The cotton MYB108 forms a positive feedback regulation loop with CML11 and participates in the defense response against Verticillium dahliae infection

Highlight Cotton MYB108 interacts with CML11 and acts as a positive regulator in defense against V. dahliae infection.


Introduction
Increasing numbers of transcription factors (TFs), including members of the WRKY, NAC, bHLH, bZIP, ERF/AP2, and MYB families, have been reported to play crucial roles in plant immunity (Singh et al., 2002;Buscaill and Rivas, 2014). In the defense against pathogen attack, TFs modulate the transcription of defense-related genes, and they often execute such functions through interaction with other regulators (Tsuda and Somssich, 2015). MYB proteins represent one of the largest TF families and have a characteristic MYB domain consisting of 52 amino acid repeats that form helix-turn-helix structures required for binding to cis-elements in the promoters of target genes (Dubos et al., 2010). Based on the numbers of adjacent repeats, MYB proteins are divided into four classes: R1-MYB, R2R3-MYB, 3R-MYB, and 4R-MYB (Dubos et al., 2010). MYB proteins play important roles in plant development and responses, as shown for various species such as Arabidopsis (Arabidopsis thaliana), tobacco (Nicotiana tabacum), rice (Oryza sativa), and cotton (Gossypium hirsutum), and the molecular mechanisms by which these MYBs fulfill their functions are very well established (Lippold et al., 2009;Liu et al., 2009;Zhang et al., 2010;Walford et al., 2011;Yang et al., 2012;Lee et al., 2015). Several MYBs have been reported to function in defense against pathogens, including AtMYB30, AtBOS1 (AtMYB108), and TaPIMP1 (Vailleau et al., 2002;Mengiste et al., 2003;, yet the regulatory mechanisms and signaling processes mediated by MYB proteins in defense responses remain largely unknown. Ca 2+ is an important second messenger for the transduction of signals regulating plant development and the response to environmental cues (Hepler, 2005;Sarwat et al., 2013). Influx of Ca 2+ into the cytosol is an important early event in pathogen attack (Lecourieux et al., 2006). The major Ca 2+ sensors include calmodulin (CaM) and CaM-like proteins, which localize in various cellular compartments such as the cytoplasm, apoplast, nucleus, and peroxisome (Yang and Poovaiah, 2003). CaMs regulate a number of downstream targets involved in diverse plant processes (Bouché et al., 2005). After pathogen challenge, expression of multiple CaM genes is induced or suppressed as part of the plant defense response (Heo et al., 1999;Chiasson et al., 2005). Several studies reported that CaMs regulate gene expression by interacting with TFs such as members of the WRKY and CAMTA families, in plant innate immunity responses Galon et al., 2008). These studies have begun to reveal the molecular mechanisms by which Ca 2+ /CaM and TFs co-operate to modulate defense-related transcriptional responses.
Cotton Verticillium wilt is a highly destructive vascular disease that is mainly caused by the soil-borne fungus Verticillium dahliae, and this disease leads to severe loss of cotton yields worldwide and threatens most cotton-producing areas (Fradin and Thomma, 2006). Although long-term efforts have been made to produce wilt-resistant cotton cultivars by traditional breeding, very few varieties of upland cotton are resistant to Verticillium wilt (Cai et al., 2009). During the past years, progress has been made in exploring the molecular mechanism of the disease tolerance against V. dahliae invasion in cotton, with the ultimate aim of generating Verticillium wilt-resistant cultivars by molecular breeding. Accumulating evidence indicates that sets of V. dahliae-responsive genes, such as GhNDR1, GhNaD1, GhSSN, GbWRKY1, and GhMLP28 (Gao et al., 2011;Gaspar et al., 2014;Li et al., 2014;Sun et al., 2014;Yang et al., 2015), are functionally related to defense responses against V. dahliae infection in cotton.
In this study, we identified the V. dahliae-responsive gene GhMYB108 from upland cotton. Functional characterization indicates that it participates in the defense response through interaction with the CaM-like protein GhCML11. Moreover, the two proteins form a positive feedback loop to regulate the transcription of GhCML11. Another interesting finding of this study is that GhCML11 proteins localize in the apoplast as well as in the nucleus and cytoplasm. Apoplastic GhCML11 may be required for Ca 2+ influx in response to pathogen attack, and nuclear GhCML11 may act with GhMYB108 to activate the transcription of defense genes. Our results provide important insights into the significance of the synergetic interaction between a MYB transcription factor and Ca 2+ /CaM in plant immune responses.

Arabidopsis transformation
The ORF of GhMYB108 was cloned under control of the 35S promoter in the plant expression vector pBI121. The resulting plasmid pBI121-GhMYB108 was introduced into the Agrobacterium tumefaciens strain EHA105. Transformation of Arabidopsis plants was performed using the floral-dip method (Clough and Bent, 1998).

