CALCIUM-DEPENDENT PROTEIN KINASE38 regulates flowering time and common cutworm resistance in soybean

Abstract

Photoperiod-sensitive plants such as soybean (Glycine max) often face threats from herbivorous insects throughout their whole growth period and especially during flowering; however, little is known about the relationship between plant flowering and insect resistance. Here, we used gene editing, multiple omics, genetic diversity and evolutionary analyses to confirm that the calcium-dependent protein kinase GmCDPK38 plays a dual role in coordinating flowering time regulation and insect resistance of soybean. Haplotype 2 (Hap2)-containing soybeans flowered later and were more resistant to the common cutworm (Spodoptera litura Fabricius) than those of Hap3. gmcdpk38 mutants with Hap3 knocked out exhibited similar flowering and resistance phenotypes as Hap2. Knocking out GmCDPK38 altered numerous flowering- and resistance-related phosphorylated proteins, genes, and metabolites. For example, the S-adenosylmethionine synthase GmSAMS1 was post-translationally upregulated in the gmcdpk38 mutants. GmCDPK38 has abundant genetic diversity in wild soybeans and was likely selected during soybean domestication. We found that Hap2 was mostly distributed at low latitudes and had a higher frequency in cultivars than in wild soybeans, while Hap3 was widely selected at high latitudes. Overall, our results elucidated that the two distinct traits (flowering time and insect resistance) are mediated by GmCDPK38.

A calcium-dependent protein kinase positively regulates flowering time and negatively regulates resistance to the common cutworm in soybean and was artificially selected during domestication.

Introduction

Millions of years of selection pressure from insects has led to the evolution of complex plant defenses (Erb and Reymond, 2019). Plants show strong spatiotemporal variation in the expression of defense metabolites, resulting in changes in the defense capabilities of plants by orders of magnitude throughout development (Barton and Boege, 2017). Particularly in annual plants that reproduce only once per year, reproduction often has a higher resource priority than defense. When they reach the reproductive stage, flowers and fruits may demand resources for production and defense, while the defense-related demands of vegetative parts decrease (Boege and Marquis, 2005). For example, in Arabidopsis (Arabidopsis thaliana), the expression levels of immunity genes correlate positively with duration of the vegetative state, and winter and summer annuals differ in their response to biotic and abiotic stresses (Davila Olivas et al., 2017; Glander et al., 2018), which illustrates that flowering strategies are linked to plant defense strategies.

To date, great progress has been made in determining the molecular mechanisms of plant flowering and insect resistance. Five major flowering pathways have been proposed in Arabidopsis. The photoperiod and vernalization pathways monitor seasonal changes in day length and temperature, and the autonomous, gibberellic acid (GA), and age pathways integrate signals of the plant development state (Simpson and Dean, 2002; Boss et al., 2004; Bernier and Périlleux, 2005; Baurle and Dean, 2006). These five pathways form a complex flowering regulatory network including many key genes, such as GIGANTEA (GI), CONSTANS (CO), FLOWER LOCUS T (FT), and APETALA1 (AP1) (Yasushi and Detlef, 2007; Turck and Fornara, 2008; Adrian et al., 2009). Multiple plant insect resistance-related genes have been identified in the jasmonic acid (JA) signaling pathway, including allene oxide synthase AtAOS, F-box gene AtCOI1, allene oxide cyclase NtAOC and VQ motif-containing gene GmVQ58 (Bodenhausen and Reymond, 2007; Goodspeed et al., 2012; Fragoso et al., 2014; Li et al., 2020), and among secondary metabolite biosynthesis-related genes, including myrosinases AtTGG1/2, terpene synthase AtTPS08, putrescine N-methyl transferase NtPMT, and S-adenosylmethionine synthase GmSAMS1 (Steppuhn et al., 2004; Barth and Jander, 2006; Vaughan et al., 2013; Fan et al., 2018). However, little is known about the relationship between plant flowering and insect resistance.

Among the many transduced signals, the calcium ion (Ca²⁺) plays a key role in the circadian regulation of photoperiod-controlled flowering in morning glory (Dodd et al., 2010) and is thus characterized as a part of the circadian clock output signal (Yakir et al., 2007). Calcium-dependent protein kinase (CDPK) is one of the main receptors in the Ca²⁺ signaling pathway and transduces the signal by phosphorylation (Harmon et al., 2000). In Arabidopsis, AtCPK6 and AtCPK33 are responsible for the phosphorylation of the basic region/leucine-zipper transcription factor AtFD, which is essential for the formation of FT–FD florigen complexes. Loss of function of AtCPK33 results in a late flowering phenotype (Kawamoto et al., 2015a, 2015b). In rice (Oryza sativa), mutation at a potential phosphorylation site abolishes the ability of OsFD1 to interact with the rice FT homolog Hd3a through 14-3-3 proteins (Taoka et al., 2011). In morning glory (Pharbitis nil), PnCDPK1 was reported to participate in evocation and flower morphogenesis. The transcriptional level and enzyme activity of PnCDPK1 increase transiently and specifically after the transformation of leaf buds into flower buds (Jaworski et al., 2012). Ca²⁺ is also a ubiquitous second messenger that is involved in early defense signaling in plants (Yan et al., 2018). Arabidopsis CPK3 and CPK13 activate the wound- and herbivore-induced network by the accumulation of plant defensin PDF1.2 (Kanchiswamy et al., 2010). In native tobacco (Nicotiana attenuata), silencing of NaCDPK4 and NaCDPK5 strongly upregulates JA accumulation and increases plant resistance to Manduca sexta (Yang et al., 2012). In addition to enhanced resistance to Manduca sexta, mutant lines of NaCDPK4 and NaCDPK5 double-knockdown exhibit delayed flowering (Yang et al., 2012). All these results suggest that CDPKs may have potential roles in the regulation of plant flowering and resistance to insects.

Soybean (Glycine max) is a photoperiod-sensitive crop that is constantly exposed to various herbivorous insects during its life-cycle. The high pest incidence period coincides with the flowering period of soybean (Gai and Cui, 1997). The relationship between flowering time and insect resistance was previously unknown in soybean and other photoperiod-sensitive plants. In this study, we isolated and characterized the soybean CDPK gene GmCDPK38, which is regulated by photoperiod. We identified the favorable GmCDPK38 haplotype and knocked out this gene in a susceptible haplotype using CRISPR/Cas9 technology. The gmcdpk38 mutants flowered later under both long-day (LD) and short-day (SD) conditions and exhibited enhanced resistance to the common cutworm (CCW) (Spodoptera litura Fabricius). Comparative transcriptome, metabolome, and phosphoproteome analyses of the control and gmcdpk38 mutants revealed many differentially expressed genes (DEGs), metabolites and proteins related to flowering time or insect resistance. Among them, one potential GmCDPK38 target was confirmed by protein–protein interaction assays. Furthermore, we identified the distribution frequencies of GmCDPK38 haplotypes at different latitudes and the sequence diversity of this gene in wild soybeans, landraces, and improved cultivars. Our studies suggested that GmCDPK38 plays a dual role, as it regulates photoperiod-induced soybean flowering and resistance to CCW in a coordinated manner.

Results

GmCDPK38 expression is regulated by photoperiod

Fifty soybean CDPK members (denoted GmCDPK1 to GmCDPK50) were identified via a domain search against all predicted proteins in the soybean genome (Glycine max Wm82.a1.v1, http://www.phytozome.net/soybean) and phylogenetically clustered into four different subgroups (subgroups I–IV) by Hettenhausen et al. (2016). Using the gene sequence accession numbers of these GmCDPKs in the Phytozome database as keywords to search the RNA sequencing (RNA-seq) data (GSE51007, NCBI) of two soybean varieties (Clark and Williams 82) at three times (06:30, 14:30, and 22:30) under LD conditions (06:45–22:45) for 26 days after planting and SD conditions (06:45–16:45) for 24 days after planting (Wu et al., 2014), we obtained the transcript data of 42 soybean CDPK genes (Supplemental Figure S1 and Supplemental Table S1). The expression levels of GmCDPK1, GmCDPK8, GmCDPK9, GmCDPK31, GmCDPK38, and GmCDPK45 were relatively high, but the expression levels of the other GmCDPKs were relatively low, with an average reads per kilobase per million (RPKM) value under 20. Additionally, these six genes exhibited a diurnal rhythm (Supplemental Figure S1 and Supplemental Table S1). Under LD conditions, GmCDPK38 was upregulated from 06:30 to 22:30, while the other five genes were downregulated. Under SD conditions, GmCDPK38 was downregulated from 06:30 to 14:30 and upregulated from 14:30 to 22:30, which was different from the pattern for the other five genes. Among the four genes in subgroup IV, GmCDPK1, GmCDPK31, and GmCDPK38 all showed rhythmic expression patterns, suggesting that the genes in subgroup IV are more likely to be involved in soybean flowering. GmCDPK38 was chosen for further expression pattern analysis.

Time course-dependent expression pattern analysis indicated that GmCDPK38 responded to the photoperiod, as greater accumulation of GmCDPK38 transcripts was observed under LD conditions than under SD conditions (Figure 1A). We then analyzed the diurnal expression patterns of GmCDPK38. Under LD conditions, GmCDPK38 expression was decreased during the day and increased during the night (Figure 1B). The same expression pattern of this gene was observed under SD conditions; however, the amplitude and abundance of its transcripts were lower than those under LD conditions. These results suggested that GmCDPK38 expression is regulated by photoperiod and is abundant under LD conditions.

