Introduction

Expression profiling studies in different organisms have suggested that proteins with unknown functions play important roles in many biological processes (Gollery et al. 2006). These proteins have been divided into two types: one includes proteins with obscure features that lack defined motifs or domains (POFs) and the other includes proteins with defined features that contain at least one previously defined domain or motif (PDFs). Among the latter, a group of proteins containing the cystathionine-β-synthase (CBS) domain might play important roles in stress response/tolerance in Arabidopsis under various stress conditions (Kushwaha et al. 2009). Since the CBS domain was first identified in the Archaebacterium Methanococcus jannaschii (Bateman 1997), CDCPs have been found to represent a large superfamily of evolutionarily conserved proteins. Kushwaha et al. identified CDCPs in whole-genome analyses of Oryza sativa and Arabidopsis thaliana and found that the CBS domain coexists with other functional domain(s) in most of these proteins, which may indicate their probable functions. Based on whether they have additional domain(s), these proteins were further classified into different subclasses: CBSX, CBSCLC, CBSSIS, CBSPPR, CBSIMPDH, CBSCBS, CBSCBSPB and CBSDUF. These subclasses possess various functions, including cytoplasmic targeting, subcellular localization of chloride channels (CLC), protein–protein interaction, protein regulation, sensing of cellular energy status, and maintenance of intracellular ion gradients (Bateman 1997). For example, the highly conserved structure of CBS domains from CLC plays a role in regulating the common gate (Estevez et al. 2004). AKINbc, a CDCP containing four CBS domains, contributes to SnRK1 heterotrimeric complexes and interacts with two proteins implicated in plant pathogen resistance (Gissot et al. 2006). OsCBSX4, a CDCP, could improve abiotic stress tolerance in plants (Singh et al. 2012). OsBi1, a CDCP, could be induced by BPH and is related to resistance to brown plant hopper in rice plants (Wang et al. 2004). OsCBSX3, a CDCP, is involved in rice resistance to M. oryzae (Singh et al. 2012).

However, very few studies have been reported on the CBSDUF subgroup. The CBSDUF subgroup protein contains one domain of unknown function (DUF21) (PF01595) and an N terminus that is adjacent to two intracellular CBS domains. This transmembrane region has no known function. Many of the sequences in this family are annotated as hemolysins because of their similarity to Q54318 (HLYC_BRAHO), which does not contain this domain. Therefore, the functions of DUF21 are still unknown. DUF21 often exists together with CBS domains and plays important roles in plant growth and development. The characteristics of the CBSDUFs in this subgroup are not yet clear. In our previous study, we identified CDCPs in soybean, but there was no detailed analysis of the CBSDUF subgroup. We found that overexpression of soybean GmCBS21, which belongs to the CBSDUF subgroup, possesses a novel function to improve low nitrogen tolerance in A. thaliana in our previous study (Hao et al. 2016). In addition, Sinharoy et al. found that a protein containing the CBS-DUF21 domain from Medicago truncatula is required for rhizobial infection and symbiotic nitrogen fixation (Sinharoy and Liu 2016). Therefore, considering the above studies, we speculate that proteins in the CBSDUF subgroup may play an important role in regulating biotic and abiotic stress, especially in legumes, and are worthy of further exploration. Soybean is one of the most important oil crops in the world and provides a large proportion of the protein used by humans and animals (Kereszt et al. 2007). However, to date, few data (Hao et al. 2016) are available about proteins in the CBSDUF subgroup in soybean. In this study, we took advantage of bioinformatics and publicly available data to identify and analyze soybean CBSDUF genes on a genome-wide scale. A total of 18 CBSDUFs were identified, and their phylogenetic relationships, gene structures, protein structures, conserved motifs, and expression patterns were analyzed in detail. Furthermore, the expression of CBSDUFs in response to various abiotic stresses as well as low nitrogen treatments in a low N-tolerant soybean variety (Pohuang) was determined. Our results provide a basis for further investigation of the evolution and functions of CBSDUFs.