Pathogen cultivation and inoculation
The V. dahliae strain V991 originally isolated from an infected upland cotton, which is a strong pathogenic defoliating isolate (W.W. , was used as the pathogen. Fungal colonies were cultured on potato dextrose agar plates for 1 week at 26 °C. For V. dahliae infection, the roots of cotton seedlings grown under hydroponic conditions for 12 d were inoculated with a spore suspension (10 6 spores ml −1 ), and then harvested at the indicated time for RNA extraction. To infect VIGS (virus-induced gene silencing) cotton plants, the spore suspensions were stem-inoculated into cotton plants at a position 1 cm under the cotyledons with a syringe needle (Bolek et al., 2005), at a dose of 3 μl per plant. For Arabidopsis infection, roots of 4-week-old plants were incubated in spore suspensions for 3 min. Subsequently, plants were transplanted into fresh steamsterilized vermiculite. The disease index was calculated according to the following formula: disease index=[(∑disease grades×number of infected plants)/(total checked plants×4)]×100. Seedlings were classified into five grades (grade 0, 1, 2, 3, and 4) based on the disease severity after V. dahliae infection, as described by Wang et al. (2004).
Pseudomonas syringae pv. tomato strain DC3000 was grown in King's B medium at 28 °C. Overnight culture cells were resuspended in 10 mM MgCl 2 . The cell density was adjusted to 2 × 10 5 colonyforming units (cfu) ml −1 for inoculation, and the bacterial growth was detected 3 d after inoculation. Botrytis cinerea strain BO5-10 was grown on potato dextrose agar at 23 °C for 10-14 d. Spores were harvested and adjusted to a concentration of 10 5 spores ml −1 with distilled water. A 6 μl aliquot of spore suspension was dropped on Arabidopsis leaves and the lesion size was measured at 3 d after inoculation.
Hormone, CaCl 2 , and LaCl 3 treatments Cotton roots were treated with 0.1 mM salicylic acid, 0.15 mM jasmonic acid, 1 mM ethylene, and different concentration of CaCl 2 . Cotton roots were treated with 300 μM LaCl 3 before and after V. dahliae infection. Roots treated with sterile water were used as mock control.

RNA extraction and qRT-PCR analysis
Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. The quantitative real-time PCR (qRT-PCR) assay was conducted using the SYBR Green Real-Time PCR Master Mix (Toyobo, Japan) and the DNA Engine Opticon 2 Real-Time PCR Detection System (MJ Research). The cotton Histone3 gene or Arabidopsis EF-1α gene was used as the internal control. The expression levels of genes were calculated by using the 2 -ΔΔCT or 2 -ΔCT method, where CT is the cycle threshold. ΔCT=CT Target -CT Histone3/EF-1α . ΔΔCT=ΔCT sample -ΔCT control . The amplification efficiency of the samples was quantified and adjusted. All reactions were conducted in triplicate. The primers used in qRT-PCR are listed in Supplementary Table S1 at JXB online.

Electrophoretic mobility shift assay
The ORF of GhMYB108 was cloned into the pMAL-p2X vector to produce maltose-binding protein (MBP)-GhMYB108 fusion protein. The ORF of GhCML11 was fused to the pGEX6P-1 vector to produce glutathione S-transferase (GST)-GhCML11 fusion protein. Fusion proteins were expressed in Escherichia coli strain BL21 and purified. EMSA was performed using biotin-labeled probes and a Pierce LightShift Chemiluminescent EMSA kit (Thermo). The binding reaction was carried out in a 20 μl reaction mixture at room temperature for 30 min and then separated on a native 6% polyacrylamide gel in 0.5× Tris-borate/EDTA buffer. To test the effect of GhCML11 on the DNA binding activity of GhMYB108, the reaction mixtures were supplemented with CaCl 2 (10 μM) or EGTA (0.5 mM), or purified GST-GhCML11 fusion proteins. The labeled probes were detected based on the manufacturer's instructions.

Dual-luciferase reporter (DLR) assay
The TF activity of GhMYB108 was examined by DLR assay as described by Ohta et al. (2001). The reporter plasmid contains the Luc gene, controlled by the minimal TATA region of the 35S promoter with five GAL4-binding elements upstream. The Renilla luciferase gene driven by the 35S promoter was used as an internal control. To construct effector plasmids, the ORF of GhMYB108 was introduced into the pRT-BD vector to generate 35S-BD-GhMYB108. The Arabidopsis protoplasts were transfected with a mixture of 6 μg of effector plasmid, 6 μg of reporter plasmid, and 1 μg of internal control plasmid by polyethylene glycol (PEG) transformation. After 16 h under dark conditions, the luciferase assay was conducted using the DLR assay system (Promega) and a GloMax 20-20 luminometer (Bio-rad).

