CmMLO17 and its partner CmKIC potentially support Alternaria alternata growth in Chrysanthemum morifolium

The Mildew Resistance Locus O (MLO) gene family has been investigated in many species. However, there are few studies on chrysanthemum MLO genes. We report in this study that CmMLO17 in Chrysanthemum morifolium was upregulated after Alternaria alternata infection. Silencing of CmMLO17 by artificial microRNA resulted in reduced susceptibility of chrysanthemum to A. alternata infection. Genes in the abscisic acid (ABA) and Ca2+ signaling pathways were upregulated in the CmMLO17-silenced line R20 compared to the wild-type plants. We speculated that CmMLO17-silenced plants had a faster and stronger defense response that was mediated by the ABA and Ca2+ signaling pathways, resulting in reduced susceptibility of chrysanthemum to A. alternata infection. In addition, a candidate gene, CmKIC, that may interact with CmMLO17 was discovered by the yeast two-hybrid assay. The interaction between CmMLO17 and CmKIC was confirmed using the yeast two-hybrid assay and bimolecular fluorescence complementation (BiFC) analysis. CmMLO17 and CmKIC were both located on the plasma membrane, and CmKIC was also located on the nucleus. CmKIC overexpression increased the susceptibility of chrysanthemum to A. alternata, whereas CmKIC silencing resulted in reduced susceptibility. Therefore, CmMLO17 and CmKIC may work together in C. morifolium to support the growth of A. alternata. The results of this study will provide insight into the potential function of MLO and improve the understanding of plant defense responses to necrotrophic pathogens.


Introduction
Chrysanthemum (Chrysanthemum morifolium) is one of the most common cut flowers in the world and possesses ornamental and economic value. Leaf black spot disease, caused by the necrotrophic fungus Alternaria alternata, is one of the most serious diseases during chrysanthemum production. After A. alternata infection, the chrysanthemum leaves exhibit round black spots, which gradually expand to round, nearly round, or irregular spots, and dark mildew spots are formed under humid conditions. Recently, the incidence of black spot disease has increased, which has severely affected the output and ornamental quality of chrysanthemum and caused great losses in flower production. At present, black spot disease is mainly controlled by the spraying of fungicides; however, this often causes environmental pollution and increases the resistance of fungal pathogens to pesticides. Therefore, breeding resistant varieties are the most economical, safe, and effective way to control black spot disease in chrysanthemum.
One method of breeding disease-resistant chrysanthemum is based on the introduction of the R gene, which encodes a protein that recognizes the effectors of pathogens and triggers a series of defense responses 1 . Activation of the R gene produces reactive oxygen species (ROS) and a hypersensitive response at the sites of pathogenic infection. This is a programmed cell death response that can prevent further invasion by pathogens [2][3][4][5] . However, new mutants of pathogens can overcome R gene-mediated resistance, and the persistence of R genes is limited 6 . An alternative method is to modify susceptibility genes (S genes), the absence or silencing of which reduces plant susceptibility to pathogens 7 .
The Mildew Resistance Locus O (MLO) is a typical class of S genes that contributes to the infection and growth of powdery mildew (PM). First discovered in barley, loss-of-function mutants of the MLO gene have broad-spectrum resistance to almost all known pathovars of the PM pathogen Blumeria graminis f. sp. hordei 8 . Many studies have revealed that MLO genes are highly conserved throughout the plant kingdom, and their loss-of-function mutants make plants, such as Arabidopsis thaliana 9 , tomato 10 , pea 11 , pepper 12 , wheat 13 , apple 14 , and grapevine 15 , resistant to PM. MLO proteins are categorized into seven phylogenetic branches 16,17 , of which only two clades include the S genes, and clades IV and V contain all S genes of monocots and dicots, respectively 9,10,18-21 . Transcription levels of S genes are upregulated during the early stages of pathogen invasion; therefore, not all members belonging to these two clades are S genes. Based on this characteristic, candidate genes can be identified, as has been reported in barley 22 , tomato 10 , pepper 12 , grape 15,20,21 , rose 23 , and apple 14 .
Topological analysis has shown that MLO is a plasma membrane protein with seven transmembrane domains; this protein is concentrated at the plasma membrane and has an extracellular amino terminus and an intracellular carboxy terminus 24 . The cytoplasmic C-terminus of MLO proteins harbors a calmodulin-binding domain (CaMBD), which is highly conserved across the protein family 25,26 . To be fully active, barley MLO requires the binding of CaM and CaMBD under Ca 2+ -dependent conditions 26 .
Previous studies have shown that a possible function of MLO proteins is to negatively regulate vesicle-related and actin-dependent defense responses at sites where pathogens attempt to penetrate 18 . Broad-spectrum, non-racespecific resistance based on mlo is associated with the formation of callose-containing cell wall appositions called papillae and the secretion of antimicrobial compounds 16 . The papillae constitute a mechanical barrier that prevents penetration by pathogens. The formation of papillae and secretion of antimicrobial compounds depend on the delivery of substances via actin-dependent vesicle transport 27,28 . In addition to susceptibility/resistance to PM disease, MLO proteins are also involved in multiple physiological functions in different tissues, such as root thigmomorphogenesis 29 and pollen tube reception by the embryo sac in A. thaliana 30 . Despite great efforts to uncover the function of MLO proteins, their most critical biochemical function remains elusive.
Calcium ions (Ca 2+ ) are essential second messengers, and increases in Ca 2+ concentration in the cytosol are among the earliest signaling events that occur when plants are challenged by pathogens. In plant-pathogen interactions, the plant immune system is a two-tiered system, consisting of the pathogen-associated molecular pattern (PAMP)triggered immunity (PTI) and effector-triggered immunity (ETI) 31 . PTI activation can enhance the overall defense ability of plants and protect plants from pathogen attacks 32 , whereas ETI activation usually leads to local programmed cell death, also called the hypersensitivity response, to prevent invasion by pathogens. The Ca 2+ signature differs between these layers of immunity. For example, PTI activation is associated with Ca 2+ transients that return to basal levels within minutes 33 , whereas ETI involves an increase in cytosolic Ca 2+ that lasts for hours 34 . Abscisic acid (ABA), one of the major plant hormones, is also associated with plant defense responses against various pathogens 35,36 . Many studies have shown that ABA regulates defense responses by influencing the deposition of callose, the production of H 2 O 2 , and the expression of defense-related genes 37,38 . Moreover, ABA plays an important role in the regulation of stomatal closure, which acts as a physical barrier for defense against pathogen invasion 39 .
The aim of the present study was to study the function of CmMLO17 in resistance to A. alternata. CmMLO17 was upregulated after A. alternata infection. We cloned the full-length sequence of CmMLO17 and identified the interacting protein CmKIC, which was also involved in A. alternata resistance. RNA sequencing showed that CmMLO17-silenced plants had faster and stronger defense responses mediated by ABA and Ca 2+ signal transduction in the plant-pathogen interaction pathways, resulting in decreased susceptibility of chrysanthemum to A. alternata infection.

