Magnesium transporter Gene Family: Genome-Wide Identiﬁcation and Characterization in Theobroma cacao , Corchorus capsularis , and Gossypium hirsutum of Family Malvaceae

: Magnesium (Mg) is an element involved in various key cellular processes in plants. Mg transporter (MGT) genes play an important role in magnesium distribution and ionic balance maintenance. Here, MGT family members were identiﬁed and characterized in three species of the plant family Malvaceae, Theobroma cacao , Corchorus capsularis , and Gossypium hirsutum , to improve our understanding of their structure, regulatory systems, functions, and possible interactions. We identiﬁed 18, 41, and 16 putative non-redundant MGT genes from the genome of T. cacao , G. hirsutum , and C. capsularis , respectively, which clustered into three groups the maximum likelihood tree. Several segmental/tandem duplication events were determined between MGT genes. MGTs appear to have evolved slowly under a purifying selection. Analysis of gene promoter regions showed that MGTs have a high potential to respond to biotic/abiotic stresses and hormones. The expression patterns of MGT genes revealed a possible role in response to P. megakarya fungi in T. cacao , whereas MGT genes showed differential expression in various tissues and response to several abiotic stresses, including cold, salt, drought, and heat stress in G. hirsutum . The co-expression network of MGTs indicated that genes involved in auxin-responsive lipid metabolism, cell wall organization, and photoprotection can interact with MGTs.


Introduction
Magnesium (Mg) is a critical bimetal that regulates biochemical processes and provides stability to membranes in plants [1,2]. Magnesium acts as a cofactor for polymerase, kinase, and H+-ATPase, which are necessary for synthesizing proteins and nucleic acid and for generating energy [3,4]. It is also required to maintain cation-anion homeostasis in the cell [5]. Various types of adverse effects have been reported in plants during Mg deficiency, including a reduction in photosynthesis, macromolecule synthesis, and plant growth and development [6][7][8]. Therefore, plants have developed an efficient transport system for absorption, storage, and Mg translocation [2]. The Mg transporter (MGT) gene family, also known as MRS2 or CorA, has an important role in the aforementioned essential functions [9,10]. Members of the MGT family are defined by two transmembrane domains in which a tripeptide motif GMN (glycine-methionine-asparagine) occurs at the C-terminal domain of the first transmembrane [11,12]. MGTs are expressed in root tissues The Best-fit model JTT+G4 was chosen according to Bayesian Information Criterion using jModelTest 2 [50] and was employed to find relationship among genes. Finally, the iTOL, Integrative Tree of Life version 4 [51], was used for the improvement of tree representation. The MEME (Multiple Em for Motif Elicitation) server [52] was used to identify the conserved protein motifs in MGT. The gene structure of each MGT gene was constructed using the Gene Structure Display Server [53].

Gene Duplications and Synteny Analysis
The MGT genes with more than 85% identity in each species were selected as duplicated genes [54]. Then, the location diagram of duplicated genes in cacao and G. hirsutum were constructed using TBtools [55]. In addition, the synonymous (Ks) and non-synonymous (Ka) of each duplicated gene pair were calculated by DnaSP version 6 software [56]. Finally, the time of divergence of duplicated genes was determined using the following equation: T = (Ks/2λ) × 10 −6 [57], where λ is substitutions per synonymous site per year and λ = 6.5 × 10 −9 . Moreover, the synteny relationship diagrams of MGT genes among the orthologous pairs of T. cacao-G. hirsutum and T. cacao-C. capsularis were created using Circos software [58].

Structure Analyses of MGT Proteins
The transmembrane and three-dimensional structures of the candidates of MGT subgroups MRS2, NIPA, and CorA proteins in T. cacao, G. hirsutum, and C. capsularis were predicted using the Phyre2 server [59], whereas docking analysis was performed to predict the ligand-binding regions (pocket sites) using DeepSite [60] and CASTp [61] tools and finally constructed in PyMOL [62].

