Putative Medicago GLYI, GLYII and unique GLYIII (DJ-1) proteins were identified by BLASTP search at JCVI M. truncatula annotation database 4.0v1¹ (Young et al., 2011) with an e-value of 1 using previously reported Arabidopsis GLYI (GenBank: AEE28797.1), Brassica GLYII (GenBank: AAO26580.1) and rice DJ-1C (LOC_Os04g57590) protein sequence as query, respectively. Each of the identified sequences was used as secondary query subsequently until finding new members. Protein sequences of all BLAST hits were retrieved from the database and analyzed using the Hidden Markov Model (HMM) profile of Pfam² with e-value of 1. The presence of glyoxalase domain (PF00903) or Metallo-beta-lactamase domain (PF00753) or DJ-1/PfpI domain (PF01965) confirmed the identity of the proteins as GLYI or GLYII or DJ-1, respectively. Identified proteins were nomenclature as prefix “Mt” for M. truncatula, followed by GLYI or GLYII or DJ-1. They have numbered arabically for GLYI and GLYII, or alphabetically for DJ-1 depending on their chromosomal position as mentioned in previous studies (Ghosh and Islam, 2016; Ghosh et al., 2016). Detailed information about the identified genes and corresponding proteins was retrieved using locus search option of M. truncatula annotation database 4.0v1¹ (Young et al., 2011). Sub-cellular localization of all these identified proteins was predicted using CELLO v.2.5: sub-cellular localization predictor³ (Yu et al., 2006), Wolf PSORT⁴ (Horton et al., 2007) and ChloroP⁵ (Emanuelsson et al., 1999) tools with default parameters (Tables 1–3).

Chromosomal location for all the newly identified glyoxalase genes was pointed using genome browser of M. truncatula database 4.0v1¹ (Young et al., 2011) (Figure 2). Duplication events between the Medicago glyoxalase genes with that of Medicago, Arabidopsis, rice, and soybean were identified using plant genome duplication database⁶ analysis (Lee et al., 2013). The rate of synonymous (K_s) and non-synonymous substitution (K_a) was retrieved from plant genome duplication database, too (Table 4).

Phylogenetic relationship among glyoxalase proteins from Arabidopsis, rice, soybean, and Medicago were investigated using MEGA 7.0 (Kumar et al., 2016) with the Neighbor-Joining method and 1000 bootstrap replicates (Figures 3, 8A). Multiple sequence alignment was performed using ClustalW (Larkin et al., 2007) (Figures 5, 7, 8C). Protein sequences used in the phylogenetic analysis are provided as Additional Files S1, S2, S3 GLYI, GLYII and DJ-1 proteins, respectively.

All the predicted MtGLYI (35), MtGLYII (27), and MtDJ-1 (6) proteins were analyzed using Pfam² (Finn et al., 2016) to confirm the presence of their respective conserved domains and their exact positions. The position of glyoxalase domain (PF00903) or Metallo-beta-lactamase domain (PF00753) or DJ-1/PfpI domain (PF01965) were identified for all GLYI or GLYII or DJ-1 proteins and the relative figure was generated (Figures 4, 6, 8B). The presence of some additional domains along with the conserved domain was marked in the corresponding figure at an approximate position.

Genome-wide microarray data of conventional and unique glyoxalase genes were obtained from the M. truncatula gene expression Atlas (MtGEA) Project database⁷ (Benedito et al., 2008). The corresponding probesets for all MtGLYI, MtGLYII, and MtDJ-1 genes were identified using NetAffx Analysis Center online Probe Match tool⁸ with default parameters based on respective CDS sequence as query. Genes with more than one probe, probe with the highest expression value was considered for analysis; and single probe for many genes was considered as the same transcriptional profile. Normalized transcript data was obtained for 17 different tissues, including underground tissues-root and nodule; aerial tissues- stem, vegetative bud, petiole, leaf, flower, pod; seed development (seed of 10, 12, 16, 20, 24, and 36 days); and nod development (nod of 4, 10, and 14 days). These normalized expression data was used to generate heatmap (Figure 9) and hierarchically clustering based on the Manhattan correlation with average linkage using the Institute for Genomic Research MeV software package⁹ (Saeed et al., 2006).