Isolation, sequence analyses, and phylogenetic analysis of CmMLO17
Chrysanthemum transcriptome libraries were screened using sequences from Arabidopsis AtMLO2, AtMLO6, and AtMLO12, and the chrysanthemum MLO gene was isolated. Sequence alignment of this gene confirmed that it had high homology with MLO genes of other plant species, and it was named CmMLO17. Specific primers were designed based on the transcriptome sequences and used to amplify the full-length cDNAs from leaves of wild-type (WT) chrysanthemum 'Jinba'. The open reading frame (ORF) of CmMLO17 (KJ560361) is 1659 base pairs (bp) in length and encodes 552 amino acids. CmMLO17 is a typical MLO protein, containing seven transmembrane domains and one CaMBD (Fig. 1a). Peptide domain I and a tetrapeptide motif (D/E-F-S/T-F) in region II associated with PM susceptibility 18 were detected in CmMLO17, albeit with a slight residue change, with the tetrapeptide motif substituted with N-F-S-F in CmMLO17 (Fig. 1a).
Phylogenetic analysis indicated that CmMLO17 is a member of clade V (Fig. 1b), which is mainly related to disease susceptibility [9][10][11]40,41 . Therefore, we speculate that CmMLO17 may be involved in the response of plants to pathogens.
The CmMLO17 gene is highly expressed in leaves and induced by A. alternata infection CmMLO17 is differentially expressed in the root, stem, leaf, and flower tissues of the WT. The transcriptional level of CmMLO17 was highest in the chrysanthemum leaves, high in stems, and lowest in the roots and flowers (Fig. 1c). At 1 h after A. alternata infection, the expression level of CmMLO17 in the infected plants was approximately two times higher than that in the control. From 12 h to 36 h after invasion by A. alternata, the transcription level of CmMLO17 remained significantly greater than that in the uninfected plants (Fig. 1d).
CmMLO17 silencing resulted in decreased susceptibility of chrysanthemum to A. alternata infection Plants with putative CmMLO17 artificial interference were identified using polymerase chain reaction (PCR) based on the HptII sequences of the transformed vector (Fig. 2a), and the abundance of CmMLO17 transcripts was analyzed by qRT-PCR (Fig. 2b). Three independent transgenic lines, namely, R20, R21, and R22, that accumulated much lower amounts of CmMLO17 transcripts  (Table S1). d Expression pattern of CmMLO17 in leaves of the WT after A. alternata infection revealed via qRT-PCR. CK: Uninfected WT were selected for evaluating resistance against A. alternata. After inoculation with A. alternata, the old leaves in the middle and lower parts of the WT became yellow, and some of the leaves showed black necrotic spots, whereas the leaves of the CmMLO17-silenced lines R20, R21, and R22 showed only yellowing at the edge of the leaves, and the degree of yellowing was lower than that in the WT plants (Fig. 2c). The average number of diseased leaves of the WT and CmMLO17-silenced lines inoculated with A. alternata was counted (Table S2). Only a small part of the leaves of the R20 line, which had the lowest abundance of the CmMLO17 transcript, appeared slightly yellow, and some of the plants were not diseased. Therefore, transgenic CmMLO17-silenced plants were less susceptible to A. alternata infection.
Differentially expressed genes (DEGs) of the calcium and ABA signaling pathways were upregulated in CmMLO17silenced plants To study the variations in gene expression in chrysanthemum leaves after inoculation with A. alternata, RNA sequencing was conducted on the WT and a highly resistant line (R20) of CmMLO17 with silencing at 0, 1, 6, 12, 24, and 36 h post inoculation (hpi). DEGs were identified when their expression levels changed twice or more and the p-adjusted P value (padj) was <0.05. The number of DEGs was higher in the comparison between the R20-6 hpi treatment and WT-6 hpi treatment (Fig. 3a). However, the number of DEGs was relatively low in the comparison between the R20-1 hpi treatment and WT-1 hpi treatment. Among the DEGs, many plant defenserelated genes were induced, such as calcium and ABA signaling pathway genes. Calcium signaling genes were upregulated during the early and middle stages of A. alternata infection, whereas ABA signaling genes showed great changes in transcription levels during the middle and late stages of inoculation (Fig. 3b, c). The calcium and ABA signaling pathways may have contributed to the defense responses against A. alternata. Therefore, CmMLO17-silenced plants had faster and stronger defense responses, mediated by ABA and Ca 2+ signaling pathways, than WT plants, resulting in enhanced resistance of chrysanthemum to A. alternata infection.

