Background

Banana is an important staple and economic crop in tropical and subtropical regions. However, its growth and yield are constantly affected by severe abiotic and biotic stresses, such as cold in winter, drought and water-logging, as well as various diseases and pests [1]. These environmental perturbations often lead to the increased generation of reactive oxygen species (ROS) in plant cells [2]. Excess ROS can attack virtually all cellular macromolecules. This usually results in membrane damage, protein oxidation and DNA lesions, and can even lead to irreparable metabolic dysfunction and cell death [3, 4]. Thus, to cope with ROS toxicity, plants have developed efficient and complex antioxidative response systems, including many non-enzymatic and enzymatic components. Among these enzymatic components, superoxide dismutases (SODs), acting as the first line of antioxidant systems in plant, play important roles in catalyzing the dismutation of superoxide radicals to protect cells from oxidative damage [5].

In plants, there exist multiple SOD isozymes, which are classified into three types based on their metal co-factors: Cu/ZnSOD, FeSOD and MnSOD [6]. Although these SOD proteins are encoded by nuclear genes, they are distributed to different cellular compartments. Cu/ZnSODs are mainly located in the cytosol, chloroplasts, peroxisomes and/or the extracellular space, while FeSODs are mainly in chloroplasts and possibly the cytosol, and MnSODs are in the mitochondria [7]. Owing to their important roles in the antioxidant system, a considerable number of SOD genes are cloned from various monocot and dicot plants [8–12]. However, Arabidopsis thaliana and Populus trichocarpa are the only two plants whose SOD gene families have been characterized in the genome-wide level at present, and the numbers of the three SOD-type genes varies among them [8, 9].

SODs contribute to various environmental stimuli responses in plants, such as cold, drought, salinity, auxin and ethylene [7, 13, 14]. Different SOD genes exhibited different expression patterns. The responses of SODs to environmental changes or stresses were dramatically different, depending on the different SOD members present, the stress and the species. For instance, under ozone fumigation, the transcriptional levels of chloroplastic CSD2 and FSD1 were transiently decreased, while chloroplastic FSD2 mRNAs remained somewhat constant in A. thaliana [8]. In contrast, FSD2 mRNAs dramatically increased in response to UV-B, while CSD2 or FSD1 mRNAs remained constant. In addition, cytosolic CSD1 could be involved in responses to both ozone fumigation and UV-B illumination. Even so, the SOD genes of the same SOD type did not always exhibit uniform functions in different species. MnSOD showed no altered expression when subjected to a series of oxidative stress treatments in Arabidopsis, but it responded positively to salt stress in pea [15], and cold and drought stress in wheat [16, 17]. This indicated that the regulation of SOD genes is complicated in response to oxidative stress. Cis-element and transcription factor analyses of SOD promoters from Arabidopsis [6], wheat [12] and longan [10] provided some clues on how the SOD genes are modulated. Additionally, alternative splicing (AS) and miRNAs have also found to be involved in the regulation of SOD expression [18, 19]. Studies using over-expressing or knocked-out plant SOD genes have confirmed their functions in improving stress tolerance [20–22].

In previous reports, Zhou et al. studied the SOD isoenzymes in banana using biochemical methods and revealed that cold stress led to the accumulation of different SOD isoenzymes [23]. A quantitative proteomic analysis confirmed the existence of Cu/ZnSOD, MnSOD and FeSOD in banana [24]. However, these studies focused only on the proteins and changes in activity, which were unable to effectively elucidate the exact roles of banana SODs under adverse conditions. Recently, the whole-genome sequences of Musa acuminata var. DH-Pahang (wild banana, AA group) and Musa balbisiana var. Pisang Klutuk Wulang (PKW; wild banana, BB group) were made available to the public [25, 26], facilitating molecular studies on the expression and regulatory mechanisms of banana SODs in response to oxidative stress. Using these genomes, we performed a genome-wide identification of the SOD gene family in M. acuminata cv. Tianbaojiao (Cavendish banana, AAA group) to analyze the transcript types, protein motifs, exon-intron organizations, chromosomal locations and phylogenetic relationships. Then, the putative promoters of banana SODs were also isolated and cis-elements involved in stress responses were analyzed to further illuminate their transcriptional regulatory mechanisms. Finally, we studied the expression profiles of MaSOD family genes under abiotic (cold, heat, drought and salt) and hormonal [abscisic acid (ABA), gibberellin A3 (GA3), indole-3-acetic acid (IAA) and salicylic acid (SA)] stresses using real-time quantitative polymerase chain reaction (qPCR), which should help determine the functions of each MaSOD gene under adverse conditions.

Methods

Plant materials and stress treatments

Sterile plantlets of M. acuminata cv. Tianbaojiao (Cavendish banana, AAA group) were generated by inducing suckers as Zhang et al. described [27]. Twenty-five-day-old sterile plantlets were used in abiotic stress treatments, except the drought treatment. The plantlets were cultivated in Murashige and Skoog liquid solution with 200 mM NaCl for salt treatment, 100 μM IAA, 100 μM GA3 and 100 μM SA for hormonal treatments, and sprayed with 100 μM ABA in 0.02 % (v/v) Tween 20 for the ABA treatment. These treated plantlets were sampled at 4, 8, 12, 24 and 48 h, except that salt treatment, which was sampled at 4, 12, 24, 48 and 72 h. For the drought treatment, 2-month-old plants grown in soil were cultivated without watering and sampled at 1 d, 2 d, 3 d, 4 d and 5 d. All the treatment conditions were at 28 °C with 3300 lux continuous light, except the cold stress, which was conducted in 4 °C growth chambers with 400 lux continuous light, and the heat stress, which was conducted in 40 °C growth chambers with 3300 lux continuous light. Control and treated plantlets were immediately frozen in liquid nitrogen and stored at −80 °C until nucleic acid extraction.