Figure 1.

Expression analysis of GmCDPK38 and phenotypic identification of T₃gmcdpk38 mutants. A, Time course-dependent expression patterns of GmCDPK38 in soybean leaves under LD or SD treatment (n = 3). The leaves were sampled 4 h after light exposure at 4, 7, 12, 15, 19, 23, 29, and 42 DAE. B, Diurnal expression patterns of GmCDPK38 in soybean leaves under LD or SD treatment (n = 3). The leaves were sampled every 4 h over a 24-h time period at 15 and 16 DAE. C, Target site for CRISPR/Cas9 editing in the first exon of GmCDPK38. UTR, untranslated region. D, Sequences of WT and two homozygous knockout (KO) lines. Red letter, target site sequence; blue letter, protospacer adjacent motif; underline, insertions; dashes, deletions. E, Comparison of the flowering phenotypes of WT, KO#1 and KO#2 under LD and SD conditions. Upper, an overview of WT and the two gmcdpk38 mutant lines KO#1 and KO#2 (scale bar = 10 cm). Lower panel, a close-up view of the areas framed by the boxes. F, Flowering time statistics of WT plants and gmcdpk38 mutants. G and H, The leaves and PIs of WT plants and gmcdpk38 mutants after 24 h of CCW feeding under LD conditions. Twenty-third instar CCW larvae were released into the center circle of the paper and removed after 24 h of free selection. The red line represents a PI of 1, indicating no preference. PI >1 indicates resistance, and PI < 1 indicates susceptibility in the gmcdpk38 mutants. I and J, Representative size comparison and average larval weight of CCW on WT plants and gmcdpk38 mutants after 4 days of feeding under LD conditions. Scale bar = 1 cm. Two-tailed t tests were used for all statistical analysis: *P < 0.05; **P < 0.01; ***P < 0.001; ns: not significant. All error bars denote ±sd.

Genetic variation in GmCDPK38 correlates with soybean flowering time and resistance to common cutworm

We analyzed an approximately 6.8-kb sequence of GmCDPK38 from 272 cultivated soybeans (Supplemental Table S2). Among them, the flowering time of 219 cultivated soybeans in 10 environments was investigated, and 37 SNPs and InDels (minor allele frequency >0.05) were detected on the basis of whole-genome sequencing (Supplemental Table S3). Eighteen polymorphic sites were significantly associated with variations in flowering time in at least seven environments (Figure 2A). Because sites 1, 3–5, and 8 showed complete linkage (r² = 1), sites 7 and 9 showed complete linkage (r² = 1), sites 10, 12–15, and 18 showed complete linkage (r² = 1), and sites 16 and 17 showed complete linkage (r² = 1), the 18 polymorphic sites were subdivided into seven groups, and nine haplotypes were defined in these cultivated soybeans (Figure 2B). The relationship between flowering time and the representative polymorphic sites (1, 2, 6, 7, 10, 11, and 16) of seven groups was analyzed. Cultivated soybeans with sites 1-A, 2-0, 6-A, 7-C, 10-C, 11-TC, and 16-G showed significantly later flowering than those with sites 1-G, 2-GAGAGAAATTATATTAA, 6-T, 7-G, 10-T, 11-0, and 16-0, respectively, in at least seven environments (Figure 2C). In addition, Hap2 cultivated soybeans containing all late-flowering-related polymorphic sites exhibited significantly later flowering than Hap3 cultivated soybeans with all early-flowering-related polymorphic sites in the 10 environments (Figure 2D;Supplemental Table S4).

Figure 2.

Haplotype analysis of GmCDPK38 in cultivated soybeans in our population. A, Association of 37 polymorphic sites (minor allele frequency >0.05) in the GmCDPK38 gene with cultivated soybean flowering time phenotypes from 10 environments and pairwise linkage disequilibrium analysis. The x-axis represents the polymorphic sites, shown as follows from left to right: SNP_-1652, InDel_-1644, SNP_-1599, SNP_-1582, SNP_-1561, SNP_-1521, SNP_-1511, SNP_-1497, SNP_-1493, SNP_-1480, SNP_-1474, SNP_-1403, InDel_-1304, SNP_-1260, InDel_-1273, SNP_-1198, SNP_-1186, SNP_-1160, SNP_-1156, SNP_-1102, SNP_-1010, SNP_-985, SNP_-790, SNP_-676, SNP_-542, SNP_-512, SNP_-71, InDel_-5, SNP_249, SNP_657, SNP_832, SNP_2254, SNP_2793, InDel_3546, SNP_3744, SNP_3974, and SNP_4666. Polymorphic sites were defined as being significantly associated with flowering time by comparison with the Bonferroni threshold [−log₁₀(P) > 1.30 or P < 0.05] by using a general linear model in Tassel 5.0 software. Yellow dots represent significant polymorphic sites (P < 0.05), and purple dots represent nonsignificant polymorphic sites. B, Haplotypes (Hap) of GmCDPK38 in cultivated and wild soybeans. The numbers 1–18 represent these polymorphic sites. The green and yellow squares show major and minor polymorphic sites, respectively. The numbers below the “Wild soybeans” and “Cultivated soybeans” columns represent the number of accessions containing each haplotype in wild soybeans and cultivated soybeans, respectively. C, Boxplots of flowering time for seven representative polymorphic sites in cultivated soybeans; 1, 2, 6, 7, 10, 11, and 16 represent SNP_-1561, InDel_-1304, SNP_-1010, SNP_-985, SNP_-71, InDel_-5, and InDel_3546, respectively. There were 140, 30, 140, 30, 110, 60, 111, 59, 137, 33, 137, 33, 136 and 34 1-A, 1-G, 2-0, 2-17 bp, 6-T, 6-A, 7-G, 7-C, 10-T, 10-C, 11-TC, 11-0, 16-G, and 16-0 plants, respectively. D, Boxplots of flowering time for four main haplotypes in cultivated soybeans. n = 76, 33, 29, and 25 Hap1, Hap2, Hap3, and Hap4 plants, respectively. E, Boxplots of larval weight of CCW feeding on haplotypes Hap2 (n = 33) and Hap3 (n = 29) in cultivated soybeans. F, Comparative analyses of GmCDPK38 expression between haplotypes Hap2 (n = 15) and Hap3 (n = 15). G and H, GUS staining of soybean hairy roots exposed to LD and SD conditions. Hap2_pro, Hap3_pro, and empty control represent soybean hairy roots transformed with recombinant plasmids containing the GmCDPK38 promoters from accessions containing Hap2, Hap3, and the control vector, respectively. FT_2011nj, FT_2011nt, FT_2011yz, FT_2012nj, FT_2012nt, FT_2012yz, FT_2013nj, FT_2013nt, FT_2013yz, and FT_2016nt represent the flowering times of cultivated soybeans grown in 2011, 2012, 2013, and 2016 in Nanjing (nj), Nantong (nt), and Yangzhou (yz), respectively. LW_2009nj, LW_2013nj, and LW_2014nj represent the evaluated resistance to CCW of cultivated soybeans grown in 2009, 2013, and 2014 in Nanjing (nj), respectively. For all boxplots, center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Two-tailed t tests were used for all statistical analysis: *P < 0.05; **P < 0.01; ***P < 0.001; ns: not significant.

We further evaluated the resistance of these 219 cultivated soybeans to CCW. The flowering time and larval weight had a slight negative correlation trend (r = −0.105, not significant; Supplemental Figure S2A). In addition, the larval weight of CCW fed Hap2 was significantly lower than that of larvae reared on Hap3 in one environment (Figure 2E). A similar trend was found in the other two environments, although the difference in the larval weight of CCW feeding on the soybeans was not significant. That is, cultivated soybeans with the late-flowering Hap2 were more resistant to CCW.

More than half of the polymorphic sites were located in the promoter of GmCDPK38, so we examined GmCDPK38 transcript abundance in 30 cultivated soybeans containing Hap2 or Hap3. The expression of GmCDPK38 in Hap2 was significantly lower than that in Hap3 under both LD and SD conditions (Figure 2F). This is also supported by the lower promoter activity of Hap2 than that of Hap3 (Figure 2, G and H). All these results indicated that GmCDPK38 might play a role in flowering time and insect resistance, which might be affected by different expression levels of this gene.