Subcellular localization
The ORF of GhMYB108 was fused to GFP (green fluorescent protein) and the ORF of GhCML11 was fused to mCherry under the control of the 35S promoter in the expression vector pPZP111 (Hajdukiewicz et al., 1994) to generate pPZP-GhMYB108-GFP and pPZP-GhCML11-mCherry constructs, respectively. Agrobacterium cells (strain GV3101) containing recombinant plasmids were infiltrated into N. benthamiana leaves. The infiltrated plants were incubated for 40 h at 23 °C under dark conditions. The cells expressing GFP proteins were stained with DAPI to indicate the nucleus. The signals were visualized with a confocal microscope (Leica TCS SP8, Germany).
For visualization of GhCML11 in onion epidermal cells, 2 μg of plasmid DNA was used to coat gold particles, and the plasmid harboring GFP alone was used as the control. The inner layer of the onion epidermis was bombarded by a gene gun (PDS-1000/He, Bio-Rad). For plasmolysis experiment, cells were treated with 20% sucrose. Signals were visualized by confocal microscopy.
Virus-induced gene silencing pTRV1 and pTRV2 (Liu et al., 2002) vectors were used for VIGS experiments. The gene-specific fragment for GhMYB108 ( Supplementary  Fig. S1) or GhCML11 was inserted into pTRV2. Agrobacterium cultures (OD 600 =1.5) harboring pTRV1 and pTRV2-GhMYB108 or pTRV2-GhCML11 were mixed at a 1:1 ratio and agroinoculated into cotton plants by vacuum infiltration as described by Qu et al. (2012). Alternatively, cotyledons of seedlings were transfected with the mixture using a needle-less syringe as described by Gao et al. (2011). GhCLA1 was used as the positive control (Gao et al., 2011).
Yeast two-hybrid (Y2H) assay Y2H assay was performed according to the instructions of the manufacturer of the Matchmaker Gold Yeast Two-Hybrid System (Clontech). The ORF of GhMYB108 was cloned into the BD vector pGBKT7 to construct BD-GhMYB108 as bait. The ORF of GhCML11 was inserted into the AD vector pGADT7 to produce the prey constructs. Yeast strain AH109 cells co-transformed with the bait and prey constructs were plated onto SD/-Leu/-Trp DO (DDO) medium. After growth at 30 °C for 72 h, the independent colony with the same size was transferred to SD/-Leu/-Trp/-Ade/-His DO (QDO) medium supplemented with 5 mM 3-aminotriazole (3-AT). The interactions were also visually detected using an X-gal filter assay.
Pull-down assay MBP-GhMYB108 and GST-GhCML11 fusion proteins were expressed and purified as described above. The in vitro protein-protein interaction assay was carried out according to the ProFound Pull-Down GST Protein Kit. The eluted proteins were separated on an SDS-PAGE gel and detected by western blot using anti-GST or anti-MBP antibodies (1:4000; Sungene Biotechnology).

Firefly luciferase complementation imaging (LCI) assay
The ORF of GhMYB108 was inserted into the pCAMBIA-NLuc vector. The ORF of GhCML11 was cloned into the pCAMBIA-CLuc vector. Equal amounts of Agrobacterium cultures containing CLuc and NLuc constructs were mixed, and then co-infiltrated into N. benthamiana leaves. The infiltrated leaves were analyzed for relative Luc activity 48 h after infiltration using a low-light cooled charge-coupled device camera (Night owl LB985, Germany). Quantitative analysis was performed using the IndiGo software (Berthold Technologies).

Transient expression assay
The transient expression assay was performed as described by Shang et al. (2010). The GhPR4, GhPR5, and GhPDF1.2 promoters were fused with the Luc reporter gene in the plant binary vector pGWB435 (Invitrogen) to generate the reporter constructs GhPR4 pro :Luc, GhPR5 pro :Luc, and GhPDF1.2 pro :Luc, respectively. The ORF of GhMYB108 was cloned into the pBI121 vector to generate the effector construct 35S pro :GhMYB108, and the ORF of GhCML11 was cloned into the pPZP111 vector to generate the effector construct 35S pro :GhCML11. The Agrobacterium strains containing different constructs were infiltrated into N. benthamiana leaves. The infiltrated plants were incubated for 48 h at 23 °C. Image capture used the low-light cooled charge-coupled device camera, and calculation of the relative Luc activity was performed using the IndiGo software.
Ca 2+ -dependent mobility shift assay The Ca 2+ -dependent mobility shift assay was conducted according to Garrigos et al. (1991). The GST-tag in GST-GhCML11 was cleaved using PreScission Protease (GE Healthcare), and GhCML11 proteins were electrophoresed on a 15% SDS-PAGE gel in the presence of either 5 mM CaCl 2 or 5 mM EGTA, and then visualized by Coomassie brilliant blue staining.
[Ca 2+ ] cyt staining Staining of cytosolic Ca 2+ was performed as described by Zhang et al. (1998). Cotton seedlings were grown under hydroponic conditions. Agrobacterium cultures harboring pTRV1 and pTRV2 (control), pTRV2-GhMYB108, or pTRV2-GhCML11 were mixed at a 1:1 ratio and agroinoculated into cotton plants by vacuum infiltration, and then the plants were transferred to steam-sterilized vermiculite. After 2 weeks, seedlings were gently uprooted and rinsed with sterile water, and then placed in sterile water for 24 h to adapt to hydroponic conditions. The roots were infected by spore suspensions (10 6 spores ml −1 ). The cotton roots were then loaded with Ca 2+ -sensitive fluorescent dye Fluo-4/AM (Invitrogen) at 4 °C for 2 h followed by 2 h at 25 °C in the dark. The fluorescence of the cotton root cells was visualized with a confocal microscopy. The fluorescence intensity of root cells was determined using Leica LAS AF Lite software.

Transcriptome analysis
For transcriptome analysis, total RNAs were extracted from control (TRV:00) and GhMYB108-silenced (TRV:GhMYB108) plants. The library construction and Illumina sequencing were conducted by BGI (http://www.genomics.cn/en/index). After eliminating the adaptors and low-quality sequences, the sequence reads were used for further analysis. Genes with differentially expressed transcripts [fold change ≥2 and false discovery rate (FDR) <0.001] in GhMYB108-silenced plants compared with control plants were identified. The accession number of the raw transcriptomic data is SRP067059.