Identification of DEGs upregulated in CmMLO17-silenced plants
Weighted gene coexpression network analysis was performed to identify the genes related to phenotypes and to investigate the coexpression networks to elucidate the difference between WT and CmMLO17-silenced plants. Ultimately, ten gene coexpression modules, i.e., 'turquoise', 'green', 'blue', 'brown', 'black', 'red', 'magenta', 'pink', 'yellow', and 'gray', were discovered ( Fig. 4a, b). The genes from the red and black modules were highly correlated with the traits of each sample group in the CmMLO17silenced plants compared with the WT after A. alternata inoculation (Fig. 4b). In the red and black modules, a series of genes were upregulated in CmMLO17-silenced plants compared with WT plants (Fig. 4c, d).
Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation analyses and coexpression network analyses were performed to further explore the pathways in which the genes were involved. DEGs in the red module were annotated into multiple KEGG pathways, as shown in Tables S3 and S4. The regulatory network between the multiple signal transduction pathways was visualized using Cytoscape (Fig. 4e, f), which indicated the regulatory relationship among multiple signal transduction pathways. ABA-responsive element binding factor (ABF) homologs were upregulated and were involved in the ABA signal transduction pathway and the plant-pathogen interaction pathway, and 22 genes, including HSF90A and HSF90B homologs, were identified in the red module ( Fig. 4e). In the black module, ABF, calcium-dependent protein kinase, and calcium-binding protein were identified as being related to other genes ( Fig. 4f) that might play essential roles in the CmMLO17-mediated response to A. alternata invasion.