CRISPR/Cas9-mediated targeted mutagenesis of GmCDPK38 Hap3 causes delayed flowering in soybean and enhanced resistance to CCW

The sequence of GmCDPK38 in the soybean cultivar Jack belongs to Hap3 (Supplemental Texts S1 and 2). We knocked out GmCDPK38 in Jack using CRISPR/Cas9. One target site in the first exon of the gene was designed for Cas9 cleavage (Figure 1C). Nine T₀ transgenic lines harboring the T-DNA of the small guide RNA (sgRNA)/Cas9 vector were obtained, four of which had heterozygous-targeted mutations of GmCDPK38. Two homozygous T₃gmcdpk38 loss-of-function mutant lines (KO#1 and 2) were used for phenotypic analysis (Figure 1D;Supplemental Texts S1 and 2). No mutations occurred in the gmcdpk38 mutants at the two most likely off-target sites predicted by CRISPR-P (Supplemental Figure S3). We compared the flowering time of the wild-type (WT) plants and gmcdpk38 mutants. Under LD conditions, the WT plants flowered at 39.9 days after emergence (DAE), whereas the gmcdpk38 mutants flowered at 45.3 DAE (KO#1) and 42.7 DAE (KO#2). The flowering time of the gmcdpk38 mutants was significantly later than that of the WT plants (Figure 1, E and F). When plants were grown under SD conditions, knockout of GmCDPK38 delayed soybean flowering by 4.1 days (KO#1) and 2.8 days (KO#2; Figure 1, E and F). Overall, the Hap3 gmcdpk38 mutants exhibited a late flowering phenotype.

We then evaluated the insect resistance of the gmcdpk38 mutants and WT plants under LD conditions. In the dual-choice trial with CCW, the leaf area losses of the gmcdpk38 mutants were less than those of the WT plants after 24 h of free selection (Figure 1G). The preference index (PI) values calculated from the leaf area loss were significantly >1 (Figure 1H), indicating that most of the larvae preferred the WT plants to the gmcdpk38 mutants. In the no-choice assay, the average larval weight of CCW feeding on WT plants was significantly higher than that of larvae reared on the gmcdpk38 mutants (Figure 1, I and J), suggesting that mutation of GmCDPK38 elevated soybean resistance to CCW.

Identification of the possible substrates of GmCDPK38 by global phosphoproteomics

Considering that GmCDPK38 is annotated as a protein kinase, we performed a quantitative phosphoproteomic analysis to examine phosphoproteins in both the WT plants and gmcdpk38 mutants. A total of 5,399 quantified phosphopeptides spanning 3,475 proteins with 6647 quantified sites of phosphorylation were identified (Figure 3A), among which 1,374 phosphopeptides with 1,491 phosphosites in 1,197 proteins exhibited differential phosphorylation (Figure 3B;Supplemental Table S5).

Figure 3.

Summary of phosphoproteomic data. A, Statistical results of the identified and quantified phosphoproteomes. B, Number of upregulated and downregulated phosphosites. C, Number of sites per protein that were phosphorylated only in WT plants. P, phosphorylation modification. D, Number of serine, threonine, and tyrosine sites that were phosphorylated only in WT plants. E, Number of peptides containing phosphorylation sites in each motif that were phosphorylated only in WT plants. F, Heatmap showing the relative intensity of phosphosites in the gmcdpk38 mutants and WT plants. The numbers in the parentheses represent the positions of phosphosites in proteins. The pathways that were also enriched by RNA-seq analysis in Figure 4B are labeled with a red box. G, Number of members in nine families that were phosphorylated only in WT plants.

Of these differentially phosphorylated sites, 811 sites corresponding to 692 proteins were phosphorylated only in the WT plants but not in the gmcdpk38 mutants, which suggested that GmCDPK38 may directly or indirectly affect their phosphorylation. Among the 692 proteins, 88.78% exhibited single phosphorylation, 8.75% exhibited double phosphorylation, and 1.97% exhibited triple phosphorylation, and only 0.49% of the phosphoproteins had >3 phosphorylation modifications (Figure 3C). A majority of the phosphorylated amino acids in these 692 proteins were serine (88.09%), followed by threonine (10.48%), whereas phosphotyrosine accounted for 0.62% of quantified phosphorylation sites (Figure 3D). In addition, thirteen phosphoserine motifs, including [……_S_P….], [….S._S_……], and […R._S_……], and one phosphothreonine motif, [……_T_P….], were identified in these 692 phosphoproteins (Figure 3E;Supplemental Figure S4). We identified at least 32 proteins, including nine IQ-domain-containing proteins, three heat shock proteins, a respiratory burst oxidase homolog D (RBOHD), involved in regulating plant insect resistance, a flowering time control protein FCA in the autonomous flowering pathway, 14 sugar metabolism-related proteins, an acetyl-CoA acetyltransferase, a dihydroflavonol reductase, and the S-adenosylmethionine synthase GmSAMS1 among these 692 phosphoproteins (Figure 3F). A total of 59 protein kinases were identified, including 28 serine/threonine kinases, 13 receptor protein kinases, 4 CDPKs (including GmCDPK38), 2 CDPK-related kinases, and 2 casein kinases (Supplemental Table S6). Additionally, the members of nine families, including the ABC transporter family related to ion transport, E3 ubiquitin-protein ligase family related to protein degradation, serine/arginine-rich splicing factor family related to precursor mRNA splicing, and zinc finger CCCH domain-containing protein family, which is widely involved in various physiological processes, appeared more than five times (Figure 3G).

Four hundred thirteen sites derived from 360 proteins were phosphorylated only in the gmcdpk38 mutants but not in the WT plants (Figure 3B;Supplemental Table S5). Among the 360 proteins, although no flowering-related proteins were found, we identified 18 resistance-related proteins, including 4 IQ-domain-containing proteins, 2 lipoxygenases, 2 heat shock proteins and 2 WRKY transcription factors, and 8 sugar metabolism-related proteins (Supplemental Figure S5A). In addition to the phosphosites detected only in WT or mutant plants, 267 sites were phosphorylated in both types of plants but exhibited significantly altered abundances (Figure 3B;Supplemental Table S5). Of these, 75 phosphosites derived from 67 proteins were upregulated in the gmcdpk38 mutants compared with the WT plants, among which a calcium-binding EF-hand family protein, a VQ domain-containing protein and a universal stress protein, related to insect resistance, and five sugar metabolism-related proteins were found (Supplemental Figure S5B). Moreover, 192 phosphosites in 187 proteins were downregulated in response to GmCDPK38 mutation. The 187 proteins included two IQ-domain-containing proteins, a WRKY transcription factor and a lipoxygenase, related to insect resistance, the basic region/leucine-zipper transcription factor GmFDL07, related to photoperiod flowering, and four sugar metabolism-related proteins (Supplemental Figure S5C).

Taken together, these data showed a series of possible downstream substrates of the GmCDPK38 kinase. GmCDPK38 may mediate soybean flowering and insect resistance by affecting the phosphorylation of these proteins.

RNA-seq analysis revealed GmCDPK38-affected gene expression

RNA-seq was used to explore differential gene expression between the WT plants and gmcdpk38 mutants. Four treatments were performed, including at 29 DAE under LD conditions (LD29), 29 DAE under SD conditions (SD29), 42 DAE under LD conditions (LD42), and 42 DAE under SD conditions (SD42). At SD29 or LD42, the WT plants began to bloom, while the gmcdpk38 mutants remained at the vegetative growth stage (Figure 1, E and F). A total of 4201 DEGs were identified in the gmcdpk38 mutants compared with the WT plants (Figure 4A;Supplemental Table S7). Of these, 2,498 DEGs were from LD29, 1,578 DEGs were from SD29, 600 DEGs were from LD42, 570 DEGs were from SD42, and 81 DEGs were common in the four treatments.

Figure 4.

GmCDPK38 induces changes in the expression of genes in T₃gmcdpk38 mutants. A, Venn diagram showing the overlap of differentially expressed genes among the four treatments. B, Enriched KEGG pathways of differentially expressed genes in the four treatments/represents undetectable data. C, RNA-seq was performed to identify flowering-related genes that were differentially expressed. The color in each cell represents the value of the log₂ fold change (mutant/WT). LD29, 29 DAE under long-day conditions; SD29, 29 DAE under short-day conditions; LD42, 42 DAE under long-day conditions; SD42, 42 DAE under short-day conditions.

A total of 10 pathways were enriched by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Figure 4B). Most pathways were related to plant resistance to insects. The plant–pathogen interaction pathway was a common pathway in the four treatments. The plant hormone signal transduction pathway was enriched at LD29 and SD29. Four pathways, namely, Fc gamma R-mediated phagocytosis, phospholipase D signaling, starch and sucrose metabolism and pentose and glucuronate interconversions, were uniquely enriched at LD29. The remaining pathways, namely, isoflavonoid biosynthesis, diterpenoid biosynthesis, phenylpropanoid biosynthesis, and biosynthesis of secondary metabolites, were enriched at SD29. In these pathways, several biotic-stress-related genes were upregulated in the gmcdpk38 mutants, including 15 disease resistance resistance to Pseudomonas syringae pv. maculicola and 4 pathogenesis response pathogenesis-related involved in the plant–pathogen interaction pathway, a JA-responsive PDF1.2 (plant defensin) and an alcohol dehydrogenase involved in the hormone signal transduction pathway, and two isoflavone synthases and two cytochrome P450s involved in the biosynthesis of secondary metabolites pathway (Supplemental Table S8). In addition, five of the ten pathways had proteins that were phosphorylated only in WT plants (Figure 3F).