Expression Analysis of TcMGTs and GhMGTs Using RNA-Seq Data
The publicly available RNA-seq data of cacao transcriptome with accession number GSE116041 [63] were used to find a possible differential expression of cacao MGTs concerning a fungi disease caused by Phytophthora megakarya after inoculation of the different time courses of 0 h, 6 h, 24 h, 48 h, and 72 h in two contrasting cultivars: Nanay (NA-32) as a susceptible cultivar and Scavina (SCA-6) as a fungal-resistant cultivar. Finally, the expression patterns of TcMGTs were illustrated in heatmaps based on log2 transformed using TBtools [55]. Furthermore, the expression profile of GhMGTs in various tissues (ovule, fiber, anther, bract, filament, leaf, pental, root, sepal, stem, and torus) and in response to various abiotic stresses (cold, heat, salt, and drought) was retrieved from available RNA-seq data of the cotton genome database (https://cottonfgd.org/, accessed on 9 March 2021) under project PRJNA490626 using the gene ID of each gene of the newly assembled genome as query [64]. The expression of GhMGTs in various tissues was analyzed and represented as a heatmap based on percentage expression of each gene using TBtools [55], while the data of abiotic stresses were analyzed and represented in heatmap after log2 transformation through TBtools [55].

Sequence and Structure of MGT Genes
Altogether 18, 41, and 16 putative non-redundant MGT genes were identified from the genomes of T. cacao, G. hirsutum, and C. capsularis, respectively. All sequences (genomic, amino acids, coding sequences) of identified MGT genes are shown in Table S1. MGTs were characterized based on their sequences structure (Table S2) and three MGT sub-groups, including MRS2, NIPA, and CorA, were recognized according to the specific domain distribution (Table S2). Our findings revealed that MGTs in the three studied plant species are diverse in sequence length, molecular weight (MW), isoelectric point (pI), and exon number (Table S2). For instance, protein length varied from 321 amino acids (aa) to 632 aa in T. cacao, from 210 aa to 474 aa in G. hirsutum, and from 262 aa to 2417 aa in C. capsularis (Table 1). In addition, the predicted MW ranged from 32.75 kDa to 70.91 kDa in T. cacao, from 32.66 kDa to 53.95 kDa in G. hirsutum, and from 29.82 kDa to 268.42 kDa in C. capsularis (Table 1). Moreover, the pI of MGTs was between 4.48 and 8.57 in T. cacao, between 4.76 and 9.57 in G. hirsutum, and between 4.79 and 8.60 in C. capsularis (Table 1). Based on pI value, 75% of MGTs in C. capsularis, 56% in T. cacao, and 49% in G. hirsutum were predicted to be acidophilic proteins (Table S2). In addition, the prediction of subcellular localization illustrated that most MGTs are located in the endomembrane or the plasma membrane (Table S2). The exon number of MGT genes varied between 4 and 15 in T. cacao and G. hirsutum, while the exon number varied between 4 and 21 in C. capsularis (Table 1).

Phylogenetic Analysis and Classification of the MGT Gene Family
A phylogenetic tree of MGT proteins was constructed, comprising 18 TcMGT proteins from T. cacao, 41 GhMGTs from G. hirsutum, and 16 CcMGTs from C. capsularis. The MGT proteins clustered into three groups (groups I, II, and III) that group III included three sub-groups, 3a, b, and c ( Figure 1). Six MGTs, including a CorA of Jute, namely CcMGT08, a MRS2 protein from cacao, namely TcMGT03, and four MRS2 proteins of cotton, namely GhMGT11, GhMGT12, GhMGT32, and GhMGT33, with similar structure contained 11 exons and were located in group I (Figure 1a,b). In addition, five MGT proteins were located in group II and four CorA proteins, namely TcMGT01, TcMGT16, CcMGT02, and CcMGT16, clustered in group III-a. In addition, 16 MRS2 proteins along with two CorA proteins were located in group III-b and all NIPA-type proteins were located in group III-c (Figure 1a,b). In addition, MGTs were analyzed based on distribution of conserved motifs in their protein sequence. Thirty-five conserved motifs were identified, and MGTs from group III showed more diversity than other groups ( Figure 2). Motifs 12, 5, 18, 7, 10, and 2, frequently observed in MGTs and NIPA proteins, illustrated different patterns of conserved motif distribution. Moreover, CorA proteins also showed various conserved motifs ( Figure 2).