Accession numbers
Sequence data for the genes described in this study can be found in the GenBank/EMBL database under the following accession numbers:

Expression of GhMYB108 responds to V. dahliae infection
In our ongoing studies of the defense-related genes acting in the response against cotton Verticillium wilt, we frequently noticed the presence of MBS (MYB-binding site) cis-elements in the promoters of the defense-responsive genes. To investigate the role of cotton MYB genes in defense against V. dahliae infection, we first conducted a database search and randomly selected six candidate MYB genes from different subfamilies to compare the pathogen-responsive expression of the MYB genes in upland cotton. Among these MYB genes, one gene (GhMYB108) showed strong induction of transcription upon pathogen inoculation ( Supplementary  Fig. S2). Since two members of this subfamily of MYB genes were shown to participate in defense against fungus infection in Arabidopsis or wheat (Mengiste et al., 2003;Z. Zhang et al., 2012), we focused our study on the functional mechanism of the GhMYB108 gene in protection against V. dahliae infection in cotton.
qRT-PCR analysis was performed to measure the time course of pathogen-responsive expression of GhMYB108. As shown in Fig. 1A, the expression of GhMYB108 increased in roots after V. dahliae infection and reached a maximal level at 6 h post-inoculation. Next, GhMYB108 expression was analyzed after treatment with the defense-related signaling molecules salicylic acid, jasmonic acid, and ethylene. The results showed that these three signaling molecules enhanced the accumulation of GhMYB108 transcripts to different extents (Fig. 1B), supporting the idea that GhMYB108 could be involved in defense against V. dahliae invasion in cotton plants. Expression of GhMYB108 was also examined in various organs of the cotton plant. GhMYB108 transcripts accumulated to a higher level in the root, which is the site of the V. dahliae invasion, as compared with the stem and leaf (Fig. 1C). The expression of GhMYB108 was the highest in flowers, implying that GhMYB108 may also function in flower development.

GhMYB108 is a functional transcription activation factor
EMSA was used to test the DNA-binding activity of GhMYB108. The results showed that GhMYB108 proteins and labeled probe could form a complex, and addition of non-labeled probes dramatically reduced the observed DNA binding activity, indicating that GhMYB108 could bind specifically to the MBS cis-element ( Fig. 2A).
The TF activity of GhMYB108 was examined using the DLR assay in Arabidopsis protoplasts. Isolation and transformation of Arabidopsis protoplasts were carried out as described by He et al. (2007). Compared with the negative control, the protoplasts harboring GhMYB108 showed significantly higher luciferase activity (Fig. 2B), indicating that GhMYB108 can activate the transcription of the Luc reporter gene in vivo.

The region containing the R2R3 domain is required for the nuclear localization of GhMYB108
To examine the nuclear distribution of GhMYB108, Agrobacterium cells transformed with the GhMYB108-GFP fusion and GFP control constructs were infiltrated into N. benthamiana leaves. Transiently expressed GhMYB108-GFP proteins were mainly localized in the nucleus, whereas GFP control was diffusely localized throughout the cytoplasm and nucleus (Fig. 2C).
As no nuclear localization signal was found in the GhMYB108 protein sequence, we wished to know which region of the protein might be responsible for its nuclear distribution. To this end, plasmids harboring cDNA fragments encoding either C-terminus-deleted GhMYB108-GFP (GhMYB108ΔC-GFP) or N-terminus-deleted GhMYB108-GFP (GhMYB108ΔN-GFP) were constructed, and Agrobacterium cells transformed with these constructs were separately infiltrated into N. benthamiana leaves. GhMYB108ΔC-GFP proteins were localized in the nucleus, while GhMYB108ΔN-GFP proteins were distributed in the cytoplasm without entry into the nucleus (Fig. 2C). These results indicate that the region containing the R2R3 domain of GhMYB108 is required for the nuclear localization of GhMYB108.

