Background

Members of the Esi3/RCI2/PMP3 gene family have been shown to be up-regulated by environmental stresses and to contribute to stress tolerance in studies using model species. The Early Salt-Induced 3 (Esi3) gene was initially identified as a salt-stress induced gene in Lophopyrum elongatum, a salt tolerant and close relative of bread wheat, Triticum aestivum [1]. It was reported to be more strongly induced by stress in L. elongatum than the less tolerant T. aestivum and the intermediately salt tolerant amphiploid derived from a cross between L. elongatum and T. aestivum [2]. A close homolog was also identified as a low temperature induced gene in Hordeum vulgare (barley), Blt101 [3]. The genes encode small molecular weight hydrophobic proteins that are members of gene families that have been identified in a wide range of eukaryotic and prokaryotic organisms including more than 150 plant species [4]. The homologous gene family in Arabidopsis thaliana, referred to as Rare Cold Inducible 2 (RCI2) [5], has been shown to be up-regulated by environmental stresses such as low temperature, salt and dehydration [6,7,8]. The eight gene family members in Arabidopsis have been divided into two groups; Group I, which encode proteins of 52 to 64 amino acids, and Group II which encode longer proteins, of approximately 60 to 89 amino acids, with longer C terminal hydrophobic extensions which contain charged amino acids [4]. A homologous gene has also been described in Saccharomyces cerevisiae, known as Plasma Membrane Protein 3 (PMP3), the deletion of which resulted in a decrease in plasma membrane potential and caused sensitivity to Na+ and cytotoxic cations such as tetramethylammonium and hygromycin B. This suggests that this protein contributes to the regulation of intercellular ion homeostasis by controlling plasma membrane potential and preventing excessive Na+ influx [7]. The Arabidopsis RCI2A gene was shown to complement the deletion of yeast PMP3, indicating that they are functionally interchangeable [7,8,9]. The Δpmp3 mutation in S. cerevisiae can be complemented by many but not all members of the gene families from other plant species including Zea mays [10], Medicago truncatula [11], Oryza sativa [8] and the alkali grass, Puccinellia tenuiflora [12].

Overexpression of RCI2A in transgenic Arabidopsis plants was shown to decrease Na+ uptake, mitigate the salt-induced damage and enhance growth under salt stress conditions [13], whereas the disruption of RCI2A led to over-accumulation of Na+ and increased salt sensitivity [14] which suggests that RCI2 proteins play major roles in ion homeostasis in response to salt stress. These results were confirmed by experiments in which the MpRCI2 from plantain, Musa paradisiaca, was overexpressed in Arabidopsis [15]. Overexpression of RCI2 in Nicotiana tobacum increased tolerance to low temperatures [16]. The overexpression of the Z. mays gene ZmPMP3–1 in the A. thaliana rci2 mutant background resulted in higher levels of salt tolerance than either the rci2 mutant or wild type (WT) plants [10]. In response to NaCl treatment, the transgenic plants showed lower levels of Na+ accumulation and enhanced K+ accumulation relative to the mutant or the WT, though the increased K+ accumulation phenotype was not observed in lines that overexpressed AtRCI2A [13].

While the mechanisms by which these proteins act are not well known, seven of the A. thaliana RCI2 proteins [4], eight RCI2-like proteins identified in Zea mays [10] as well as several RCI2-like genes from M. truncatula [11] have been shown to localize to the plasma membrane when they are transiently expressed as green fluorescent protein (GFP) fusion proteins in onion epidermal cells. The Arabidopsis and M. truncatula studies also reported single members of the gene family that were also localized to internal cellular membranes [4, 11]. The Esi3/RCI2/PMP3 proteins are not thought to directly affect sodium transporters; combining the Δpmp3 deletion with the deletion of the two major Na+ expelling transporters in S. cerevisiae, Δpmr2 and Δnha1, increased salt sensitivity of the mutants which suggests that PMP3 does not likely act directly in conjunction with those ion pumps. The Esi3/RCI2/PMP3-encoded proteins are predicted to have two transmembrane domains which alone would not be sufficient to form transmembrane pores, though oligomeric complexes have been hypothesized to have a potential to do so [5]. Recent studies carried out with PMP3 in Saccharomyces suggest that it plays a role in plasma membrane organization and is part of the regulatory mechanism for vesicle movement between the plasma membrane and vacuole [17]. Additionally, it is involved in the regulation of levels of phosphoinositides and sphingolipids [17].

