The Cotton Wall-Associated Kinase GhWAK7A Mediates Responses to Fungal Wilt Pathogens by Complexing with the Chitin Sensory Receptors

Identification of WAK Family Genes in Cotton

WAKs represent a class of RLKs characterized by an aminol (N)-terminal conserved signal peptide, an extracellular WAK galacturonan binding domain, an EGF-like domain, an EGF calcium binding (EGF-Ca²⁺) domain, a transmembrane domain, and a carboxyl (C)-terminal cytosolic Ser/Thr protein kinase domain (Verica and He, 2002). The Arabidopsis genome encodes five WAKs and 21 WAKL genes, which bear different extracellular domain configurations and show only 18% to 22% identities in their extracellular regions to WAKs (Verica et al., 2003). Our study here is mainly focused on the genome-wide identification of cotton WAK genes. Using five Arabidopsis AtWAK proteins as queries, we performed a genome-wide identification of cotton WAK genes in the genome database of representative cotton species including an allotetraploid cotton (G. hirsutum) and two diploid cotton species (G. arboreum and G. raimondii). The databases used in this study included the Nanjing Agricultural University (NAU) assembly for G. hirsutum (Zhang et al., 2015), the Joint Genome Institute (JGI) assembly for G. raimondii (Paterson et al., 2012), and the Cotton Research Institute (CRI) assembly for G. arboreum (Li et al., 2014).

To identify WAK genes in these cotton species, we implemented a pipeline wherein BLASTp (https://cottonfgd.org/blast/) search results were refined manually by evaluating protein domain predictions for subject amino acid sequences (Figure 1A). First, the full-length amino acid sequences of five AtWAKs were used as the queries for the BLASTp search against the three cotton genome databases G. hirsutum (AD₁)-NAU, G. raimondii (D₅)-JGI, and G. arboreum (A₂)-CRI, with a threshold of E-value < 1e-50. Second, proteins annotated as non-WAK proteins or WAKL proteins were removed from the list. Furthermore, we verified the presence of the conserved abovementioned various WAK domains of candidates. The domain prediction was performed using the protein sequence analysis tools from the programs SMART (http://smart.embl-heidelberg.de/) and Interpro (https://www.ebi.ac.uk/interpro/). Proteins, which lack essential domains of WAKs, were eliminated from the data set. After manually evaluating the protein predictions of each candidate, we identified 13 putative GhWAKs from the G. hirsutum (Supplemental Table 1), 16 GaWAKs from the G. arboreum (Supplemental Table 2), and 11 GrWAKs from the G. raimondii (Supplemental Table 3). G. arboreum has more WAK members than G. hirsutum, suggesting massive gene deletions during tetraploid G. hirsutum speciation (see below for examples). The phylogenic analysis of these candidates with AtWAKs and AtWAKLs revealed that they are phylogenetically closer to AtWAKs than AtWAKLs, indicating that the cotton proteins we identified here are WAKs, but not WAKLs (Supplemental Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Identification and Phylogenetic Analysis of WAK Family Proteins in Cotton.

(A) Genome-wide identification pipeline of WAKs in cotton genomes. Five Arabidopsis WAK genes annotated in TAIR10 were used as the query sequences for a BLASTp (cutoff: E-value < 1.0e-50) against three cotton genomes. The results from this BLAST search were refined using protein domain prediction tools (e.g., SMART, InterPro) to eliminate non-WAKs by manually screening proteins without any conserved WAK motifs.

(B) A phylogenetic view of GhWAK family proteins, gene structures, protein motifs, and clade partitions. The five AtWAK proteins were included to depict the evolutionary distance of the WAK proteins between G. hirsutum and A. thaliana. The phylogenetic tree was generated by the neighbor-joining method with 1,000 bootstrap replicates in the program MEGA X. The five AtWAKs are shown in blue. The NAU accession ID of each GhWAK is labeled adjacent to the GhWAK name curated in this study. Gene structures are further diagrammed next to each GhWAK using purple boxes to indicate exons and black shrunken lines with the same length to indicate introns. Schematics depicting WAK protein motifs were manually confirmed using the tool SMART (http://smart.embl-heidelberg.de/). A legend for the protein domain schematics is included at the bottom. Alignments corresponding to these analyses can be found in Supplemental Files 1 and 2.

(C) Genomic locations of GhWAK genes on their corresponding chromosomes. White bars represent chromosomes from the A-subgenome of G. hirsutum while gray bars represent the D-subgenome chromosomes. Chromosomes that were not found to contain WAK genes were not included in this figure.

(D) The evolutionary trajectory of WAK family in A. thaliana (At), G. arboreum (Ga), G. raimondii (Gr), and G. hirsutum (Gh). Numbers in circles represent the number of family members in each genome or subgenome, and the gene number for the most recent common ancestor of two diploid cotton species is shown in rectangles. Numbers on the line connecting each node with plus (+) and minus (−) signs indicate the numbers of duplicated and deleted genes, respectively.

Phylogenetic Analysis and Evolution of WAK Family Genes in Cotton

To elucidate the evolutionary relationship of WAK family proteins in Arabidopsis and cotton, we performed a phylogenetic analysis of cotton and Arabidopsis WAK proteins using the tool MEGA X (Kumar et al., 2018). No cotton WAKs from G. hirsutum, G. arboreum, and G. raimondii were found to interleave with the isolated clade formed by the five Arabidopsis AtWAKs (Supplemental Figure 2). This result remains consistent when the WAKs from G. hirsutum, G. arboreum, or G. raimondii database were analyzed separately (Figure 1B and Supplemental Figures 3A and 3B). The phylogeny using the kinase domains also indicated that WAKs from different cotton species form a separate clade from AtWAKs (Supplemental Figure 4). The five AtWAK proteins share 60% to 80% identities while the identities among AtWAK and GhWAK proteins are 30% to 50% (Supplemental Table 4). Collectively, this suggests that members of WAKs in cotton and Arabidopsis have likely evolved independently. Based on the phylogenetic relationships, we divided the cotton WAK members into five clades (I to V) with GhWAK1D, GhWAK2D, and GhWAK3D in clade I; GhWAK4D in clade II; GhWAK5A and GhWAK5D in clade III; GhWAK6D, GhWAK7A, GhWAK8A, and GhWAK8D in clade IV; and GhWAK9D, GhWAK10A, and GhWAK10D in clade V in G. hirsutum (Figure 1B; Supplemental Figure 2; Supplemental Table 5). Because this study is focused on GhWAKs, we named the individual members of WAKs from G. hirsutum. In consideration of the allotetraploid nature of G. hirsutum (AADD), the names of GhWAKs were followed with an uppercase “A” or “D” indicating the gene derived from the A- or D-subgenome (Figure 1B; Supplemental Table 6). For the homoeologous genes in A- and D-subgenomes, we gave the same name followed by an “A” or “D,” such as GhWAK10A and GhWAK10D.