Silencing of GhMYB108 in cotton plants impairs the tolerance to V. dahliae
The VIGS strategy was employed to study the role of GhMYB108 in response to V. dahliae. GhCLA1 (produces an albino phenotype when silenced) was used as the positive control. At 14 d after agroinfiltration with the GhCLA1 construct, the cotton leaves displayed the expected photobleaching phenotype ( Supplementary Fig. S3), showing that the VIGS system worked well under our experimental conditions. Compared with the control, expression of GhMYB108 was obviously reduced in GhMYB108-silenced plants (Fig. 3A). To see if off-target silencing of other MYB genes was caused in the VIGS plants, we examined the transcription levels of six potential off-target genes, and the results showed that expression of these genes was not affected (Supplementary Figs S1, S4).
GhMYB108-silenced and control plants were challenged with V. dahliae. Before inoculation of the pathogen, no significant phenotype change was observed in VIGS plants as compared with the control plants (Fig. 3B, left panel). At 20 d after infection, GhMYB108-silenced plants began to show increased susceptibility to V. dahliae, with more severe wilting and yellowing symptoms than control plants. After a further 10 d, the disease symptoms of GhMYB108-silenced plants became more evident (Fig. 3B, right panel). The rate of diseased plants and the disease index were higher for the GhMYB108-silenced plants than for the control plants (Fig. 3C). In addition, the vascular tissue of GhMYB108silenced cotton plants turned brown 20 d after infection with V. dahliae, but that of the control plants was not chlorotic when tested at the same time (Fig. 3D). These observations indicated that silencing of GhMYB108 enhanced the susceptibility to V. dahliae. The fungal recovery assay from stem sections of inoculated cotton plants also confirmed the role of GhMYB108 in cotton defense against V. dahliae (Fig. 3E).
Overexpression of GhMYB108 enhances tolerance to V. dahliae in transgenic Arabidopsis plants A gain-of-function approach was also employed to study the function of GhMYB108 in the defense response. Due to technical difficulties and long duration of cotton transformation, we used the model plant Arabidopsis, which is also a host for V. dahliae. Three lines (7-4, 35-3, and 39-2) of transgenic plants with different expression levels of GhMYB108 were selected for further study (Fig. 4A). Arabidopsis plants at ~4 weeks old were inoculated with V. dahliae. Fifteen days after inoculation, the leaves of Arabidopsis began to show wilting and yellowing symptoms, and the plants grew stunted and short. Compared with the wild type, the transgenic plants showed much weaker symptoms at 22 d post-inoculation (Fig. 4B). The rate of diseased plants and disease index of the transgenic plants were significantly lower than those of the wild-type plants (Fig. 4C, D), showing that ectopic overexpression of GhMYB108 conferred increased disease tolerance to V. dahliae in Arabidopsis plants. To verify the observed phenotype further, the fungal biomass was measured by realtime PCR. Less fungal DNA was measured in transgenic plants than in wild-type plants (Fig. 4E), supporting the conclusion that GhMYB108-transgenic plants were more tolerant to V. dahliae infection.
In addition to V. dahliae, we also inoculated the GhMYB108-overexpressing Arabidopsis plants with two other pathogens, the bacterium Pst DC3000 and the fungus B. cinerea. The results showed that these plants were less susceptible to B. cinerea as compared with the wild type, but similar disease symptoms were found between the wild-type and transgenic plants infected with Pst DC3000, indicating that GhMYB108 overexpression rendered the transgenic Arabidopsis plants specifically more tolerant to the fungal pathogen ( Supplementary Fig. S5).

GhMYB108 interacts with GhCML11
The Y2H system was employed to identify protein(s) that may interact with GhMYB108. Screening the cDNA library of cotton roots infected by V. dahliae identified a cDNA that encodes a CaM-like protein (designated GhCML11). Direct Y2H assays confirmed the interaction between the two proteins (Fig. 5A). A pull-down assay was performed to verify further the interaction of the two proteins (Fig. 5B). Equal amounts of lysates containing GST-GhCML11 were incubated with immobilized MBP or MBP-GhMYB108 proteins. As expected, GhCML11 bound to GhMYB108, but not to the control MBP proteins. Subsequently, lysates containing MBP-GhMYB108 were incubated with immobilized GST or GST-GhCML11 proteins. GhMYB108 bound to GhCML11, but not to the control GST proteins. These results confirmed that GhMYB108 and GhCML11 could interact.
To verify the interaction of the two proteins in planta, an LCI assay (Chen et al., 2008) was conducted. As shown in Fig. 5C and D, strong Luc activity was detected in N. benthamiana leaves, but no significant Luc activity was detected in the negative controls.
Since GhCML11 interacts with GhMYB108, we investigated whether the subcellular localization of GhCML11 was similar with GhMYB108. Agrobacterium cells containing GhMYB108-GFP and GhCML11-mCherry were co-infiltrated into N. benthamiana leaves. Indeed, GhCML11 co-localized with GhMYB108 in the nucleus (Fig. 6A). In addition to the nucleus, we also noticed GhCML11 in the periphery of the N. benthamiana pavement cells (Fig. 6A). To see this subcellular localization of GhCML11 more clearly, we bombarded the GhCML11-GFP construct into onion epidermal cells and used plasmolysis to examine the plasma membrane and apoplast. GhCML11-GFP fluorescence was observed in both the nucleus and cytoplasm (Fig. 6B). Interestingly, we found that some GhCML11 proteins remained in the apoplast after plasmolysis. However, no free GFP signal was detected in the extracellular region after plasmolysis in the cells transformed with GFP alone. Thus, as reported for some CaMs in other plants (Cui et al., 2005;Wang et al., 2013), GhCML11 is probably also an apoplastic protein. As a protein that lacks a signal peptide but can be secreted from the cell independent of the endoplasmic reticulum/Golgi system can be defined as a non-classically secreted protein (Nickel and Rabouille, 2009;Drakakaki and Dandekar, 2013), GhCML11 belongs to such a protein group based on its sequence and localization. Indeed, GhCML11 is predicted to be a non-classically secreted protein by the online software http://www.cbs.dtu. dk/services/SecretomeP-1.0/.