CmMLO17 interacts with CmKIC in yeast and in planta
The C-terminus of CmMLO17 was used as bait to screen the cDNA library of Alternaria-infected chrysanthemum using the yeast two-hybrid assay, and the candidate gene KIC was found. KIC is a Ca 2+ -binding protein with one EFhand motif that interacts with kinesin-like calmodulinbinding protein (KCBP), which is a microtubule motor protein and plays important roles in the regulation of microtubules 42 . Yeast two-hybrid analysis was performed to determine whether CmMLO17-C interacted with CmKIC. The combinations pGBKT7-53 + pGADT7-T and pGBKT7-Lam + pGADT7-T acted as positive and negative controls, respectively. All yeast colonies with different plasmid combinations grew well on synthetic dropout medium without leucine and tryptophan. The results showed that the yeast colonies transformed with the plasmid combination pGBKT7-CmMLO17-C + pGADT7-CmKIC grew well on synthetic dropout medium lacking adenine, histidine, leucine, and tryptophan (SD-AHLT) and turned blue on SD-AHLT plates supplemented with X-αgal, similar to the positive control (Fig. 5a). Yeast cells transformed with the plasmid combination pGBKT7-CmMLO17-C + pGADT7 could not grow on SD-AHLT plates, similar to the negative control, suggesting that the C-terminus of CmMLO17 had no autoactivity in yeast cells (Fig. 5a). Together, these results suggest that the Cterminus of CmMLO17 binds with CmKIC in yeast.
Bimolecular fluorescence complementation (BiFC) analysis was performed to further confirm the interaction of CmMLO17 and CmKIC in plant cells (Fig. 5b, c). The pSPYNE and pSPYCE vectors containing YFP N and YFP C , respectively, were used to construct CmMLO17-YFP N and CmKIC-YFP C , or vice versa. The fusion proteins were transformed into Agrobacterium tumefaciens and injected into Nicotiana benthamiana leaves, and the fluorescence signals were observed under a scanning confocal microscope. When CmMLO17-YFP N and CmKIC-YFP C were transiently coexpressed, or vice versa, yellow fluorescence was visualized at the plasma membrane, suggesting the interaction of CmMLO17 and CmKIC. No yellow fluorescence was detected upon coexpression of YFP N with YFP C in N. benthamiana leaves (Fig. 5c).
We also confirmed the interaction between CmMLO17 and CmKIC in onion epidermal cells. The results of the BiFC analysis revealed that the combinations CmMLO17-YFP N + CmKIC-YFP C and CmMLO17-YFP C + CmKIC-YFP N displayed yellow fluorescence signals at the   (Fig. 5b), indicating that CmMLO17 interacted with CmKIC at the plasma membrane.

Subcellular localization of CmMLO17 and CmKIC
To analyze the subcellular localization of CmMLO17 and CmKIC, the fusion plasmids 2×35S::CmMLO17-GFP, 2×35S::CmKIC-GFP, and a positive control, 2×35S::GFP, were separately bombarded into onion epidermal cells. In the cells transformed with 2×35S::GFP, the GFP signals were visualized in the plasma membrane, cytoplasm, and nucleus (Fig. 6a). GFP signals were detected on the plasma membrane of the cells expressing 2×35S::CmMLO17-GFP (Fig. 6a). GFP signals were detected in the plasma membrane and nucleus in cells transformed with 2×35S:: CmKIC-GFP (Fig. 6a), indicating that CmKIC localized to both the plasma membrane and nucleus. GFP-tagged CmMLO17 and CmKIC were separately transformed into Agrobacterium strain EHA105 and injected into 5-weekold leaves of N. benthamiana. The fluorescence signals were observed using a confocal laser scanning microscope (Fig. 6b), and the results were consistent with the subcellular localization in the onion epidermal cells.
Overexpression of CmKIC increased the sensitivity of chrysanthemum to A. alternata, and gene silencing resulted in decreased susceptibility Putative transgenic plants were identified by PCR using specific primers (Fig. 7a, b). The relative expression of CmKIC in transgenic chrysanthemum was analyzed using  (Fig. 7e, f). Therefore, overexpression of CmKIC increased the sensitivity of chrysanthemum to A. alternata, whereas silencing of CmKIC significantly reduced the susceptibility to A. alternata.