Several DEGs involved in flower development were also identified (Figure 4C;Supplemental Table S9). In the photoperiod pathway, 11 genes in gmcdpk38 mutants presented differences in expression in comparison to WT plants, including 2 flowering suppressors, GmFT1a and CYCLIC DOF FACTOR (CDF2; Fornara et al., 2009; Liu et al., 2018), and 2 flowering promoters, LUXARRHYTHMO (LUX) and EARLY FLOWERING 3-like (ELF3) (Lu et al., 2017; Fang et al., 2021). Most DEGs in the GA pathway were mainly found at LD29 and SD29. Nine of the 10 genes were downregulated at LD29, while 6 out of the 7 genes were upregulated at SD29. Three age pathway-related DEGs were identified, including a flowering promoter squamosa promoter-binding-like (SPL) and 2 flowering suppressors APETALA2s (AP2s; Cao et al., 2015; Zhao et al., 2015). Two flowering suppressors SHORT VEGETATIVE PHASEs (SVPs) involved in the vernalization pathway showed differential expression between the WT plants and gmcdpk38 mutants (Andrés et al., 2014). Seven DEGs associated with floral development were identified, among which six DEGs were downregulated, including a FRUITFULL (GmFULb), a MADS-box, a GLOBOSA, and two AGAMOUSs (AGs). In addition, we identified 30 DEGs in the sugar metabolism pathway, including eight sucrose synthases, three glucose-6-phosphate isomerases and three sugar transporters.

In conclusion, a large number of resistance- and flowering-related DEGs were identified by RNA-seq. Mutation of GmCDPK38 induced the expression of most flowering suppressor genes, while most flowering promoter genes were downregulated in the gmcdpk38 mutants. To confirm the RNA-seq results, the expression of seven flowering-related DEGs from the four treatments was analyzed by reverse transcription quantitative PCR (RT-qPCR); the results showed that the expression of these genes was consistent with the RNA-seq results (Supplemental Figure S6).

Increased accumulation of multiple metabolites in gmcdpk38 mutants

To investigate the role of GmCDPK38 in metabolic regulation, we performed a widely targeted metabolomic analysis between the WT plants and gmcdpk38 mutants. At LD42, a total of 762 metabolites were detected, and 288 of them had significantly different levels (Supplemental Table S10). Compared with the levels in the WT plants, the levels of 65.28% of the metabolites in the gmcdpk38 mutants were significantly increased. These differential metabolites were classified into 16 different compound classes, among which most were classified as flavonoids, amino acids, organic acids, phenylpropanoids and lipids (Figure 5A). Most of the differential metabolites, including flavonoids, phenylpropanoids, polyphenols, phenolamides, alkaloids, and terpenoids, are known plant endogenous defense metabolites (Sosa et al., 2004; Onkokesung et al., 2012; Bosch et al., 2014). At SD29, 233 of the 742 metabolites were identified as differential metabolites and were also classified into 16 compound classes (Figure 5B;Supplemental Table S10). More metabolites (48.93%) showed decreased levels in SD29 than in LD42. All of these results suggested that mutation of GmCDPK38 induces the accumulation of multiple metabolites, especially under LD conditions.

Figure 5.

Number of differential metabolites in each compound class in T₃gmcdpk38 mutants. A, Number of differential metabolites at LD42. B, Number of differential metabolites at SD29. The differential metabolites associated with insect resistance are labeled with a red box. LD42, 42 DAE under long-day conditions; SD29, 29 DAE under short-day conditions.

GmCDPK38 post-translationally regulates S-adenosylmethionine synthase

The phosphoproteomic data showed that the S-adenosylmethionine synthase GmSAMS1, with a [……_S_P….] motif, was phosphorylated only in WT plants (Figure 3F). This phosphosite in GmSAMS1 was also predicted to exist in its Arabidopsis homolog AtSAM1 (AtMAT1) by Jin et al. (2017) (Figure 6A). We then asked whether GmCDPK38 could physically interact with GmSAMS1. We first carried out a yeast two-hybrid (Y2H) assay in which we cloned GmCDPK38 into the bait vector pBT3-SUC and GmSAMS1 into the prey vector pPR3-N. When both constructs were cotransformed into yeast cells, the transformants grew well on plates containing selective medium lacking histidine (His), indicating an interaction between GmCDPK38 and GmSAMS1 (Figure 6B). The Y2H assay also showed an interaction between GmCDPKSK5 (GmCDPK21) from soybean subgroup I and the translationally controlled tumor protein GmTCTP, which was in agreement with that reported by Wang et al. (2017). However, coexpression of GmCDPK38 with GmTCTP or empty prey could not activate the expression of the His reporter gene. In addition, transient expression of green fluorescent protein (GFP)-tagged GmCDPK38 and GmSAMS1 in Nicotiana benthamiana cells showed that GmCDPK38 and GmSAMS1 were mainly located in the nucleus and cytoplasm (Figure 6C). We then examined these interactions in N. benthamiana cells by a bimolecular fluorescence complementation (BiFC) assay. When GmCDPK38-YFP^N (the N-terminal half of yellow fluorescent protein [YFP]) and GmSAMS1-YFP^C (the C-terminal half of YFP) were coexpressed, YFP signals were detected in the nucleus and cytoplasm (Figure 6C). The observation of an interaction between GmCDPKSK5 and GmTCTP at the plasma membrane was consistent with the previous results (Wang et al., 2017), whereas no YFP fluorescence was detected in the presence of GmCDPK38 and GmTCTP (Figure 6C). We further confirmed the correct gene sequences or protein sizes in these analyses (Supplemental Figure S7). All these results suggested that GmCDPK38 interacts with and may phosphorylate GmSAMS1.

Figure 6.

S-adenosylmethionine synthase was upregulated post-translationally in T₃gmcdpk38 mutants. A, Potential phosphosites of GmSAMS1 based on the predicted results of its Arabidopsis homolog AtSAM1 by Jin et al. (2017). The highlights of black and gray represent the background color of the same and conserved amino acid sequence, respectively. The predicted residues are labeled in green, and the residue detected in this study is labeled in purple. B, Y2H analyses showing GmCDPK38 interaction with GmSAMS1. +His, synthetic dropout medium lacking tryptophan and leucine; −His + 3-AT, synthetic dropout medium lacking tryptophan, leucine, histidine but supplemented with 20-mM 3-amino-1,2,4-triazole. C, Subcellular localization of GmCDPK38-GFP, GmSAMS1-GFP, and their complex (BiFC) in N. benthamiana leaves. 35S:GFP, empty control. Scale bars = 50 μm. D, Quantitative analysis of SAMS content between the WT plants and gmcdpk38 mutants by an ELISA (n = 3). The ordinate represents the ratio of the S-adenosylmethionine synthase content to the total protein content. E, Quantitative analysis of the S-adenosylmethionine content between the WT plants and gmcdpk38 mutants by metabolome profiling (n = 3). F, Quantitative analysis of coumaroylspermidine content between the WT plants and gmcdpk38 mutants by metabolome profiling (n = 3). G, Quantitative analysis of GmSAMS1 expression between the WT plants and gmcdpk38 mutants by transcriptome profiling (n = 3). The ordinate represents the fragments per kilobase of transcript per million values. Two-tailed t tests were used for all statistical analysis: *P < 0.05; **P < 0.01; ns: not significant. All error bars denote ±sd.

To further investigate the regulation of S-adenosylmethionine synthase by GmCDPK38, the abundance of S-adenosylmethionine synthase was measured. Compared with the WT plants, GmCDPK38 knockout enhanced the accumulation of S-adenosylmethionine synthase (Figure 6D). In addition, the gmcdpk38 mutants showed overproduction of S-adenosylmethionine and the downstream coumaroylspermidine (Figure 6, E and F), while the transcript abundance of GmSAMS1 did not vary significantly between the WT plants and gmcdpk38 mutants (Figure 6G), indicating that S-adenosylmethionine synthase was regulated post-translationally in the gmcdpk38 mutants. Moreover, the upregulated accumulation of defense-related S-adenosylmethionine and coumaroylspermidine in the gmcdpk38 mutants also suggested that mutation of GmCDPK38 enhances soybean resistance to CCW.

GmCDPK38 is a domesticated gene, and the resistant Hap2 is likely an artificial selection target

Wild relatives of crops are a rich source of genetic diversity (TreuRen et al., 2017). The flowering time and larval weight were significantly negatively correlated in 121 wild soybeans (r = −0.344, P < 0.001; Supplemental Figure S2B). Based on the whole-genome sequencing results, 18 significant polymorphic sites identified in the cultivated soybeans were also found in 121 wild soybeans (Figure 2A;Supplemental Tables S2 and S11), and 27 haplotypes were defined (Figure 2B). Compared to wild soybeans with Hap3, wild soybeans with Hap2 also exhibited significantly later flowering in three environments and significantly higher CCW resistance in two environments (Figure 7, A and B). Hap2 was derived via two SNPs from wild Hap25 (Figure 7C), which was present only once and exhibited later flowering and stronger resistance (Figure 2B;Supplemental Table S4). Hap3 was derived from wild Hap15 via three SNPs and one InDel and was different from Hap7 by only one nucleotide. Hap3, Hap7, and Hap15 all had early flowering and low resistance phenotypes (Supplemental Table S4).

Figure 7.