Results

Identification and Phylogenetic Analysis of the Soybean DUF21- and CBS-Domain-Containing Proteins

Eighteen putative GmCBSDUF members were found in the NCBI database and used as queries to conduct BLAST searches against the public genome database (https://phytozome.jgi.doe.gov/pz/portal.html#). If more than one transcript existed, the primary transcript was selected as a representative. Using the same approach, 8, 10, 10, 4, 9, 4, and 4 putative CBSDUF members were identified from common bean (Phaseolus vulgaris), M. truncatula, Lotus japonicus, sorghum, Arabidopsis, rice, and maize, respectively. Table 1 shows the information of CBSDUF genes. Based on available information in the Phytozome 12 database, functional annotations for soybean CBSDUFs were obtained. Less information about the functions of the CBSDUF genes was found. The main functional annotations showed that most of the CBSDUF genes were predicted to be ancient conserved domain protein-related, metal transporter CNNM, or hemolysin-related. The specific functions of these genes remain to be discovered.

Table 1 CBSDUFs gene information
Full size table

A phylogenetic tree was built with 67 protein sequences from eight plant species to investigate the phylogenetic relationships among CBSDUFs from soybean, three other legumes, Arabidopsis, and three gramineous plants (Fig. 1). The soybean CBSDUFs were named GmCBSDUF1 to GmCBSDUF18 according to their chromosomal positions. The genes from the other plant species were named by the same method. Based on the results of phylogenetic tree analysis, we divided these CBSDUFs into eight groups: Group A to Group H (Fig. 1). Group A included 21 members, and it covered eight species. All members of Group B and Group E were dicotyledonous plants. Group C was monocot-specific. Group D did not include legume members. Group F and Group G were legume-specific. The legume CBSDUFs show a very close evolutionary relationship, and the CBSDUFs from gramineous plants show a close evolutionary relationship. Compared to other species, the soybean CBSDUF gene family is extensively expanded. The number of soybean CBSDUFs was almost as many as those from rice, maize, sorghum, and Arabidopsis combined (Table 1). The number of GmCBSDUF genes is approximately two times more than those of Arabidopsis, common bean, M. truncatula, or L. japonicus and four times more abundant than those of rice, maize, or sorghum. The reason for this increase may be the multiple whole-genome duplication events of the soybean genome (Schmutz et al. 2010). The number of CBSDUF genes in dicotyledonous plants is much greater than that in monocotyledonous plants. Therefore, we speculate that CBSDUF plays an important role in dicots than monocots. The phylogenetic relationships may reflect some distinction between legume plant CBSDUFs and the four nonlegume plant CBSDUFs and indicate that the potential biological functions of some CBSDUFs are specific to legume plants.

Fig. 1
figure 1

Phylogenetic relationships of the CBSDUFs. Phylogenetic relationships of the CBSDUFs from soybean (Gm), common bean (Pv), Medicago truncatula (Mt), Lotus japonicus (Lj), Arabidopsis (At), rice (Os), maize (Zm), and sorghum (Sb). The phylogenetic tree was constructed using Mega 6.0. The 67 CBSDUF proteins from eight plant species can be divided into eight groups (a–h); the branches are shown in different colors (Color figure online)

Full size image

Gene Structure and Protein Structure of GmCBSDUFs

Exon–intron structural diversity often plays a key role in the evolution of gene families. To investigate the exon–intron organization of GmCBSDUFs, gene structures were mapped on the basis of the genomic and coding region sequences. The results showed that GmCBSDUFs have 8–15 exons and highly similar gene structures in the conserved region (Fig. 2). The size of GmCBSDUF genes is mainly affected by their intron size. GmCBSDUF12 is the largest gene and has the longest total intron length.

Fig. 2
figure 2

Phylogenetic relationships and gene structures of GmCBSDUFs. The phylogenetic tree (left panel) was constructed using MEGA 6.0, and the gene structures (right panel) were drawn using the gene structure display server

Full size image

The soybean genome has undergone significant changes in the long-term evolutionary process. Some CBSDUF proteins are highly homologous in the terminal nodes, suggesting that they are putative paralogous pairs. In the study, a total of seven putative paralogous pairs (4/6, 10/14, 11/13, 5/17, 2/3, 8/16, 1/12) were identified, with sequence identities ranging from 60.47 to 99.26%.