Nucleic acid extraction and cDNA synthesis

Genomic DNAs were extracted from ‘Tianbaojiao’ banana using the modified CTAB method [28]. Total RNAs were isolated using Column Plant RNAOUT 2.0 Kit (TIANDZ, China). Then, the quality was checked using 1.0 % agarose gel electrophoresis and quantified using spectrophotometry. Total RNAs were reverse transcribed using a Thermo Scientific RevertAid First Strand cDNA Synthesis Kit (Fermentas, EU) for 3’ UTR (untranslated region) and open reading frame (ORF) cloning, and a SMART™ RACE cDNA Amplification kit (Takara, Japan) for 5’ UTR cloning. A GeneRacer™ kit (Invitrogen, USA) was used for the alternative transcriptional start site analysis.

Isolation of the SOD genes from M. acuminata cv. Tianbaojiao

“Superoxide dismutase” and published SOD sequences in NCBI were used as the index words and probes, respectively, to search for SOD genes in the wild banana genome databases (http://banana-genome.cirad.fr/) [29]. Full-length cDNAs of ‘Tianbaojiao’ SOD genes were cloned using 5’/3’ RACE-PCR and reverse transcription PCR (RT-PCR), with primers designed from known SOD sequences in the NCBI and the banana genome databases. Genomic sequences of MaSODs were obtained from DNA templates using PCR. 5’- flanking regions of MaSOD genes, 1.0–2.0 kb in length, were generated using PCR with forward primers (designed to regions 2.0 kb upstream of the start codon of ‘DH-Pahang’ SOD genes) and reverse primers (designed to regions downstream of the start codon of each MaSOD). All of the primer information is listed in Additional files 1 and 2.

PCR reactions were performed according to the instructions of ThermoScientific DreamTaq Green PCR Master Mix (2×) (USA) and Takara LA Taq (Japan). Then, PCR products with acceptable sizes were purified, subcloned into the vector pMD18-T (Takara, Japan), and confirmed by DNA sequencing. The correct sequences of SOD genes were abbreviated as MaCSD, MaMSD and MaFSD for Cu/ZnSOD, MnSOD and FeSOD, respectively, with a prefix representing the genus and species.

Sequence analysis and subcellular localization

Multiple sequence alignments were performed using ClustalX (version 1.83). The molecular weights and isoelectric points of the MaSODs were calculated using the ExPASy ProtParam tool [30]. Syntenic relationships of SOD genes were searched in the Plant Genome Duplication Database (PGDD, http://chibba.agtec.uga.edu/duplication/) [31]. Conserved motifs were detected using MEME with the default settings, except that the maximum number of motifs was defined as nine, and the minimum and maximum motif width was set to 20 and 150, respectively. Exon-intron organizational analyses were carried out using the Gene Structure Display Server (http://gsds2.cbi.pku.edu.cn/). Transcriptional response elements of the promoters were predicted using the PlantCARE tool [32].

Subcellular localizations and putative transit peptides were predicted by SoftBerry (http://linux1.softberry.com/), ChloroP1.1 (http://www.cbs.dtu.dk/services/ChloroP/) [33] and MITOPROT (http://ihg.gsf.de/ihg/mitoprot.html) [34]. The coding regions of MaCSD1D were amplified using specific primers. The forward primer was 5’-CATGCCATGGATGGTTAAGGCTGTAGCTGTG-3’ and the reverse primer was 5’-GGACTAGTCTCCTGAAGCCCAATGACAC-3’ (restriction enzyme sites are in bold). The amplicons were then ligated into the N-terminus of the green fluorescent protein (GFP) sequence of the pCAMBIA1302 vector to generate the pCAMBIA1302-35S::MaCSD1D-GFP::NOS construct. The recombinant plasmid was introduced into onion epidermal cells by Agrobacterium-mediated transformation according to a reported method [35]. Two days later, the subcellular localization of the MaCSD1D protein was detected using an A1R/A1 laser confocal scanning microscope (Nikon, Japan).

Chromosomal locations and phylogenetic analysis

MaSOD genes were mapped to the chromosomes by performing a BLASTn search against the banana genome databases. Unrooted phylogenetic trees based on SOD protein sequences were constructed using the maximum likelihood method of the MEGA 5.02 software with the Poisson model and a bootstrap of 1000 replicates [36].

Real-time quantitative PCR and data analysis

Total RNAs from different tissues (leaf, pseudostem and root) and stress-treated leaves were reverse transcribed with PrimeScript™ RT Master Mix (Perfect Real Time) kit (Takara, Japan) into cDNAs for qPCR analysis. Primers, specific to every MaSOD gene, were designed based on the specific 3’ UTR sequences, except that of MaFSD1B, which were designed on the 5’ end sequences. A clathrin adaptor complexes medium gene (MaCAC) was used as an internal control because of its relatively stable expression level [37]. The specificity of primers was tested and verified by analyzing the melting curve. A gene expression analysis was performed using a LightCycler480 Real-time PCR detection instrument (Roche, Switzerland) and SYBR® Premix Ex Taq™ II (Tli RNaseH Plus; Takara, Japan) with three biological replicates and technical replicates. PCR reactions included an initial denaturation at 95 °C for 3 min, followed by 40 cycles at 95 °C for 10 s, 57 °C for 20 s, and 72 °C for 30 s. Relative expression levels were determined using the 2-∆∆Ct method. Then, the expression patterns of the 12 MaSOD genes were clustered by MeV (version 4.8) using the average linkage hierarchical clustering method. The details of the specific primers for qPCR are listed in Additional file 3.