Normalized and curated perturbation expression data of Medicago glyoxalase genes were retrieved from Genevestigator Affymetrix Medicago genome array¹⁰ (Hruz et al., 2008; Zimmermann et al., 2008) using the already identified corresponding probe. The log₂ fold change ratio was retrieved using Genevestigator with default parameters using experiments id: MT-00011: Effects of salt stress on M. truncatula seedlings (GSE14029)¹¹ and MT-00013: Global reprogramming of transcription and metabolism in M. truncatula during progressive drought and after rewatering (E-MTAB-2681) (Zhang et al., 2014). Detailed informations about the gene expression experimental design is provided as Additional File S4. These experiments were performed with minimum three replicates and fold change values were calculated in comparison to the untreated control 0 h seedlings in case of salinity and an untreated 2 days control sample of respective tissue in the case of drought. Generation of heat maps (Figure 10) and hierarchical clustering with Manhattan distance metric method was performed using Institute for Genomic Research MeV software package⁹ (Saeed et al., 2006).

Glyoxalase proteins have been well characterized from various plant as well as animal species. The genome-wide analysis identified a total of 35 unique Medicago GLYI proteins encoded by 29 GLYI genes. These genes were found to be located on different chromosomes of Medicago (Figure 2). They were nomenclature as MtGLYI-1 to MtGLYI-29 according to the order of their chromosomal location (Table 1). The number of Medicago GLYI genes was found to be greater than previously reported Arabidopsis (10), rice (11), and soybean (24) GLYI genes (Mustafiz et al., 2011; Ghosh and Islam, 2016). Similarly, a total of 27 Medicago GLYII proteins encoded by 14 GLYII genes were identified. Both the number of GLYII genes and proteins were found to be greater than previously reported Arabidopsis (5 genes and 9 proteins), rice (3 genes and 4 proteins), and soybean (12 genes and 23 proteins) (Mustafiz et al., 2011; Ghosh and Islam, 2016). They were symbolized as MtGLYII-1 to MtGLYII-14 like MtGLYI genes (Table 2).

In the same way, genome-wide searching of unique GLYIII proteins in Medicago identified six DJ-1 proteins encoded by five genes. They were named as MtDJ-1A to MtDJ-1E like previously reported members from other species (Ghosh et al., 2016) (Table 3). Although the number of genes was almost similar, but the number of MtDJ-1 proteins was less than previously reported Arabidopsis (6 genes and 11 proteins), rice (6 genes and 12 proteins) (Ghosh et al., 2016) and soybean (7 genes and 11 proteins) (unpublished data) counterparts. In all cases of MtGLYI, MtGLYII, and MtDJ-1 families, proteins number was found to be greater than that of genes (Tables 1–3), that indicate the presence of alternate splicing event in Medicago glyoxalase transcripts.

All the newly identified glyoxalase members were analyzed further in terms of the length of transcripts, coding DNA sequence (CDS), and polypeptide; molecular weight, isoelectric point and sub-cellular localization of the corresponding protein. CDS length of MtGLYI members was found to be fluctuating from 258 bp (MtGLYI-26) to 2466 bp (MtGLYI-7) with an average of 1533 bp. Thus, the largest member of MtGLYI protein family is MtGLYI-7 with a polypeptide length of 821 aa and molecular weight of 92.26 kDa. Subsequently, the smallest member is MtGLYI-26 with 85 aa in length and 9.63 kDa in molecular weight (Table 1). MtGLYI protein family members showed an average length of 510 aa and predicted molecular weight of 57.04 kDa. Isoelectric point (pI) is an important parameter of protein that determined the net charge of a protein under certain physiological conditions. Members of MtGLYI family showed the broad difference in their pI-value ranging from 4.69 (MtGLYI-2) to 9.62 (MtGLYI-29). Out of 35 MtGLYI proteins, 24 of them showed a pI-value less than 7.0 (neutral pH) and the rest 11 showed more than 7.0. Thus both positively and negatively charged MtGLYI proteins co-exist in the system under particular physiological condition. MtGLYI proteins were predicted to be localized at cytosol, chloroplast, mitochondria, nucleus, and extracellular spaces. The majority of MtGLYI family members are found to be localized in the cytosol. Chloroplast localization of MtGLYI-21, MtGLYI-24.1, and MtGLYI-24.2 were confirmed by all three tools namely CELLO, Wolf PSORT, and ChloroP. Localization signal for mitochondria, nucleus or extra-cellular matrix were found to be predicted by either CELLO or Wolf PSORT (Table 1).