To some extent, functional information can be derived from structural similarity. Knowledge of the structure is often essential for interpreting functional data. GmCBSDUF protein structures are shown in Fig. S1. It is clear that GmCBSDUF proteins have a highly conserved hydrophobicity profile, with one hydrophobic segment located at the N terminus. SMART allows the identification and annotation of genetically mobile domains and the analysis of domain architectures. The results are shown in Fig. 3. The major domains are the DUF21 and CBS domains. The DUF21 domain is found in the N terminus of each protein, adjacent to two intracellular CBS domains, and has no known function. In addition, most GmCBSDUF proteins possess 3–4 transmembrane helices except for GmCBSDUF10, GmCBSDUF11, and GmCBSDUF13, which have 2, 1, and 5, respectively. Interestingly, all GmCBSDUFs transmembrane domains pass through the DUF21 domain. Therefore, we speculate that the domain of unknown function DUF21 may play a role in ion channel or signal transduction. In this study, the secondary and tertiary structures of GmCBSDUF proteins were predicted (Fig. 4). The structures were analyzed and compared to the results of Fig. 2. Proteins with high identities also have similar secondary structures, such as GmCBSDUF4/6, GmCBSDUF11/13, GmCBSDUF10/14, GmCBSDUF5/17, GmCBSDUF2/3, GmCBSDUF8/16, and GmCBSDUF1/12. Interaction with a ligand molecule is essential for many proteins to carry out their biological function. This interaction is generally specific, not only in terms of the molecules involved in the interaction but also in the location (i.e., the site of ligand binding) in which the interaction takes place. The results showed that although most GmCBSDUF proteins have similar structures, they have different binding sites, suggesting that they may display different functions.

Fig. 3
figure 3

Main domains detected in soybean CBSDUF proteins by SMART. The blue rectangle represents the transmembrane region; the gray rectangle represents the DUF21 domain; the pink pentagon represents the CBS domain; the green hexagon represents the CorC_HlyC domain; and the orange rectangle represents the SCOP domain (Color figure online)

Full size image
Fig. 4
figure 4figure 4

Protein structure analysis of soybean CBSDUF proteins. a The secondary structure analysis of soybean CBSDUF proteins. protein binding region, polynucelotide-binding region, helix, strand, disordered region, buried, exposed, helical transmembrane region. b The tertiary protein structures were predicted by using Phyre2 (Color figure online)

Full size image

Tissue-Specific Expression Profiling of GmCBSDUFs

Based on the publicly available soybean RNA-Seq data (Libault et al. 2010), the expression patterns of 18 GmCBSDUFs were investigated in various tissues, including (1) root hair cells isolated at 84 h after sowing (HAS), (2) root hair cells isolated at 120 HAS, (3) root tips, (4) roots, (5) mature nodules, (6) leaves, (7) shoot apical meristems, (8) flowers, and (9) green pods. An expression heat map was constructed (Fig. 5a). The results showed that (1) all GmCBSDUFs were expressed in at least one tissue; (2) GmCBSDUF2/3/5 were expressed in all tissues, and their expression levels were relatively high; (3) GmCBSDUF9 had the lowest expression under all conditions; (4) GmCBSDUF8 was expressed only in the underground tissues; and (5) GmCBSDUF9 was expressed only in one shoot apical meristem. In addition, GmCBSDUF1/12 as well as GmCBSDUF16/13 showed similar expression patterns. Moreover, based on the publicly available soybean RNA-Seq data (Libault et al. 2010), expression heat maps of 14 GmCBSDUFs (except GmCBSDUF7/11/13/16, which were not or barely expressed in roots) in root hairs harvested at 12, 24, and 48 h after Bradyrhizobium japonicum inoculation (HAI), in mock-inoculated root hairs at 24 HAI, and in stripped roots at 48 HAI were also constructed (Fig. 5b). Based on the rhizobial inoculation method according to Libault et al. (2010), a B. japonicum cell suspension or water (mock inoculation) was sprayed on soybean seedlings growing on B&D agar medium. The results showed that inoculation with B. japonicum significantly increased the expression of GmCBSDUF8/9, but not other GmCBSDUFs, in root hairs. Therefore, we suspect that GmCBSDUF8/9 may be required for bacterial recognition, nodulation, and nitrogen fixation.