GhCML11 promotes the transcriptional function of GhMYB108
Since GhMYB108 acts as a TF, the interaction between GhCML11 and GhMYB108 may have an effect on its activity. To test this possibility, EMSA was performed in the presence of GhCML11. As shown in Fig. 7A, GhMYB108 bound to the MBS cis-elements and formed a band representing the DNA-protein complex; when GhCML11 and Ca 2+ were present in the reaction simultaneously, a supershifted band with markedly enhanced intensity appeared. When GhCML11 was included in the reaction without addition of Ca 2+ , no effect was observed on the DNA binding activity of GhMYB108 either. The result indicated that the DNA binding activity of GhMYB108 was enhanced by its interaction with GhCML11 in a Ca 2+ -dependent manner in vitro. The EMSA was conducted to determine the Ca 2+ binding property of GhCML11. It is known that CaMs undergo conformational changes and exhibit an increase in their electrophoretic migration rates after binding Ca 2+ (Garrigos et al., 1991;Wang et al., 2015). As shown in Supplementary Fig.  S6, the mobility of GhCML11 was increased in the presence of Ca 2+ , demonstrating that GhCML11 is a functional Ca 2+binding protein.
We next conducted an in vivo test to see if the effect of GhCML11 on GhMYB108 DNA binding activity reflects its role in the TF activity of GhMYB108. As it was reported that a plant MYB could bind to the promoter sequence of PR5 (thaumatin-like protein) and regulate its transcription (Kenton et al., 2000;Z. Zhang et al., 2012), we performed a transient expression assay by using the promoter sequence of a cotton PR5 gene to drive the expression of the reporter gene with or without the presence of GhCML11 (Fig. 7B-D). First, the binding of GhMYB108 to the GhPR5 promoter was tested by EMSA. As shown in Supplementary  Fig. S7C, GhMYB108 bound to the GhPR5 promoter efficiently. The GhPR5 promoter was then fused to the Luc reporter gene (GhPR5 pro :Luc) and infiltrated into N. benthamiana leaves. Two days later, the expression of GhMYB108 and GhCML11 was confirmed by qRT-PCR (Fig. 7B) and Luc expression was examined. The results showed that the GhPR5 promoter drove Luc expression weakly on its own, but co-expression of GhPR5 Pro :Luc with GhMYB108 created an obvious increase in Luc activity, indicating that GhMYB108 activated the expression of Luc driven by the PR5 promoter. Luc activity was also enhanced when 35S:GhCML11 was co-transformed with GhPR5 Pro :Luc, probably caused by endogenous GhMYB108 homolog(s) in N. benthamiana, which might act co-operatively with GhCML11 and promote the GhPR5 promoter activity. Co-expression of the GhPR5 Pro :Luc reporter with GhMYB108 and GhCML11 led to much stronger Luc intensity than in the cells injected with the GhPR5 Pro :Luc reporter and GhMYB108 (Fig. 7C,  D), demonstrating that GhCML11 could promote the transcriptional activation activity of GhMYB108 in plant cells.

GhMYB108 regulates the transcription of GhCML11
In GhMYB108-silenced cotton plants, the expression of GhCML11 was also suppressed (Fig. 8A). This raised the possibility that GhMYB108 may affect the transcription of GhCML11. To test this, the promoter sequence of GhCML11 was isolated and subjected to EMSA analysis. As shown in Fig. 8B, GhMYB108 protein could form a complex with the promoter sequence of GhCML11, and non-labeled probes considerably decreased the binding of GhMYB108 proteins to the labeled probes, indicating that GhMYB108 could specifically bind to the tested sequence of the GhCML11 promoter. As it was observed that binding of GhMYB108 to MBS ciselements could be increased by GhCML11, we also tested if GhCML11 could enhance the binding of GhMYB108 to its own promoter. As shown in Fig. 8C, a supershifted signal with enhanced intensity appeared when GhCML11 proteins and Ca 2+ were added in the reaction; when an anti-GST antibody was added to the binding mixture, the band was shifted even further, indicating that GhCML11 was present in the supershifted band.
To investigate further whether GhMYB108 could activate the transcription of GhCML11 in plant cells, the promoter of GhCML11 was inserted into the vector pGWB435 with the Luc reporter gene. Co-expression of the GhCML11 promoter fused to the Luc reporter gene with 35S:GhMYB108 showed an obvious increase in Luc activity as compared with the control (Fig. 8D, E), indicating that GhMYB108 activated the expression of Luc driven by the GhCML11 promoter.

Calcium signaling is active in response to V. dahliae infection in cotton
Ca 2+ plays an important role in plant immune responses (Lecourieux et al., 2006). To gain more insight into the involvement of Ca 2+ -mediated signaling in the cotton defense response against V. dahliae, we performed a time course experiment to assess the change of [Ca 2+ ] cyt in response to V. dahliae infection. The cotton root cells were loaded with the Ca 2+ indicator Fluo-4/AM. The fluorescence intensity in the root cells of control plants increased significantly after inoculation with the pathogen, reaching a peak at 4 min and then decreasing quickly (Fig. 9). The result indicated that Ca 2+ influx into the cytosol occurred in response to V. dahliae infection.
The fluorescence intensity in the root cells of GhMYB108silenced and GhCML11-silenced plants was compared with that of the control plants. Before V. dahliae infection, the fluorescence intensity in GhMYB108-and GhCML11-silenced root cells was similar to that of control root cells, but it increased relatively less upon pathogen inoculation, indicating that the influx of [Ca 2+ ] cyt upon V. dahliae infection was influenced in these cells (Fig. 9). These results show that Ca 2+ influx into the cytosol occurs in response to V. dahliae invasion and the expression levels of GhCML11 and GhMYB108 had an impact on this process.