Duplication Events and Synteny Analysis
The TcMGT genes were mapped onto 5 of 10 chromosomes in the cacao genome (Figure 3a), while GhMGTs were distributed over 22 of 26 chromosomes in the G. hirsutum genome (Figure 3b). Due to the incomplete physical map of the C. capsularis plant, the mapping of CcMGT genes is not provided. In the cacao genome, chromosome 2 encompassed the most significant number of TcMGTs, with five genes, while in G. hirsutum genome, the most significant number of GhMGTs were located on chromosomes A04 and D01 ( Figure 3). In addition, the duplication events of MGT genes in selected plant species were investigated. Eight segmental duplication gene pairs were identified between 12 TcMGT genes of cacao ( Figure 3a and Table S3). Tandem duplication events seemed to occur on chromosome 6 between three TcMGT genes of cacao, including, TcMGT013, TcMGT14, and TcMGT15 ( Figure 3a). In addition, four segmental duplication events were predicted for the TcMGT03 gene (Table S3). Notably, a duplication event occurred around 10 MYA between two CorA-like genes in cacao, namelyTcMGT01 and TcMGT16. In addition, five segmental duplication gene pairs were recognized between CcMGT genes, and a triplication event was predicted between CcMGT05 and CcMGT06, CcMGT07, and CcMGT04 (Table S3). According to the Ka/Ks ratio, the first duplication was approximately 103 million years ago (MYA) between CcMGT05 and CcMGT04. The most duplication events were observed between GhMGT genes in the G. hirsutum genome, with 22 segmental duplication gene pairs. Moreover, four GhMGT genes, namely GhMGT02, GhMGT05, GhMGT23, and GhMGT26, had a common ancestor and probably the first duplication event occurred approximately 68 MYA between GhMGT23 and GhMGT26 (Table S3). The intraspecies synteny of MGT genes was constructed between T. cacao and G. hirsutum and between T. cacao and C. capsularis ( Figure 4). The 18 TcMGT genes in T. cacao illustrated 15 and 16 syntenic block relationships with MGT genes in G. hirsutum and C. capsularis, respectively (Figure 4a,b). Interestingly, TcMGT genes of cacao showed more syntenic relationships with GhMGTs of D-genome than A-genome.

Duplication Events and Synteny Analysis
The TcMGT genes were mapped onto 5 of 10 chromosomes in the cacao genome ( Figure 3a), while GhMGTs were distributed over 22 of 26 chromosomes in the G. hirsutum genome (Figure 3b). Due to the incomplete physical map of the C. capsularis plant, the mapping of CcMGT genes is not provided. In the cacao genome, chromosome 2 encompassed the most significant number of TcMGTs, with five genes, while in G. hirsutum genome, the most significant number of GhMGTs were located on chromosomes A04 and D01 ( Figure 3). In addition, the duplication events of MGT genes in selected plant species were investigated. Eight segmental duplication gene pairs were identified between 12 TcMGT genes of cacao ( Figure 3a and Table S3). Tandem duplication events seemed to occur on chromosome 6 between three TcMGT genes of cacao, including, TcMGT013, TcMGT14, and TcMGT15 ( Figure 3a). In addition, four segmental duplication events were predicted for the TcMGT03 gene (Table S3). Notably, a duplication event occurred around 10 MYA between two CorAlike genes in cacao, namely TcMGT01 and TcMGT16. In addition, five segmental duplication gene pairs were recognized between CcMGT genes, and a triplication event was predicted between CcMGT05 and CcMGT06, CcMGT07, and CcMGT04 (Table S3). According to the Ka/Ks ratio, the first duplication was approximately 103 million years ago (MYA) between CcMGT05 and CcMGT04. The most duplication events were observed between GhMGT genes in the G. hirsutum genome, with 22 segmental duplication gene pairs. Moreover, four GhMGT genes, namely GhMGT02, GhMGT05, GhMGT23, and GhMGT26, had a common ancestor and probably the first duplication event occurred approximately 68 MYA between GhMGT23 and GhMGT26 (Table S3). The intraspecies synteny of MGT genes was constructed between T. cacao and G. hirsutum and between T. cacao and C. capsularis ( Figure 4). The 18 TcMGT genes in T. cacao illustrated 15 and 16 syntenic block relationships with MGT genes in G. hirsutum and C. capsularis, respectively (Figure 4a,b). Interestingly, TcMGT genes of cacao showed more syntenic relationships with GhMGTs of D-genome than A-genome.