Variation analysis of GmCDPK38 in cultivated soybeans and wild soybeans in our population. A, Boxplots of flowering time for haplotypes Hap2 (n = 7) and Hap3 (n = 18) in wild soybeans. B, Boxplots of larval weight of CCW feeding on haplotypes Hap2 (n = 7) and Hap3 (n = 18) in wild soybeans. C, Haplotype network of GmCDPK38 in cultivated and wild soybeans. Black circles represent unobserved, inferred haplotypes. The number of hash marks between haplotypes represents the number of different polymorphic sites between alleles. ft_2011nj, ft_2012nj, and ft_2013nj represent the flowering time of wild soybeans grown in 2011, 2012, and 2013 in Nanjing (nj), respectively. lw_2014nj, lw_2016nj, and lw_2019nj represent the evaluated resistance to CCW of wild soybeans grown in 2014, 2016, and 2019 in Nanjing (nj), respectively. For all boxplots, center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Two-tailed t tests were used for all statistical analysis: *P < 0.05; **P < 0.01; ***P < 0.001; ns: not significant.

In our population of cultivated soybeans and wild soybeans, Hap2 appeared at a higher frequency in cultivated soybeans than in wild soybeans, while Hap3 was more abundant in wild soybeans than in cultivated soybeans (Figure 8A). In addition, Hap2 occurred more frequently in low latitudes south of the Qinling Mountains–Huaihe River, whereas Hap3 was mainly distributed in high latitudes north of the line (Figure 8B).

Figure 8.

Selection of GmCDPK38 during soybean domestication. A, Distribution of haplotypes Hap2 and Hap3 in cultivated and wild soybeans. B, Geographical distribution of haplotypes Hap2 and Hap3 in China. I indicates the northern single cropping, spring planting area; II indicates the Huang-huai-hai double cropping, spring and summer planting area; III indicates the middle and lower Changjiang valley double cropping, spring and summer planting area; IV indicates the central south multiple cropping, spring, summer, and autumn planting area; V indicates the southwest plateau double cropping, spring and summer planting area; VI indicates the south China tropical multiple cropping, all season planting area. C, π and Tajima’s D values of the GmCDPK38 gene in the three groups. D, F_ST in the GmCDPK38 gene among the three groups. The black line represents the threshold of the whole-genome level.

Furthermore, the sequencing data of 302 soybeans with more cultivars were used for frequency analysis of GmCDPK38 haplotypes in the cultivars (Supplemental Table S12; Zhou et al., 2015). Fourteen of 18 significant polymorphic sites identified in our population were also observed in the database population (Figure 2A;Supplemental Table S13). Based on these results, Hap2 could be isolated independently, while Hap3 and Hap7 could not be distinguished. Hap2 was present in most cultivars (51/85) and a few landraces (23/102) but was not detected in the 39 wild soybeans in the population (Figure 8A). Instead, the frequency of Hap3 (including Hap7) in cultivars (14/85) and landraces (15/102) was relatively low. The high frequency of Hap2 in cultivars further indicated that this allele is likely an artificial selection target.

To determine whether GmCDPK38 is a domesticated gene, we calculated the nucleotide diversity of GmCDPK38 in our population and the database population. The wild soybeans had relatively high π values, followed by the landraces and cultivars. Moreover, the Tajima’s D value for GmCDPK38 was negative in the cultivars (Figure 8C). In the coding sequence and immediate downstream region of GmCDPK38, the F_ST of wild soybeans versus landraces was substantially higher than that of landraces vs. cultivars (Figure 8D). All these results suggested that GmCDPK38 was selected during soybean domestication.

Discussion

GmCDPK38 may link soybean flowering time and resistance to common cutworm

Plant defense has an ontogeny trajectory, and the expression of defense metabolites shows strong spatiotemporal variability (Barton and Boege, 2017). The peak CCW attack period in the Nanjing area of China is from August 10 through September 20, which coincides with or partially coincides with the flowering stage of most soybean varieties (Gai and Cui, 1997). Here, regardless of whether cultivated or wild soybeans were grown, the CCW larvae fed with late-flowering accessions were lighter than those fed with early-flowering accessions, and the correlation between flowering time and larval weight in wild soybeans was extremely significant (Supplemental Figure S2). These results indicated that there is a relationship between flowering and resistance.

Expression analysis showed that soybean GmCDPK38 responded to the photoperiod and was highly induced by LD conditions (Figure 1, A and B). The genetic variation of the gene was correlated with soybean flowering time (Figure 2A). The late-flowering Hap2 with lower GmCDPK38 expression had higher CCW resistance than the early-flowering Hap3 of two species of the Glycine genus (Figures 2, D–H and 7, A and B). CRISPR/Cas9-mediated mutagenesis of GmCDPK38 Hap3 resulted in delayed flowering of soybean under both LD and SD conditions (Figure 1, E and F). Moreover, we confirmed the resistance of the gmcdpk38 mutants to CCW under LD conditions (Figure 1, G–J). Compared with the WT plants, metabolome analysis showed that the gmcdpk38 mutants accumulated more defense-related metabolites, including flavonoids, phenylpropanoids, and polyphenols (Sosa et al., 2004; Onkokesung et al., 2012; Bosch et al., 2014), under LD conditions than under SD conditions (Figure 5). This result indicated that the gmcdpk38 mutants may have a better insect resistance phenotype under LD conditions. Additionally, CCW is one of the major herbivorous insect pests at low latitudes in China, feeding naturally on soybean (Cui et al., 1997). In these areas, the summer sunshine in the field belongs to LD conditions. Therefore, it is more suitable to determine the resistance of gmcdpk38 mutants to CCW under LD conditions. Overall, these results indicated that GmCDPK38 is a common locus that regulates flowering time and insect resistance in soybean.

CDPKs are plant-specific protein kinases that consist of multigene families in many plants. Fifty CDPK genes were identified in soybean and phylogenetically clustered into four different subgroups (I–IV; Hettenhausen et al., 2016). GmCDPK38 is a member of subgroup IV, which is closely related to subgroup IV orthologs AtCPK28, NaCDPK4/5, OsCPK4/18, and SlCDPK28 of Arabidopsis, native tobacco, rice, and tomato (Solanum lycopersicum), respectively (Cheng et al., 2002; Asano et al., 2005; Yang et al., 2012; Hettenhausen et al., 2016; Hu et al., 2016). In Arabidopsis, overexpression of AtCPK28 by T-DNA inserted into the 5′-untranslated region (UTR) of the gene confers the gain-of-function mutant an early flowering phenotype (Gao et al., 2014; Li et al., 2021), while no alteration in flowering time was observed in two atcpk28 mutants, which carry T-DNA insertions in exons 8 and 11, respectively (Matschi et al., 2013). In addition, AtCPK28 is linked to vegetative-to-reproductive transition. In seedlings, loss of function of AtCPK28 results in enhanced resistance against bacterial infection (Monaghan et al., 2014), while in adult plants, the accumulation of JA in the atcpk28 mutants does not contribute to JA-mediated resistance against Trichoplusia ni, Spodoptera littoralis, and Alternaria brassicicola (Matschi et al., 2013; 2015). In native tobacco, simultaneous silencing of NaCDPK4 and NaCDPK5 results in a significant delay in flowering, massive accumulation of JA and defense-related metabolites, and increased resistance to Manduca sexta (Yang et al., 2012). In rice, OsCPK4 and OsCPK18 have not been shown to be involved in flowering but are implicated as negative regulators of immunity against pathogens (Xie et al., 2014; Wang et al., 2018a). In our study, the gmcdpk38 mutants exhibited late flowering and enhanced resistance to CCW (Figure 1, E–J). All these results indicated that GmCDPK38 orthologs from subgroup IV broadly function as positive regulators of flowering time and negative regulators of defense across plant species, potentially through similar molecular mechanisms. Moreover, AtCPK28, OsCPK4, and SlCDPK28 function in the osmotic stress response, salt and drought tolerance, and plant thermotolerance, respectively (Campo et al., 2014; Gao et al., 2014; Hu et al., 2021). Whether GmCDPK38 has a role in abiotic stresses needs to be studied in the future.

In addition, both GmCDPK38 and GmSAMS1 localized in the nucleus and cytoplasm of N. benthamiana leaves (Figure 6C). S-adenosylmethionine synthases in Arabidopsis and rice are located in the nucleus and cytoplasm (Chen et al., 2013a; Mao et al., 2015). However, the glycine in position 2 of GmCDPK38 is conserved, suggesting that GmCDPK38 might have a predicted N-myristoylation site involved in membrane targeting. In the future, the localization pattern of GmCDPK38 needs to be further examined in soybean.

Determination of the potential mechanism by which GmCDPK38 regulates flowering and resistance by multiple omics analyses

In flowering, GmCDPK38 may specifically phosphorylate the FCA protein in the autonomous pathway and regulate gene expression in the photoperiod, GA, age, vernalization, and sugar metabolism pathways (Figures 3, F and 4C). The photoperiod pathway may be the main pathway via which GmCDPK38 helps to mediate soybean flowering time. In the photoperiod pathway, the FT protein starts the flowering process by forming a complex with the basic region/leucine-zipper transcription factor FD (Abe et al., 2005). The phosphorylation of FD is essential for the formation of FT–FD florigen complexes (Kawamoto et al., 2015b). In our results, the flowering suppressor GmFT1a was upregulated and its downstream gene GmFULb was downregulated in the gmcdpk38 mutants (Figure 4C;Liu et al., 2018). An FD (GmFDL07) exhibited a significant change in phosphorylation abundance between the WT plants and gmcdpk38 mutants [fold change (mutant/WT) = 0.5623] (Supplemental Figure S5C and Supplemental Table S5). To confirm this result, more work is needed in the future.