Members of the Esi3/RCI2/PMP3 gene families studied in A. thaliana and Z. mays displayed differential expression in developing tissues and in response to various abiotic stresses, which suggests that they are involved in separate signaling pathways and that those gene family members may play divergent roles [5, 10, 18]. The Esi3s that were first identified in the highly salt tolerant L. elongatum are thought to be members of a small gene family in the genomes of the closely related Triticeae species and to have gene family members represented in the three genomes of hexaploid bread wheat, T. aestivum. The allohexaploid genome of T. aestivum is the result of two polyploidization events. The first event occurred between the diploid Triticum urartu and a species thought to be closely related to Aegilops speltoides, which donated the A and B genomes respectively, and which resulted in the tetraploid species Triticum turgidum. The second allopolyploid speciation event involved a cross between T. turgidum and Aegilops tauschii, which contributed the D genome and occurred approximately 8000 years ago [19,20,21]. The three genomes have a high degree of synteny and DNA sequence similarity.

Here we report the identification of all homologous sequences of Esi3 genes within the genome of T. aestivum, and analyze the differential expression patterns of these genes. Esi3 gene family members were identified in cDNA and genomic databases. In addition, gene expression levels in response to stress treatments were characterized from microarray and transcriptome sequence sources. To elucidate the evolutionary relationships between the Esi3 homologues across plant species, the gene sequence similarity was compared to those of other monocot species including Ae. tauschii, Sorghum bicolor, Brachypodium distachyon, Hordeum vulgare, Secale cereale, O. sativa, Z. mays, and to the dicot A. thaliana.

Results

Esi3 genes in Triticum aestivum

Twenty-nine Esi3/RCI2/PMP3 genes were found in the hexaploid genome of T. aestivum. These represent ten paralogous loci in the haploid genome, each with three homeologous copies in the A, B and D genomes with the exception of Esi3–2 for which the B copy was not identified (Table 1). In most cases, the sequences of the gene family members were confirmed in three independent sequence databases, the International Wheat Genome Sequencing Consortium (IWGSC) database of genomic chromosomal survey sequences [22], and the transcriptome shotgun assembly (TSA) and expressed sequence tag (EST) databases for T. aestivum at the National Center for Biological Information (NCBI). Additional confirmation of genome assignment and delineation of coding sequences (CDS) were obtained from the databases for other species of the Triticeae tribe. There was no evidence for an Esi3–2-B gene copy in the B genome of T. aestivum found in the IWGSC genomic sequence database, the EST or TSA databases at NCBI nor in the databases for other B genome-containing species, T. turgidum and Aegilops speltoides. All members of the gene family had two exons and one intron. Annotations for twenty-four of these gene family members agreed with the most current annotation of the T. aestivum genome assembly at the Ensembl Plants database [23] with the exception of five genes (Esi3–6-B and -D, 7-B, 10-A and 10-D). The genes that did not agree with our annotations have either extended N terminal domains, mismatches in the protein sequence, or a lack of annotations as protein coding regions; these differences are listed in Additional file 1: Table S1.