Transcriptomic analysis of genes affected in GhMYB108-silenced cotton plants
Comparative transcriptome analysis was employed to identify genes possibly regulated by GhMYB108. A total of 391 differentially expressed genes (fold change ≥2 and FDR <0.001) were identified, of which 181 genes were up-regulated and 210 genes were down-regulated (Supplementary Table  S2). Among the differentially expressed genes, a large number were involved in the biological processes of transcriptional regulation, signal transduction, developmental process, biosynthesis, and metabolism (Fig. 10A). In accordance with the above results on the relationship between GhMYB108 and Ca 2+ /GhCML11, several calcium signaling genes were downregulated in GhMYB108-silenced cotton plants (Fig. 10B).
Among the identified differentially expressed genes, 23 defense-related genes were inhibited in GhMYB108-silenced plants (Supplementary Table S3). The expression of these genes in GhMYB108-silenced cotton plants was then evaluated by qRT-PCR, which verified the down-regulation of these genes ( Supplementary Fig. S8). We also analyzed the expression of these genes in GhMYB108-overexpressing Arabidopsis ( Supplementary Figs S7B, S9), and observed that 14 out of 23 genes were up-regulated. These results indicate that there is a correlation between expression of GhMYB108 and these defense-related genes. As a previous study suggested that MYB may bind to the promoter sequences of some defenserelated genes, we analyzed the promoter sequences of these 14 genes for the presence of MBS cis-elements, and found that several of these genes harbored the MBS motifs in their promoters. Of these, we chose PDF1.2 (defensin-like gene), PR4, and PR5 (see above), which showed altered expression in both GhMYB108-overexpressing and GhMYB108-silenced plants ( Supplementary Fig. S7A, B), and tested the binding of GhMYB108 to their promoter sequences by EMSA ( Supplementary Fig. S7C, D). GhMYB108 could bind to the promoter fragments of these three genes. In addition, GhMYB108 activated expression of Luc driven by the PDF1.2, PR4, and PR5 promoters ( Supplementary Fig. S10). These results suggest that GhMYB108 may also be able to bind to the promoters of these genes and activate their transcription.

Discussion
GhMYB108 and GhCML11 form a positive feedback loop to regulate GhCML11 transcription MYB proteins are one of the largest families of TFs and have been implicated to function in plant defense by regulating defense-related transcriptional responses (Buscaill and Rivas, 2014). In this study, we show that GhMYB108 is a pathogen-responsive gene and that inhibition of its expression by VIGS resulted in increased disease susceptibility of cotton plants, indicating that GhMYB108 is involved in the defense response against V. dahalie in cotton. By Y2H assay, we found that GhMYB108 interacted with GhCML11 and the two proteins had a synergetic relationship. On the one hand, GhMYB108 functions as a TF to activate GhCML11 expression. On the other hand, GhCaM11 acts as a transcriptional activator to enhance the activity of GhMYB108. Thus, the two proteins form a positive feedback loop to enhance GhCML11 transcription. Previous work reported that the Arabidopsis MYB2 interacted with Glycine max CaMs for abiotic stress tolerance in Arabidopsis . Here, we provide a novel line of evidence showing the interaction and co-operative function of MYB and CaM, which contributed to the biotic stress tolerance in cotton.
The V. dahliae-induced redistribution of Ca 2+ may depend on the apoplastic GhCML11 Ca 2+ plays a critical role in plant innate immunity, and Ca 2+ influx is an early event in plant defense response against pathogen attack (Ma, 2011). In our study, we observed a clear Ca 2+ influx into the cytosol of cotton root cells upon V. dahliae infection. Through comparative transcriptome analysis, we found altered expression of a number of genes encoding calcium-binding proteins. These results show that Ca 2+ influx is tightly associated with the defense against V. dahliae infection in cotton.
The apoplast is a large Ca 2+ pool in plants, and some cytosolic CaMs can be secreted into the apoplast. These apoplastic CaMs promote the entry of Ca 2+ from the apoplast into the cytosol (Wang et al., 2009;Jiang et al., 2014). It was demonstrated that Ca 2+ redistribution across the plasma membrane is required for pollen tube growth (Wang et al., 2013). Using onion epidermis as an experimental system, we found that a portion of GhCML11 proteins is distributed in the apoplast. It will be interesting to investigate whether the apoplastic localization is involved in modulating the Ca 2+ influx, which contributes to subsequent defense responses in cotton cells. In support of this notion, we found that the pathogeninduced Ca 2+ influx was disturbed in root cells in GhCML11silenced cotton plants, which was coupled with the increased disease susceptibility. It is likely that when expression of GhCML11 was reduced, less GhCML11 protein was secreted into the apoplasts, resulting in reduced influx of Ca 2+ into the cytosol and, as a consequence, disturbed defense responses. This result provides novel hints on the function of apoplastic CaMs in the plant immune response. Further study is required to assess the links between dynamic redistribution of Ca 2+ and GhCML11 in defense response.
In GhMYB108-silenced cotton root cells, Ca 2+ influx was also altered upon pathogen attack (Fig. 9). This could be due to reduced expression of GhCML11, which was caused by silencing of GhMYB108. In this regard, GhMYB108 is also functionally linked to the Ca 2+ redistribution during responses to pathogen infection.