Among 13 GhWAKs, nine are in the D-subgenome and four are in the A-subgenome (Figure 1C). The phylogenic analysis showed all the four GhWAK genes from the A-subgenome are orthologous to genes from the G. arboreum genome (Supplemental Figures 5A and 6) and three of them have corresponding D-subgenome homeologs (Figure 1B; Supplemental Figure 6). Interestingly, three GhWAKs (GhWAK7A, GhWAK8A, and GhWAK10A) on chromosome 2 of G. hirsutum A-subgenome are orthologous to WAKs on chromosome 3 of G. arboreum A-genome (Figure 1C; Supplemental Figure 5A), which may be due to the reciprocal translocation during the speciation of tetraploid G. hirsutum (Hu et al., 2019). Apparently, 12 out of 16 GhWAKs from G. arboreum were lost during G. hirsutum speciation (Supplemental Figures 5A and 6). Of the nine GhWAKs from the D-subgenome, seven genes have corresponding orthologs from G. raimondii (Supplemental Figures 5B and 6). Two other genes, GhWAK1D and GhWAK3D, are tandemly arranged and share high protein identity with GhWAK2D (92% and 87% respectively; Figure 1C; Supplemental Table 4), indicating they might have recently evolved as a result of tandem gene duplication in G. hirsutum. One pair of orthologs in G. raimondii and G. arboreum (Ga03G0824.1; Gorai.005G086200.1) do not have associated G. hirsutum counterparts (Supplemental Figure 6), suggesting that either the G. hirsutum ortholog was lost during the speciation, or the G. hirsutum ortholog was not able to be resolved with the genome assembly used in this study.

We attempted to further understand how WAK genes have evolved in cotton by comparing the number and relatedness of WAKs in A. thaliana, G. arboreum, G. raimondii, and G. hirsutum. Inferences made using the phylogenetic analysis between GhWAKs from the A-subgenome and GaWAKs, and GhWAKs from the D-subgenome and GrWAKs (Supplemental Figures 5A, 5B, and 5C, and 6), suggested that the trajectory of WAK gene number changes during the cotton genome evolution (Figure 1D). We propose that there might be seven WAK genes in the most recent common ancestor of two diploid cotton species (G. arboreum and G. raimondii) with massive gene duplications and losses during the evolution of G. arboreum, G. raimondii, and G. hirsutum (Figure 1D; Supplemental Figure 5C).

The chromosomal distribution of AtWAK genes indicates a tandemly arrayed gene (TAG) cluster on chromosome 1 (He et al., 1999). We investigated the physical positions of GhWAK genes in the G. hirsutum genome and found that GhWAKs distribute unevenly among six chromosomes with many members as TAG clusters (Figure 1C). GhWAK1D, GhWAK2D, GhWAK3D, and GhWAK4D present as a cluster of TAGs on the D10 chromosome. Notably, GhWAK1D, GhWAK2D, GhWAK3D, and GhWAK4D are clustered together on the phylogenetic tree with 83% to 93% identities between each other, suggesting these four paralogous GhWAKs likely were derived from one ancestor. Surprisingly, there are no corresponding homoeologous genes of these four GhWAKs on the A subgenome (Figure 1C). However, there are seven GaWAKs as a cluster of TAGs on chromosome 10 of G. arboreum (A-genome ancestor; Supplemental Figure 5A), suggesting that this cluster was lost during speciation of tetraploid G. hirsutum.

The other two separate clusters are composed of GhWAK6D, GhWAK8D, and GhWAK10D on the D02 chromosome and GhWAK7A, GhWAK8A, and GhWAK10A on the A02 chromosome. GhWAK8A-GhWAK8D and GhWAK10A-GhWAK10D are likely two pairs of positional homeologs with 89.46%% and 94.33% protein identities, respectively (Figure 1C). GhWAK6D and GhWAK7A are also possible homoeologous genes with relatively high divergence (63.11% protein identity). Thus, they were given different names.

Interestingly, the closest orthologs of GhWAK1D, GhWAK2D, GhWAK3D, and GhWAK4D in G. raimondii (D-genome ancestor) are Gorai.011G218500 and Gorai.011G218100, two TAGs on chromosome 11 (Supplemental Figure 5B), in accordance with the observation that the D-subgenome chromosome 10 of G. hirsutum has a high collinear relationship with the D-genome chromosome 11 of G. raimondii (Hu et al., 2019). The orthologs of GhWAK7A, GhWAK8A, and GhWAK10A in G. arboreum are Ga03G0825.1, Ga03G0822.1, and Ga03G0831.1, three TAGs on chromosome 3, respectively (Supplemental Figure 5A). Regarding the collinear relationship between G. hirsutum A-subgenome chromosome 2 and G. arboreum chromosomes 2 and 3 (Hu et al., 2019), we infer these three genes (GhWAK7A, GhWAK8A, and GhWAK10A) were translocated from chromosome A02 to A03 during the speciation of G. hirsutum. Together, these analyses indicate that WAKs are highly evolved genes with extensive gene duplications, losses, and translocations.

Next, we analyzed the gene and protein domain structures of GhWAKs and AtWAKs. Like their Arabidopsis orthologs, most GhWAKs contain three exons and two introns, except GhWAK6D, which contains four exons and three introns (Figure 1B). The six motifs, including the signal peptide, and GUB_WAK, EGF_like, EGF_Ca²⁺, transmembrane region, and the kinase domain, are conserved and common to both GhWAKs and AtWAKs (Figure 1B). Alignments of the amino acid frequencies of the EGF_like domains in GhWAKs revealed six highly conserved Cys residues (xCxxxxxxxxxxxCxxxSxCxxxxxxxGYxCKCxxGxxGxxGxPYxxxGCx, where x is any amino acid; Supplemental Figure 7A), which are implicated in the formation of disulfide bonds (Appella et al., 1988). A similar analysis of GhWAK EGF_Ca²⁺ domains revealed the expected conservation of five amino acid residues (DxD/NECxxxxxxxCxxxxxCxNxxGxYxCxCxxxxxGDxxxxxxGCx) required for calcium binding (Rao et al., 1995; Supplemental Figure 7B). The WAK domain is the most variable of all the domains contained within a WAK protein; however, our results indicated several conserved or semi-conserved residues near the N- and C-terminal ends (N terminus: xxxCxxxCGxxxIPVPFGxxxxxCxxxxxFxIxCxxxxxxPx; C terminus: xNKFxAxGCDT) of this domain in GhWAK proteins (Supplemental Figure 7C).

GhWAK7A Is Involved in Cotton Resistance Against Fungal Wilt Pathogens

WAK family proteins have been recognized to play an important role in disease resistance against fungal pathogens by several independent studies (Zuo et al., 2015; Shi et al., 2016; Hu et al., 2017; Saintenac et al., 2018; Yang et al., 2019). To unravel the potential involvement of GhWAKs in response to cotton fungal pathogens, the transcriptional changes of GhWAK genes upon fungal pathogen infection were investigated. We first analyzed the published RNA-seq data sets from the cotton G. hirsutum ‘YZ-1’, which has been widely used for cotton transformation and regeneration, infected with the Vd V991 isolate prevalently present in China (Zhang et al., 2018). This analysis revealed that eight out of 13 GhWAKs were upregulated at 6-, 12-, or 24-h post-inoculation (hpi), with GhWAK7A showing 12-fold induction at 6 hpi (Figure 2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

GhWAK7A Is Involved in Cotton Defense Responses against Fungal Wilt Diseases.

(A) Heat map of GhWAK family genes in RNA-seq data derived from cotton roots infected with the Vd V991 isolate (Zhang et al., 2018). The map was generated using the program Morpheus (https://software.broadinstitute.org/morpheus). Red color indicates relatively high expression and blue indicates relatively low expression. The corresponding value of fragments per kilobase of transcript per million is shown in the colored box.

(B) The expression patterns of GhWAK5A and GhWAK7A upon Vd infection. Two-week-old cotton seedlings were inoculated with the Vd ‘King’ isolate using a root-dipping method. The root samples were harvested at the indicated time-points for RT-qPCR. GhUBQ1 was used as an internal control. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using one-way ANOVA (*P < 0.05; **P < 0.01).