Results

Identification of the SOD gene family in M. acuminata cv. Tianbaojiao

In the wild banana genome databases, 15 sequences in ‘DH-Pahang’ (AA group) and 14 sequences in ‘PKW’ (BB group) were identified in silico as SOD genes based on annotations and a BLASTn search using known SOD genes from NCBI. Excluding the remnant sequences and chimeric genes, there were 13 potentially functional SOD genes in ‘DH-Pahang’ and 11 genes in ‘PKW’ (Additional file 4). Two of these genes (GSMU_Achr10T27190_001 and GSMU_Achr10T27220_001 of ‘DH-Pahang’, and ITC1587_Bchr10_T31275 and ITC1587_Bchr10_T31280 of ‘PKW’) were tandemly located on chromosome 10 and shared similar ORF and gDNA sequences, which indicates that they were recently tandemly duplicated. Sequence analysis revealed that the SOD ORF lengths and sequence identities of the two wild bananas were somewhat different (Additional file 4). In this study, the presence and the structure of the SOD family genes were experimentally verified in the triploid cultivated banana ‘Tianbaojiao’.

A total of 25 different SOD cDNAs with intact ORFs were obtained from banana ‘Tianbaojiao’ using 5’ and 3’ RACE-PCR and confirmed by RT-PCR. Two of the cDNAs were described previously [38] and renamed here as MaCSD1A-1 (GenBank: JQ411718) and MaCSD1A-2 (GenBank: JQ411719) for the sake of consistency. A sequence alignment and BLAST results using known SODs/SODs in the NCBI database as queries revealed that these 25 transcripts were transcribed from 12 different MaSOD genes as a result of alternative transcription start sites (ATSSs), AS and alternative polyadenylation (APA). These transcripts, which encoded the same SOD protein, shared an identical ORF but differed in their UTRs. The major MaSOD features are summarized in Table 1.

Table 1 Characteristics of the SOD genes from M.acuminata cv. Tianbaojiao
Full size table

The ORF of MaCSD1A-1 was 98.14 % identical to that of MaCSD1A-2. A further analysis of their corresponding gDNAs (GenBank: KM017525 and KM017524) revealed they were alleles. Three independent transcripts encoding MaCSD1B, named MaCSD1B-1,-2 and −3, shared the same 459 bp ORF but varied in the 5’ UTR length or nucleotide composition owing to AS and ATSSs (Fig. 1a-1). Two other transcripts contained a 459-bp ORF but encoded two different amino acid polypeptides, which were named as MaCSD1C and MaCSD1D. The two transcripts, designated as MaCSD2A-1 and MaCSD2A-2, contained the same ORF of 684 bp but with different 3’ and 5’ UTRs (Fig. 1a-2 and b-1). The 996-bp long cDNA MaCSD2B had a 675-bp ORF with a 45-bp 5’ UTR and a 276-bp 3’ UTR.

Fig. 1
figure 1

Alignment of 5’-and 3’-ends of MaSOD cDNAs. a Alignment of 5’-ends of MaSOD cDNAs and their corresponding gDNAs. Coordinates are nucleotide positions relative to the translational start site (ATG). b Alignment of 3’-ends of MaSOD cDNAs

Full size image

Ten independent transcripts of MnSOD were classified into four MaMSD genes (named MaMSD1A, 1B, 1C and 1D) based on the sequence differences of their ORFs and deduced amino acids. Each MaMSD gene contained two to four types of transcripts with different 3’ UTR lengths (Fig. 1b-2, b-3 and b-4), resulting from APA, except MaMSD1D, which had a unique 3’ UTR (Table 1). APA sites were also found in the MnSOD genes of Larix gmelinii [39].

Multiple sequence alignments of the five MaFSD transcripts showed that they encoded two FeSOD genes with distinct ORFs. MaFSD1A-1,-2 and −3, co-encoded a 301-amino acid polypeptide, contained the same 906 bp ORF but had different 3’ UTR lengths (Fig. 1b-5). The other two cDNAs shared 100 % identity, except that one of them contained two extra sequences of 87 bp and 172 bp in the ORF region (Additional file 5). Aligning with gDNA sequences showed that these two extra sequences were introns retained in the mRNA during splicing, forming an alternative splicing transcript named MaFSD1B-variant1. Comparing MaFSD1B-variant1 with the normal transcript MaFSD1B revealed that its ORF was shorter with a premature stop codon.

In addition, the isoelectric points of the 12 MaSOD proteins ranged from 4.87 to 7.90, and their molecular weights varied from 15.0 to 34.2 kDa (Table 1). Pairwise similarities among the 12 MaSOD genes were performed and are listed in Additional file 6. The analysis revealed that the MaMSD genes shared 80.9–86.5 % identity at both the nucleotide and deduced amino acid levels, followed by MaCSD genes at 39.7–87.5 % and MaFSD genes at 33.3–42.5 %, whereas the similarity between genes from different MaSOD types, such as MaCSD1A and MaMSD1A, was below 35.6 %. A BLASTp search of the NCBI database revealed that the putative polypeptides of the MaSOD genes shared about 62.0–89.0 % sequence identity with orthologous SOD proteins in other plants.