Discussion
MLO is a family of integral membrane proteins that are highly conserved across the plant kingdom. First discovered in barley, recessive mutants with the loss of function of WT MLO proteins show broad-spectrum resistance to PM fungi 8 . MLO genes have been identified in various monocots and eudicots, including barley, Arabidopsis, tomato, and pea [8][9][10]40 . However, not all MLO genes have been suggested to be related to plantpathogen interactions. Moreover, the function of MLO in chrysanthemum has not been previously reported. Cloning and determining the functions of the MLO gene in chrysanthemum can provide a new direction for breeding disease-resistant chrysanthemum.
The genomic sequence of C. morifolium has not been published; therefore, CmMLO17 was isolated based on chrysanthemum transcriptome databases. The expression patterns in leaves inoculated with A. alternata showed that CmMLO17 might play a role in the response to A. alternata. We chose the artificial microRNA interference approach to knock down CmMLO17. In transgenic plants, the expression of CmMLO17 was significantly reduced (Fig. 2b); however, it was not completely silenced. When inoculated with A. alternata, the CmMLO17-silenced plants showed less severe symptoms than the WT plants (Fig. 2c). Therefore, CmMLO17 may have a function similar to that of its orthologs in the MLO clade V subfamily in A. thaliana.
Although mlo alleles have durable broad-spectrum resistance to PM fungi, they may not function similarly in response to other pathogens. Barley mlo plants exhibited increased susceptibility to the hemibiotrophic fungus Magnaporthe grisea 43 and increased sensitivity to toxincontaining culture filtrates of Bipolaris sorokiniana compared to WT barley 44 . Furthermore, the susceptibility of the barley mlo genotype to Ramularia leaf spot caused by the necrotrophic ascomycete Ramularia collo-cygni was apparently affected by environmental conditions 45 . Under field conditions, there seemed to be no difference between mlo and WT barley infested with Fusarium spp. and R. collo-cygni 46 . The barley mlo genotype was less sensitive to Phytophthora palmivora but only in young leaf tissues 47 . Our study showed that CmMLO17-silenced chrysanthemum plants exhibited attenuation of the infection by the necrotrophic fungus A. alternata. These data indicate that mlo alleles have a trade-off between PM resistance and sensitivity to hemibiotrophic or necrotrophic pathogens, and whether this trade-off is due to pleiotropic phenotypes of mlo mutants or environmental conditions still needs further exploration.
When challenged with pathogens, plants produce a series of defense responses to pathogen invasion. When induced by pathogens, different Ca 2+ sensors, such as CaM, CaMlike proteins (CMLs), Ca 2+ -dependent protein kinases, and calcineurin B-like proteins, can recognize Ca 2+ signatures and convert the signals to a variety of plant immune responses, consisting of defense-related gene induction and the production of ROS and nitric oxide (NO) [48][49][50][51] . The generation of Ca 2+ , ROS, and NO not only induces cell wall reinforcement and a hypersensitive response 48 but also stimulates stomatal closure in guard cells 52,53 . Several studies have reported that stomata can act as barriers to invasion by pathogens, and PAMP-induced stomatal closure is one of the defense responses adopted by vascular plants to limit pathogenic infection 39 .
In addition to regulating the calcium signaling pathway, ABA is an important mediator of the regulation of stomatal opening and closing 54 . The ABA signaling pathway is mainly composed of three parts, namely, ABA receptor proteins (PYR/PYL/RCAR), type 2C protein phosphatases (PP2C), and SNF1-related protein kinase 2 (SnRK2) 55,56 . Complex formation by PYR/PYL/RCAR and PP2C results in inhibition of the dephosphorylation activity of PP2C, which activates SnRK2 and leads to closure of the stomatal pore. As a physical barrier, stomatal closure can reduce the chance of cell invasion by A. alternata hyphae. ABA also plays an important role in regulating stress responses 57 . ABA acts as a positive regulator of the defense against some necrotrophic pathogens, such as A. brassicicola and Plectosphaerella cucumerina 58 . In our study, the results of the DEG analysis showed that the genes involved in Ca 2+ and ABA signaling pathways in CmMLO17-silenced plants responded faster and more strongly to A. alternata infection than the WT plants, thus improving the resistance of transgenic chrysanthemum.
The C-terminus of the MLO protein contains a conserved CaMBD domain, and the CaM protein binds to the CaMBD of MLO in the presence of Ca 2+ to regulate MLO-dependent disease susceptibility 26 . The interaction between CaMBD at the C-terminus of CaMLO2 and CaCaM1 has been confirmed in pepper; this interaction led to the repression of cell death and defense responses triggered by Xanthomonas AvrBsT 41 , suggesting that the C-terminus of MLO may be the region where MLO interacts with other proteins. Therefore, to further explore the function of CmMLO17, the C-terminus of CmMLO17 was used to screen putative interacting proteins. Interestingly, we found the calcium-binding protein KIC.
KIC, which binds and negatively regulates KCBP, represents a novel Ca 2+ -binding protein 42 . KCBP is a member of the kinesin superfamily and plays important roles in the regulation of microtubule organization, microtubule dynamics, and directional vesicle transport 42,59 . Vesicle trafficking in plant-pathogen interactions plays a critical role in secretion-related defense responses, which are an important aspect of mlo-based resistance. Barley Ror2 and its ortholog PEN1 in Arabidopsis are required molecular components for full resistance of mlo 9,60 . Ror2 and PEN1 were found to encode plasma membrane-located syntaxin proteins that are members of the SNARE (soluble Nethylmaleimide-sensitive factor attachment protein receptor) superfamily and possibly participate in the secretion of antimicrobial compounds at the sites of attempted fungal infection 61 . We hypothesize that KIC, the interacting protein of CmMLO17, binds to KCBP to inhibit the regulation of KCBP in vesicle transport, which may affect secretion-related defense responses and support fungal growth. In this study, the phenotypes of CmKIC transgenic plants indirectly confirmed this hypothesis, but direct evidence is needed in further research. Moreover, we observed that the KIC protein is localized in both the plasma membrane and nucleus (Fig. 6), which may be related to its dynamics. In addition to interacting with CmMLO17 on the plasma membrane, CmKIC also binds to transcription factors in the nucleus to regulate defense responses (unpublished). In the future, analysis of biological processes regulated by CmKIC will reveal the biochemical function of MLO proteins.
In conclusion, we found that CmMLO17 interacted with CmKIC at the plasma membrane. Transgenic plants with silencing of CmMLO17 or CmKIC were less susceptible to A. alternata infection, indicating that CmMLO17 and CmKIC are involved in pathways that support fungal growth. RNA sequencing showed that ABA and Ca 2+ signaling pathway genes were altered in CmMLO17-silenced plants. Exploring the mechanism by which CmMLO17 and its partner CmKIC are involved in the response of chrysanthemum to A. alternata will provide insight into the functions of MLO and defense responses of plants to necrotrophic pathogens.

