Molecular characterization and expression dynamics of MTP genes under various spatio-temporal stages and metal stress conditions in rice

Metal Tolerance Proteins (MTPs) are the class of membrane proteins involved in the transport of metals, mainly Zn, Mn, Fe, Cd, Co and Ni, and confer metal tolerance in plants. In the present study, a comprehensive molecular analysis of rice MTP genes was performed to understand the evolution, distribution and expression dynamics of MTP genes. Exploration of the whole genome re-sequencing information available for three thousand rice genotypes highlighted the evolution and allelic diversity of MTP genes. Based on the presence of non-synonymous single nucleotide polymorphism (SNP), MTP1, MTP6, MTP8 and MTP9 were found to be the most conserved genes. Furthermore, results showed localization of MTP1, MTP8.1 and MTP9, and MTP11, respectively with QTLs/m-QTLs for Zn and Cd accumulation, making these genes promising candidates to understand the QTL regulation. Expression profiling of the entire set of 10 MTP genes revealed root and shoot specific expressions of MTP9 and MTP8.1, respectively, under all tested vegetative stages. Expression of seed-specific MTPs increased as seed maturation progressed, which revealed their potential role in transporting metals during seed filling. Upon exposure to harmful heavy metals, expression of most MTP genes decreased in root and increased in shoot, suggests that different mechanisms are being employed by MTPs in different tissues. Contrastingly, only a few MTPs were found to be responsive to Fe and/or Zn starvation conditions. The extensive analysis of MTPs presented here will be helpful in identifying candidate MTP genes for crop biofortification and bioremediation purposes.


Introduction
Plants, being sessile, are continuously exposed to several biotic and abiotic stresses, including nutritional constraints. Among the abiotic stresses, heavy metal stresses imposed by many contaminants present in the environment adversely affect plant growth [1]. Some heavy metals, PLOS

Identification and structural organization of MTP genes
In this study, search was performed for term 'Metal Tolerance Protein' at RAP database and Oryzabase, we found the rice genes which are annotated as Metal Tolerance Proteins. To further identify new MTP genes in the rice genome, we used OsMTP1 amino acid sequences as a query to perform a database search using BLASTP against predicted proteins in the Oryza sativa genome derived from the RAP database. BLAST hits with less than 100 bitscore were removed. With these criteria, we didn't find any new MTP genes in the rice genome. Genomic DNA sequences, CDS and protein sequences for all the genes, except for MTP12, were retrieved from the RAP Database. The exon/intron organization of MTP was visualized with the Gene Structure Display Server (GSDS) program (http://gsds.cbi.pku.edu.cn/).

Phylogenetic and evolutionary analysis of MTP genes
Multiple sequence alignments were conducted with amino acid sequences. A phylogenetic tree was constructed using the Maximum-likelihood method (MLM) provided in the MEGA 7.0 software tool. Robustness of the branch nodes was confirmed with bootstrap analysis performed with 10000 replications. To study the molecular evolution of MTP genes, the publically available SNP dataset generated from the whole genome resequencing of 3000 rice genotypes was used [18]. All the SNP data for ten MTP genes were converted into hapmap format and diversity analysis was performed using Tassel V5.0 software (http://www.maizegenetics.net/ tassel). Functional annotation of SNPs was performed using SnpEff tool, and the effect of the amino acid variation was predicted using PROVEAN tool [19]. PROVEAN scores equal to or below -2.5 were considered deleterious, and the score above -2.5 was considered neutral.

Identification of conserved motifs, subcellular location and structural prediction
The protein sequences were analyzed to identify conserved protein motifs (motif scan) using the MEME (multiple EM for motif elicitation) program. Transmembrane domains in MTP proteins were identified using TOPCONS software tools. Subcellular localization of MTP genes was performed using CELLO server (http://cello.life.nctu.edu.tw/). The protein sequences of all MTP proteins were submitted to the Phyre2 protein-modeling server (www. sbg.bio.ic.ac.uk/ � phyre2). The protein 3D structure were further refined using ModRefiner (https://zhanglab.ccmb.med.umich.edu/ModRefiner/).