(C) Expression of GhWAK genes in control and silenced plants measured by RT-PCR. Cotyledons from 2-week-old cotton were syringe-infiltrated with Agrobacterium carrying a VIGS construct to silence GhWAK7A (VIGS-GhWAK7A), GhWAK5A (VIGS-GhWAK5A), or a GFP vector control (Ctrl). Three weeks after VIGS, the second true leaves were harvested for RNA isolation. GhActin9 was used as an internal control.

(D) The VIGS efficiency and specificity of GhWAK7A and GhWAK5A in cotton roots by RT-qPCR analysis. Three weeks after VIGS infiltration, cotton roots were harvested for RNA isolation and RT-qPCR analysis. GhUBQ1 was used as an internal control. Data are shown as mean ± se from three independent repeats. Asterisks indicate significant differences from Ctrl using a two-tailed Student’s t test (*P < 0.05; **P < 0.01).

(E) Disease symptoms of leaves and stems in Vd-infected VIGS-GhWAK genes and control plants. Three weeks after VIGS, plants were inoculated with Vd ‘King’ isolate at an inoculum of 1 × 10⁶ spores/mL using the root-dipping method. Images depicting the severity of plant wilting were taken at 14 dpi (left representation). Close-up views from the left representations are shown in the middle. Scale bar = 2 cm for plants. Vascular discoloration was observed in the stem 1.5 cm above the soil-line at 16 dpi under an optical microscope (right representation). Scale bar = 0.5 mm for stems.

(F) The DIs after Vd infection. The data are shown as the mean ± se from three independent repeats. An asterisk indicates a significant difference using one-way ANOVA (*P < 0.05).

(G) Disease symptoms of leaves and stems in Fov-infected VIGS-GhWAK genes and control plants. Three weeks after VIGS, plants were inoculated with Fov ‘CA10’ at 1 × 10⁷ spores/mL using the root-dipping method. Images depicting the severity of foliar and plant wilting symptoms were taken at 20 dpi (left representation). Close-up views from the left representations are shown in the middle. Scale bar = 2 cm for plants. Vascular discoloration was observed in the stem, 1.5 cm above the soil-line, at 22 dpi (right representation). Scale bar = 0.5 mm for stems.

(H) The DIs after Fov infection. The data are shown as the mean ± se from three independent repeats. An asterisk indicates a significant difference using one-way ANOVA (*P < 0.05).

The above experiments were repeated at least three times with similar results.

We then confirmed the transcriptional changes of GhWAK genes using reverse transcription- quantitative PCR (RT-qPCR) with gene-specific primers in G. hirsutum ‘CA4002’, a U.S. cultivar, infected by the Vd ‘King’ isolate, a prevalent field isolate in Texas, USA. In addition to GhWAK7A, GhWAK5A—which shares a relatively high protein identity (61.16%) with GhWAK7A, but displayed no induction within 24 hpi—was selected for comparison. Consistent with the RNA-seq result, GhWAK7A showed enhanced expression at 1-d post-inoculation (dpi), whereas GhWAK5A was only induced at 7 dpi (Figure 2B).

The transcriptional changes of GhWAK5A and GhWAK7A in response to Vd infections prompted us to examine their functional involvement in cotton defense against fungal diseases. We silenced the individual members of GhWAK5A and GhWAK7A by VIGS. Three weeks after Agrobacterium infiltration, RT-PCR analysis confirmed the reduced transcripts of GhWAK genes in the true leaves of cognate VIGS plants compared with the control vector-inoculated plants (VIGS-Ctrl; Figure 2C). In addition, RT-qPCR was performed to examine the transcript levels of individual GhWAK homologs in roots, where vascular pathogens first infect the plant. The results showed that the transcripts of GhWAK7A and GhWAK5A, but not any of the other eight GhWAK homologs, were reduced in the cognate VIGS plants compared with the control plants (Figure 2D), confirming the efficiency and specificity of VIGS silencing in cotton roots.

Three weeks after VIGS, the plants were inoculated with spores of the Vd ‘King’ isolate by a root-dipping method (Gao et al., 2013a). Compared with control, GhWAK7A-silenced plants displayed more wilting and yellow leaves, and deeper vascular discoloration in stem tissues (Figure 2E). We quantified the disease severity with disease index (DI) via five scales of foliar symptoms (Xu et al., 2012). Quantification of DI further supported this conclusion (Figure 2F).

We next tested whether GhWAK7A and GhWAK5A contribute to cotton defense against another fungal wilt pathogen, Fov, which causes Fusarium wilt. Interestingly, VIGS-GhWAK7A plants became more susceptible to the Fov CA10 isolate with more wilting leaves and plants and darker vascular discoloration in stem tissues compared with control vector-inoculated plants (Figure 2G), whereas VIGS-GhWAK5A plants were more resistant to the Fov CA10 infection with fewer wilting leaves and less vascular discoloration compared with control plants (Figure 2G). The DI also indicated that VIGS-GhWAK7A plants were more susceptible, while VIGS-GhWAK5A plants were more resistant to the Fov CA10 infection than the control plants (Figure 2H).

Several WAKs and WAKLs have been reported to link to plant responses against abiotic stresses (Hou et al., 2005; Xia et al., 2018; Wang et al., 2019). However, VIGS-GhWAK7A plants showed comparable responses to VIGS-Ctrl plants during water-deprivation and 200 mM of NaCl salt treatment (Supplemental Figure 8). Taken together, the data indicate that GhWAK7A may not be involved in cotton tolerance to drought or salt stresses but plays a positive role in cotton resistance to both Vd and Fov, implying a potentially conserved mechanism in cotton responses to different fungal pathogens.

GhWAK7A Contributes to Chitin-Induced Immune Responses in Cotton

Chitin, a conserved fungal cell wall component, triggers immune responses in plants (Shinya et al., 2015; Kawasaki et al., 2017). The contribution of GhWAK7A to defense against two different fungal pathogens prompted us to test the involvement of GhWAK7A in chitin-induced immune responses. RT-qPCR analysis indicated that GhWAK7A, but not GhWAK5A, was induced as early as 0.5 h upon chitin treatment (Figure 3A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The Involvement of GhWAK7A in the Chitin-Triggered but Not OG-Triggered Immune Responses in Cotton.

(A) The expression pattern of GhWAK5A and GhWAK7A upon chitin treatment. The leaf discs from cotton true leaves were soaked in distilled water overnight to eliminate wounding stress and treated with 100 μg/mL of chitin solution. Samples were harvested at the indicated time-points for RT-qPCR analysis. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using a two-tailed Student’s t test (**P < 0.01).

(B) Silencing GhWAK7A in cotton reduces chitin-induced MAPK activation. Cotton seedlings were silenced with GhWAK7A, GhWAK5A, or Ctrl by VIGS. The leaf discs from the second pair of true leaves were treated with 100 μg/mL of chitin at the indicated time-points. Phosphorylated GhMPK3 (pGhMPK3) and GhMPK6 (pGhMPK6) were detected by immunoblotting with an anti-phosphorylated extracellular-regulated protein kinase 1/2 (α-pERK1/2) antibody. The protein levels of GhMPK6 and GhMPK3 were detected with α-MPK6 and α-MPK3 antibodies, respectively. Protein loading is indicated by Ponceau S staining for Rubisco (RBC). The molecular mass (kDa) was labeled on the left of images (same for other immunoblots).

(C) The quantification of the relative intensities of immunoblot bands in (B). The values indicate the densitometry units of phosphorylated GhMPK3/6 (pGhMPK3/6) bands normalized to the protein inputs at 15 or 30 min after chitin stimulation. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using one-way ANOVA (**P < 0.01).

(D) Silencing GhWAK7A in cotton dampens chitin-induced ROS production. Three weeks after VIGS, the leaf discs from the second true leaves were soaked in distilled water overnight to eliminate wounding stress and treated with 100 μg/mL of chitin. ROS production was measured at the indicated time-points. The data are shown as the mean ± se (n ≥ 8).