Plant materials and growth conditions
The WT chrysanthemum cultivar 'Jinba' used in this study was obtained from the Chrysanthemum Germplasm Resource Conservation Center, Nanjing Agricultural University, China. The plants were cultivated in a 1:2 (v/v) mixture of soil and vermiculite and were grown in a greenhouse held at 25°C/22°C in 70% humidity and a 14 h light/10 h dark cycle.

A. alternata inoculation and disease severity assessment
A. alternata was cultured on potato dextrose agar medium at 28°C for 4-5 days. Ten disks (~4 mm in diameter) obtained from the edges were ground in a tissue macerator, transferred to potato dextrose broth medium, and cultured in a shaker at a temperature of 28°C for 2 days. Mycelia of A. alternata were collected from 1 mL of mycelial culture medium and inoculated on the back of the third fully unfolded leaf with a brush. Each leaf had two inoculation points. Each line was inoculated with ten plants, and the experiment was repeated three times independently. The inoculated seedlings were placed in an incubator and cultured in darkness at 25°C under a relative humidity of 90% for 48 h. Then, the infected area of the inoculated leaves was measured at 2 days post inoculation using ImageJ software. Disease severity was evaluated by the average number of leaves showing disease symptoms and the infected area. Variance analysis was employed to determine the significance of test data according to Tukey's multiple range test (P < 0.05). SPSS v19.0 software was applied for statistical analyses.