Table 1 Esi3 genes of Triticum aestivum
Full size table

All members of the Esi3 gene family have a core conserved protein sequence encoding two transmembrane domains. The gene family can be delineated into three groups based on the presence or absence of C terminal or N terminal extensions of the CDS. Group I includes five of the paralogous gene sets, Esi3–1, −2, −4, −5, and −8, which encode proteins of the classical Esi3/RCI2/PMP3 lengths of 54–57 amino acids with two conserved transmembrane domains. Group II contains four paralogous sets of genes, Esi3–3, −6, −7, and −10 which encode proteins between 71 to 79 amino acids in length that include C terminal extensions between 14 and 20 amino acids similar to the Group II genes described in Arabidopsis and other species [5]. These extensions are rich in hydrophobic amino acids but also contain charged amino acids. Group III contains Esi3–9-A, -B and -D genes that encode N terminal extensions of 49–79 amino acids relative to the other members of the gene family. These N terminal extensions were rich in hydrophobic amino acids and have charged amino acids distributed throughout, a pattern that is similar to the C terminal extensions seen in Esi3–3, −6, −7, and −10. At the junction between the N terminal extensions and the conserved domains the proteins had runs of five consecutive valines followed by five prolines. Esi3–9-like proteins with long N terminal extensions similar to those described above are not found in Arabidopsis nor in the Esi3/RCI2/PMP3 gene families in other well characterized species, and so we have designated them as Group III.

The degree of nucleotide sequence similarity for paralogous copies of the Esi3 genes of T. aestivum within the core conserved sequences, excluding the N or C terminal extensions, for paralogous members of the gene family ranged from 62 to 95% (Additional file 2: Table S2). Amino acid sequences among paralogous gene copies ranged between 47 to 93% identity. The three homeologous gene copies for each of the paralogous groups had amino acid sequence identity ranging from 95 to 100%. The Group I genes, Esi3–1, −2, −4, −5, and −8, had 100% amino acid sequence identity among homeologous copies (Additional file 2: Table S2). Homeologous copies of Esi3 genes were localized to homeologous chromosomes. The ten paralogous Esi3 genes were localized to five different chromosomes. Two pairs of genes appear to be in tandem arrangement. The two most similar pairs of paralogs, Esi3–1 and Esi3–2, which share approximately 92% nucleic acid identity, were both localized to chromosome 4. The two genes are approximately 109 kb and 27 kb apart on the short arm of chromosome 4A and the long arm of chromosome 4D, respectively. Esi3–3 and Esi3–10 which are approximately 90% identical are roughly 4 kb, 54 kb and 109 kb apart on chromosomes 5A, 5B and 5D, respectively [22]. In the Ae. tauschii genome assembly Esi3–1 and Esi3–2 are 20 kb apart on chromosome 4, and Esi3–3 and Esi3–10 are approximately 57 kb apart on chromosome 5 [24, 25].

Genes with high sequence similarity to the Esi3/RCI2/PMP3 gene family members are found in a wide range of plant species including monocots, dicots, and mosses. Esi3–9-like genes, which have long N terminal extensions with high sequence similarity, were only identified in other species in the Triticeae. Searches in the GenBank protein databases identified homologous gene family members with similar N termini in Ae. tauschii, H. vulgare and Secale cereale (XP_020160007.1, BAK07288.1, GCJW01018134.1, respectively; Additional file 3: Table S3) though an Esi3-like gene with a long N terminal extension but with low sequence similarity in this domain was identified in the salt tolerant dicot Eutrema salsugineum (formerly Thellungiella halophile, GBKH01000241.1). Esi3s encoding proteins with C terminal extensions, like those of Esi3–3, −6, −7 and −10 were found in the EST databases for many other monocotyledonous and dicotyledonous species but not in the moss, Physcomitrella patens.