Protein Structure and Docking Analysis
The 3D structures of all candidates of three types of MGTs, namely NIPA, MRS2, and CorA, were predicted in T. cacao, G. hirsutum, and C. capsularis ( Figure S1). Nine α-helices were observed in the predicted 3D structure of NIPAs in three studied plants, while fewer α-helices were predicted in the structure of MRS2 and CorA proteins ( Figure S1). Furthermore, nine transmembrane helices with eight pores were predicted in the structure of NIPAs in three plants, while in candidate MRS2 proteins two transmembrane helices were observed in all studied plant species. However, both N-terminal and C-terminal of candidate MRS2 proteins from T. cacao and C. capsularis were predicted in the extracellular part, while in G. hirsutum, both N-and C-terminal were observed in the cytoplasmic part. In the candidate CorA protein, three transmembrane helices were predicted in G.

Protein Structure and Docking Analysis
The 3D structures of all candidates of three types of MGTs, namely NIPA, MRS2, and CorA, were predicted in T. cacao, G. hirsutum, and C. capsularis ( Figure S1). Nine α-helices were observed in the predicted 3D structure of NIPAs in three studied plants, while fewer α-helices were predicted in the structure of MRS2 and CorA proteins ( Figure S1). Furthermore, nine transmembrane helices with eight pores were predicted in the structure of NIPAs in three plants, while in candidate MRS2 proteins two transmembrane helices were observed in all studied plant species. However, both N-terminal and C-terminal of candidate MRS2 proteins from T. cacao and C. capsularis were predicted in the extracellular part, while in G. hirsutum, both N-and C-terminal were observed in the cytoplasmic part. In the candidate CorA protein, three transmembrane helices were predicted in G.

Protein Structure and Docking Analysis
The 3D structures of all candidates of three types of MGTs, namely NIPA, MRS2, and CorA, were predicted in T. cacao, G. hirsutum, and C. capsularis ( Figure S1). Nine αhelices were observed in the predicted 3D structure of NIPAs in three studied plants, while fewer α-helices were predicted in the structure of MRS2 and CorA proteins ( Figure S1). Furthermore, nine transmembrane helices with eight pores were predicted in the structure of NIPAs in three plants, while in candidate MRS2 proteins two transmembrane helices were observed in all studied plant species. However, both N-terminal and C-terminal of candidate MRS2 proteins from T. cacao and C. capsularis were predicted in the extracellular part, while in G. hirsutum, both N-and C-terminal were observed in the cytoplasmic part.
In the candidate CorA protein, three transmembrane helices were predicted in G. hirsutum, and two transmembrane helices were predicted in T. cacao and C. capsularis ( Figure 5).
Moreover, pocket sites of MGT proteins related to the active binding site were predicted in structures of candidate proteins. The results illustrated that sub-groups of MGT proteins are different based on the residues present in predicted pocket sites ( Figure 6). Phenylalanine (PHE) amino acid was frequently observed in binding sites of NIPA proteins from T. cacao and G. hirsutum, while in C. capsularis isoleucine (ILE) and lysine (LYS) were more often observed in pocket sites. In candidate MRS2 proteins, proline (PRO), PHE, glutamine (GLN), asparagine (ASN), glutamic acid (GLU), and glycine (GLY) were frequently predicted in T. cacao as binding sites, while in G. hirsutum tyrosine (TYR), leucine (LEU), GLU, GLY, and GLN were more repeated in pocket sites. In addition, LEU residue was highly observed as a key binding site in candidate CcMGT of C. capsularis. In candidate CorA proteins, GLU amino acid was more often predicted in pocket sites of T. cacao and G. hirsutum, while PHE was frequently observed in pocket sites of candidate CorA protein in C. capsularis.

Distribution of Cis-Regulatory Elements in Promoter Region of MGT Genes
The promoter regions of MGTs in three plant species, comprising T. cacao, G. hirsutum, and C. capsularis, were analyzed and compared based on type and frequency of cis-regulatory elements. All recognized elements were classified into five groups: hormone-responsive elements, stress-responsive elements, light-responsive elements, growth-responsive elements, and binding sites of transcription factors. Our results revealed that MGT' promoters contain cis-regulatory elements related to stress response ( Figure 7 and Table S4). In addition, all stress-responsive elements were grouped in six classes related to drought, wounding, anaerobic, low temperature, biotic stress, and