GhMYB108, calcium, and GhCML11 function interdependently to mediate defense responses
A mechanism by which TFs, CaM, and Ca 2+ function cooperatively to de-repress the expression of the immune system has been proposed based on studies on the Arabidopsis TF CAMTA3 . According to this model, plant TFs such as CAMTA3 bind to CaM and repress target gene expression prior to pathogen attack (Du et al., 2009;Nie et al., 2012). Upon pathogen infection, with the elevation of nuclear Ca 2+ that binds to the CaM-TF complex, the TF is dissociated from CaM and degraded by ubiquitin-mediated destruction and, as a consequence, expression of the immune system is de-repressed Fromm and Finkler, 2015). Here, we found that GhMYB108 is a transcriptional activator and GhCML11 enhances its activity in the presence of Ca 2+ . The expression of defense genes upon pathogen attack is by a mechanism of activation in this case, thus different from the mechanism involving CAMTA3.
On the other hand, we found that the pathogen-responsive Ca 2+ influx was interrupted upon silencing of GhCML11 or GhMYB108, indicating that these proteins in turn had an effect on Ca 2+ uptake which may affect the calcium-mediated signaling in the defense response. Taken together, our results demonstrate that GhMYB108, Ca 2+ , and GhCML11 can form a functional unit to regulate gene expression in cotton's response to V. dahliae invasion.
GhMYB108 and GhCML11 contribute to protection against V. dahliae invasion in cotton Verticillium wilt is the most serious disease affecting cotton production. In our study we observed that the expression of GhMYB108 is induced by pathogen attack and by defense-related signaling molecules; knock down of GhMYB108 expression by VIGS impaired the disease tolerance to V. dahliae in cotton plants, and ectopic overexpression of GhMYB108 enhanced disease tolerance to V. dahliae in transgenic Arabidopsis plants. Also, expression of a number of defense-related genes was inhibited when GhMYB108 was silenced, and this may be the cause of the impaired disease tolerance to V. dahliae invasion, by either a direct or an indirect mechanism. Based on our results, we speculate that GhMYB108 is a positive regulator of the cotton defense response to V. dahliae.
Some CaM genes have been reported to be involved in defense responses (Yamakawa et al., 2001;Park et al., 2004). For example, overexpression of GmCaM-4/-5 in wild-type Arabidopsis enhances disease resistance and induces PR gene expression (Park et al., 2004). In this study, we found that the expression of GhCML11 was highest in the root compared with the stem and leaves, and its expression was also induced by V. dahliae invasion (Supplementary Fig.  S12). Cotton plants with reduced expression of GhCML11 showed decreased disease tolerance compared with control plants ( Supplementary Fig. S13). These results indicate that GhCML11 is also an important contributor in defense against Verticillium wilt in cotton. It should be mentioned that in addition to the nucleus and apoplast, GhCML11 proteins are also present in the cytoplasm. It is known that CaM in the cytosol acts as a calcium sensor and transmits the Ca 2+ signal by interacting with target proteins (Yang and Poovaiah, 2003). Thus, apart from its roles in the nucleus and apoplast, GhCML11 may also participate in calcium signaling in the cytosol as do other CaMs.
Due to the difficulty in generating Verticillium-resistant cotton cultivars by traditional breeding, it is desirable to make breakthroughs in this field through genetic manipulation. Based on our data, we suggest that GhMYB108 and GhCML11 may be suitable candidate genes for molecular breeding of upland cotton cultivars with high tolerance to Verticillium wilt.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Multiple sequence alignments of GhMYB108 and potential off-target MYB genes. Figure S2. Expression pattern of GhMYB genes upon V. dahliae infection in cotton plant. Figure S3. Photobleaching phenotype in GhCLA1-silenced cotton plants. Fig.ure S4. qRT-PCR analysis of expression levels of six potential off-target MYB genes in control and GhMYB108silenced plants. Figure S5. Disease symptoms of GhMYB108overexpressing Arabidopsis plants inoculated with B. cinerea or Pst DC3000. Figure S6. The Ca 2+ -dependent mobility shift assay of GhCML11. Figure S7. Expression of PDF1.2, PR4, and PR5 genes in GhMYB108-silenced and GhMYB108-overexpressing plants, and binding of GhMYB108 to their promoter sequences. Figure S8. Verification of transcriptomic data in GhMYB108-silenced cotton plants. Figure S9. Expression levels of defense-related genes in GhMYB108 transgenic Arabidopsis plants. Figure S10. Transient expression analysis of GhCML11enhanced transcriptional activation activity of GhMYB108. Figure S11. Effects of Ca 2+ on the expression of GhMYB108 and GhCML11. Figure S12. Expression pattern of GhCML11 in cotton plants. Figure S13. Increased susceptibility of GhCML11-silenced cotton plants to V. dahliae. Table S1. Primers used in this study. Table S2. Genes differentially expressed in GhMYB108silenced plants determined by comparative transcript profiling analysis. Table S3. Defense-related genes down-regulated in GhMYB108-silenced cotton plants.