Esi3/RCI2/PMP3 gene family members were also identified in other monocotyledonous species in which the complete Esi3/RCI2/PMP3 gene families were not previously described. Using the nucleotide collection, EST and TSA databases at GenBank, ten Esi3 gene family members were identified in two other species of the Triticeae: Secale cereale and Ae. tauschii. The gene family in H. vulgare is comprised of nine members and includes two genes that had previously been reported [3, 26]. The complete Esi3 gene family composition is summarized in Additional file 3: Table S3. Seven Esi3/RCI2/PMP3 gene family members were also identified in Sorghum bicolor, one of which was reported by Khurana et al., 2015 [26] (Additional file 3: Table S3). Two novel Esi3/RCI2/PMP3 gene family members were identified in B. distachyon in addition to the six gene family members previously described [5]. A small number of discrepancies were found between the previously described Esi3/RCI2/PMP3 gene families for O. sativa and Z. mays and the Ensembl Plants database. There were also a number of differences between our compiled amino acid sequences and the annotation of the genes in BLAST at Ensembl Plants [23]; these are summarized in Additional file 3: Table S3 and Additional file 1: Table S1. Protein sequences for the Esi3/RCI2/PMP3 gene families for these species can be found in Additional files 4 and 5: Sequences S4 and S5.

Tissue-specific expression of Esi3 genes

The members of the Esi3 gene family showed varied tissue-specific patterns of gene expression in both transcriptome and microarray datasets. One transcriptomic analysis of five tissues including developing seed, root, leaf, stem and inflorescence [27] was assayed for the 29 gene family members; expression levels ranged from 151.8 RPKPM to undetectable (Fig. 1 and Additional file 6: Table S4). Homeologous groups within Esi3–1, −2, −3, −4, −5 and −10 had relatively high level of expression compared to Esi3–6, −7, −8, and −9. Most paralogous groups that were highly expressed had high levels of expression in more than one tissue type. For example, Esi3–5, the most highly expressed paralogous group, was highly expressed in the root, leaf and stem. Esi3–5 was also expressed in the seed and inflorescence but at lower levels. In tissues in which Esi3 gene expression was detected, all three members of the homeologous groups were detected, though at different levels, with the exception of Esi3–10-A. The genome of origin of the most highly expressed homeolog varied from among the paralogs, for example for both Esi3–1 and Esi3–2 the A homeolog was more highly expressed than the B or D copy, whereas for Esi3–10 the D homeolog was detected at higher levels than its corresponding homeologs in all tissues with the exception of the seed (Fig. 1). The extensive transcriptome analysis of seventy-one tissue types in Azhurnaya Spring Wheat also showed that each paralog has varying expression across the tissue panel (Additional file 7: Table S5) [28]. Anther tissue showed relatively high levels of expression of Esi3–1-A, Esi3–4-A, -B, −D, Esi3–5-A, −D and Esi3–9-A, -B (Additional file 7: Table S5). Esi3–8-A had high levels of expression in different tissues of grains at various stages of development but was undetected in most other tissue types. Esi3–6 had low levels of expression across all tissue types assayed (Additional file 7: Table S5). Esi3–5-A, -B and -D had high expression levels in root tissues at different stages of development with the highest levels in the root at the seedling stage (Additional file 7: Table S5). This extensive analysis indicates that the Esi3 gene family members have very dynamic tissue specific and developmentally regulated patterns of expression.

Fig. 1
figure 1

Tissue-specific gene expression. Esi3 gene expression in tissues of T. aestivum, the tissues analyzed included the seed at the whole plant ripening stage, the root at the cotyledon emergence stage, the leaf when the seed was 30 to 50% developed, the stem at the two nodes or internodes visible stage, and the inflorescence when maximum stem length was reached. Esi3–6 and Esi3–9 are not graphed because the RPKPM values were lower than 1.0 across all tissue types for all homeologous copies. Quantification is from RNA-seq data and values were normalized to reads per kilobase per million (RPKPM)