Distribution of Cis-Regulatory Elements in Promoter Region of MGT Genes
The promoter regions of MGTs in three plant species, comprising T. cacao, G. hirsutum, and C. capsularis, were analyzed and compared based on type and frequency of cis-regulatory elements. All recognized elements were classified into five groups: hormone-responsive elements, stress-responsive elements, light-responsive elements, growth-responsive elements, and binding sites of transcription factors. Our results revealed that MGT' promoters contain cis-regulatory elements related to stress response (Figure 7 and Table S4). In addition, all stress-responsive elements were grouped in six classes related to drought, wounding, anaerobic, low temperature, biotic stress, and general stresses. However, cis-regulatory elements responsive to low temperature, anaerobic stresses, and biotic stresses were observed most often in the promoter region of TcMGT genes (Figure 7b). In addition, regulatory elements related to response to abscisic acid (ABA), salicylic acid (SA), auxin, gibberellin (GA), and methyl jasmonate (MeJA) were observed among hormone-responsive elements (Figure 7b,f,i). We found that MGT genes might be more frequently induced in response to ABA hormone.
Agronomy 2021, 11, x FOR PEER REVIEW 10 of 18 general stresses. However, cis-regulatory elements responsive to low temperature, anaerobic stresses, and biotic stresses were observed most often in the promoter region of TcMGT genes (Figure 7b). In addition, regulatory elements related to response to abscisic acid (ABA), salicylic acid (SA), auxin, gibberellin (GA), and methyl jasmonate (MeJA) were observed among hormone-responsive elements (Figure 7b,f,i). We found that MGT genes might be more frequently induced in response to ABA hormone.

Expression Profile of TcMGT Genes
The expression levels of TcMGTs were also provided in response to P. megakarya after 0 h, 6 h, 24 h, and 72 h after infection using available RNA-seq data of two contrasting genotypes: T. cacao Nanay (fungal-susceptible cultivar) and Scavina (fungal-resistant cultivar) (Figure 8a,b). According to expression heatmaps, most TcMGT genes were less induced by a fungal infection, P. megakarya. A NIPA gene, TcMGT02, showed an upregulation after 48 h of fungal infection in both cultivars (Figure 8). In addition, two MRS2 genes, TcMGT18 and TcMGT12, were more expressed after 72 h in the fungalresistant cultivar (Figure 8b).

Expression Profile of TcMGT Genes
The expression levels of TcMGTs were also provided in response to P. megakarya after 0 h, 6 h, 24 h, and 72 h after infection using available RNA-seq data of two contrasting genotypes: T. cacao Nanay (fungal-susceptible cultivar) and Scavina (fungal-resistant cultivar) (Figure 8a,b). According to expression heatmaps, most TcMGT genes were less induced by a fungal infection, P. megakarya. A NIPA gene, TcMGT02, showed an upregulation after 48 h of fungal infection in both cultivars (Figure 8). In addition, two MRS2 genes, TcMGT18 and TcMGT12, were more expressed after 72 h in the fungal-resistant cultivar (Figure 8b).

Expression Profile of GhMGT Genes
The expression profile of GhMGT genes was investigated in different tissues of G. hirsutum and under abiotic stresses, including cold, heat, drought, and salinity ( Figure 9). The results illustrated that GhMGTs are involved in early and late responses to abiotic stresses (Figure 9a). For instance, a MRS2 gene, GhMGT32, showed an upregulation in response to the temperature stresses of cold and heat after one hour (Figure 9a). Furthermore, GhMGT24, as a NIPA gene, was more upregulated in response to all studied abiotic stresses after 6 and 12 h. In addition, the expression profile of GhMGT genes showed that six MRS2 genes, i.e., GhMGT12, GhMGT33, GhMGT06, GhMGT41, GhMGT13, and GhMGT20, are more expressed after 24 h of drought and salt stress. In the first hours of heat stress, NIPA genes are more expressed than MRS2 genes in cotton. In addition, CorA-like genes of GhMGT35 and GhMGT14 showed more upregulation in response to drought and salt stress. Expression levels of GhMGTs were also evaluated in different tissues and organs of G. hirsutum (Figure 9b). The results show that GhMGTs were expressed in different organs for the proper distribution of magnesium throughout the cotton plant. In root tissues, NIPA genes are more expressed, while in leaf and torus tissues, MRS2 genes are more expressed (Figure 9b). Furthermore, two CorA-like genes showed high expression in filament tissues. In addition, most GhMGTs are expressed in ovule tissues in 10 days post-anthesis (DPA) (Figure 9b).