(E) Silencing GhWAK7A reduces chitin-induced GhWRKY30 and GhMPK3 expression. Three weeks after VIGS, true leaves were treated with 100 μg/mL of chitin and collected at 0 and 30 min for RT-qPCR analysis. GhUBQ1 was used as an internal control. The data are shown as mean ± se from three independent repeats. An asterisk indicates a significant difference using a two-tailed Student’s t test (*P < 0.05).

(F) Silencing of GhWAK7A or GhWAK5A by VIGS does not alter OG-induced MAPK activation in cotton leaves. Three weeks after VIGS, leaf discs from the second true leaves were treated with 100 μg/mL OGs at the indicated time-points. MAPK activation assays were done as in (B).

(G) Silencing GhWAK7A in cotton does not affect OG-induced ROS production. After VIGS infiltration, the cotton leaf discs were treated with 100 μg/mL of OGs, and ROS production was measured as described in (D). The data are shown as the mean ± se (n ≥ 8).

(H) Silencing GhWAK7A in cotton does not affect OG-induced GhWRKY30 and GhMPK3 expression. Three weeks after VIGS, the cotton leaf discs were treated with 100 μg/mL of OGs and RT-qPCR analysis was performed as in (E). The data are shown as mean ± se from three independent repeats.

(I) Neither silencing GhWAK7A nor GhWAK5A alters FovCWE-induced MAPK activation in cotton. Three weeks after VIGS, the cotton leaf discs were treated with 1 mg/mL of FovCWE isolated from Fov ‘CA10’. The MAPK activation was detected as in (B).

The above experiments were repeated at least three times with similar results.

The activation of mitogen-activated protein kinases (MAPKs) is one of the early immune responses triggered by the perception of chitin and other MAMPs (Yu et al., 2017). Chitin treatment induced rapid and transient activation of GhMPK3 and GhMPK6 in cotton detected by immunoblots using an anti-phosphorylated extracellular-regulated protein kinase 1/2 antibody (Supplemental Figure 9A). The activation of GhMPK3 and GhMPK6 was confirmed using VIGS-GhMPK3 or VIGS-GhMPK6 plants in which the protein levels of GhMPK3 or GhMPK6 were reduced (Supplemental Figures 9A and 9B). Compared with control, the silencing of GhWAK7A, but not GhWAK5A, resulted in compromised GhMPK6 and GhMPK3 activation without affecting the GhMPK6 or GhMPK3 protein levels (Figures 3B and 3C).

The production of ROS is another well-characterized immune response triggered by chitin and other MAMPs (Yu et al., 2017). The chitin-induced ROS production was lower in GhWAK7A-silenced cotton plants than that in the control plants (Figure 3D). Chitin treatment induced expression of GhWRKY30 and GhMPK3 in cotton, and the expression levels of both GhWRKY30 and GhMPK3 after chitin treatment were reduced in GhWAK7A-silenced cotton plants relative to control plants (Figure 3E). Collectively, these data indicate that GhWAK7A is involved in the chitin-induced immune responses.

Earlier studies suggested that AtWAKs bind pectin in plant cell walls and are involved in cell expansion during development and the responses to pathogens and wounding (Brutus et al., 2010; Kohorn, 2016; Riese et al., 2018). To examine whether the sensing of cell-wall–derived pectin may account for the role of GhWAK7A and GhWAK5A in defense responses, we examined MAPK activation triggered by OGs, pectin fragments derived from the partial hydrolysis of plant cell wall pectin (Galletti et al., 2009). OGs were generated by the digestion of polygalacturonic acid (Na⁺ salt) with Aspergillus niger endopolygalacturonase. The RT-qPCR data showed that the induction of AtFRK1 by OGs was similar to that by flg22 in Arabidopsis (Supplemental Figure 10A). High-performance anion-exchange chromatography with pulsed amperometric detection indicated OG oligomers with a distribution of the degree of polymerization between 6 and 20 and an enrichment in the degree of polarization around 12 (Supplemental Figure 10B). This is consistent with the analysis using electrospray ionization mass spectrometry (Supplemental Figure 10C). Cotton leaves treated with OGs triggered rapid GhMPK6 and GhMPK3 activation with a peak at 15 min (Supplemental Figure 10D), and ROS production (Supplemental Figure 10E) and transcriptional upregulation of GhWRKY30 and GhMPK3 at 30 min in cotton (Supplemental Figure 10F).

However, compared with control plants, the silencing of GhWAK7A did not affect OG-induced MAPK activation (Figure 3F), ROS production (Figure 3G), and GhWRKY30 and GhMPK3 upregulation (Figure 3H). In addition, the fungal cell wall extracts from Fov ‘CA10’ (FovCWE) triggered immune responses, including MAPK activation. However, compared with control plants, the silencing of GhWAK7A or GhWAK5A did not affect FovCWE-induced MAPK activation (Figure 3I). This contrasts with the requirement of GhWAK7A in chitin-induced immune responses, suggesting that the function of GhWAK7A in plant defense might not be due to its role in response to cell-wall–derived components.

Identification and Functional Analysis of Chitin-Sensing Receptor Complex in Cotton

The Arabidopsis LYKs with the extracellular LysM motif compose a family of proteins with five members (Tanaka et al., 2013). AtCERK1 (AtLYK1), AtLYK4, and AtLYK5 were reported to function in chitin perception and signaling activation (Miya et al., 2007; Liu et al., 2012; Wan et al., 2012; Cao et al., 2014). Chitin-induced MAPK activation and ROS production were abolished in the atcerk1 mutant and atlyk4/5 double mutant and were reduced in the atlyk4 mutant compared with wild-type Columbia-0 (Col-0) plants (Supplemental Figures 11A and 11B; Cao et al., 2014; Xue et al., 2019). However, the atlyk5 (atlyk5-2) mutant exhibited a slight reduction in chitin-induced ROS production and no observable difference in chitin-induced MAPK activation compared with wild-type plants (Supplemental Figure 11). AtLYK4 and AtLYK5 function as the chitin receptor and AtCERK1 is likely a coreceptor or signaling partner of AtLYK4/AtLYK5 (Liu et al., 2012; Cao et al., 2014; Xue et al., 2019). We further tested whether the AtLYK4/5-AtCERK1 chitin sensory complex is involved in Arabidopsis response to Verticillium and Fusarium wilt pathogens. It has been reported that Vd could infect Arabidopsis (Veronese et al., 2003), and F. oxysporum (Fo) ‘5176’ is a pathogenic fungus on Arabidopsis (Thatcher et al., 2012). The atcerk1 and atlyk4/5 mutants were more susceptible to Vd and Fo ‘5176’ infections in terms of disease symptom development (Figure 4A) and DI (Figure 4B) compared to wild-type plants. Thus, Arabidopsis AtCERK1 and AtLYK4/5 play a positive role in resistance against Verticillium and Fusarium fungal pathogens.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Arabidopsis LYKs Mediate Resistance against Fungal Wilt Diseases and Phylogenetic Analysis of Cotton LYK Family.

(A) atcerk1 and atlyk4/5 are susceptible to Vd and F. oxysporum infections. Three-week-old Arabidopsis plants were inoculated with Vd ‘King’ isolate at 1 × 10⁶ spores/mL or F. oxysporum (Fo ‘5176’) at 1 × 10⁷ spores/mL via the root-dipping method. The images were taken 17 d after Vd infection (top) or Fo ‘5176’ infection (bottom). Scale bar = 2 cm.