Full size image

Tissue-specific expression was also analyzed in a panel of thirteen tissue types (Additional file 8: Figure S1 and Additional file 9: Table S6) assayed by the Affymetrix 61 k wheat microarray dataset described by Schreiber et al., 2009 [29]. All paralogous gene family groups were represented on the microarray with the exception of Esi3–6. We note that the microarray cannot distinguish between homeologous gene copies in wheat and expression levels were taken as those of homeologous groups that include the A, B and D copy of each paralogous group. The most striking result from the microarray data was the high level of expression of Esi3–9 in the anthers and its low levels of detection in any other tissue (Additional file 8: Figure S1 and Additional file 9: Table S6). Gene family members Esi3–1 to Esi3–5 have relatively high levels of expression in all tissues, a finding that was similar to the RNA-seq data noted above. In addition low levels of expression of Esi3–1, −8 and −10 were detected in the anthers, and there were low levels of expression of Esi3–2, −3 and −9 in the immature inflorescence. Esi3–3 showed low levels of expression in the caryopsis 3–5 days after pollination. Esi3–7 and Esi3–10, overall have very low levels of expression across all tissue types assayed.

Expression of Esi3 genes under drought and temperature stress conditions

The Esi3 gene family’s expression in leaf tissue in response to osmotic stress and high temperature identified five homeologous groups of the gene family that had significant changes in expression (Fig. 2 and Additional file 10: Table S7). Transcriptomic data from Liu et al., 2015 [30] characterized global changes in gene expression in response to short term osmotic stress treatment with PEG and high temperature treatments in seedlings grown on petri dishes. The experiments revealed a strong induction of gene expression for members of the homeologous groups Esi3–1 and −3, and modest induction of Esi3–2, −4, and −7. All three homeologs of Esi3–1 were up-regulated more than two fold by one hour of osmotic stress and up-regulated over 20 fold by six hours of stress. The Esi3–1-A copy had the highest level of expression in control conditions and was induced 20 fold with six hours of osmotic stress, whereas the B and D copies of the gene had much lower levels of expression in the control plants but were induced 44 and 89 fold, respectively. Nevertheless, the Esi3–1-A copy had the highest absolute level of expression in the stressed plants. The Esi3–1 homeologs were down-regulated by heat treatment and by combined osmotic stress and heat treatment of one hour, however the A copy of the gene was induced two fold by six hours of the combined heat and osmotic stress treatment (Fig. 2 and Additional file 10: Table S7). Similarly, the Esi3–3-B copy was highly up-regulated under six hours of osmotic stress. All homeologs of Esi3–5 had relatively high levels of expression in control conditions, and were down-regulated by drought, up-regulated by heat treatment and up-regulated by combined drought and heat treatment (Additional file 10: Table S7). Alternatively, all homeologs of Esi3–2 were highly induced after 6 h of drought meanwhile the homeologs are unaffected by the other stress conditions (Fig. 2). Esi3–10-D was up-regulated 5 fold after six hours of mixed heat and drought stress, compared to the 0.5–2-fold induction under the other conditions (Additional file 10: Table S7). Esi3–4 and Esi3–7 were moderately up-regulated by drought, heat and combined drought and heat treatment (Fig. 2). Most of the homeologous members of the Esi3 gene family responded similarly to stress conditions, though the absolute level of expression and the degree of change in response to stress varied a few fold among the members of the homeologous groups.

Fig. 2
figure 2

Short term osmotic and heat stress. Esi3 gene expression in seedling leaves in response to drought, heat, and combined drought and heat at 1-h and 6-h treatment times. Measurements are based on RNA-seq data. Gene expression differences for a Esi3–1, b Esi3–2, c Esi3–3, d Esi3–4, and e Esi3–7 are represented in reads per kilobase per million (RPKPM). Esi3s with significant expression changes are shown. Letters signify Duncan’s multiple range values