Expression Profile of GhMGT Genes
The expression profile of GhMGT genes was investigated in different tissues of G. hirsutum and under abiotic stresses, including cold, heat, drought, and salinity ( Figure 9). The results illustrated that GhMGTs are involved in early and late responses to abiotic stresses (Figure 9a). For instance, a MRS2 gene, GhMGT32, showed an upregulation in response to the temperature stresses of cold and heat after one hour (Figure 9a). Furthermore, GhMGT24, as a NIPA gene, was more upregulated in response to all studied abiotic stresses after 6 and 12 h. In addition, the expression profile of GhMGT genes showed that six MRS2 genes, i.e., GhMGT12, GhMGT33, GhMGT06, GhMGT41, GhMGT13, and GhMGT20, are more expressed after 24 h of drought and salt stress. In the first hours of heat stress, NIPA genes are more expressed than MRS2 genes in cotton. In addition, CorAlike genes of GhMGT35 and GhMGT14 showed more upregulation in response to drought and salt stress. Expression levels of GhMGTs were also evaluated in different tissues and organs of G. hirsutum (Figure 9b). The results show that GhMGTs were expressed in different organs for the proper distribution of magnesium throughout the cotton plant. In root tissues, NIPA genes are more expressed, while in leaf and torus tissues, MRS2 genes are more expressed (Figure 9b). Furthermore, two CorA-like genes showed high expression in filament tissues. In addition, most GhMGTs are expressed in ovule tissues in 10 days post-anthesis (DPA) (Figure 9b).

Co-Expression Network of MGT Genes
A co-expression network of TcMGTs was formed using their orthologs in the diploid model plant, Arabidopsis ( Figure 10). All nodes in the co-expression network of MGTs were classified into four groups (I to IV) ( Figure 10). In group I, three auxin-responsive genes, namely SAUR4, 7, and 12, two genes affecting lipid metabolism, PLC genes, two genes involving in disease resistance-responsive, namely AT4G38700 and AT2G21110, and CYP21-1 affecting protein folding showed high co-expression scores with MGT genes. In group II, a protein phosphatase 2C, AT2G20050, was found with high co-expression with three MGT genes. In group III, MGT genes showed high co-expression with AT5G42070, a hypothetical protein, RPL15 (ribosomal protein), AT2G23390 (acyl-CoA protein), and CH1. In group IV, high co-expression connections were observed between MGT genes and genes involving in cell wall organization/modification processes such as S2LB, AGM1, AGM2, EXPA13, GXM2, GXM1, and QRT1. In addition, LrgB involved in response to water deprivation and MHX encoding a mg/proton exchanger showed string coexpression with MGTs. Gene ontology (GO) analysis illustrated that the biological processes, including mg ion transporter, photoprotection, and cell wall pectin biosynthetic process, xylan biosynthetic process, and chloroplast organization, were significantly enriched based on all nodes of co-expression networks of MGTs.

Co-Expression Network of MGT Genes
A co-expression network of TcMGTs was formed using their orthologs in the diploid model plant, Arabidopsis ( Figure 10). All nodes in the co-expression network of MGTs were classified into four groups (I to IV) ( Figure 10). In group I, three auxin-responsive genes, namely SAUR4, 7, and 12, two genes affecting lipid metabolism, PLC genes, two genes involving in disease resistance-responsive, namely AT4G38700 and AT2G21110, and CYP21-1 affecting protein folding showed high co-expression scores with MGT genes. In group II, a protein phosphatase 2C, AT2G20050, was found with high co-expression with three MGT genes. In group III, MGT genes showed high co-expression with AT5G42070, a hypothetical protein, RPL15 (ribosomal protein), AT2G23390 (acyl-CoA protein), and CH1. In group IV, high co-expression connections were observed between MGT genes and genes involving in cell wall organization/modification processes such as S2LB, AGM1, AGM2, EXPA13, GXM2, GXM1, and QRT1. In addition, LrgB involved in response to water deprivation and MHX encoding a mg/proton exchanger showed string co-expression with MGTs. Gene ontology (GO) analysis illustrated that the biological processes, including mg ion transporter, photoprotection, and cell wall pectin biosynthetic process, xylan biosynthetic process, and chloroplast organization, were significantly enriched based on all nodes of co-expression networks of MGTs.