The length of MtGLYII CDS varied from 723 bp (MtGLYII-6.2) to 2853 bp (GmGLYII-8.1) with an average of 2579 bp (Table 2). The largest member (MtGLYII-8.1) has a polypeptide length of 950 aa with a molecular weight of 104.98 kDa; whereas MtGLYII-6.2, the smallest member has polypeptide length and molecular weight of 240 aa and 26.47 kDa, respectively (Table 2). In terms of pI, MtGLYII-2 and MtGLYII-6.2 showed the most acidic (4.82) and basic (8.93) value, respectively. Like MtGLYI members, the majority of MtGLYII proteins (17 out of 27) showed acidic pI-value, while rest 10 MtGLYII members showed basic pI-value (Table 2). Similar to MtGLYI, most of the MtGLYII proteins showed cytosolic localization. Chloroplast localization of MtGLYII-9.1 and MtGLYII-13.1 was confirmed by all three tools-CELLO, Wolf PSORT, and ChloroP; and mitochondrial localization of MtGLYII-6.1 was confirmed by both CELLO and Wolf pSORT. Nucleus and extracellular localization of few members were predicted by either CELLO or Wolf pSORT.

Like conventional glyoxalase enzymes (GLYI + GLYII), different physio-chemical properties of unique glyoxalase III proteins were analyzed. The length of MtDJ-1 CDS was varied from 279 bp (MtDJ-1B) to 1356 bp (MtDJ-1E.1) with an average of 898 bp (Table 3). Consequently, the smallest MtDJ-1 member (MtDJ-1B) is 91 aa in length with a molecular weight of 10.43 kDa; while the largest one is MtDJ-1E.1 with a length of 450 aa and molecular weight of 48.20 kDa (Table 3). Isoelectric value (pI) varies from 5.69 (MtDJ-1D) to 9.42 (MtDJ-1E.1); with four acidic and two basic members at neutral pH. Chloroplast localization of MtDJ-1A, MtDJ-1E.1, and MtDJ-1E.2 was confirmed by all three tools-CELLO, Wolf PSORTand ChloroP; followed by cytosolic localization of MtDJ-1C and MtDJ-1D, and mitochondrial localization of MtDJ-1E.2 was confirmed by both CELLO and Wolf pSORT tools. Localization of extracellular matrix and plasma membrane of few members were only predicted by either CELLO or Wolf pSORT tool (Table 3).

All glyoxalase genes were found to be distributed randomly in seven different chromosomes of Medicago out of total eight chromosomes (Figure 2). A detailed chromosome map was constructed to exactly point out the newly identified MtGLYI, MtGLYII, and MtDJ-1 genes on different Medicago chromosomes. Five of MtGLYI (MtGLYI-25 to MtGLYI-29) and one of the MtGLYII genes (MtGLYII-14) were present in scaffold region and thus unable to point out them in chromosomes. Rest 24 MtGLYI genes were found to be unevenly distributed in seven Medicago chromosomes. Chromosome 4 contains a maximum number of eight GLYI genes, followed by chromosome 5 with six and chromosome 2 with four members. However, chromosomes 1 and 8 have two GLYI genes each, and only one GLYI gene each is present in chromosomes 3 and 6 (Figure 2). Similarly, 13 MtGLYII genes were found to be located on six different chromosomes (Figure 2). The gene density per chromosome is highly uneven, where chromosome 2 contains a maximum of five GLYII genes, followed by three genes in chromosome 1 and two genes in chromosome 4. However, chromosomes 3, 5, and 8 have only one GLYII gene each (Figure 2). Out of 5 MtDJ-1 genes, maximum three genes present in chromosome 3 and one member each in chromosomes 2 and 5. All these genes were found to be distributed in a highly uneven fashion in terms of chromosome number and position. Chromosome 4 was found to be the highly dense with 10 genes in total; while chromosomes 2, 3, and 5 contain at least one member of each gene family (Figure 2).