Database searches, cloning, and sequence analysis of CmMLO17
Arabidopsis MLO protein sequences of clade V were downloaded from The Arabidopsis Information Resource database and acted as query sequences to identify CmMLO susceptibility genes in chrysanthemum. Homology searches of the MLO sequences were performed, to avoid repetition, with the BLASTX tool at the National Center for Biotechnology Information (NCBI, USA) (http://www.ncbi.nlm.nih.gov). Sequences were fully annotated by taking advantage of prediction programs, including SMART (http://smart. emblheidelberg.de/) 62 , Pfam 63 , NCBI-CDD 64 , and Inter-Proscan (http://www.ebi.ac.uk/interpro/scan.html) 65 . The full ORF sequence of CmMLO17 was amplified using a primer pair (CmMLO17-ORF-F/R) (Table S1), with cDNA from the WT used as a template. The AxyPrep DNA Gel Extraction Kit (Axygen, China) was used to purify the putative amplicons, and the purified fragments were ligated into the pMD19-T (Takara, Japan) vector for sequencing.
To study the evolutionary relationships, an unrooted neighbor-joining phylogenetic tree of CmMLO17 with its homologs in several other species was constructed using the MEGA7 software program 66 , and a bootstrap test with 1000 replicates was performed. The DNAMAN and ClustalW software programs were used to align the homology of MLO peptide sequences.

Yeast two-hybrid screening
For yeast two-hybrid screening, the C-terminal subclone of CmMLO17 (amino acids 405 to 552), including the CaMBD, was amplified by PCR from a full-length CmMLO17 clone in pMD19-T using the primers BD-CmMLO17-C-F and BD-CmMLO17-C-R (for specific primers, refer to Table S1). After purification, the amplified PCR product and pGBKT7 empty plasmid were digested by EcoR I and Sal I, respectively, and were then ligated. The pGBKT7-CmMLO17-C plasmid was verified by sequencing and acted as bait to screen the interaction library of chrysanthemum and Alternaria in the pGADT7 vector. Transformation of the Y2H yeast strain was performed according to the Matchmaker Gold Yeast Two-Hybrid Kit (Clontech, Mountain View, CA, USA). Putative positive clones were obtained and sequenced. The BLAST program was used to search the homologous genes of the obtained sequences, and SMART and InterProscan were applied to predict the domains of the sequences. The results showed that the KIC protein was the most frequently targeted. Interaction assays in yeast were performed using plasmids carrying pGADT7-CmKIC and pGBKT7-CmMLO17-C.

Bimolecular fluorescence complementation analysis
Genes encoding the tested proteins were cloned into the multiple cloning sites of different pSAT4A BiFC vectors 67 as follows. The ORFs of CmMLO17 and CmKIC were PCRamplified using the primers BiFC-MLO17-F1/BiFC-MLO17-R1 and BiFC-KIC-F1/BiFC-KIC-R1 (Table S1) with the restriction sites of EcoR I and Sma I. The amplicons were subsequently digested using EcoR I and Sma I, and ligated into pSAT4A-cEYFP/nEYFP-N1 vectors digested with the same enzymes, generating pSAT4A-nEYFP/cEYFP-CmMLO17 and pSAT4A-cEYFP/nEYFP-CmKIC, respectively. For the transformation experiments, the mixing of plasmids encoding cEYFP and nEYFP fusion proteins and microbombardment into onion epidermal cells were performed as previously described 67 .

Subcellular localization
The coding region of CmKIC was PCR-amplified, with pGADT7-CmKIC used as a template. According to the multiple cloning sites, the ORF sequences of CmMLO17 and CmKIC were ligated into the vector pENTR1A (Invitrogen, USA) to generate the vectors pENTR1A-CmMLO17 and pENTR1A-CmKIC, using the BamH I/Sal I and Not I restriction enzymes, respectively. LR Clonase II enzyme mix (Invitrogen, USA) was used to recombine pENTR1A-CmMLO17 or pENTR1A-CmKIC with pMDC43-GFP to construct GFP-CmMLO17 and GFP-CmKIC fusion vectors driven by the 2×35S promoter. A particle gun (PDS-1000; Bio-Rad, USA) was used to bombard the plasmids pMDC43-CmMLO17 and pMDC43-CmKIC into onion epidermal cells 69 . We also transferred the constructed fusion plasmids into the Agrobacterium strain EHA105 and then injected the leaves of N. benthamiana to observe the localization of CmMLO17 and CmKIC. The GFP signals in the transformed cells were detected under a laser scanning confocal microscope (Leica, Germany).