Discussion
Cocoa (Theobroma cacao L.) is an economical plant due to its wide use in producing chocolate, a popular commodity worldwide. Scientific research is underway to better understand the genome and metabolomes of this plant. In addition, several gene families, including GASA [40], sucrose transporter [65], WRKY [66], NAC [67], desaturase [68], and sucrose synthase [69], have been investigated and characterized in cacao. Due to the important role of magnesium ions in regulating plant growth and development, magnesium transporters (MGTs) have been investigated in cacao, upland cotton (Gossypium hirsutum), and white jute (Corchorus capsularis). We characterized 18 putative non-redundant MGT genes in cacao (TcMGTs) along with 41 MGTs in G. hirsutum (GhMGTs), and 16 MGTs from the genome of C. capsularis (CcMGTs). In previous studies, 62 MGTs in Camelina sativa [12], 41 MGTs in Triticum turgidum [12], 12 MGTs in Zea mays [70], 16 MGTs in Pyrus bretschneideri [23], 36 MGTs in Brassica napus [25], 12 MGTs in Fagaria vesca [27], and 8 MGTs in Poncirus trifoliata [24] were characterized. The number of MGTs is probably correlated with polyploidy events and genome size [12,71]. The prediction of the pI value of MGT proteins illustrated that CcMGTs are more acidophilic proteins than TcMGTs and GhMGTs, indicating that CcMGTs are mostly active under acidic conditions (pI < 6.50). This can be related to the optimal growing environment of white jute.
According to the phylogenetic analysis, MGT family members from T. cacao, G. hirsutum, and C. capsularis can be classified into three groups, and CcMGTs showed close