Intragenome syntenic relationship analysis indicated that MaCSD2A and 2B, and MaMSD1A and 1C or 1D are derived from banana whole genome duplications (WGDs). Cross-genome syntenic analysis revealed that another five MaSOD genes (MaCSD1A, MaCSD1D, MaMSD1B, MaFSD1A and MaFSD1B) are contained in blocks shared among different plant species, suggesting that the other duplicated copies of them in banana genome are lost after WGDs.

Conserved motifs and clustering analysis of MaSODs

To elucidate the domain features and phylogenetic relationship of the MaSODs, an unrooted phylogenetic tree and a linear distribution map of the conserved motifs in the deduced MaSODs were generated (Fig. 2). MaSODs were clustered into three major clades, which showed a good accordance with their metal cofactor types. The three clades were designated as groups I, II and III. Group I contained two subgroups (Ia and Ib) and harbored motif 1 and motif 5, which contained Cu/ZnSOD signatures and conserved Cu2+ and Zn2+ binding sites (Additional file 7). Compared with Group Ia, Group Ib included an additional motif 8, which contained chloroplast peptide signal sequences, suggesting MaCSD2A and 2B were located in chloroplasts, whereas MaCSD1A-1D were cytosolic Cu/ZnSODs as predicted by SoftBerry. The MaCSD1D protein fused with GFP to co-express on onion epidermal cells, confirming the cytosolic localization (Fig. 3). Four MaMSDs formed group II, which contained the motifs of 2, 3 and 6. Motif 2 included the conserved metal-binding domain “DVWEHAYY” and five residues (Gly, Gly, Phe, Gln and Asp) [9] that were present in MnSOD but absent from FeSOD, while motif 6 included mitochondrial location signal sequences, suggesting they targeted to the mitochondria. Group III was made up of MaFSD1A and MaFSD1B, which shared motifs 3, 4 and 7. Residues (Ala, Gln, Trp, Phe and Ser) responsible for recognizing iron ion and active sites for FeSOD [40] were found in motif 4 and the conserved metal-binding domain “DVWEHAYY” was detected in motif 7.

Fig. 2
figure 2

Conserved motifs and clustering analysis of MaSOD proteins. Clustering tree resulting from the amino acid alignments of MaSODs is shown on the left of the figure. The distribution of conserved motifs among the deduced MaSOD proteins is shown on the right of the figure. Different motif types are marked by different color blocks as indicated at the bottom of the figure. Group Ia and Ib are copper/zinc superoxide dismutases, Group II are manganese superoxide dismutases, and Group III are iron superoxide dismutases

Full size image
Fig. 3
figure 3

Subcellular localization analysis of MaCSD1D. The upper panel presents the merged fluorescence, fluorescence and bright filed image of GFP control. The lower panel presents the merged fluorescence, fluorescence and bright filed image of MaCSD1D-GFP

Full size image

Gene structures of MaSODs

The gDNA lengths of MaSODs from the start to stop codons varied between 1807 and 4720 bp (Table 1). A gene-structure map was produced by aligning ORF gDNA sequences with their corresponding cDNA sequences (Fig. 4). Interestingly, the MaSOD gene structures could also be divided into four groups (Ia, Ib, II and III) based on their exon numbers, which is in agreement with the motif-based classification. MaSODs in the same group contained an equal number of exons, with one exception (group III). The four MaCSD1s in group Ia possessed seven exons. However, the size of the first exon in MaCSD1A was the same as the one in AtCSD3, while the size of the first exon in the other three MaCSD1s (MaCSD1B, 1C and 1D) was the same as that in AtCSD1. In group Ib, both MaCSD2A and MaCSD2B had eight exons, just like AtCSD2. Similar to previous reports of AtCSDs and PtCSDs [9], the size of the second exon in MaCSD1s (102 bp) corresponded to that of the second and third exons in MaCSD2s (62 + 40 bp). This indicates that the inferred exon merge (or split) was shared by monocot and dicot species and should thus to be a defining characteristic of angiosperms. The four MaMSD genes belonging to group II contained six exons. In contrast to the conserved genomic structures of groups Ia, Ib and II, the genes in group III exhibited different exon-intron organizational patterns. MaFSD1A contained eight exons (one less than its ortholog AtFSD2), while MaFSD1B possessed nine (one more than its ortholog AtFSD3) and MaFSD1B-variant1 harbored seven.

Fig. 4
figure 4

Gene structures of MaSODs and AtSODs. Exons are shown as black boxes, introns are shown as thin lines, and UTRs are shown as white boxes

Full size image

Additionally, many of the introns in banana SOD genes were longer than those in their Arabidopsis homologs, resulting in longer MaSOD gDNAs (Fig. 4). All MaSOD introns were spliced in accordance with the eukaryotic GU-AG splice junction site rules [41] with two exceptions: the fourth intron of MaCSD2A and the second of MaMSD1D, which had splice boundaries of GC-AG. Additionally, one intron was found in the 5’ UTRs of MaCSD1B and MaCSD1D gDNAs, while two were found in the 5’ UTR of MaCSD1C gDNA (Fig. 1a-1, a-3 and a-4).

Chromosomal locations and phylogenetic analysis

The chromosomal locations of MaSOD genes were determined by performing a BLASTn search against the banana genome databases. Eight out of the 11 chromosomes harbored MaSOD genes (Fig. 5), four (chr 2, 3, 8 and 10) of which possessed two MaSOD genes, while the others (chr 4, 7, 9 and 11) contained only one.