(B) The DIs of Arabidopsis plants infected by Vd and Fo ‘5176’. The DIs were calculated at 22 dpi for Vd infection and 20 dpi for Fo ‘5176’ infection. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using one-way ANOVA (*P < 0.05; **P < 0.01).

The experiments in (A) and (B) were repeated at least three times with similar results.

(C) Phylogenetic analysis of GhLYK family proteins and corresponding protein motifs. Five Arabidopsis LYK proteins were included to represent the different clades of LYK family. The phylogenetic tree was constructed as described in Figure 1B. The GhLYK name generated in this study was labeled adjacent to the gene NAU accession ID. Gene structures are diagrammed next to each GhLYK using red boxes to indicate exons and black lines with the same length to indicate introns. A schematic diagram of LYK protein motifs was constructed based on the predictions in the tool SMART. A legend for the protein motif is included at the bottom. Alignments corresponding to these analyses can be found in Supplemental Files 3 and 4.

(D) The evolutionary trajectory of LYK family in A. thaliana (At), G. arboreum (Ga), G. raimondii (Gr), and G. hirsutum (Gh). Numbers in circles represent the number of family members in each genome or subgenome, and the gene number for the most recent common ancestor of two diploid cotton species is shown in the rectangle. Numbers on the line connecting each node with plus (+) and minus (−) signs indicate the numbers of duplicated and deleted genes, respectively.

To uncover the chitin perception and signaling mechanism in cotton, we first performed an in silico analysis to identify the LYK family members in cotton using a similar method as we did for cotton WAKs. The Arabidopsis five LYK full-length amino acid sequences were used as BLASTp queries against three cotton databases in CottonFGD (https://cottonfgd.org/) with a threshold E-value < 1e-50. The candidates were further manually screened to retain only those proteins that contain a predicted signal peptide, LysM motif (or the extracellular CERK1 structure; Protein Data Bank [PDB] ID is PDB:4EBZ), transmembrane domain, and Ser/Thr kinase domain. Seventeen GhLYKs from the G. hirsutum genome were identified (Supplemental Table 7). Eight GaLYKs and 10 GrLYKs were identified from the genomes of G. arboreum and G. raimondii, respectively (Supplemental Tables 8 and 9).

To understand the evolutionary history of LYK gene family members, the 35 cotton LYKs and five Arabidopsis LYKs were subjected to a phylogenetic analysis using the program MEGA X. Interestingly, cotton LYKs can be classified into five clades with each of five Arabidopsis LYK proteins in one clade (Figure 4C, Supplemental Figure 12; Supplemental Table 10). This is different from the phylogenetic relationship of WAKs in cotton and Arabidopsis, in which none of the Arabidopsis WAKs interleave with cotton WAKs (Supplemental Figure 2). This indicates that LYK members may have evolved before cotton and Arabidopsis speciation, whereas WAKs may have evolved independently after cotton and Arabidopsis speciation.

GhLYKs were named considering the orthology to AtLYKs and homoeology between A- and D-subgenomes (Figure 4C; Supplemental Table 11). The number (1 to 5) is to indicate the clade relativeness to the five AtLYKs (I to V). Lowercase letters (a to c) are added after the number to differentiate genes in the same clade. An uppercase “A” or “D” is followed to indicate from either the A- or D-subgenome. In G. hirsutum, four proteins, including GhCERK1aA, GhCERK1aD, GhCERK1bA, and GhCERK1bD, are putative AtCERK1 orthologs with 50% to 60% protein identity (Figure 4C; Supplemental Table 12). Five GhLYK3, including GhLYK3aA, GhLYK3aD, GhLYK3bA, GhLYK3bD, and GhLYK3cD, are putative orthologs of AtLYK3; and two GhLYK2—GhLYK2A and GhLYK2D—could be inferred as AtLYK2 orthologs. Four proteins with 30% to 50% identity to AtLYK4, including GhLYK4aA, GhLYK4bA, GhLYK4cA, and GhLYK4cD, and two proteins with 56% identity to AtLYK5, including GhLYK5A and GhLYK5D, are clustered (Figure 4C; Supplemental Table 12). This is consistent with the analysis in Arabidopsis that AtLYK4 and AtLYK5 have a closer evolutional relationship than the other three AtLYKs (Cao et al., 2014). The four GhCERK1 homoeologs showed an extracellular PDB:4EBZ structure (Figure 4C), which consists of three LysM motifs (Liu et al., 2012). An alignment of the GhLYK LysM motif sequences reveals four highly conserved residues (Y_D_N_P) as those in the AtLYK LysM motif (Supplemental Figures 13A and 13B). Gene structure analysis indicated that unlike WAKs, CERK1 and LYK3 in both Arabidopsis and cotton exhibit relatively complex exon–intron organization while LYK2, LYK4, and LYK5 have no or a few introns (Figure 4C). The phylogenetic analyses also indicate that CERK1 and LYK3 are more closely related, whereas LYK2, LYK4, and LYK5 are quite related in both Arabidopsis and cotton (Figure 4C).

The physical positions of GhLYKs distribute among eight chromosome pairs with nine genes in the A-subgenome and eight in the D-subgenome (Supplemental Figure 14). Unlike GhWAKs in which most of the genes are TAGs, only GhLYK4aA and GhLYK4bA are tandemly arrayed among all GhLYKs (Supplemental Figure 14). The phylogenetic analyses between GhLYKs from the A-subgenome and GaLYKs, and between GhLYKs from the D-subgenome and GrLYKs, suggest that seven out of nine GhLYK A-subgenome genes have the corresponding orthologs in G. arboreum, and all eight GhLYK D-subgenome genes have the corresponding orthologs in G. raimondii (Supplemental Figures 15A and 15B, and 16). The analyses revealed the trajectory of LYK gene number changes during the cotton genome evolution (Figure 4D). In addition, one gene (Ga05G2087.1) from G. arboreum, and two genes (Gorai.011G012100.1 and Gorai.006G245600.1) from G. raimondii, were lost during G. hirsutum speciation (Supplemental Figures 15A and 15B, and 16).

There are six orthologous LYK genes between G. arboreum and G. raimondii (Supplemental Figure 15C), which were considered as ancestor LYK genes in cotton (Figure 4D). Additionally, GhCERK1aA and GhLYK4bA are unique in G. hirsutum (Supplemental Figure 15A). GhLYK4bA clusters with GhLYK4aA on chromosome A11 and they share a high degree of sequence identity (85.9%; Supplemental Figure 14; Supplemental Table 12), suggesting that GhLYK4bA may have emerged via a local duplication event from GhLYK4aA, which is derived from Ga11G3609.1. Thus, LYKs and WAKs represent two unique family genes for studying genome evolution, gene birth, death, and divergence.

Silencing GhCERK1 or GhLYK5 Enhances Cotton Susceptibility to Fungal Wilt Diseases

The evolutionary conservation of cotton LYKs and Arabidopsis LYKs suggests that GhLYKs may exhibit a function that is similar to that found in their AtLYK orthologs. To investigate the potential contribution of GhLYKs to cotton fungal diseases, we analyzed the expression pattern of GhLYK genes in the RNA-seq data of cotton upon infection with the Vd V991 isolate (Zhang et al., 2018). All the GhLYK genes, except GhLYK3aD and GhLYK3aA, were upregulated to different extents upon Vd infection at different time-points (Figure 5A). We further deployed VIGS to silence GhLYK1 (GhCERK1 hereafter), GhLYK2, GhLYK3, GhLYK4, and GhLYK5. Given that there are several copies of each GhLYK in the cotton genome, we selected conserved regions of the given gene for VIGS and attempted to silence all the copies of the corresponding genes; for example, VIGS-GhCERK1 would silence GhCERK1aD, GhCERK1aA, GhCERK1bD, and GhCERK1bA. Three weeks after Agrobacterium-mediated VIGS infiltration, RT-PCR analysis was performed to confirm the silencing efficiency (Figure 5B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Silencing GhCERK1 or GhLYK5 Increases Cotton Susceptibility to Fungal Wilt Pathogens.