Gene duplication and divergence are two most important phenomenon of plant evolution. Two major angiosperms (eudicots and monocots) are calculated to diverge from 125–140 Mya to 170–235 Mya ago (Lee et al., 2013). This largely results in subsequent chromosomal duplication and gain/loss of genes. To identify glyoxalase gene duplication among four plants (Arabidopsis, Rice, Soybean, and Medicago) genome, plant genome duplication database¹² were analyzed (Lee et al., 2013) and the data is presented as Table 4. In this study, five MtGLYI and one MtGLYII sister-gene pairs have been observed (Table 4). The gene pairs are MtGLYI-4/MtGLYI-14, MtGLYI-5/MtGLYI-8, MtGLYI-7/MtGLYI-22, MtGLYI-12/MtGLYI-8, MtGLYI-19/MtGLYI-23, and MtGLYII-7/MtGLYII-10. Moreover, orthologous of MtGLY genes were found in three other well-characterized plant genomes of Arabidopsis, Soybean, and Rice. Non-synonymous to synonymous substitutions (K_a/K_s) ratio indicates the evolutionary history and direction of selection (Li et al., 1981; Chen et al., 2014). A pair of genes with K_a/K_s < 1 indicates purifying selection, K_a/K_s = 1 implies neutral drifting, and lastly K_a/K_s > 1 implies positive or Darwinian selection (Juretic et al., 2005; Chen et al., 2014). K_a/K_s ratio of all gene pairs was found to be below 1 indicating the influence of purifying selection in the evolution of glyoxalase genes among Arabidopsis, rice, soybean, and Medicago (Table 4).

Glyoxalase genes from Arabidopsis, rice, soybean, and Medicago were analyzed further for their sequence similarities and evolutionary divergence through phylogenetic analysis. Three major classes GLYI family members were identified in the phylogenetic tree (Figure 3A). Based on various previous studies, these classes may belong to Ni²⁺-dependent GLYI, Zn²⁺-dependent GLYI, and functionally diverse inactive GLYI. Clade-I had the largest GLYI members from different plant species, while clade-II had the lowest number of members only from Arabidopsis and rice genome (Figure 3A). Clade-I contains three rice GLYI members OsGLYI-2, OsGLYI-7, and OsGLYI-11; and two members of Arabidopsis GLYI family AtGLYI-3 and AtGLYI-6 those have been considered as Ni²⁺-dependent GLYI enzyme (Kaur et al., 2013; Mustafiz et al., 2014). Thus, members of this clade are supposed to show Ni²⁺-dependent GLYI activity. Similarly, clade-II has previously reported Zn²⁺-dependent rice (OsGLYI-8) and Arabidopsis (AtGLYI-2) member which indicated to be Zn²⁺-dependent GLYI clusters. This confirmed the dominance of Ni²⁺-dependent GLYI over Zn²⁺-dependent isoform in plant species (Figure 3A) which had been recently proved in Arabidopsis, too (Jain et al., 2016). Members of clade-III are considered as functionally diverged GLYI members as it contains OsGLYI-10, which did not possess conventional GLYI activity (unpublished data).

Similarly, phylogenetic analysis of GLYII proteins could be divided into three major classes (I–III) (Figure 3B). One of rice GLYII protein (OsGLYII-1) and one Arabidopsis GLYII protein (AtGLYII-3) was previously reported to possess sulfur dioxygenase (SDO) activity (Kaur et al., 2014b). Both of them were present in the clade-III, representing the functional divergence members of GLYII family. As AtGLYII-2 was found to be localized in mitochondria (Marasinghe et al., 2005), other members of clade II (one from rice, OsGLYII-2), four proteins from soybean and one from Medicago, MtGLYII-12) could be mitochondrial localized. Lastly, Clade-I might contain all the active and cytoplasmic localized GLYII members from four plant species.