With RNA-Seq, DEGs involved in plant resistance to biotic stress were identified between the WT plants and gmcdpk38 mutants (Figure 4B). The plant–pathogen interaction pathway is important for plant defense against pathogens, and there is signal crosstalk between plant–pathogen interactions and plant–insect interactions (Kessler and Baldwin, 2002). The plant hormone signal transduction pathway is well established as the core pathway that regulates plant resistance to insects (Bodenhausen and Reymond, 2007). The biosynthesis of secondary metabolites pathway is downstream of the plant–pathogen interaction pathway and plant hormone signal transduction pathway and is emerging as a defense regulator that can modulate defense deployment (Erb and Reymond, 2019). Here, the levels of many defense-related metabolites were upregulated in gmcdpk38 mutants, including flavonoids, phenylpropanoids, polyphenols, phenolamides, alkaloids, and terpenoids (Figure 5;Sosa et al., 2004; Onkokesung et al., 2012; Bosch et al., 2014). Furthermore, in these significantly enriched pathways (Figure 4B), many proteins were phosphorylated only in WT plants (Figure 3F). Because overexpression of S-adenosylmethionine synthase GmSAMS1 in tobacco (Nicotiana tabacum) increased plant resistance to CCW (Fan et al., 2018), we focused on characterizing the association between GmCDPK38 and GmSAMS1. Y2H and BiFC assays were used to demonstrate that GmCDPK38 can interact with GmSAMS1 in yeast and N. benthamiana cells, respectively (Figure 6, B and C). Arabidopsis AtCPK28 interacts with and phosphorylates S-adenosylmethionine synthases (AtMAT1, AtMAT2, and AtMAT3) for degradation through the ubiquitin/26S proteasome pathway (Jin et al., 2017). Mutagenesis of AtCPK28 decreases the phosphorylation of S-adenosylmethionine synthases and increases the contents of total MAT proteins and S-adenosylmethionine. Here, the lower phosphorylation of GmSAMS1 in the gmcdpk38 mutants also led to more S-adenosylmethionine synthase and its metabolic end products, without influencing the mRNA level of GmSAMS1 (Figure 3F and Figure 6, D–G). Interestingly, six E3 ubiquitin ligases were identified in our phosphoproteomic analysis (Figure 3G). These results suggested that GmCDPK38 may phosphorylate and destabilize GmSAMS1 for proteasomal degradation as its Arabidopsis ortholog does. In future work, we will further explore whether the phosphorylation of S-adenosylmethionine synthase by GmCDPK38 impacts its protein stability and whether these E3 ligases are involved in the degradation of S-adenosylmethionine synthase. S-adenosylmethionine synthase catalyzes the conversion of methionine and ATP to S-adenosylmethionine, which serves as the precursor for polyamines and ethylene (Chiang et al., 1996; Binet et al., 2011). Polyamines and ethylene play important roles in plant flowering regulation and resistance to insects (Kakkar and Rai, 1993; Kesy et al., 2011; Lu et al., 2014; Mo et al., 2015). Ethylene is also one of the major signaling molecules in the plant hormone signal transduction pathway.

In addition to S-adenosylmethionine synthase, we identified one RBOHD, two receptor-like cytoplasmic kinases (a PTI1-like kinase and a BSK2), six E3 ligases and nine serine/arginine-rich proteins by global phosphoproteomics (Figure 3, F and G). In Arabidopsis, AtCPK28 has been reported to be associated with RBOHD, BIK1, PUB25/26, ATL31/6, and IRR (Monaghan et al., 2014; Wang et al., 2018b; Dressano et al., 2020; Liu et al., 2022). RBOHD regulates plant stress response by triggering a burst of reactive oxygen species and interacts with AtCPK28 (Monaghan et al., 2014; Block et al., 2018). BIK1 belongs to the receptor-like cytoplasmic kinase family and functions as a signaling hub of plant immunity (Lin et al., 2014). The phosphorylation of BIK1 and E3 ligases PUB25/26 by AtCPK28 enhances ubiquitin ligase activity and promotes BIK1 degradation (Monaghan et al., 2014; Wang et al., 2018b). To maintain BIK1 homeostasis, AtCPK28 is targeted by the E3 ligases ATL31/6 for proteasomal degradation (Liu et al., 2022). In addition, a retained intron variant of AtCPK28 encoding a truncated protein with impaired kinase activity is induced by dephosphorylation of the serine/arginine-rich protein IRR (Dressano et al., 2020). Among these proteins associated with AtCPK28, the E3 ligase ATL6 has a putative ortholog identified in our phosphoproteomic analysis (Supplemental Table S14). We also found several other potential substrates of GmCDPK38, including nine IQ proteins related to plant insect resistance, a trehalose-phosphate synthase in sugar metabolism and an acetyl-CoA acetyltransferase in diterpenoid biosynthesis (Figure 3F). Moreover, a large number of kinases were identified by phosphoproteomic analysis (Supplemental Table S6), and these kinases may form a kinase cascade with GmCDPK38 to mediate flowering time and insect resistance. The function of these phosphoproteins needs to be further confirmed in soybean, and it may be other pathways involved in GmCDPK38.

Artificial selection of GmCDPK38 during soybean domestication affected both flowering time and insect resistance

Flowering time and insect resistance are key domestication traits for artificial selection (Zhou et al., 2015; Lu et al., 2020). The analysis of the nucleotide diversity of this gene indicated that there was a strong selective molecular footprint in GmCDPK38 (Figure 8, C and D). Haplotype analysis revealed that there were 27 alleles in wild soybeans and 9 alleles in cultivated soybeans (Figure 2B). The decrease in haplotypes in cultivated soybeans further indicated that this gene was selected during soybean domestication. Hap2 showed late flowering and high CCW resistance compared with Hap3 in both cultivated soybeans and wild soybeans (Figure 2, D and E and Figure 7, A and B). The difference between them was more obvious in wild soybeans than in cultivated soybeans, indicating the existence of other domesticated genes associated with flowering time and insect resistance.

The high frequency of Hap2 in cultivars is the hallmark indicating that Hap2 in cultivars is a beneficial haplotype and has been strongly favored during soybean selection (Figure 8A). The introduction of Hap2 into cultivars without this haplotype may improve the resistance of the targeted soybean accessions to CCW. Hap2 originated from wild Hap25 (Figure 7C). It is worth noting that wild Hap25 exhibited the latest flowering time and conferred higher resistance to CCW, and it appeared only once in our population (Figure 2B;Supplemental Table S4). Since wild Hap25 has not yet been fixed in cultivated soybeans, this rare allele of GmCDPK38 may be introduced into those lacking it to breed new insect-resistant cultivars during the next wave of selection.

CCW is a common pest at low latitudes and sporadic at high latitudes in China (Cui et al., 1997). It is a polyphagous insect that likes warmth. From high to low latitudes, the annual generation of CCW increases gradually (Shi, 2013). Analysis of the geographical distribution showed that the cultivated soybeans and wild soybeans with resistant Hap2 were mainly distributed at low latitudes in China (Figure 8B). At these latitudes, soybean is usually planted in intercropping systems. Due to the complex cultivation mode, the farmland environment is quite complicated, with rampant pests and changing conditions (Gao et al., 2018). In addition, the photothermal time during the soybean growth period is long enough at low latitudes in China. These factors could enable soybean to appropriately delay flowering time and improve insect resistance in these areas, which is beneficial for soybean. Cultivated and wild soybeans with susceptible Hap3 were mainly distributed at high latitudes in China (Figure 8B). Soybean is usually concentrated planting in large acreages at these latitudes. The farmland ecosystem and pest community in the field are both relatively simple, and the photothermal time is insufficient (Gao et al., 2018). The different environmental pressures may be the reason for the different distributions of the two haplotypes. The coincidence of the insect-affected region and the insect-resistant haplotype suggested that Hap2 may be a major haplotype selected in the distribution and utilization of soybean, especially in regions with severe CCW attack.

Materials and methods

Plant materials and growth conditions

The soybean (Glycine max) cultivar Jack (WT plants), gmcdpk38 mutants and 30 soybean accessions carrying Hap2 or Hap3 were grown in a growth room at 25°C under LD (16-h light/8-h dark) and SD (12-h light/12-h dark) photoperiodic conditions with a light intensity of 500 μmol m⁻² s⁻¹.

The flowering time of 219 cultivated soybeans was screened 10 times in Nanjing (32°12ʹN, 118°37ʹE), Yangzhou (32°23ʹN, 119°25ʹE), and Nantong (31°58ʹN, 120°53ʹE) from 2011 to 2013, as well as in Nantong in 2016. All accessions were arranged in a completed randomized block design with three replicates. The length of each row was 2-m long, and the row spacing was 0.5 m. The number of plants per row was limited to 20 at the seedling stage (approximately 2 weeks after emergence). The flowering time was calculated based on the number of days from emergence to the beginning of flowering (50% of the plants in a plot had an open flower at any node on the main stem; Fehr and Caviness, 1977). Twelve individuals for each accession were randomly screened.