(A) Heat map of GhLYK family genes in RNA-seq data derived from cotton roots infected with the Vd V991 isolate (Zhang et al., 2018). The map was generated from the program Morpheus (https://software.broadinstitute.org/morpheus). Red color indicates the relatively high gene expression level, and blue indicates a low expression. The corresponding fragments per kilobase of transcript per million value are shown in the colored box.

(B) The expression of genes in control and silenced plants upon VIGS. Cotyledons from two-week-old cotton plants were syringe-infiltrated with Agrobacterium carrying a VIGS construct to silence GhCERK1 (VIGS-GhCERK1), GhLYK2 (VIGS-GhLYK2), GhLYK3 (VIGS-GhLYK3), GhLYK4 (VIGS-GhLYK4), GhLYK5 (VIGS-GhLYK5), or a GFP control (Ctrl). The second true leaves were harvested at 3 weeks after VIGS infiltration for RNA isolation for RT-PCR analysis. GhActin9 was used as an internal control.

(C) Cotton leaf wilting and stem vascular discoloration symptoms upon Vd infection. Three weeks after VIGS, plants were inoculated with the Vd ‘King’ isolate at an inoculum of 1 × 10⁶ spores/mL using a root-dipping method. Images depicting the severity of foliar wilt symptoms were taken at 13 dpi (left). Close-up views from the left are shown in the middle. Scale bar = 2 cm for plant images. Vascular discoloration was observed in the stem, 1.5 cm above the soil-line, at 19 dpi (right). Scale bar = 0.5 mm for stem images.

(D) Quantitative measurements of disease severity after Vd infection. The data are shown as the mean DI ± se from three independent repeats. An asterisk indicates a significant difference using one-way ANOVA (*P < 0.05).

(E) Cotton plant wilting and stem vascular discoloration upon Fov infection. After VIGS, plants were inoculated with Fov ‘CA10’ at 1 × 10⁷ spores/mL using the root-dipping method. Images depicting the severity of plant wilt symptoms were taken at 20 dpi (left). Close-up views from the left are shown in the middle. Scale bar = 2 cm for plant images. Vascular discoloration was observed in the stem, 1.5 cm above the soil line at 27 dpi (right). Scale bar = 0.5 mm for stem images.

(F) Quantitative measurements of disease severity after Fov infection. The data are shown as mean DI ± se from three independent repeats. An asterisk indicates a significant difference using one-way ANOVA (*P < 0.05).

The above experiments (B) to (F) were repeated at least three times with similar results.

Compared with control plants, VIGS-GhCERK1 plants showed markedly more severe disease symptoms that included leaf chlorosis and wilting, stem vascular discoloration (Figure 5C), and an elevated DI after Vd ‘King’ isolate infection (Figure 5D). Similar results were consistently obtained in all seven biological repeats (Supplemental Table 13). We also observed that VIGS-GhLYK2 plants were more susceptible than control plants in five out of seven repeats, and VIGS-GhLYK5 plants were much susceptible in four out of seven repeats relative to control plants (Supplemental Table 13). Similar to Vd, Fov CA10 infection caused more severe infection in VIGS-GhCERK1 or VIGS-GhLYK5 plants compared with control plants as shown with more wilting leaves and plants, darker stem discoloration, and elevated DI (Figures 5E and 5F). Among six repeats with Fov infections, VIGS-GhLYK5 plants showed susceptibility in five repeats and VIGS-GhCERK1 plants were susceptible in all repeats in terms of disease symptoms and indexes (Supplemental Table 14).

GhCERK1 and GhLYK5 Are Essential for Chitin-Triggered Immune Responses in Cotton

Given the result that GhCERK1 and GhLYK5 positively contribute to cotton defense against fungal wilt diseases, we further explored whether plants silenced for these genes showed altered responses to chitin treatment. Silencing of GhCERK1 (but not GhLYK2, GhLYK3, or GhLYK4) compromised chitin-induced MAPK activation compared with control plants (Figures 6A and 6B). In addition, silencing of GhLYK5 resulted in a slight, but reproducible, reduction of chitin-induced MAPK activation (Figures 6A and 6B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

GhCERK1 and GhLYK5 Are Essential for Chitin Perception in Cotton.

(A) Chitin-induced MAPK activation in GhCERK1-, GhLYK2-, GhLYK3-, GhLYK4-, GhLYK5-, and GhLYK4/5-silenced cotton plants. Three weeks after VIGS, the leaf discs collected from the second true leaves were treated with 100 μg/mL of chitin at the indicated time-points for immunoblot with an α-pERK1/2 antibody. Protein loading is indicated by Ponceau S staining for RBC.

(B) The quantification of the relative intensities of immunoblot bands in (A). The values indicate the densitometry units of phosphorylated MPK3/6 (pMPK3/6) bands normalized to the protein inputs at 15 or 30 min after chitin stimulation. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using one-way ANOVA (**P < 0.01).

(C) Chitin-induced ROS production in GhCERK1-, GhLYK5-, and GhLYK4/5-silenced cotton plants. Three weeks after VIGS, the leaf discs from the second true leaves were treated with 100 μg/mL of chitin and ROS production was detected at the indicated time-points. The data are shown as the mean ± se (n ≥ 8).

(D) Chitin-induced GhWRKY30 and GhMPK3 expression in GhCERK1, GhLYK5, and GhLYK4/5 silenced cotton plants. Three weeks after VIGS, the true leaves were treated with 100 μg/mL of chitin and collected at 0 and 30 min upon treatment for RT-qPCR. GhUBQ1 was used as an internal control. The data are shown as mean ± se from three independent repeats. Asterisks indicate a significant difference using a two-tailed Student’s t test (*P < 0.05; **P < 0.01).

(E) GhLYK5 complements atlyk4/5 double mutant on chitin-induced MAPK activation. GhLYK5-HA was expressed in wild-type (WT) and atlyk4/5 mutant protoplasts, respectively. Immunoblot was performed for MAPK activation with an α-pERK1/2 antibody (top), and GhLYK5-HA were detected with an α-HA antibody (middle). Protein loading is shown by Ponceau S for RBC (bottom).

(F) The chitin binding activity of GhCERK1 and GhLYK5 in cotton protoplasts. GhCERK1-HA, GhLYK5-HA, GhWAK7A-HA, or vector control (Ctrl) were expressed in cotton protoplasts. The proteins were immunoprecipitated (IP) with chitin magnetic beads and immunoblotted (IB) with an α-HA antibody (top for IP; middle for input). Protein loading is indicated by Ponceau S staining for RBC (bottom).

(G) and (H) Chitin-induced GhCERK1 (G) and GhLYK5 (H) phosphorylation in cotton protoplasts. Protoplasts expressing GhCERK1-FLAG or GhLYK5-FLAG were treated with 100 μg/mL of chitin for 0, 15, and 30 min. Total proteins were separated in an 8% SDS-PAGE gel supplemented with 50 μM of Phos-tag (NARD Institute).

(I) GhLYK5 associates with GhCERK1 upon chitin elicitation in cotton protoplasts. GhLYK5-HA and GhCERK1-FLAG were co-expressed in protoplasts, followed by 100 μg/mL of chitin treatment for 10 min. The association was detected by co-IP with α-FLAG agarose (IP: α-FLAG) and immunoblotting with α-HA (IB: α-HA) or α-FLAG (IB: α-FLAG) antibody (top two panels). The protein inputs are shown with immunoblots before immunoprecipitation (bottom two panels).