Discussion
Cocoa (Theobroma cacao L.) is an economical plant due to its wide use in producing chocolate, a popular commodity worldwide. Scientific research is underway to better understand the genome and metabolomes of this plant. In addition, several gene families, including GASA [40], sucrose transporter [65], WRKY [66], NAC [67], desaturase [68], and sucrose synthase [69], have been investigated and characterized in cacao. Due to the important role of magnesium ions in regulating plant growth and development, magnesium transporters (MGTs) have been investigated in cacao, upland cotton (Gossypium hirsutum), and white jute (Corchorus capsularis). We characterized 18 putative non-redundant MGT genes in cacao (TcMGTs) along with 41 MGTs in G. hirsutum (GhMGTs), and 16 MGTs from the genome of C. capsularis (CcMGTs). In previous studies, 62 MGTs in Camelina sativa [12], 41 MGTs in Triticum turgidum [12], 12 MGTs in Zea mays [70], 16 MGTs in Pyrus bretschneideri [23], 36 MGTs in Brassica napus [25], 12 MGTs in Fagaria vesca [27], and 8 MGTs in Poncirus trifoliata [24] were characterized. The number of MGTs is probably correlated with polyploidy events and genome size [12,71]. The prediction of the pI value of MGT proteins illustrated that CcMGTs are more acidophilic proteins than TcMGTs and GhMGTs, indicating that CcMGTs are mostly active under acidic conditions (pI < 6.50). This can be related to the optimal growing environment of white jute.
According to the phylogenetic analysis, MGT family members from T. cacao, G. hirsutum, and C. capsularis can be classified into three groups, and CcMGTs showed close relationships to TcMGTs. However, the MGT family proteins of Arabidopsis and rice were divided into five clusters based on phylogenetic analysis [11,72]. In the current study, MRS2 sub-group proteins showed more diversity than the other two sub-groups, indicating that NIPA and CorA genes may be derived from MRS2 genes during the process of evolution. In addition, more MGT genes were identified in cotton, which is due to polyploidy in G. hirsutum. The analysis of gene structure showed that TcMGTs and GhMGTs contained 4 to 15 exons, while CcMGTs contained 4 to 21 exons. This suggests that under evolution events more insertions and deletions of introns occurred in MGT genes, especially in CcMGTs. In addition, the first duplication event was estimated to occur approximately 103 MYA between CcMGT05, as a MRS2 gene, and CcMGT04, as a NIPA gene.
Several segmental/tandem duplication events were estimated to occur between TcMGTs. Notably, a gene cluster of three MRS2/TcMGT genes, namely TcMGT013, TcMGT14, and TcMGT15, was observed in chromosome 6, providing an avenue for further molecular investigations in cacao. In addition, the K a /K s ratios in most duplicated genes were less than one, suggesting that MGTs have evolved slowly under a purifying selection [54,71]. The comparisons between structures of MGT proteins revealed that NIPA proteins include the conserved structures with more transmembrane regions. More transmembrane regions could indicate a more important role for the NIPA protein group in transport of magnesium within the plant cell [12]. However, the transmembrane structure of MSR2 and CorA proteins in T. cacao and C. capsularis is highly conserved, while these proteins possess a different transmembrane structure in G. hirsutum, which may affect the ability of the magnesium transmembrane transport process [73,74]. Moreover, sub-groups of MGTs in the three studied plants showed diversity in the predicted pocket sites in the 3D structure. Overall, our findings suggest that PHE, GLU, LEU, GLY, and ILE, as the key binding sites, are associated with the function and interaction of MGTs in response to environmental stimuli and changes in ion/Mg concentration [54,75,76].
Magnesium transporters as well as other ion transporters are not only involved in response to Mg concentrations, but also their activity can be affected by changes in environmental conditions [10,12]. Analysis of gene promoter regions is one of the strategies to predict the response of target genes to various environmental factors [77,78]. MGT genes have a high potential to respond to stresses, both biotic and abiotic, as well as ABA based on the distribution of corresponding cis-regulatory elements in the promoter region. However, investigations that are more molecular are needed to confirm their functions. The black rod disease caused by the genus Phytophthora is a production limiting factor, reducing cacao production by around 20-25% [79]. On the other hand, expression studies of genes in beans of bulk cultivars, like disease resistance and fine flavor cocoa (disease susceptibility), may be of interest [80,81]. In the current study, the expression level of TcMGT genes was investigated using RNA sequencing data responses to P. megakarya in two contrasting cultivars of T. cacao, susceptible and fungal-resistant cultivars. Two TcMGTs, i.e., TcMGT12 and TcMGT18, were identified as P. megakarya responsiveness genes by their upregulation specifically in the cacao-tolerant cultivar. These may be a good target for further molecular studies related to introducing the new cocao-resistant cultivars with high-quality, delicious chocolate. Furthermore, the expression profile of GhMGTs suggested that MGT genes are involved in response to abiotic stresses such as temperature stresses (cold and heat stress), drought, and salinity stress. Nevertheless, GhMGTs can be expressed in different plant tissues to regulate Mg homeostasis. Previous studies have indicated that MGT genes are associated with maintaining ion homeostasis in plant tissues during adverse conditions [20,21].
Mg transporter genes via affecting Na + transporters and K + transporters (HKTs) can improve salinity tolerance in plant species [82]. In addition, MGTs have been speculated to regulate the downstream pathways related to response to abiotic stresses by interacting with Ca 2+ sensors [83]. Our findings revealed that the MGT duplicated gene pair could have diverse expression patterns, suggesting that these genes probably under some modifications or insertion/deletion in their sequence, CDS, or promoter regions have received novel functions [71,84]. Modifications, such as gains and losses of cis elements in promoters between duplicated gene pairs, e.g., parent and daughter genes, could occur after duplication events, affecting the expression levels [85,86]. By constructing a network of co-expression genes, it is possible to identify other molecular pathways in which target genes are involved to gain a better understanding of the function of genes [54,87]. In this study, MGT genes showed diverse co-expressions with genes involved in auxin-responsive processes, lipid metabolism, cell wall organization, photoprotection, and chloroplast organization. Magnesium is a critical element of chlorophyll, affecting photosynthesis rate and biomass production [88,89]. Overall, MGTs appear to be involved in various pathways to control plant growth and development and response to adverse conditions.

Conclusions
A genome-wide analysis of MGT family genes was performed in the genomes of Theobroma cacao, Corchorus capsularis, and Gossypium hirsutum. Our findings provide insight into certain aspects of the sequence structure, evolutionary events, regulatory systems, and function of MGT genes in three species of Malvaceae. Furthermore, our results show that MGTs are involved in diverse cellular pathways, and they can interact with proteins associated with growth and development as well as with response to environmental stimuli. Further functional-molecular analyses are required to improve our understanding of the role of MGTs in cacao resistance to stress and to increase the quality of chocolate.