Fig. 5
figure 5

Chromosomal locations of MaSOD genes. The positions of SOD genes are marked with red lines on the chromosomes. Locus names of ‘DH-Pahang’ SODs are in brackets; for simplicity, the prefix “GSMU_” and suffix “_001” are omitted from each gene name

Full size image

To investigate the phylogenetic relationships of MaSODs with other plant SODs, an unrooted phylogenetic tree was constructed from an aligned dataset of MaSODs and 58 homologous SODs from M. acuminata var. DH-Pahang, Zea mays, Oryza sativa, Phalaenopsis equestris, A. thaliana, and P. trichocarpa. The two major clades of SODs were well supported in this tree (Fig. 6a). All Cu/ZnSODs formed a large clade comprising three subgroups (groups a, b and c). FeSODs were clustered with MnSODs into another large clade, indicating, as previous reports [6], that these two subgroups (groups d and e) originated from a common ancestor. MaSODs were found in all five subgroups, where they were clustered with ‘DH-Pahang’ SODs with strong bootstrap support; this close relationship is consistent with their high sequence identities (Additional file 4) and implies that they originated from the same gene.

Fig. 6
figure 6

Phylogenetic trees of the SOD genes from banana and other plants. a A maximum likelihood-based phylogeneitc tree of SOD families. GenBank accession numbers or locus IDs are in brackets. a–e refer to different gene clusters. b A phylogeneitc tree limited to the FeSODs in group e

Full size image

MaCSD1B, 1C and 1D grouped with AtCSD1 and other cytosolic Cu/ZnSODs in group a, while MaCSD1A was placed into another subgroup along with AtCSD3 (group c). The position of these MaCSD1s is consistent with the size of their first exons (Fig. 4), suggesting that two copies of CSD1 existed before the monocot/dicot split, with only one undergoing expansion in banana (to MaCSD1B, 1C and 1D). MaCSD2A and MaCSD2B clustered in group b with other plant chloroplastic Cu/ZnSODs.

Interestingly, all four MaMSDs, like PtMSDs and ZmMSDs, were found on a species-specific cluster of the MnSOD clade; this placement indicates that MnSOD gene number amplifications were mainly due to recent WGDs or segmental duplications [9, 31]. Two MaFSDs (MaFSD1A and 1B) fell into different subgroups in group e, only one of which was well supported (85 % bootstrap support). To confirm this result, a small phylogenetic tree limited to FeSODs of group e was constructed (Fig. 6b). In this tree, each MaFSD was separately grouped with FeSOD members from all considered species, revealing the existence of two ancestral FeSOD genes before the monocot/dicot split.

Isolation and bioinformatics analysis of putative MaSOD promoters

To further understand and determine the regulatory roles of MaSODs under various stresses, the regions upstream of the start codons were isolated by PCR. We obtained a length that varied from 1084 to 2114 bp of 5’-flanking sequences for all MaSOD genes except for MaCSD2B (Additional file 2). Potential regulatory cis-elements that were related to stress and light responses are predicted and summarized in Fig. 7. All 11 putative MaSOD promoters possessed typical TATA and CAAT boxes, which are the core elements of the promoters.

Fig. 7
figure 7

Cis-element analysis of putative MaSOD promoters related to stress responses. Different cis-elements with the same or similar functions are shown in the same color

Full size image

As shown in Fig. 7, light-responsive elements accounted for the majority of elements in every MaSOD promoter, excluding MaMSD1A, MaMSD1B and MaMSD1C. There were 25 different types of light-responsive elements present in the 11 banana SOD promoters. Every promoter possessed two to 10 types, which indicated that MaSODs might be differentially regulated when subjected to light. Ten kinds of hormone-responsive regulatory elements were found in the MaSOD promoters, such as ABRE, ERE, CGTAC-motif, TGACG-motif, GARE-motif, P-box, TATC-box, AuxRR-core, TGA-box and TCA element, which were associated with ABA, ethylene, methyl jasmonate (MeJA), GA, auxin and SA responses, respectively. Moreover, six types of stress-responsive regulatory elements, TCA elements, AREs, LTRs, HSEs, TC-rich repeats, MBSs and Box-W1s, with responses to anaerobic induction, low-temperature, heat stress, defense and stresses, drought inducibility and fungal elicitors, respectively, were identified in the MaSOD promoters. Different types and numbers of regulatory elements were present in distinct MaSOD promoters, indicating that MaSOD genes should have different regulatory mechanisms in response to various stress and hormone treatments.

Expression pattern of MaSOD genes in different tissues

qRT-PCR analysis was performed to assess the expression pattern of MaSOD genes in different organs of ‘Tianbaojiao’ (Fig. 8). Eleven of the 12 MaSOD genes were expressed in all tested tissues (leaf, pseudostem and root), whereas MaCSD2B was expressed only in leaf and root tissues. Different SOD-type genes in banana had similar expression patterns. MaCSD1D and MaFSD1A exhibited the highest expression levels in leaves, followed by pseudostems and roots. MaCSD2A and MaFSD1B were expressed strongly in pseudostems, moderately in leaves and weakly in roots. At mRNA level, MaCSD1B and MaMSD1B showed their maximum expression in roots, followed by pseudostems and leaves. Expression levels of MaCSD1C and MaMSD1A were low in leaves and pseudostems and, high in roots.