Fig. 5
figure 5

Tissue-specific expression profiles of GmCBSDUF genes. a Gene expression patterns of GmCBSDUF genes in nine different tissues, according to RNA-Seq data (Libault et al. 2010). SAM shoot apical meristem, HAS hours after sowing. b Comparison of the expression of soybean GmCBSDUF genes in root hairs (RH) and stripped roots inoculated (IN) and mock-inoculated (UN) with B. japonicum at 12, 24, and 48 h after B. japonicum inoculation (HAI). HAI IN RH: Root hair inoculated with B. japonicum; HAI UN RH: Root hair not inoculated by B. japonicum. Stripped roots: A soybean root after the strip** of root hairs. The color scale above the heat map indicates gene expression levels. The green color indicates a low expression level, and the red color indicates a high expression level (Color figure online)

Full size image

Furthermore, the soybean (Glycine max) genome database (Phytozome 12) provides high-resolution gene expression data for a diverse set of 17 soybean GeneAtlas tissue samples, such as flower (open and unopened), lateral root (standard), leaf (ammonia, nitrate, urea, standard and symbiotic condition), nodule (symbiotic condition), root tip (standard), root (ammonia, nitrate, urea, standard and symbiotic condition), shoot tip (standard), stem (standard), and 9 soybean normal tissue samples (flower, leaf, nodule, pod, root, root hair, seed, SAM, and stem). These data were also analyzed and represented as heat maps (Fig. S3). Expression analyses of all GmCBSDUF genes revealed that the different members have different tissue-specific expression. Among all 18 analyzed genes, GmCBSDUF5 showed the highest level of constitutive expression in all tissues, followed by GmCBSDUF3, GmCBSDUF2, and GmCBSDUF12. This high level of constitutive expression indicates a significant role in all these soybean tissues (Fig. S3). A cluster of genes showed low levels of expression in all tissues. They are GmCBSDUF8/9/11/13. GmCBSDUF16 is highly expressed only in root nodules, but its expression is very low in symbiotic conditions. These results are basically consistent with the results in Fig. 5, which makes the analysis of tissue expression patterns of GmCBSDUF genes more sufficient and meaningful. Analysis of the expression patterns of these genes will be helpful to the study of their function. All these expression profiles suggest functional redundancy and divergence among the soybean GmCBSDUFs during plant growth and development.

Promoter Analysis

Based on the soybean genome database (https://www.phytozome.net/soybean), the promoter regions located 2 kb upstream of the translation start codons of the GmCBSDUF genes were analyzed using the PlantCARE promoter analysis program (https://bioinformatics.psb.ugent.be/webtools/plantcare/html/). Multiple elements were identified, and the stress and hormone signaling-related sites are shown in Table 2. The table describes information pertaining to functions, such as elements in response to hormones, including abscisic acid (ABRE, CE1, and MRE) (Narusaka et al. 2003), salicylic acid (TCA element) (Liu et al. 2020), ethylene (ERE) (Song et al. 2009) reported that some AtCBS genes, such as AtCBSX2, AtCBSX3, and AtCBSCBS1, were stably expressed under any stress conditions, while some, such as AtCBSX1 and 15, were more sensitive to all stress conditions in both roots and shoots, and some, such as AtCBSDUFCH2, AtCBSDUF1, AtCBSDUF2, and AtCBSCBS2, were sensitive to stress conditions only in roots. In this study, the expression patterns of soybean CBSDUF genes under abiotic stresses were analyzed (Fig. 6). In contrast to other subgroup members, the results showed that GmCBSDUF7/8/11/16 was upregulated after exposure to cold, drought, salt, and H2O2, while GmCBSDUF17/18 was downregulated by cold, H2O2, salt and ABA, suggesting that these GmCBSDUF genes may play a role in crosstalk between signaling pathways responding to drought, H2O2, salinity, cold, and ABA. The results presented here will be helpful for future studies of the biological functions of GmCBSDUF proteins. Remarkably, we found that GmCBSDUF7/8/11/13/16 showed significant differences in expression under stress treatments. Therefore, we speculate that these genes are inducible and may play an important role in stress response. We will further examine this prospect in subsequent studies.