The above experiments were repeated at least three times with similar results.

Because AtLYK5 has an overlapping function with AtLYK4 in chitin perception and signaling (Cao et al., 2014), we designed a VIGS construct to silence both GhLYK4 and GhLYK5 simultaneously (VIGS-GhLYK4/5) and the VIGS efficiency was confirmed by RT-PCR (Supplemental Figure 17A). A significant reduction of chitin-induced MAPK activation was detected in VIGS-GhLYK4/5 plants, compared with VIGS-GhLYK4 or -GhLYK5, suggesting that cotton GhLYK4 and GhLYK5 exert redundant functions in chitin perception and signaling (Figures 6A and 6B).

Consistent with these observations, chitin-induced ROS production was largely compromised in the VIGS-GhCERK1 or VIGS-GhLYK4/5 plants and partially reduced in the VIGS-GhLYK5 plants (Figure 6C). Chitin-induced marker gene expression exhibited a similar trend in these GhLYK-silenced cotton plants. Compared with control plants, the chitin-induced expression of GhWRKY30 and GhMPK3 was significantly compromised in GhCERK1- or GhLYK4/5-silenced plants (Figure 6D). Expression of GhLYK5 in atlyk4/5 mutant protoplasts restored chitin-induced MAPK activation (Figure 6E), implying a conserved function of GhLYK5 and AtLYK4/5 mediating the chitin-induced MAPK activation.

Chitin Is Perceived by the GhCERK1 and GhLYK5 Complex in Cotton

Arabidopsis AtLYK5-AtCERK1 is a sensory complex for chitin perception (Cao et al., 2014). We examined whether GhCERK1 and GhLYK5 function similarly in chitin perception and signaling in cotton. GhCERK1 (GhCERK1bD), GhLYK5 (GhLYK5D), or GhWAK7A with a hemagglutinin (HA)-tag (GhCERK1-HA, GhLYK5-HA, or GhWAK7A-HA) was expressed in cotton protoplasts and chitin magnetic beads were used to pull down the chitin binding proteins from total proteins isolated from protoplasts. Chitin pulled down GhLYK5 and GhCERK1, rather than GhWAK7A (Figure 6F), suggesting that GhCERK1 and GhLYK5 bind to chitin in cotton. Although GhWAK7A is involved in chitin signaling, it may not bind to chitin directly. Chitin pulled down more GhLYK5 than GhCERK1 (Figure 6F, top), suggesting that GhLYK5 might have a higher affinity to chitin than GhCERK1, as was observed in Arabidopsis (Cao et al., 2014).

Perception of chitin induces the association of AtCERK1 and AtLYK5, and the phosphorylation of AtCERK1 and AtLYK5 in Arabidopsis (Cao et al., 2014; Erwig et al., 2017). Using a Phos-tag gel, we could detect a clear mobility shift of GhCERK1 and GhLYK5 after chitin treatment, indicating chitin-induced phosphorylation of GhCERK1 and GhLYK5 (Figures 6G and 6H). A co-immunoprecipitation (co-IP) assay using cotton protoplasts expressing GhLYK5-HA and GhCERK1-FLAG revealed that GhLYK5 associated with GhCERK1 upon chitin elicitation (Figure 6I). The chitin-induced GhCERK1 and GhLYK5 phosphorylation were evidenced as mobility shifts in a regular SDS-PAGE analysis (Figure 6I, bottom two panels).

GhWAK7A Complexes with GhCERK1 and GhLYK5

Given that GhWAK7A is involved in chitin-induced early responses and the important roles of GhCERK1 and GhLYK5 in cotton chitin signaling, we tested the potential relationship of GhWAK7A with GhCERK1 and GhLYK5. A co-IP assay in cotton protoplasts co-expressing GhCERK1-FLAG with GhWAK7A-HA indicated that GhWAK7A-HA coimmunoprecipitated with GhCERK1-FLAG before and after chitin elicitation (Figure 7A). The association of GhWAK7A-HA and GhLYK5-FLAG independent of chitin treatment was observed in a similar assay (Figure 7B) and confirmed using reciprocally switched epitope tags in co-IP assays with GhWAK5A and GhRLP2, a cotton LRR-RLP, as specificity controls (Supplemental Figure 17B). A split-luciferase assay, in which GhWAK7A and GhLYK5 were fused to the N-terminal and C-terminal half of the luciferase gene respectively, further demonstrated the association of GhWAK7A and GhLYK5 in vivo (Figure 7C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

GhWAK7A Interacts with GhCERK1 and GhLYK5.

(A) GhWAK7A associates with GhCERK1 in cotton protoplasts. GhWAK7-HA and GhCERK1-FLAG were co-expressed in protoplasts, followed by 100 μg/mL of chitin treatment for 10 min. co-IP was performed with α-FLAG agarose (IP: α-FLAG), and the proteins were analyzed by immunoblotting with an α-HA (IB: α-HA) or α-FLAG (IB: α-FLAG) antibody (top two panels). The input controls are shown in the bottom two panels.

(B) GhWAK7A associates with GhLYK5 independent of chitin elicitation in cotton protoplasts. GhWAK7A-HA and GhLYK5-FLAG were co-expressed in protoplasts followed by 100 μg/mL of chitin treatment for 10 min. The co-IP was performed as in (A).

(C) GhWAK7A associates with GhLYK5 with a split-luciferase assay. GhLYK5-Cluc-FLAG or GFP-Cluc-FLAG was co-expressed with GhWAK7A-Nluc-HA in protoplasts, and luciferase (LUC) activity was detected 12 h after transfection. The data are shown as mean ± se (n = 8). Asterisks indicate a significant difference using a two-tailed Student’s t test (**P < 0.01). Protein expression is shown on the bottom with immunoblots.

(D) FLIM-FRET analysis of GhLYK5 and GhWAK7A interaction in Arabidopsis protoplasts. The indicated proteins were transiently expressed in Arabidopsis protoplasts for 12 h before imaging analysis. Localization of the GhLYK5-GFP/AtBAK1-GFP and GhWAK7A-mCherry/AtBIR2-mCherry is shown with the first (green) and second column (red), respectively. The lifetime (τ) distribution (third column) and FRET efficiency (fourth column) are presented as pseudo-color images according to the scale. Paired GhLYK5-GFP and AtBIR2-mCherry was used as a negative control, and paired AtBAK1-GFP and AtBIR2-mCherry as a positive control (Halter et al., 2014). Scale bar = 10 μm.

(E) Quantification of lifetimes of GFP signals in (E). The GFP mean fluorescence lifetime (τ) values, ranging from 2.2 to 2.6 ns, were statistically analyzed and are shown as mean ± sd. “n” denotes the number of independent protoplasts. Asterisks indicate statistically significant differences (**P < 0.01), according to a two-tailed Student’s t test.

(F) GhWAK7A interacts with GhLYK5 in vitro. HA-tagged MBP-GhWAK7A^CD (MBP-GhWAK7A^CD-HA) was incubated with glutathione beads coupled with GST or GST-GhLYK5^CD. The beads were collected and washed for immunoblots with α-HA or α-GST antibodies (left and middle). The protein inputs were determined by Coomassie Brilliant Blue (CBB) staining (right).

(G) GhWAK7A interacts with GhCERK1 in vitro. HA-tagged MBP-fusion proteins (MBP-GhWAK7A^CD-HA or MBP-GhLYK5^CD-HA) were incubated with glutathione beads coupled with GST or GST-GhCERK1^CD. The pull-down assay was performed as in (F).