Construction of hierarchical clustering and heatmap of expression data
The heatmap of log2 transformed values of MTP genes was generated using MeV4.9 software tool (http://mev.tm4.org/). The hierarchical clustering was performed using HCL option in MeV software tool with covariance values for distance metric selection. The average linkage clustering method was selected for clustering.

RNA extraction and cDNA synthesis
Total RNA was extracted from different samples using TRI Reagent (Sigma), as per the supplier's protocol. Briefly, after powdering the sample with pestle-mortar, the sample was homogenized in Tri Reagent. The extract was mixed with chloroform and centrifuged, and the upper aqueous phase was mixed with 2-propanol to pellet RNA. Afterwards the RNA pellet was washed twice with 70% ethanol. The concentration and RNA quality were checked on Nanodrop (Thermo Scientific) and by agrose gel electrophoresis. Before proceeding to cDNA synthesis RNA was treated with DNase I (RNase free) (New England Biolabs) to remove DNA contamination. After that, DNase was heat inactivated by incubating at 75˚C. First strand cDNA was synthesized using 1μg total RNA using iScript-cDNA synthesis kit (Bio-rad Inc.) as per the manufacturer's instructions.

Q-PCR analysis
Quantitative real-time PCR was performed using 5× diluted cDNAs and iTaq Universal SYBR Green Supermix (Bio-rad Inc.) on CFX96 Touch Real-Time PCR machine (Bio-rad Inc.) In 10 μl reaction, 5 μl 2x SYBR Green Supermix, 0.5 μl 10 μM primers (Forward + Reverse), 1 μl diluted cDNA, and remaining amount of water was used. Rice UBQ5 and eEF-1a were used as internal reference genes for seedling and seed tissues, respectively. The sequence of the primers used for Q-PCR analysis is provided in S2 Table. Delta-delta CT-method (2−ΔΔCT) was used to calculate relative gene expression levels [20]. The statistical significance of expression data was determined using two-tailed paired Student's t test.