In conclusion, we performed a comprehensive bioinformatics analysis and provided detailed information on the soybean CBSDUF gene subgroup. Specifically, our results show that the soybean genome contains 18 CBSDUF genes, the largest subgroup among the identified CBSDUF gene subgroups in the study. Our analysis revealed the possible function of each GmCBSDUF gene in response to cold, salt, H2O2, ABA, dehydration, and low nitrogen, identified their potential clients and functional interactions, and revealed the specific responses of some GmCBSDUF genes to specific stresses. By interaction network prediction, some candidate interacting genes were found. At the same time, we preliminarily explored the function of GmCBSDUF3, which might improve the ability to resist abiotic stress in plants. This result provides an impetus for additional investigation of the biological roles and interacting proteins of the CBSDUF protein family in soybean, and a functional analysis of the genes in this family will be carried out systematically. In the future, we will use functional genomics in combination with a transgenic approach to verify the utility of those proteins with defined features as tools to improve stress tolerance in crop plants. Based on the present research and the characteristics of each family member, the research on functional analysis was classified and summarized. We will use gene knockout and transgenic technology to study the functions of the GmCBSDUFs. At the same time, the functions of the two domains, CBS and DUF21, will be studied by site-directed mutagenesis. In addition, due to the lack of information about this family of proteins, the biological pathways involving these genes are still unknown. We will screen for interacting proteins with yeast two-hybrid technology and provide evidence for their mechanisms of action. We will also determine the expression of transgenic plants under specific conditions by high-throughput sequencing technology and infer the gene regulatory network. The ideas provided here would also have a way for expounding the definite role of CBSDUF proteins in plants.

Materials and Methods

Identification of DUF21 and CBS Domain-Containing Proteins in Soybean

The known DUF21 and CBS domain-containing protein sequences from soybean, Arabidopsis, common bean, M. truncatula, L. japonicus, rice, maize, and sorghum were obtained from the NCBI database and used as queries to conduct BLAST searches against the public genome database (https://phytozome.jgi.doe.gov/pz/portal.html#) and L. japonicus genome database (https://www.kazusa.or.jp/lotus/). Sequences with an E value < 1.0 were selected for further analysis. A search with the keywords PF00571 for the CBS domain and PF01595 for the DUF21 domain was conducted for putative soybean CBSDUFs by searching ontologies against the Phytozome (v12.0) database (https://www.phytozome.net). If more than one transcript existed, the primary transcript was selected as a representative.

Phylogenetic, Gene, and Protein Structure Analyses

Multiple alignment analysis was performed with ClustalX 1.83 software (Thompson et al. 1997). Phylogenetic trees were generated by the neighbor-joining (NJ) method and bootstrap analysis (1000 replicates), and phylogenetic analysis was performed using MEGA6 software (Hall 2013). The exon/intron structures of the CBS genes were determined by comparing the coding sequences and corresponding genomic sequences in the gene structure display server (GSDS, https://gsds.cbi.pku.edu.cn/) (Guo et al. 2007). The protein transmembrane topology was predicted using TMHMM Server v2.0, and tertiary protein structures were predicted using Phyre. Domain architecture was analyzed by SMART (a Simple Modular Architecture Research Tool).