Fig. 8
figure 8

qPCR analysis of MaSOD genes in different tissues. Gene names are shown on the x-axis and the expression levels on the y-axis. Different tissues are shown in different colors

Full size image

Differential expression of MaSOD genes in response to various abiotic stresses

The expression patterns of the 12 MaSOD genes as seen by qPCR were detected under cold, heat, drought and NaCl treatments. The results showed that their expression modes were complex (Fig. 9). During cold treatment, the expression levels of most MaSOD genes exhibited slight variations. However, there were still four MaCSD genes showing obvious differential expression in response to cold stress (MaCSD2A and MaCSD1C were up-regulated, while MaCSD1D and MaCSD2B were dramatically down-regulated; Fig. 9a). The heat-treatment-responsive genes were found in all three SOD-types. Most of the MaSOD genes (MaCSD1B, MaCSD1D, MaMSD1A, MaMSD1B, MaMSD1C and MaFSD1A) were increased at the transcriptional levels, except MaCSD2A, which was down-regulated at 12 h (Fig. 9b). Obviously, most MaSOD genes were down-regulated in response to drought stress (Fig. 9c). Among them, five members (MaCSD1A, MaCSD1D, MaMSD1A, MaMSD1B and MaFSD1B) exhibited more than 2- to 10- fold decreases. In contrast, three Cu/ZnSOD genes, MaCSD1B, MaCSD1C and MaCSD2A, were strongly induced at 1 d, 5 d and 3 d, respectively. Under the NaCl treatment (Fig. 9d), MaCSD1D, MaMSD1A and MaMSD1B shared similar expression patterns, which increased gradually to high levels as the treatment continued. The expression of MaCSD1A and MaCSD1B was dynamic, increasing quickly at the 4-h point, then decreasing gradually, but finally increasing again. In addition, MaCSD1C, MaCSD2A, MaCSD2B and MaFSD1B were first up-regulated at 4 h, and then down-regulated over the duration of the treatment.

Fig. 9
figure 9

qPCR analysis of differential gene expression under abiotic treatments. The colors of the bar vary from green to red representing the scale of relative expression levels. Each column represents a sampling time point, and each row represents a MaSOD member. The expression level of the control (at 0 h/d; marked in black) in every treatment for each gene was used as the rescaled value when calculating the relative expression levels. The clustering results of gene expression patterns are shown on left. a–d refer to different abiotic treatments

Full size image

Differential expression of MaSOD genes in response to hormonal treatments

The expression of MaSOD genes in response to hormonal stimuli (ABA, GA3, IAA and SA) was investigated to further analyze the possible functions of MaSODs involved in phytohormone signaling pathways. Most MaSOD genes showed an up-regulation in response to hormonal treatments (Fig. 10). Ten out of the 12 MaSOD genes were up-regulated to different degrees, whereas the other two (MaCSD1C and MaCSD2A) were down-regulated under ABA treatment (Fig. 10a). Among the up-regulated genes, MaCSD1D exhibited a continuously high-level of transcript abundance over the 48 h time-course with an 8.9-fold peak at 12 h. During the GA3 treatment, only two genes (MaCSD1D and MaMSD1A) were strongly induced at 4 h and 48 h, respectively, while the other genes were slightly induced or repressed with no significant changes (Fig. 10b). The IAA treatment dramatically induced three MaCSDs and one MaMSD (MaCSD1A, MaCSD1D, MaCSD2B and MaMSD1A), but didn’t obviously repress any member (Fig. 10c). Likewise, the expression levels of most MaSOD genes were up-regulated in response to the SA treatment, but only MaCSD1D was greatly transcribed after 24 h when treated. In addition, the transcription of MaCSD1C remained at a low level without a significant change during the treatment time compared with the control (Fig. 10d).

Fig. 10
figure 10

qPCR analysis of differential gene expression under hormonal treatments. The colors of the bar vary from green to red representing the scale of relative expression levels. Each column represents a sampling time point, and each row represents a MaSOD member. The expression level of the control (at 0 h/d; marked in black) in every treatment for each gene was used as the rescaled value when calculating the relative expression levels. The clustering results of gene expression patterns are shown on left. a–d refer to different hormonal treatments

Full size image

Discussion

Expanded SOD family in banana

In this study, we identified 12 SOD genes representing three major types of plant SODs (Cu/ZnSOD, MnSOD and FeSOD) with 25 distinct transcripts from cultivated banana ‘Tianbaojiao’ (AAA group). In contrast to wild bananas, which possess two tandem FeSOD genes on chromosome 10, only one FeSOD, as assessed by mRNA levels, was isolated from ‘Tianbaojiao’. Further sequence analysis revealed that the two tandem FeSOD genes had different putative promoters (1.5 kb upstream of ATG) even thought their ORF and 3’ UTR sequences were similar. Forward primers specific to the two promoters and the same reverse primer (FSD1B-ORFR; Additional file 1) were used to amplify the two tandem FeSODs in ‘Tianbaojiao’. PCR amplification of each FeSOD using 10 forward primers at different starting positions still yielded only one sequence; this sequence, corresponding to GSMA_Achr10T27190_001 or ITC1587_Bchr10_T31275, was that of the MaFSD1B gene. Because the two tandem FeSOD genes are present in both the AA and BB genome of Musa, the tandem duplication likely predates the A and B genome independent evolution. We propose that the other FeSOD on chromosome 10 (corresponding to GSMA_Achr10T27220_001 or ITC1587_Bchr10_T31280) was lost in ‘Tianbaojiao’ over the course of evolution. MaSOD genes had more similar ORF lengths and higher sequence identities with their homologs in the AA genome than in the BB genome (Additional file 4). This observation provides additional evidence to support crosses within diploid M. acuminata as the source of the edible triploid M. acuminate [42].