Results
Identification, phylogenetic relationship, gene structure, protein structure and molecular evolution of MTP genes in the rice genome A total of 10 MTP genes were identified consistently in both the rice sub-species, indica and japonica. The DNA sequence details for the rice MTP genes are summarized in Table 1.
Phylogenetic tree developed using Maximum-likelihood methods (MLM) showed expectedly a cluster of MTP8, MTP8.1, MTP9, MTP11, and MTP11.1, classified as Mn-CDF, distinct from other MTPs (Fig 1A). Analysis of gene intron-exon organization revealed more conserved gene structure for these Mn-CDF members, with 5-6 introns present in each gene ( Fig  1A). Similarly, MTP1, MTP5 and MTP12 classified as Zn-CDFs, were found to be clustered together in the phylogenetic tree. Gene intron-exon organization is also very similar to the Zn-CDF genes, particularly for MTP1 and MTP12, which were found to have a single exon ( Fig  1A). The MTP6 and MTP7 respectively having 11 and 12 introns clustered apart from the rest of MTPs in phylogenetic tree ( Fig 1A).
Exploration of whole genome re-sequencing data available for three thousand rice genotypes showed high level of genetic diversity for the MTP genes ( Table 2). Besides having high frequency of single nucleotide polymorphism (SNP), four MTP genes MTP1, MTP6, MTP8, and MTP9 showed high level of protein sequence conservation ( Table 2). Unusually high frequency of non-synonymous SNPs was observed in MTP11.1 and MTP12. Further investigations performed to verify the facts revealed that all six non-synonymous SNPs represent a rare allele present in less than five percent of total re-sequenced genotypes. For both the MTP11.1 and MTP12 genes the minor allele is present exclusively in indica lines. Tajima's D, a powerful population genetics test, performed with the SNPs identified within the genic region of MTP genes showed positive Tajima's D value for all the genes ( Table 3). The observed range of Tajima's D was from 0.3 to 4.8, suggesting a more balancing selection for most of the MTPs.
Evaluation of two-dimensional structure predicted using TMHMM tool showed a wide range of transmembrane (TM) domains (S1 Table). Surprisingly, no TM domain was observed in MTP6. To confirm the discrepancies in TM domain prediction, homology based threedimensional (3D) structures were predicted for the entire set of rice MTP proteins. Unlike TMHMM results, homology based 3D structure showed typical MTP structure with 10 to 12 transmembrane domains for all the rice MTPs including MTP6 (Fig 2). Further search for conserved motifs in rice MTPs was performed with the MEME (multiple EM for motif elicitation) program to gain additional insights into their diversity. As shown in Fig 1B, eight conserved motifs, designated as motif 1 to motif 8, were identified across the MTP family. Interestingly, all the MTPs belonging to Mn-CDF (MTP8, MTP8.1, MTP9, MTP11, MTP11.1) have the same motifs in their protein structure ( Fig 1B). However, the remaining MTP proteins didn't have any of the motifs present in the Mn-CDF family MTPs. Some of these motifs have TYR phosphorylation site, ligand binding domains or have role in cytoskeleton reorganization ( Fig 1B)A systematic survey of previously identified QTL and meta-QTLs (m-QTLs) associated with accumulation of metals or tolerance to various metal stress conditions was performed to check whether any of the MTP genes overlap with them (Table 4). Co-localization of MTP8.1 with two m-QTLs, and overlap of MTP1, and MTP9 with another QTL/m-QTL associated with Zn accumulation was observed. MTP11 was also found to be overlapping with QTL associated with Cd tolerance (Table 4).

Gene expression dynamics of MTP genes during vegetative growth
To examine the expression pattern of MTP genes during the vegetative phase, we analyzed their expression separately in root and shoot tissues in one-week, two-week and three-week- old rice seedlings through Quantitative-PCR (Q-PCR) ( Fig 3A). Additionally, we also examined the expression pattern of MTP genes in public transcriptomic data available at Rice Expression Database (Fig 3B). The Rice Expression Database hosts most of the publicly available RNA-Seq datasets for rice, with normalized gene expression in the form of FPKM (fragments per kilo-base of gene per million mapped reads) values [21]. Significantly, we have found an excellent correlation between our Q-PCR analysis and public transcriptomic data, (Fig 3A and 3B). For instance, in both Q-PCR and public transcriptome data, expressions of MTP6, MTP11.1 and MTP12 was found lowest among all MTPs under tested conditions. Similarly, MTP9 showed root specificity during all the three-development stages in both the datasets (Fig 3A and 3B). Furthermore, MTP8 was found to have intermediate expression levels among all MTPs in both root and shoot tissues of all three developmental stages in both Q-PCR and public transcriptomic data (Fig 3A and 3B). There are some examples in which we found differences between Q-PCR and public transcriptomic data. For example, MTP1 and MTP7 are consistently expressed at all these developmental stages at the highest levels in our Q-PCR analysis, but in public transcriptomic data only MTP1 was expressed at the highest level ( Fig 3A and 3B). To understand temporal patterns of expression, comparison of gene expression was performed between different developmental stages. Some MTP genes, like MTP5 and MTP8.1, have the least expression at one-week stages in both root and shoot tissues, however at later stages (two-and three-week) their expression increases in both the tissues (Fig 3A). MTP8.1 shows shoot specific expressions in all the three development stages in public transcriptomic data, and only at 2-week-old stage in our Q-PCR analysis (Fig 3A and 3B). In conclusion, combining Q-PCR and public transcriptomics results provides a detailed description of the expression dynamics of MTP genes during vegetative stages. To gain insight about relative expression levels of MTP genes in these tissues through Q-PCR experiments, we have calculated expression levels of MTP genes with relation to expression of OseEF-1a gene ( Fig  3A). As shown in Fig 3A, the expression of MTP genes in descending order in roots was MTP9>MTP1>MTP7>MTP8>MTP11>MTP5>MTP8.1>MTP6> MTP12>MTP11.1. In the shoots a similar pattern was observed.