Plant Materials and Treatments

For low nitrogen treatment, seeds of a low N-tolerant soybean variety (Pohuang) were germinated. After 7 days, the seedlings were grown hydroponically in half-strength modified Hoagland solution until the first trifoliate leaf was fully developed and then grown in normal nitrogen solution (2 mM Ca(NO3)2·4H2O, 2.5 mM KNO3, 0.5 mM NH4NO3, 0.5 mM KH2PO4, 1 mM MgSO4·7H2O, 0.05 mM Fe-EDTA, 0.005 mM KI, 0.1 mM H3BO3, 0.1 mM MnSO4·H2O, 0.03 mM ZnSO4·7H2O, 0.0001 mM CuSO4·5H2O, 0.001 mM Na2MO4·2H2O, 0.0001 mM CoCl2·6H2O) or low nitrogen solution (0.2 mM Ca(NO3)2·4H2O, 1.8 mM CaCl2·2H2O, 0.25 mM KNO3, 1.125 mM K2SO4, 0.05 mM NH4NO3, 0.5 mM KH2PO4, 1 mM MgSO4·7H2O, 0.05 mM Fe-EDTA, 0.005 mM KI, 0.1 mM H3BO3, 0.1 mM MnSO4·H2O, 0.03 mM ZnSO4·7H2O, 0.0001 mM CuSO4·5H2O, 0.001 mM Na2MO4·2H2O, 0.0001 mM CoCl2·6H2O) at 25 °C in a chamber with a 12-h light and 12-h dark photoperiod. All treatments were performed over a continuous time course (0 h, 0.5 h, 2 h, 6 h, 12 h, and 3, 6, and 9 days). Roots, stems, and leaves from control and stress-treated plants (five plants were collected as mixed samples at each time point) were collected as samples in three biological replicates for RNA preparation, and the samples were quickly frozen in liquid nitrogen and stored at − 80 °C until use.

Soybean seeds were geminated in water at 25 °C in the dark under conditions of a 12--h light and 12-h dark photoperiod and 70% humidity. Salt, dehydration, cold, H2O2, and abscisic acid (ABA) stresses were applied to 2-week-old soybean seedlings. For salt stress, the roots of seedlings were dipped into solutions of 200 mM NaCl. For dehydration, the root systems of whole plants were placed onto filter paper with 70% humidity at room temperature for induction of a rapid drought treatment (Feng et al. 2015). For H2O2 stress, the roots of seedlings were dipped into solutions of 25 mM H2O2. For ABA treatment, soybean seedlings were sprayed with 100 μM ABA. For cold treatment, soybean seedlings were subjected to 4 °C. All stress treatments lasted from 0 to 12 h. Each treatment contained three independent replicates. At 0, 0.5, 5, and 12 h after each treatment, soybean seedlings were harvested, and five plants were collected as mixed samples at each time point, frozen in liquid nitrogen, and stored at − 80 °C until extraction of total RNA for qRT-PCR assays.

Expression Analysis of GmCBSDUFs

Total RNA was isolated from soybean tissues using TRIzol reagent (Invitrogen) and treated with DNase I (Invitrogen) to avoid genomic DNA contamination. First-strand cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen). Gene-specific primers were designed according to gene sequences using Primer 5.0 software (Table S1). The quantitative RT-PCR was performed with a CFX96TM real-time system (Bio-Rad) in a 20 μl system containing 2 μl of a tenfold diluted cDNA, 10 μl of 2 × SYBR green real-time PCR master mix (Takara), and 1 μl each of 10 μM forward and reverse primers. β-actin was used as the internal control. Statistical analyses were performed using the t-test, and p < 0.05 and < 0.01 were considered significant and extremely significant differences, respectively.

Vector Construction, Arabidopsis Transformation, and Stress Treatment

The full-length coding sequence (the primers 5′ ATGGCGGCAGAGATACCG 3′ and 5′ CTATTGATTCCTTAGTGACTCACT 3′.) of GmCBSDUF3 was TA cloned into the plant expression vector pCXSN. The recombinant construct containing the 35S::GmCBSDUF3 (Fig. S2A) cassette was introduced into Agrobacterium tumefaciens strain GV3101 and then transformed into Arabidopsis (Columbia) via the floral dip method. The transgenic plants were screened on MS medium with 100 mg/L hygromycin and confirmed by PCR analyses. The expression levels of GmCBSDUF3 in transgenic plants were determined by qPCR (Fig. S2B).

Seeds of transgenic overexpressing Arabidopsis and WT plants were grown on 10 × 10 cm MS agar plates. They were routinely kept for 2 days in darkness at 4 °C to break dormancy and transferred in a light growth chamber under a day/night 16/8 h cycle at 23 °C. For stress treatment, the seeds of transgenic lines or WT were kept on MS media supplemented with 50 mM NaCl, 2% PEG, or 1.5 μM ABA. Each treatment contained three independent replicates.