The presence of conserved glyoxalase domain (PF00903) in all the predicted 35 MtGLYI proteins was analyzed using Pfam. Analyses revealed that all MtGLYI protein has at least one GLYI domain, while some of them have two. Out of 35, nine MtGLYI proteins contained two consecutive GLYI domains (Figure 4). Single GLYI protein with two domains had been reported from G. max (Ghosh and Islam, 2016), S. cerevisiae (Frickel et al., 2001), O. sativa (Mustafiz et al., 2014), and Plasmodium falciparum (Deponte et al., 2007). Apart from conserved GLYI domain, some proteins have addition domains, such as WD domain (PF00400), Alpha/beta hydrolase family (PF12697) and Enoyl-CoA hydratase/isomerase (PF00378). The role of these domains in association with GLYI proteins is still not clear. GLYI domain (only N-terminal one in a case of two domain containing members) sequences of all MtGLYI proteins were aligned with that of Ni²⁺-dependent OsGLYI-11.2 and Zn²⁺-dependent OsGLYI-8 (Kaur et al., 2013) proteins to predict on their metal ion dependency and catalytic activity (Figure 5). Enzymatic activity of GLYI proteins has been found to be dependent on metal ions (Kaur et al., 2013). This dependency could be easily predicted based on the size of GLYI domain; domain length of ∼120 aa showed Ni²⁺-dependence and length of 142 aa showed Zn²⁺-dependence (Kaur et al., 2013). According to the alignment, Zn²⁺-dependent OsGLYI-8 has an extra stretch of 12 aa (Figure 5, marked with rectangle), that is completely absent in Ni²⁺-dependent OsGLYI-11.2 proteins. Proteins with this stretch of amino acids such as MtGLYI-10.1 and MtGLYI-10.2 are found to be Zn²⁺ dependent, while proteins like MtGLYI-4, MtGLYI-7, MtGLYI-9, MtGLYI-11, MtGLYI-14, MtGLYI-15, MtGLYI-18, MtGLYI-22, and MtGLYI-24 might be Ni²⁺ dependent. Some of the proteins possess partial regions, thus difficult to predict (Figure 5).

The active site of GLYI proteins had a conserved motif of H/QEH/QE irrespective of their metal ion dependency. These four residues were found to be conserved in both OsGLYI-8 and OsGLYI-11.2 proteins, and marked with red stars (Figure 5). Among these four residues, the position of first glutamate residue showed maximum divergence in GmGLYI proteins. MtGLYI-11.1 and MtGLYI-11.2 proteins showed variation in all the four conserved active side residues. Proteins with the presence of all these four highly conserved residues are expected to have functional GLYI enzyme activity as per suggestion of previous studies (Mustafiz et al., 2014; Ghosh and Islam, 2016). Eight MtGLYI proteins such as MtGLYI-4.1, MtGLYI-7.1, MtGLYI-10.1, MtGLYI-10.2, MtGLYI-22.1, MtGLYI-24.1, MtGLYI-24.2, and MtGLYI-28.1 possess all four residues and expected to have function GLYI enzyme activity.

Pfam analysis confirmed the presence of conserved Metallo-beta-lactamase domain (PF00753) in all MtGLYII proteins either alone or with some other domains (Figure 6). Out of total 27 proteins, 22 MtGLYII proteins had only Metallo-beta-lactamase domain, while rest 5 contained additional domain such as Beta-Casp domain (PF10996), Zn-dependent metallo-hydrolase RNA domain (PF07521), Cleavage and polyadenylation factor 2 (PF13299), Myb/SANT-like DNA-binding domain (PF13837) (Figure 6). The significance of these additional domains is not clear in relation to GLYII enzyme activity. Similar to GLYI, GLYII proteins contained two well-conserved domain; one is metal binding motif (THXHXDH) and another one is active site motif (C/GHT) (Ghosh et al., 2014). The presence of both these motifs is crucial for a functional GLYII enzyme. Therefore, sequences of all MtGLYII proteins were aligned with that of E. coli, yeast, human, rice, Arabidopsis and Brassica GLYII proteins (Figure 7). The position of both these motifs was marked with red boxes (Figure 7). Out of 27 identified MtGLYII proteins, only two (MtGLYII-2.1 and MtGLYII-8.1) do not possess the conserved metal binding residues, but 13 of them do not have the active site motif (Figure 7). MtGLYII-1.1, MtGLYII-2.1, MtGLYII-2.2, MtGLYII-4.1, MtGLYII-5.1, MtGLYII-8.1, MtGLYII-9.1, MtGLYII-10.1-4, MtGLYII-11.1, and MtGLYII-13.1 proteins are either missing metal binding or active side residues, thus might not possess functional GLYII enzyme activity.