The above experiments were repeated at least three times with similar results.

Furthermore, fluorescence lifetime imaging (FLIM)-Förster resonance energy transfer (FRET) measurements revealed that GhLYK5-GFP proteins are in close proximity to GhWAK7A-mCherry, but not to AtBIR2-mCherry, when co-expressed in Arabidopsis protoplasts (Figures 7D and 7E). The FRET efficiency, calculated based on the GFP fluorescence lifetime of GhLYK5-GFP and GhWAK7A-mCherry (8.85% ± 2.30%), is similar to that of AtBAK1-GFP and AtBIR2-mCherry (9.68 ± 1.68%; Figures 7D and 7E).

To test whether GhWAK7A directly interacts with GhCERK1 or GhLYK5, we performed an in vitro pull-down assay. The glutathione s-transferase (GST)-fused cytosolic domain of GhLYK5 (GST-GhLYK5^CD), immobilized on glutathione-sepharose beads as the bait, specifically pulled down the GhWAK7A cytosolic domain fused to maltose binding protein (MBP) with an HA epitope at the C terminus (MBP-GhWAK7A^CD-HA; Figure 7F). Similarly, GST-GhCERK1^CD could pull down MBP-GhWAK7A^CD-HA proteins (Figure 7G), suggesting a direct interaction of the cytosolic domain of GhWAK7A with the cytosolic domains of GhLYK5 and GhCERK1. These findings imply that although GhWAK7A does not directly bind to chitin, it may modulate the cotton chitin perception complex of GhCERK1 and GhLYK5.

Interestingly, GST-GhCERK1^CD could not pull down MBP-GhLYK5^CD, supporting the specificity of an in vitro pull-down assay (Figure 7G). The data also indicate that the cytosolic domain of GhCERK1 and GhLYK5 cannot interact with each other directly in the absence of chitin treatment. This is in line with the notion that chitin binds and induces the complex formation of AtCERK1 and AtLYK5 via their extracellular domains in Arabidopsis (Cao et al., 2014; Xue et al., 2019).

GhWAK7A Modulates Chitin-Induced GhCERK1 and GhLYK5 Association and Phosphorylates GhLYK5

We further investigated the potential role of GhWAK7A in complexing with GhCERK1 and GhLYK5. When co-expressed with GhWAK7A in cotton protoplasts, the chitin-induced association of GhCERK1 with GhLYK5 was further enhanced (Figure 8A), suggesting that GhWAK7A likely promotes chitin-induced GhCERK1-GhLYK5 complex formation. Because GhWAK7A is a functional kinase (see below), we generated the GhWAK7A kinase-inactive mutant (GhWAK7A^K451M) by mutating the Lys residue at position 451 (K⁴⁵¹), the predicted ATP binding site to Met (M).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

The Phosphorylation of GhLYK5 by GhWAK7A Is Important for Chitin-Triggered Association of GhCERK1 and GhLYK5.

(A) GhWAK7A enhances the chitin-induced GhCERK1-GhLYK5 association. GhLYK5-HA, GhCERK1-FLAG, and GhWAK7A-GFP were co-expressed in protoplasts and treated with 100 μg/mL of chitin for 10 min. Proteins were immunoprecipitated with α-FLAG agarose (IP: α-FLAG) and immunoblotted with α-HA (IB: α-HA) and α-FLAG (IB: α-FLAG) antibodies (top two panels). The protein inputs are shown with immunoblots before immunoprecipitation (middle three panels). The quantification of the densitometry units of immunoprecipitated GhLYK5 normalized to the GhLYK5 inputs are shown on the bottom. The data are shown as mean ± se from three independent repeats. An asterisk indicates significant differences, according to one-way ANOVA (*P < 0.05).

(B) GhWAK7A kinase mutant (GhWAK7A^K451M) reduces the chitin-induced GhCERK1-GhLYK5 association. GhLYK5-HA, GhCERK1-FLAG, and GhWAK7A^K451M-GFP were co-expressed in protoplasts. GhWAK7A^K451M is a kinase-inactive mutant of GhWAK7A with the mutant of Lys (K) at position 451 to Met (M). The co-IP assay was performed and analyzed as in (A).

(C) GhWAK7A phosphorylates GhLYK5 in vitro. GST-GhLYK5^CD was incubated with MBP-GhWAK7A^CD for an in vitro kinase assay using [³²P]-γ-ATP. Phosphorylation was detected by autoradiography (top) and the protein loading is shown by Coomassie Brilliant Blue (CBB) staining (bottom).

(D) GhCERK1 cannot phosphorylate GhLYK5 in vitro. GST-GhLYK5^CD was incubated with MBP-GhCERK1^CD for an in vitro kinase assay using [³²P]-γ-ATP. The assay was done as in (C). Phosphorylation was detected by autoradiography (left) and the protein loading is shown by Coomassie Brilliant Blue (CBB) staining (right).

(E) GhWAK7A increases whereas GhWAK7A^K451M reduces GhLYK5 phosphorylation in vivo. GhLYK5-FLAG was co-expressed with GhWAK7A-HA or GhWAK7A^K451M-HA in protoplasts and treated with or without 100 μg/mL of chitin for 10 min. The GhLYK5 proteins were immunoprecipitated with α-FLAG agaroses (IP, α-FLAG). The immunoprecipitated and input samples were immunoblotted with indicated antibodies. α-pS/T, anti-phosphoSer/Thr antibody.

The above experiments were repeated at least three times with similar results.

Using this mutant, we performed a co-IP assay to examine the potential role of the GhWAK7A kinase activity in the chitin-induced formation of the GhCERK1-GhLYK5 complex. The co-expression of GhWAK7A^K451M dampened the chitin-induced association of GhCERK1-GhLYK5 (Figure 8B). Thus, GhWAK7A modulates chitin sensory complex GhLYK5-GhCERK1 in a kinase-dependent manner. We did not observe any notable effects of GhWAK7A or GhWAK7A^K451M on the chitin-induced mobility shift of GhCERK1 or GhLYK5 (Figures 8A and 8B).

We further examined the potential autophosphorylation and transphosphorylation among GhWAK7A, GhCERK1, and GhLYK5. Similar to their counterparts in Arabidopsis (Tanaka et al., 2013; Cao et al., 2014), GST-GhLYK5^CD had no detectable kinase activity, and GST-GhCERK1^CD had moderate kinase activity in an in vitro kinase assay (Figures 8C and 8D). In contrast, MBP-GhWAK7A^CD exhibited strong autophosphorylation activity, whereas MBP-GhWAK7A^K451M-CD showed undetectable phosphorylation activity (Figure 8C). Significantly, MBP-GhWAK7A^CD phosphorylated GST-GhLYK5^CD (Figure 8C). In addition, different from the phosphorylation of AtLYK5 by AtCERK1 in Arabidopsis (Erwig et al., 2017), neither MBP-GhCERK1^CD nor GST-GhCERK1^CD phosphorylated GST-GhLYK5^CD (Figure 8D; Supplemental Figure 17C).

To test the in vivo phosphorylation of GhLYK5 by GhWAK7A, GhLYK5-FLAG was co-expressed with vector control, GhWAK7A-HA, or GhWAK7A^K541M-HA in cotton protoplasts with or without chitin treatment, and the phosphorylated GhLYK5-FLAG proteins after immunoprecipitation with FLAG-agarose were detected by anti-phosphoSer/Thr antibody. The results indicate that GhWAK7A enhanced, whereas GhWAK7A^K451M reduced, the phosphorylation of GhLYK5 before and after chitin treatment (Figure 8E). Taken together, the data support the idea that GhWAK7A interacts with and phosphorylates chitin receptor GhLYK5.