Our group previously determined the flowering time of 121 wild soybeans in Nanjing from 2011 to 2013 (Hu et al., 2020). A 7-day CCW no-choice assay was performed with 219 cultivated soybeans in Nanjing in 2009, 2013, and 2014 (Wang et al., 2011; Liu et al., 2016) and 121 wild soybeans in Nanjing in 2014 and 2016 (Du et al., 2019). In this study, we reevaluated the insect resistance of 121 wild soybeans in Nanjing in 2019 as reported by Du et al. (2019). Insect bioassays began at the beginning of the CCW outbreak under natural conditions in Nanjing, when all accessions in our population did not bloom. During the experiments, the accessions bloomed successively. At the end of the experiments, most Hap2- and Hap3-containing accessions had bloomed.

Analysis of rhythmically expressed GmCDPK genes

Forty-two GmCDPKs were present in an analysis of the response of the soybean transcriptome to flowering (Wu et al., 2014). The gene expression data (RPKM) were downloaded from the NCBI database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE51007). The average of the expression data of three biological replicates was calculated and is listed in Supplemental Table S1.

RNA extraction and RT-qPCR

The soybean cultivar Jack was used for analysis of the expression pattern of GmCDPK38. The fully expanded leaves on the third preapical nodes were sampled 4 h after light exposure at 4, 7, 12, 15, 19, 23, 29, and 42 DAE and every 4 h over a 24-h time period at 15 and 16 DAE. All the samples were immediately frozen in liquid nitrogen and stored at −80°C.

Total RNA was isolated from soybean samples using the RNAsimple Total RNA Kit (TianGen, Beijing, China), and first-strand cDNA was reverse-transcribed with the PrimeScriptTM 1^st Strand cDNA Synthesis Kit (TaKaRa, Dalian, China). Gene expression was detected by performing RT–qPCR analysis using an ABI 7500 system (Applied Biosystems, Carlsbad, CA, USA) with ChamQTM SYBR qPCR Master Mix (Vazyme Biotech Co, Nanjing, China). The RT-qPCR data were analyzed using the 2^−ΔΔCT method in three replicates with the mRNA level of the soybean tubulin gene (Glyma.03G124400) as an internal control (Livak and Schmittgen, 2001). The primers used for expression analyses are listed in Supplemental Table S15.

CRISPR/Cas9 expression vector construction and soybean transformation

We used the CRISPR/Cas9 vector published by Du et al. (2016). The sgRNA of GmCDPK38 was designed with the web-based tool CRISPR-P (http://crispr.hzau.edu.cn/CRISPR/) based on the GmCDPK38 sequence in Jack and inserted into the LguI site of the CRISPR/Cas9 vector. The constructed vector was introduced into Agrobacterium tumefaciens strain EHA105 and subsequently transformed into Jack via the cotyledon-node method (Paz et al., 2006). Specific primers spanning the target site or likely off-target sites were designed for PCR sequencing. All the primers are listed in Supplemental Table S15.

Flowering time measurements and insect bioassay of transformed soybean

WT plants and T₃gmcdpk38 mutants were grown under both LD and SD conditions. The flowering time of at least 14 plants for each line was investigated. We recorded the number of days from emergence to flowering (Fehr and Caviness, 1977). Insect assays were performed in a growth room at 25°C and 30% relative humidity with an LD photoperiod (16-h/8-h light/dark). At 42 DAE under LD conditions, the fully expanded leaves on the third preapical nodes were fed to the CCW in both the dual-choice assay and no-choice assay. For each line, nine plants were used for the dual-choice assay, and nine plants were used for the no-choice assay. In the dual-choice trial, the leaves of the WT plants and gmcdpk38 mutants were separately placed on the side of moist filter paper in an oblong container. Each container was considered a biological replicate, and three replications were conducted in this experiment. Twenty third-instar CCW larvae were released into the center circle of the paper and removed after 24 h of free selection. The leaf area was measured using a leaf area scanner (WinFOLIA, LA2400, Canada). The PI values were calculated using the formula PI = 2C/(T + C), where C and T represent the consumed area of WT plants and gmcdpk38 mutants, respectively. PI > 1 indicated relative resistance, and PI < 1 indicated relative susceptibility of the gmcdpk38 mutants. In the no-choice assay, we fixed white net bags to the leaves of the WT plants or gmcdpk38 mutants. Four larvae were raised in one net bag and weighed after feeding for 4 days. Each net bag was considered a biological replicate, and at least six replications were conducted. In addition, the total S-adenosylmethionine synthase content was determined with three replicates according to the instructions for an enzyme-linked immunosorbent assay (ELISA) kit (JiangsuMeimian Industrial Co., Ltd., Catalog MM-62897O2). Two-tailed t tests were used for statistical analysis.

Transcriptome, metabolome, and phosphoproteome profiling

For RNA-seq experiments, the fully expanded leaves on the third preapical nodes from the WT plants and gmcdpk38 mutants (KO#1) were sampled 4 h after light exposure at 29 and 42 DAE under LD and SD conditions with three biological repeats. The cDNA libraries were constructed and sequenced on an Illumina HiSeq platform (Illumina, San Diego, USA). After removing the adapter sequence and low-quality reads, clean reads were aligned to the soybean reference genome (https://genome.jgi.doe.gov/portal/pages/dynamicOrganismDownload.jsf?organism=Gmax) by HISAT2. The clean reads mapped to each gene were counted and normalized as fragments per kilobase of transcript per million mapped reads by Cuffquant and Cuffnorm software. Differential gene expression was determined using the DESeq R package. The combination of fold change and false discovery rate (FDR) was used to identify DEGs (fold change ≥ 2 or ≤ 0.5 and FDR < 0.05). We annotated all the DEGs in the KEGG database (http://www.genome.jp/kegg) and used the KEGG pathways as units to compare pathways that were significantly associated with DEGs with the genomic background (Fisher test, adjusted P < 0.05).

The fully expanded leaves on the third preapical nodes of the WT plants and gmcdpk38 mutants (KO#1) from 29 DAE under SD conditions and 42 DAE under LD conditions were delivered to Metware Biotechnology Co., Ltd. (Wuhan, China, http://www.metware.cn/) for analysis of the widely targeted metabolome with three biological repeats. A liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) system (HPLC, Shim-pack UFLC SHIMADZU CBM30A system; MS, Applied Biosystems 6500 Q TRAP) was used for metabolomic analysis. Freeze-dried soybean leaf samples were ground for 1.5 min in a mixer mill (MM 400, Retsch) with zirconia beads at a frequency of 30 Hz. One hundred milligrams of the powder was weighed and extracted overnight with 1.0 mL of 70% (v/v) aqueous methanol at 4°C. After 10 min of centrifugation at 10,000g, the extracts were absorbed and filtered. Metabolites were quantified using a scheduled multiple reaction monitoring method (Chen et al., 2013b). Metabolites with a fold change ≥1.2 or ≤ 0.833 and variable importance in project ≥1 were considered to be differentially expressed.

The fully expanded leaves on the third preapical nodes of the WT plants and gmcdpk38 mutants (KO#1) at 29 DAE under LD conditions were sent to Applied Protein Technology Co., Ltd. (Shanghai, China) for three biological replicates of the phosphoproteome analysis. Proteins were extracted by the TCA/acetone method and digested by the filter-aided sample preparation method as previously described (Pi et al., 2018). The phosphopeptides were enriched with the High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo Fisher Scientific) and analyzed by LC–MS/MS via a label-free quantitation method (Ren et al., 2017). A Q Exactive mass spectrometer (Thermo Fisher Scientific) equipped with an Easy nLC HPLC (Thermo Fisher Scientific) was used for MS experiments. We searched for protein annotations through the UniProt database (www.uniprot.org) and quantified the MS data via MaxQuant software. P < 0.05 (two-tailed t test) together with fold change ≥1.2 or ≤0.833 was designated as a threshold for identifying the phosphosites that significantly changed in abundance. In addition, phosphosites that were phosphorylated only in the WT plants or gmcdpk38 mutants were also considered significant phosphosites. The sequence information of 13 amino acids (including the phosphosites and ±6 amino acids from the modified sites) was extracted to predict the possible specific motifs by using MEME software.

Y2H assay

Coding sequences of GmCDPK38, GmSAMS1, GmCDPKSK5, and GmTCTP were amplified by PCR from Jack cDNA (primers shown in Supplemental Table S15) and cloned into the SfiI site of the bait vector pBT3-SUC-Cub and prey vector pPR3-N-NubG to construct pBT3SUC-GmCDPK38, pBT3SUC-GmCDPKSK5, pPR3N-GmSAMS1, and pPR3N-GmTCTP recombinant plasmids, respectively. Plasmids of each pair were then cotransformed into the yeast strain NMY51. The combination of pBT3SUC-GmCDPKSK5 and pPR3N-GmTCTP was used as the positive control, while the negative control was the combination of pBT3SUC-GmCDPK38 and pPR3N-GmTCTP. Transformants were selected on plates containing synthetic dropout medium lacking tryptophan and leucine, while the selection of interactions was conducted on plates containing synthetic dropout medium lacking tryptophan, leucine, histidine but supplemented with 20-mM 3-amino-1,2,4-triazole (3-AT).