Expression profiling of rice MTP genes during seed development
No previous study has examined the expression dynamics of MTP genes during seed development. Hence, to understand the expression pattern of rice MTP genes during seed development, Q-PCR analysis was performed on developing seeds at 7, 14, 21 and 28 Days After Flowering (DAF) stages (Fig 4A and 4B). In addition, the expression pattern of MTP genes was analyzed in different reproductive tissues and seed tissues in the public transcriptomic data (Fig 4C and 4D). The Q-PCR analysis showed expression of most of the MTP genes increased as the seed development progressed, except for expression of MTP5, MTP9 and MTP11 (Fig 4A). Consistently, in the public transcriptomic data the expression of most MTP genes was higher in mature seed compared to 5 DPA (Day Post Anthesis) old seed (Fig 4C). These results suggest that MTP genes are likely to play an important role in filling metal micronutrients in developing seeds. When expression of MTP genes was compared among themselves during seed development, we found that MTP9 has maximum expression in all the four analyzed seed development stages, followed by MTP1, MTP7 and MTP8 (Fig 4B). This observation was also partially true with public transcriptomics data (Fig 4C). Expression of other MTP genes was found minimal during different stages of seed development (Fig 4B). Notably, in the public transcriptomics data, we found that MTP5 specifically expresses in pistil and MTP11.1 specifically in anthers (Fig 4C), which suggests their potential role in these tissues. Furthermore, we note that in the public transcriptomics data MTP9 is strongly expressed in aleurone layer, and up-regulated by Gibberellic Acid (GA) treatment (Fig 4D), which is consistent with maximum expression of this gene in seed tissue among all MTPs. Overall, our results identify reproductive tissue and stage specific MTP genes.

Effects of heavy metal stress on MTP gene expression
Since previous studies highlight the role of few MTP genes in tolerance to heavy metal stress [22,23], so we decided to thoroughly investigate the effect of different heavy metals on expression of all the MTP genes. To impose heavy metal stress, rice seedlings were transferred to hydroponics media containing different heavy metals for different time points. Subsequently, root and shoot tissue were collected separately and analyzed for gene expression through Q-PCR. There was no obvious effect on plant growth with 24 hour (hr) treatment of Cobalt (Co), Cadmium (Cd) and Nickel (Ni). However, after one week of treatment expression of most MTP genes was down-regulated in both root and shoot tissues (S1 Fig). This suggests that longer heavy metal treatments have altered plant physiology, so we will mainly discuss our gene expression results for only 24 hours of heavy metal treatment.
At 24 hours of treatment, most of the MTP genes showed gene expression changes either in root or shoot (Fig 5). Interestingly, we found that among the differentially expressed genes, most genes show down-regulation in the roots (Fig 5A), and up-regulation in the shoot ( Fig  5B). The exception to this notion was MTP6, which was repressed by Co in both root and shoot tissues (Fig 5). Furthermore, it was found that most genes were affected only with one out of three different heavy metal stress conditions (Fig 5), suggesting that these genes have specificity towards particular heavy metal species. The few notable exceptions to this notion includes MTP8 which shows down-regulation in roots by both Co and Cd, MTP8.1 which shows up-regulation by both Cd and Ni in shoot, and MTP6 which show down regulation by Co and up-regulation by Cd in shoot (Fig 5), suggesting that these genes have broad sensitivity towards different heavy metal species. Meticulously, in roots, only Cd and Co were found to be affecting the expression of MTP genes, as expression of none of the MTP genes was found to be affected by Ni treatment (Fig 5A). However, in the shoot we found that Cd affects the most number of the MTP genes ( Fig 5B).