The presence of unique glyoxalase III proteins has been reported in many microorganisms and plants. GLYIII activity was found in the proteins with DJ-1/PfpI domain. Like conventional glyoxalase enzymes, DJ-1 proteins from Arabidopsis, rice, soybean, and Medicago were analyzed using Mega 5.2 tool (Figure 8A). There were two major clades in the tree, named- clades I and II. Among them, AtDJ-1D and OsDJ-1C were experimentally proven to have substantial GLYIII enzymatic activity (Kwon et al., 2013; Ghosh et al., 2016). Proteins present in clade-I along with these two might have significant GLYIII activity; while clade-II members could have less activity.

All previously reported rice and Arabidopsis DJ-1 proteins were found to contain two DJ-1/PfpI domains, while that of E. coli, Drosophila, and human were found to contain only single domain (Ghosh et al., 2016). DJ-1/PfpI domain has an average size of around 140 to 150 amino acids in all species except E. coli. Sequence analysis of MtDJ-1 proteins revealed that most of them have two DJ-1/PfpI domain, too (Figure 8B). Only MtDJ-1B and MtDJ-1C has single DJ-1/PfpI domain with a size of 69 and 88 amino acids each. These two might have partial DJ-1/PfpI domain. However, the domains of other MtDJ-1 proteins are connected by a linker of 40 amino acids (MtDJ-1A and MtDJ-1E) or 10 amino acids (MtDJ-1D). The N-terminal DJ-1/PfpI domains of all MtDJ-1 proteins were aligned with that of E. coli, Drosophila, human DJ-1 proteins; and AtDJ-1D and OsDJ-1C proteins (Figure 8C). All reported DJ-1 proteins with GLYIII contain a conserved catalytic triad of aspartate/glutamate, cysteine, and tyrosine/histidine residues (Ghosh et al., 2016). All these three residues were marked with red stars (Figure 8C). Among them, the central cysteine residue is the key for GLYIII enzymatic activity due to it’s directly involvement in catalysis (Lin et al., 2012). All MtDJ-1 proteins possessed this highly conserved cysteine residues, except MtDJ-1E (Figure 8C). The absence of this important residue in MtDJ-1E protein might lead to complete lack of GLYIII activity as proven by previously for AtDJ-1C (Kwon et al., 2013). However, the absence of any of other two conserved residues led to partial loss of GLY III activity in case of AtDJ-1e and AtDJ-1f (Kwon et al., 2013). Thus, MtDJ-1A should be the most efficient GLYIII enzyme of Medicago with all three conserved residues, followed by MtDJ-1D (Figure 8C).

To investigate the expression patterns of Medicago glyoxalase genes, expression data for a diverse set of 17 Medicago tissues was retrieved and analyzed. All these tissues could be broadly divided into four classes; such as underground, aerial, seed and nod; and represented as heat maps (Figures 9A–C). Different members of MtGLY gene families showed different tissue specificity. Among all MtGLYI genes, MtGLYI-4 showed the highest level of constitutive expression in all these analyzed tissues, followed by MtGLYI-24, MtGLYI-13, MtGLYI-21, and MtGLYI-16. MtGLYI-8 also maintained a high level of expression except for some aerial tissues; while MtGLYI-14 and MtGLYI-15 showed low expression only in underground and nod tissues. There are some clusters of genes that showed expression in particular type of tissues only. MtGLYI-7, MtGLYI-9 and MtGLYI-19 showed medium to high-level expression in aerial tissues only, while MtGLYI-5, MtGLYI-6 and MtGLYI-17 showed a similar pattern of expression only in underground and nod tissues. The presence of seed-specific GLYI transcripts has been previously reported in rice, Arabidopsis, and soybean (Mustafiz et al., 2014; Ghosh and Islam, 2016). Similarly, MtGLYI-1 and MtGLYI-11 has medium to high level of expression in different seed maturation stages and low expression in other tissues (Figure 9A). However, some members maintained the high level of expression in all tissues except for seed only, such as MtGLYI-12 and MtGLYI-23. This indicates toward the presence of a complex regulatory network for MtGLYI gene expression.