The number of SOD genes varies among monocots (Additional file 8). For instance, the number of distinct SOD genes that have been in silico-identified ranges from 6 in Brachypodium distachyon (3 Cu/ZnSODs, 1 MnSOD and 2 FeSODs) and Hordeum vulgare (3 Cu/ZnSODs, 1 MnSOD and 2 FeSODs) to 7 in O. sativa (4 Cu/ZnSODs, 1 MnSOD and 2 FeSODs), Sorghum bicolor (4 Cu/ZnSODs, 1 MnSOD and 2 FeSODs) and Setaria italica (4 Cu/ZnSODs, 1 MnSOD and 2 FeSODs), 9 in P. equestris (5 Cu/ZnSODs, 1 MnSOD and 3 FeSODs), 10 in Z. mays (5 Cu/ZnSODs, 2 MnSODs and 3 FeSODs) and 18 in Triticum aestivum (10 Cu/ZnSODs, 2 MnSODs and 6 FeSODs). Excluding the SOD family of hexaploid bread wheat, banana harbors the largest monocot SOD gene family, especially with respect to MnSODs (four in total). The three whole-genome duplication rounds that have occurred during banana evolution are a major factor responsible for the amplification in gene numbers [25]. However, only five MaSOD members (MaCSD2A and 2B, and MaMSD1A and 1C or 1D) are retained in pairs after banana WGDs, while another five genes (MaCSD1A, MaCSD1D, MaMSD1B, MaFSD1A and MaFSD1B) are singletons with the loss of other duplicated copies after WGDs. MaCSD1B, 1C and 1D but not MaCSD1A clustered together in Group a (Fig. 6a) and had identical-sized first exons (Fig. 4) and high sequence similarities (Additional file 6); this indicates that MaCSD1B and 1C were probably derived from recent segmental duplications and shared the same origin with MaCSD1D.

More than one kind of transcript was obtained from seven of the SOD genes in the ‘Tianbaojiao’ banana. Genes, such as MaCSD2A, MaMSD1A, MaMSD1B, MaMSD1C and MaFSD1A, transcribe two to four different mRNAs as a result of APA. This is not surprising because APA transcripts have already been detected in Cu/ZnSOD, MnSOD and FeSOD genes from Dimocarpus longan, L. gmelinii, T. aestivum and Z. mays [10, 12, 43]. Like APA, AS is a regulatory mechanism to generate multiple transcripts from one single gene. In this study, two kinds of AS occurred in banana SODs. In one, the introns were retained in the 5’ UTR region of MaCSD1B mRNA when transcribed, resulting in a normal ORF but a longer 5’ UTR. In the other one, the introns resided in the coding region of MaFSD1B, resulting in a premature stop codon, producing a truncated protein (228 aa). Similar observations were made in other plant Cu/ZnSOD and FeSOD genes [10, 18, 44]. A previous report in rice suggested that introns that resided in the 5’ UTR region were linked to regulating spatially and temporally gene expression [45]. In addition, the rice FeSOD variant with a short ORF is sensitive to low temperatures and its truncated protein is active [46], indicating that a similar variant of the MaFSD1B gene may play a crucial role in the responses to some abiotic stresses. In spite of these, ATSS also provides a regulatory method to generate multiple transcripts. Although a previous study showed that each type of longan SOD genes has multiple transcriptional start site [10], we found only two MaSOD genes (MaCSD1B and MaCSD2A) have ATSSs.

In addition to the above findings, allelic SOD genes (MaCSD1A-1 and MaCSD1A-2, MaMSD1C-1 and MaMSD1C-2) with single nucleotide substitutions were detected in the triploid banana ‘Tianbaojiao’. This discovery demonstrates that allelic variation exists inside its three A genomes, which is apparently another mechanism promoting RNA and protein diversity in heterozygotes. Furthermore, the sequence polymorphism found between the SOD genes of ‘DH-Pahang’ and ‘Tianbaojiao’ (Additional file 4) is direct evidence of sequence variability in the A genomes.

Taken together, our data provide more detailed sequence information and transcriptional regulatory types for banana SODs. Whole genome duplication, segmental duplication and the complex regulation of transcription have contributed to the gene expansion and mRNA diversity of banana SODs.

Specific MaSOD members are involved in different abiotic stress responses

Three types of plant SODs (Cu/ZnSOD, MnSOD and FeSOD) are all reported to be in involved in ROS scavenging caused by abiotic stress [8, 16]. The banana Cu/ZnSOD subfamily comprises six members (MaCSD1A-1D and MaCSD2A-2B). An expression analysis revealed that every MaCSD gene was responsive to at least one abiotic stress treatment performed in this study (cold, heat, drought or salinity) (Fig. 9). Notably, MaCSD1D is the only member that showed obvious expression changes under all four abiotic stresses. Compared with other MaCSD promoters, the MaCSD1D promoter harbored more kinds and numbers of cis-elements involved in abiotic stresses, including one LTR motif (cis-element involved in low-temperature responsiveness), one HSE motif (cis-element involved in heat stress responsiveness) and two MBS motifs (MYB binding site involved in drought-inducibility), which could explain why MaCSD1D showed obvious responses to the four abiotic treatments. This indicates that MaCSD1D might play a predominant antioxidant role in banana. Similarly, various responses to the abiotic stress were also identified in other plant Cu/ZnSOD subfamiles [47].