Effects of zinc and iron starvation conditions on MTP gene expression
Since some of the plant MTP genes are known to transport essential heavy metals, such as Fe and Zn [24], so we asked which rice MTP genes are likely to have this function. For this purpose, we exposed the rice seedlings to media depleted with these mineral elements and performed gene expression analysis through Q-PCR. Contrary to mis-expression of most MTP genes by non-essential heavy metals, expression of only a handful of MTPs genes was found to be affected by one week Zn and Fe starvation conditions (Figs 5 and 6). For instance, expression of only MTP6 (in Zn-) and MTP12 (in Fe-) was up-regulated in roots after 1 week of starvation ( Fig 6A). However, after 2 weeks of starvation, we found that expression of MTP7, MTP11, and MTP12 (all in Zn-), MTP8 (in Fe-), and MTP8.1 (in both Zn-and Fe-) genes were also significantly up regulated in roots (Fig 6C). These results suggest the potential role of these genes in the transport of Fe and Zn in roots. In the shoots, MTP9 and MTP12 were found to be significantly up-regulated at 1 week of Zn deficiency conditions (Fig 6B). Significantly MTP9 was also found to be highly expressed in seed aleurone layer, and the aleurone layer accumulates very high levels of Zn and Fe (Fig 4D). Altogether these results very strongly suggest that MTP9 might be a good target for future Zn biofortification strategies.