Constructs of amiRNAi
Sequences of CmMLO17 and CmKIC were submitted to the artificial microRNA design program WMD (http://wmd3.weigelworld.org/) to design the candidate microRNAs. The six specific sequences were designed (I to IV, A and B; listed in Table S1) for engineering the artificial microRNAs through site-directed mutagenesis, for which the plasmid pRS300 containing the precursor of miR319a was used as a template. The amiRNA was constructed using overlap PCR as previously reported 70 . The PCR amplicons were digested with Sal I and Not I and inserted into the pENTR1A vector digested with the same enzymes. For plant transformation, pENTR1A-amiCmMLO17 and pENTR1A-amiCmKIC were digested with Nsi I and introduced into the pMDC32 vector by LR recombination.

Generation of transgenic chrysanthemum plants and molecular analysis
To clarify the function of CmMLO17 and CmKIC, the overexpression vector of CmMLO17 and overexpression and silencing vectors of CmKIC were transferred into the EHA105 strain of A. tumefaciens using the freeze-thaw method. Chrysanthemum plants were transformed using Agrobacterium-mediated methods as previously described 71 , and hygromycin was used to select putative transgenic plants. DNA was extracted from the WT and transformants and used in PCR with specific primers (Table S1) to identify successfully transformed plants. The primers Hyg-F/R were used to detect the pMDC32-amiCmMLO17 vector in transformation lines. The primers KIC-ID-F1/R1 and KIC-ID-F2/R2 were used to amplify the pMDC43-CmKIC vector fragment. The primers KIC-ID-F3/R3 and KIC-ID-F4/R4 were used for amplification of the pMDC32-amiCmKIC vector. RNA was extracted and used in qPCR experiments with the primers CmMLO17-QRT-F/CmMLO17-QRT-R and KIC-QRT-F/KIC-QRT-R (Table S1) to measure the expression levels of CmMLO17 and CmKIC, respectively.

Gene expression analysis by qRT-PCR
Tissues of roots, stems, leaves, and flowers were sampled from the WT to elucidate the expression patterns of CmMLO17 in different tissues. To determine whether CmMLO17 was induced after A. alternata infection, leaves were sampled before infection and then 1, 12, 24, and 36 hpi. To analyze the relative expression of CmMLO17 and CmKIC, leaves from the same part of the WT and transgenic plants were harvested.
According to the manufacturer's instructions, RNA was extracted using RNAiso reagent (TaKaRa) and digested with RNase-free DNase I (TaKaRa) to remove genomic DNA. Reverse transcription was performed using a Reverse Transcription Kit (TaKaRa). The transcript levels of CmMLO17 and CmKIC were detected with qRT-PCR assays using SYBR Premix Ex Taq II from the same company as mentioned above following the manufacturer's instructions. The CmMLO17-QRT-F/CmMLO17-QRT-R and KIC-QRT-F/KIC-QRT-R primer pairs (Table S1) were designed using Primer Express 3.0.1 software, and the EF1α gene was used as a reference. A Roche Lightcycler 480 (Roche, Switzerland) was used to perform qRT-PCR. Three biological replicates were performed for each sample. The qRT-PCR data were analyzed by the 2 −ΔΔCT method as previously described 72 .

RNA sequencing and DEG identification analysis
After inoculation with A. alternata, the treated leaves were harvested at 0, 1, 6, 12, 24, and 36 hpi and used for RNA sequencing, with three replicates at each time point. Total RNA from the 36 samples was extracted as mentioned above. The libraries of all samples were constructed and sequenced at the Novogene Bioinformatics Institute (Tianjin, China). High-quality reads (clean reads) were extracted from the raw reads after eliminating lowquality reads and filtering adapter sequences. The clean reads were reassembled using Trinity software and matched using RSEM software to acquire the chrysanthemum unigenes of the transcriptome. DEGs were identified using DESeq. To define the DEGs, the criteria for absolute values of log 2 (induction of expression of transgenic plants compared with WT at defined time points) >1 and padj <0.05 were met. To study the function of the DEGs, various databases were used for gene annotation, such as the clusters of orthologous groups of proteins, gene ontology, KEGG, NCBI nonredundant protein sequence, NCBI nucleotide sequence, protein family, and Swiss-Prot databases. Weighted gene coexpression network analysis was performed with the R package for the identified modules 73,74 . Cytoscape software was used to visualize the coexpression networks 75 .