate the correlation between disease se v erity and phyllospher e micr obiota, a combination of culturomics and in vitro/in planta antagonistic activity assays is needed in future research.

Figure 2 .
Figure 2. α div ersity anal ysis in differ ent degr ees of disease .T he α diversity indices including Shannon, Simpson indice , ACE and Chao in bacteria (A) and fungi (B) (Le v els of significance are indicated as follows: P < 0.05 marked as * , P < 0.01 marked as * * and P < 0.001 marked as * * * ).

Figure 4 .
Figure 4. Phyllospher e micr obial comm unities had unique or shar ed OTUs Venn dia gr ams at differ ent disease le v els.(A) Bacteria.(B) Fungi.

Figure 6 .
Figure 6.Analysis of differences in dominant genera under different disease levels.(A) Bacteria.(B) Fungi.

Figure 7 .
Figure 7. Discriminant analysis of species differences in phyllosphere community composition of cucumber powdery mildew under different degrees of incidence .T he taxon fr om the inside to the outside of the br anc hing dia gr am r epr esents phyla, class, order, family, and gen us respecti vely.The red, green, and blue nodes represent significantly different taxonomic levels in the three incidence levels (LDA threshold of 4).The y ello w node indicates that there is no significant difference in the different groups.(A, B) Bacteria (C, D) Fungi.