Discussion
Plants have evolved several molecular mechanisms involving specific gene families to tolerate toxic levels of heavy metals. Considerable work has been done to understand the function of plant MTP/CDF gene family in heavy metal stress tolerance. In rice, MTP genes were identified and named after completion of genome sequencing, based on the sequence similarity with Arabidopsis MTP genes [11,12]. Previous studies mainly looked at phylogenetic relationships among various CDFs across diverse taxa and classification of plant MTPs into various groups based on substrate specificity. Other more recent studies were aimed at functional analysis of individual rice MTP genes [14,15,24]. The present study is an attempt to comprehensively understand the evolution of rice MTP genes and their response to essential and non-essential heavy metals.
Gene intron-exon organizations have been considered as more of an efficient model to study evolutionary relations between orthologues and paralogues. Comparison of intron-exon organization with respective phylogenetic distribution showed similar pattern for the phylogenetically closer MTP genes (Fig 1A). In addition, numbers of introns in the gene have been reported to correlate with gene expression, and loss or gain of homologues. Mostly, intronless genes in plants are thought to have evolved recently [25]. Among the rice MTPs, intron-less MTP1 and MTP12 genes do not show any other characteristic (Tajima's D and SNPs) suggesting early evolution. Interestingly, MTP1 and MTP12 share a characteristic histidine rich loop towards c-terminal, which is known to have a role in Zn-binding. The phylogenetic distribution and gene intron-exon structure have been helpful in understanding the evolutionary relationship of MTP genes.
The whole genome re-sequencing information available for 3000 diverse rice genotypes is an excellent resource for the identification of allelic diversity and to study evolution. Wide range of allelic diversity for the MTP genes observed in the present study will be helpful in exploring the genetic resources for the development of heavy metal tolerant varieties as well as  varieties with efficient uptake of essential elements. The rare allele of MTP8.1 with three amino acid changes compared to the most of the japonica cultivars provide an opportunity to evaluate functional impact and subsequent exploration through breeding approaches ( Table 5). The amino acid change from A to S predicted in MTP8.1 to have a deleterious effect suggesting significant variation in the functionality of the alleles. The results look promising in view of the previous study performed in soybean, where the effects predicted with PROVEAN tool have been validated [26]. Analysis of the functional effect of different alleles of MTP genes is beyond the scope of present study but the information presented here will serve for the study aiming to explore genetic resources. The evolutionary parameters studies here suggest balancing evolution for MTP genes in rice. Earlier similar type of selection pressure suggesting balancing evolution instead of selective sweeps have been reported for the R genes [27]. Expression pattern of a gene can give clues about its function. In the present study, we have analyzed the expression profile of all the MTP genes across various tissues and developmental stages through Q-PCR and in transcriptomics data available in the public domain. Most significantly, we have found a good correlation between these two independent analyses, which gives very high confidence about these results. Since the expression pattern of some of the rice MTPs was previously analyzed [14,15,24], so we compared our results with these studies and found that our gene expression results are comparable to them. For example, Yuan et al. [23] found that MTP1 has higher expression in two-week old leaves compared to roots, our Q-PCR results also confirm this (Fig 3A). Zhnag et al. [16] and Chen et al. [14] found a higher expression of MTP8 and MTP8.1, respectively, in leaves of two-week old seedlings compared to their roots, our Q-PCR results also confirm this notion (Fig 3A). In conclusion, some of our findings have been confirmatory in nature in cases where the individual MTP gene has been previously characterized. However, for other MTP genes, which are not yet functionally characterized, our results provide novel insights about their expression patterns in different tissues and organs.
Expression analysis for any of the rice MTP genes in seed and reproductive tissues has not been performed earlier. Expression profile analysis derived from transcriptome data and Q-PCR analysis revealed that the tissue expression patterns of OsMTPs in the same group were similar but varied among groups (Fig 3). Similar results have been obtained in current study performed for MTP genes in Nicotiana tabacum [28]. Together these results indicate the conserved and essential roles of MTP genes across various growth and development stages of various plants. Analysis of expression values in public transcriptomics data revealed anther and pistil specific expression of some of the MTP genes (Fig 4C), which was not reported earlier. Furthermore, as the seed development progressed, expression of most of the MTP genes was also found to be increased ( Fig 4A). As seeds are major sink organ and sink strength of the seeds increases along with their development [29], so increased expression of MTP genes during seed development indicates their potential role in metal homeostasis during seed filling stage. Borill et al. [30] have proposed potential role of MTP genes in accumulation of essential heavy metals in vacuoles of aleurone layer. Supporting this, our results indicate that MTP9 is strongly expressed in aleurone layer (Fig 4D). Very recently it has been shown that MTP9 expresses in cell-type specific manner in rice roots, where it expresses only in exo-and endodermis cells of the roots [15]. This further supports the role of MTP9 as a cell-type specific player. Ueno et al. [15] also reported 40% yield loss and 50% reduction in total seed Mn content in brown rice in knock-down lines of MTP9 genes, which strongly support its role in essential metal filling in developing seeds. However, it was not analyzed if there were reduction in accumulation of other essential metals such as Zn and Fe.
In order to understand the response of MTP genes to heavy metals, we analyzed their gene expression either in the absence of essential heavy metals Zn and Fe or in the presence of non-essential heavy metals such as Cd, Co and Ni. Expression of most MTPs is increased in shoot and decreased in roots after 24 hr treatment with non-essential heavy metals (Fig 5). These results suggest a potential heavy metal tolerance mechanism in which plants accumulate heavy metals away from their site of exposure (roots in this case). Hyper-accumulator plants use the same strategy to accumulate heavy metals in their aboveground parts [10].
In comparison to expression of most of the MTP genes affected by non-essential heavy metals, the expression of only a few MTP genes was affected by Zn and Fe deficiency (Figs 5 and  6). Furthermore, the expression of the affected MTP genes was induced in both roots and shoots by deficiency of both Zn and Fe, which suggests that both roots and shoots induce the mechanism of sequestration. Zn deficiencies in the early stages of seedling development and iron deficiency chlorosis of leaves are two major growth challenges for rice seedlings [30,31]. A proper understanding of homeostasis of these ions in seedlings stages would help to design crops with better survival under nutrient deficient soil.