On the other hand, two distinct clades were observed in case of MtGLYII and MtDJ-1 gene expression data (Figures 9B,C). One set of genes maintained high level of constitutive expression in all tissues, while the other set of genes had very low level of expression. Out of 13 analyzed MtGLYII genes, 6 genes such as MtGLYII-4, MtGLYII-6, MtGLYII-7, MtGLYII-8, MtGLYII-10, and MtGLYII-12 showed medium to high-level expression in all the tissues with very few exceptions (Figure 9B). Among others, MtGLYII-11 showed the lowest level of expression in all the tissues, while MtGLYII-1, MtGLYII-3, MtGLYII-9, and MtGLYII-13 genes showed significant expression in the aerial tissues only. In the case of MtDJ-1 family members; MtDJ-1A and MtDJ-1D showed a high level of expression in all these tissues, while MtDJ-1B and MtDJ-1C showed the lowest level (Figure 9C). As compared to MtGLYI family members, MtGLYII and MtDJ-1 family members showed less complexity and tissue specificity.

To investigate the physiological role of Medicago glyoxalase genes in response to abiotic stress, corresponding transcript expression was analyzed under salinity and drought stresses. In response to salt stress (3 days old Jemalong A17 seedlings were treated with 180 mM NaCl) (Li et al., 2009), log₂ fold change ratio was analyzed at 6, 24, and 48 h of stress as compared to their 0 h expression level. A cluster of MtGLYI genes-MtGLYI-16, MtGLYI-8, MtGLYI-21, MtGLYI-11, MtGLYI-20, MtGLYI-6, MtGLYI-23, MtGLYI-1, and MtGLYI-3 showed early up-regulation (6 h) with further enhancement with stress duration (48 h). However, another set of MtGLYI genes-MtGLYI-17, MtGLYI-7/9, MtGLYI-14/15, MtGLYI-5, MtGLYI-18, and MtGLYI-24 showed strong down-regulation at all three-time points (Figure 10A). Rest members-MtGLYI-4, MtGLYI-10, MtGLYI-13, and MtGLYI-19 showed slight alteration as compared to the untreated control sample. On contrary, only one MtGLYII member-MtGLYII-9 showed strong up-regulation, followed by MtGLYII-4 and MtGLYII-8 (Figure 10A). Similarly, MtDJ-1C showed stable up-regulation from 6 to 48 h of salt stress, followed by MtDJ-1B and MtDJ-1A (Figure 10A). Rest of MtGLYII and MtDJ-1 members showed a different degree of down-regulation.

A complex differential expression pattern was observed for all three gene families in response to drought stress at both shoot (Figure 10B) and root (Figure 10C). Most of the genes showed less sensitivity during early days of stress (2 and 3 days). From day-4 and onward, some MtGLYI genes (MtGLYI-11, MtGLYI-8, MtGLYI-6, MtGLYI-16, MtGLYI-21, MtGLYI-4, MtGLYI-18, and MtGLYI-14/15) showed strong up-regulation; while two other members (MtGLYI-17 and MtGLYI-7/9) showed drastic down-regulation. In case of MtGLYII; MtGLYII-12, MtGLYII-6, and MtGLYII-9 showed a similar pattern of up-regulation from 4 to 14 days of drought stress. All these up/down-regulated genes maintain this pattern after 24 h of rewatering, too. Similarly, drought stress causes severe alteration of MtGLYI, MtGLYII, and MtDJ-1 genes at root tissues (Figure 10C). MtGLYI-16, MtGLYI-6, MtGLYI-3, MtGLYI-8, MtGLYI-21, MtGLYI-11, and MtGLYI-14/15 genes showed sharp up-regulation from day-3 to onward at root tissue, and come down to normal level after rewatering (Figure 10C). Likewise, MtGLYII-12, MtGLYII-13, MtGLYII-9, and MtDJ-1A, MtDJ-1D maintained up-regulated transcripts among MtGLYII and MtDJ-1 families from very early to late drought stress. As root has a direct contact with soil water, re-watering brings back the expression level of most glyoxalase genes into their ground level at root tissue (Figure 10C). All these expression data indicated the diverse role of different conventional and unique glyoxalase members in different stress modulation pathways of Medicago plant.