Banana harbors four MnSOD genes (MaMSD1A-1D) with high sequences similarities (80.9-86.5 %), which were in accord with their phylogenitic clustering results. Interestingly, individual MaMSD genes have some distinguishing features from each other, especially in the 5’ and 3’ UTR regions, indicating that they have evolved over the time and probably underwent modifications to form regulatory diversification under constantly changing environments. This was confirmed by their having different cis-elements involved in stress responses. For instance, two TC-rich repeat motifs (cis-elements involved in defense and stress responsiveness) were present in the MaMSD1A promoter but absent from MaMSD1B, 1C and 1D. A qPCR analysis showed that MaMSD1A was responsive to heat, drought and salt stress, while MaMSD1B was responsive to heat and drought stress. MaMSD1C was only responsive to heat stress, and MaMSD1D was not responsive to any stress. This suggested that the four MnSOD genes in banana play distinct roles in scavenging ROS caused by different stimuli.

Additionally, MaSOD members located in the same subcellular department also exhibited different expression patterns. The expression of chloroplastic MaCSD2B decreased during the cold treatment while that of MaCSD2A increased, and vice versa during the NaCl treatment. Likewise, drought up-regulated the expression of cytosolic MaCSD1B and MaCSD1C, whereas it down-regulated MaCSD1A and MaCSD1D. MaMSD1A and MaMSD1B exhibited increased expression levels under NaCl stress while the other two members (MaMSD1C and MaMSD1D) showed little change in expression. Our data also show that heat stress represses the expression of chloroplastic MaCSD2A but strongly induces chloroplastic MaFSD1A. In transgenic tobacco, the over-expression of chloroplastic FeSOD suppressed the expression of both chloroplastic and cytosolic Cu/ZnSODs, and the authors proposed that chloroplastic FeSOD over-expression interfered with a signal pathway regulating the Cu/ZnSODs through a low superoxide radical concentration [48]. Therefore, we hypothesize that there may exist some inter and cross-family signal pathways regulating the expression of MaSODs under various environmental stresses.

In addition, the promoters of MaCSDs and MaFSDs contain large amounts of light responsive cis-elements of 5 to 10 different types (Fig. 7). Kurepa et al. detected an increased accumulation of both Cu/ZnSOD and FeSOD transcripts in tobacco when exposed to light stress [49]. Similar observations were also detected in Arabidopsis and rice [7]. Tobacco plants harboring an over-expressed Cu/ZnSOD gene from pea exhibited an increased tolerance against high light [50]. Therefore, we propose that MaCSDs and MaFSDs might participate in light responses.

MaSODs probably participate in phytohormone signaling pathways

Hormone-responsive transcription factors are known to act via combination with their corresponding cis-elements in the promoter to regulate the expression of target genes during various stresses [51]. The PlantCARE database predicted that four MaSOD promoters (MaCSD1A, MaCSD2A, MaFSD1A, MaFSD1B) contained two to four ABREs, a cis-acting element involved in the abscisic acid responsiveness, indicating that these genes probably participate in ABA responses. The mRNA levels of MaCSD1A, MaFSD1A and MaFSD1B were induced to 1.8-fold, 1.5-fold and 1.8-fold, respectively, at 12 h under ABA treatment, whereas the transcripts of MaCSD2A were slightly reduced over the treatment’s time course. Moreover, the expression of three other members (MaCSD1D, MaCSD2B and MaMSD1A), which had no ABREs, showed >2.0-fold expression inductions during the ABA treatment, suggesting that there are other regulation mechanisms responding to ABA. In the recent study, ABA was shown to regulate the expression of miR398 [52], which was negatively correlated with that of its target genes (Cu/ZnSODs) [19, 53]. MiR398 was also reported to exist in banana [25], and a sequence analysis found binding sites for miR398 in the mRNAs of MaCSD2A, but not in MaCSD1A mRNA, which may explain why the MaCSD2A was down-regulated under ABA treatment. We therefore proposed that the expression of MaSOD genes in response to ABA may be synergistically mediated by ABREs and miRNAs. In addition to ABA, tobacco SOD genes were also reported to be responsive to auxin, gibberellin A and other substances [54]. In this study, we predicted using the PlantCARE database that there were eight other hormone-responsive cis-elements (ERE, CGTA-motif, TGACG-motif, GARE-motif, P-box, TATC-box, AuxRR-core and TCA-element) located 5’-upstream of MaSODs. We confirmed their transcriptional regulation under GA3, IAA and SA treatments using qPCR. Furthermore, H2O2 produced in the dismutation reaction by SODs is also a signal molecular that can interact with phytohormones to affect various metabolic processes in the cell under stress [55–57]. We hypothesize that MaSODs probably participate in phytohormone signaling pathways.

Conclusions

Banana harbors 12 SOD genes, including three types of plant SODs (Cu/ZnSOD, MnSOD and FeSOD). Whole genome duplication, segmental duplication, APA, AS and ATSS have contributed to the gene expansion and mRNA diversity of banana SODs. The 12 MaSODs were distributed on eight out of the 11 banana chromosomes. Based on structural characteristics, the 12 MaSODs were divided into four groups. Promoter sequence analyses revealed that there were many abiotic and hormonal-responsive cis-elements in the 5’ upstream regions of the MaSODs, but distinct members harbored different types and numbers, which suggested that the 12 MaSODs were differentially regulated. A qPCR analysis revealed that distinct MaSOD genes exhibited different expression patterns in response to abiotic and hormonal stresses, which indicated that specific MaSOD members play roles in different aspects of banana abiotic stress tolerance and hormonal signaling pathway.