BiFC assay

Coding sequences of GmCDPK38, GmSAMS1, GmCDPKSK5, and GmTCTP were cloned into BamHI and SmaI sites of the SPYNE173 vector and BamHI and SalI sites of the 35S-SPYCE (M) vector to produce the constructs GmCDPK38-YFP^N, GmCDPKSK5-YFP^N, GmSAMS1-YFP^C, and GmTCTP-YFP^C, respectively (primers shown in Supplemental Table S15). The YFP^N- and YFP^C-fused plasmids were transformed into A.tumefaciens EHA105 and cotransformed into the leaves of N.benthamiana. The combination of GmCDPKSK5-YFP^N and GmTCTP-YFP^C was used as the positive control, while the negative control was the combination of GmCDPK38-YFP^N and GmTCTP-YFP^C. Confocal laser scanning microscopy (Leica TCS SP2, Mannheim, Germany) was used to detect the YFP fluorescence signals. The YFP fluorescence was excited with a 488-nm laser and collected over 520–540 nm.

Subcellular localization

Coding sequences of GmCDPK38 and GmSAMS1 were separately cloned into XhoI and BamHI sites of the pFGC5941 vector with the GFP gene downstream of the CaMV 35S promoter (primers shown in Supplemental Table S15). Two recombinant vectors, 35S:GmCDPK38-GFP and 35S:GmSAMS1-GFP, and the empty vector 35S:GFP were transformed into A.tumefaciens EHA105 for transient expression in N.benthamiana leaves. The GFP signal was observed under confocal laser scanning microscopy (Leica TCS SP2, Mannheim, Germany). Excitation and detection wavelengths of GFP signal detection were 488 nm and 490–540 nm, respectively.

Western blot

To analyze the protein expression in subcellular localization and BiFC assays, total proteins of transformed N.benthamiana leaves for protein gel blot analysis were extracted with protein extraction buffer (50-mM Tris–HCl pH7.5, 150-mM NaCl, 5-mM EDTA, 0.1% (v/v) Triton X-100, and protease inhibitor cocktail). Immunodetection was performed using monoclonal HA (Babco, http://www.covance.com/), c-myc (Sigma, http://www.sigmaaldrich.com/), and GFP antibodies (Cell Signaling Technology, http://www.cellsignal.com/). Western blot was performed as described previously (Waadt et al., 2008).

GmCDPK38-based association analysis and sequence diversity analysis

Whole-genome resequencing was conducted for our population (121 wild soybeans, 197 landraces, 55 cultivars, and 20 uncertain accessions). Among them, the sequencing data of 262 soybean accessions have been published (Lu et al., 2020). We performed resequencing, mapping and variant calling of the remaining 131 varieties as described by Lu et al. (2020) and used the sequencing data to analyze the polymorphic sites of GmCDPK38 (including the 1,707-bp promoter region upstream of ATG and the 5,105-bp genomic region) in cultivated soybeans and wild soybeans. Forty-nine heterozygous cultivated soybeans and 28 heterozygous wild soybeans that had different bases detected at one site in the sequencing data were excluded from the haplotype analysis of our population. In the cultivated soybeans, the association of polymorphic sites with minor allele frequency >0.05 and flowering time data was calculated using a general linear model in Tassel 5.0 software. Polymorphic sites were defined as being significantly associated with flowering time by comparison with the Bonferroni threshold [−log₁₀(P) > 1.30 or P < 0.05]. Haplotype analysis and pairwise linkage disequilibrium analysis of GmCDPK38 were carried out with Haploview 4.2 software. In addition, we downloaded 302 soybean sequencing data sets (62 wild soybeans, 130 landraces, and 110 cultivars) in the database (https://figshare.com/articles/Soybean_resequencing_project/1176133;Zhou et al., 2015) and further identified the GmCDPK38 haplotypes among them, except for 76 heterozygous soybeans. The whole-genome sequencing data of our population (121 wild soybeans, 197 landraces, and 55 cultivars) and the database population (62 wild soybeans, 130 landraces, and 110 cultivars) were imported into the program VCFtools. In VCFtools, the nucleotide diversity (π), Tajima’s D and F_ST were calculated according to the sequencing data within GmCDPK38 (including the 1,707-bp promoter region upstream of ATG and the 5,105-bp genomic region) between accessions in each subpopulation.

Correlation analysis

Correlation analysis of flowering time and CCW resistance in soybean was performed with the average flowering time and the average larval weight using SPSS 20.0 software based on Pearson’s correlations (SPSS Statistics 20).

Transient expression assay of the GmCDPK38 promoter

Two natural GmCDPK38 promoters were cloned from the soybean accessions Jiangninglaoshudou (NJAU_C205, carrying Hap2) and Donghaixiaoheidou (NJAU_C146, carrying Hap3) and then inserted into EcoRI and HindIII sites of the pCAMBIA1381z vector (primers shown in Supplemental Table S15). Three constructs, namely, Hap2_pro:GUS, Hap3_pro:GUS and the empty control, were introduced into Agrobacterium rhizogenes K599. The transformation of soybean hairy roots using the Jack variety was performed as described by Du et al. (2016). After 3 weeks of cultivation in the dark, the hairy roots were transferred separately to LD and SD conditions for another 5 days. The samples were subjected to GUS staining at 4 h after light exposure on the sixth day. GUS staining was performed as described by Chao et al. (2014).

Accession numbers

The sequence data can be found in the Phytozome databases under the following accession numbers: GmCDPK38 (Glyma.16G128600), GmSAMS1 (Glyma.15G190500), GmCDPKSK5 (Glyma.08G005600), and GmTCTP (Glyma.09G044200). Other gene accession numbers involved in the transcriptome and phosphoproteome assays are provided in Supplemental Tables S5–S9, respectively.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Table S1. RPKM values of all GmCDPK genes presented in the previous RNA-seq database, downloaded from Wu et al. (2014).

Supplemental Table S2. Sequencing data of DNA polymorphic sites in the promoter and genomic regions of GmCDPK38 in our population.

Supplemental Table S3. Distribution of DNA polymorphic sites (minor allele frequency >0.05) in the GmCDPK38 promoter and genomic regions among cultivated soybeans in our population.

Supplemental Table S4. Phenotypic data of 30 haplotypes for GmCDPK38.

Supplemental Table S5. Sites of significant phosphorylation between WT plants and gmcdpk38 mutants.

Supplemental Table S6. Identification of kinases that were phosphorylated only in WT plants.

Supplemental Table S7. Differentially expressed genes between WT plants and gmcdpk38 mutants.

Supplemental Table S8. Defense-related differentially expressed genes between WT plants and gmcdpk38 mutants.

Supplemental Table S9. Flowering regulation-associated differentially expressed genes between WT plants and gmcdpk38 mutants.

Supplemental Table S10. Differential metabolites between WT plants and gmcdpk38 mutants.

Supplemental Table S11. Eighteen significant DNA polymorphic sites in the GmCDPK38 promoter and genomic regions among wild soybeans in our population.

Supplemental Table S12. Sequencing data of DNA polymorphic sites in the promoter and genomic regions of GmCDPK38 in the database population.

Supplemental Table S13. Accessions and types of haplotypes 2 and 3 for GmCDPK38 in the database population.

Supplemental Table S14. Arabidopsis putative orthologs of putative targets of GmCDPK38.

Supplemental Table S15. Primer pairs used for PCR or RT–qPCR.

Supplemental Figure S1. mRNA abundance of GmCDPK genes in soybean varieties Williams 82 and Clark as determined by RNA-seq.

Supplemental Figure S2. Regression correlations of flowering time and larval weight.

Supplemental Figure S3. Potential off-target analysis of the target site of GmCDPK38 in T₃gmcdpk38 mutants (KO#1 and KO#2).

Supplemental Figure S4. Motifs deduced from phosphopeptides with differential levels of modification in response to GmCDPK38 mutation.

Supplemental Figure S5. Heatmap showing the relative intensities of phosphosites corresponding to resistance-, flowering-, and sugar metabolism-related proteins in gmcdpk38 mutants and WT plants.

Supplemental Figure S6. RT–qPCR verification of the expression of selected flowering-related DEGs.

Supplemental Figure S7. Confirmation of correct gene sequences or protein sizes by intracellular localization and protein–protein interaction assays.

Supplemental Text S1. Gene sequence of GmCDPK38 in Jack and two T₃ homozygous mutant lines (KO#1 and KO#2).

Supplemental Text S2. Protein sequence of GmCDPK38 in Jack and two T₃ homozygous mutant lines (KO#1 and KO#2).

 

H.W. and D.Y. designed this research. X.Li and D.H. conducted soybean transformation experiments and phenotypic identification. L.C., HQ.W, and X.Liu performed the expression analysis. H.D. evaluated the insect resistance of the soybean population. H.Z. and Z.H. investigated the flowering time of the soybean population. F.H. and G.K. carried out the haplotype analysis. Z.Y. performed the evolutionary analysis. F.K. and B.L. participated in the excavation of flowering-related genes in multiple omics analyses. X.Li and D.H. wrote the manuscript. H.W. and D.Y. revised the manuscript. All authors read and approved the final manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Deyue Yu (dyyu@njau.edu.cn).