Full size image

The Esi3 gene family’s expression in leaf tissue in response to long term drought and high temperature stress was also analyzed in the microarray datasets for two durum wheat, T. turgidum, cultivars that differed in their degree of water use efficiency. The Cappelli cultivar has high water use efficiency and Ofanto has low water use efficiency [31]. The abiotic stress regimes were applied to soil-grown plants at booting stage. Though the changes of expression for the Esi3 genes were generally similar in the two cultivars, there were a few striking differences (Fig. 3 and Additional file 11: Table S8). Cappelli had higher expression of Esi3–3 under control conditions, though it had higher levels of induction in Ofanto in response to drought treatment. Nevertheless, the resulting levels of expression in Cappelli remained higher than Ofanto after being subjected to drought stress. Esi3–1 was down-regulated by drought in Cappelli but up-regulated in Ofanto. The drought tolerant cultivar Cappelli had higher levels of expression of Esi3–8 than the Ofanto cultivar under control, drought and heat treatments, but Ofanto showed Esi3–8 expression levels similar to Cappelli under mixed drought and heat treatment. Esi3–5 was more strongly induced by heat stress in Cappelli than in Ofanto, inversely Esi3–10 was down-regulated by heat in Ofanto. Esi3–8 was induced by combined drought and heat treatment in Ofanto. The response to mixed drought and heat treatments was complex, in the case of Esi3–5 the induction due to heat was similar to the gene induction seen with heat and drought in Cappelli. In other cases, the induction due to treatments appears to be partially additive, as in Cappelli’s Esi3–8, or synergistic in the case of Esi3–8 and −5 in Ofanto (Fig. 3 and Additional file 11: Table S8). The differences in changes in Esi3 expression in the two cultivars in response to heat and drought suggests that drought tolerant genotypes may have adaptive gene expression patterns.

Fig. 3
figure 3

Long term drought stress. Esi3 expression in response to drought, heat and both stresses combined compared between (a) Cappelli an efficient water use line and (b) Ofanto a water use inefficient line. Data is from Affymetrix microarray analysis. Values are in log2 units, thus an increase by 1 unit represents a two-fold increase, background of 3.4 subtracted from each value. Conditions were those described in Aprile et al., 2013 [31]; in brief, plants were stressed at booting stage by withholding water to 12.5% soil water content. Heat treatment was done by incremental increases to 40 °C, and the combined treatment used the same two conditions. Letters signify Duncan’s multiple range values

Full size image

Esi3 gene expression patterns in response to drought were also compared between the T. aestivum cultivar Chinese Spring, the tetraploid T. turgidum cultivar Creso, and a Chinese Spring genetic line with the partial deletion of the long arm of chromosome 5AL (CS_5AL-10), based on microarray datasets of Aprile et al., 2009 [32]. The two species are closely related, the tetraploid T. turgidum being an ancestral species to T. aestivum and sharing the A and B genomes and genes with approximately 99% sequence identity. The T. aestivum cultivar Chinese Spring has higher water use efficiency than the T. turgidum cultivar Creso [32] and other studies have shown a general trend for T. aestivum to be more drought tolerant than T. turgidum [33]. The comparison demonstrated different responses to drought stress for Esi3s in these two closely related species (Fig. 4 and Additional file 12: Table S9). Esi3–1 and Esi3–3 had significantly higher constitutive levels of expression in Creso than in Chinese Spring. Esi3–10 only has generally high expression levels in Creso and lacks expression in the other cultivars assayed. Esi3–1 was up-regulated 1.4 fold in Chinese Spring in response to drought treatments, and moderately down-regulated in Creso (Fig. 4 and Additional file 12: Table S9). Esi3–3 was more strongly up-regulated in Chinese Spring than in Creso. Esi3–8 has moderately higher levels of mRNA in Chinese Spring in control conditions but it was more strongly induced in Creso than it was in Chinese Spring. Comparisons between Chinese Spring and a derivative genetic line with the 5AL chromosome partially deleted showed a similar induction of Esi3–1 but showed expression for Esi3–3 that was approximately 1/10th than that observed in other genotypes (Fig. 4 and Additional file 12: Table S9). The Esi3–3 and Esi3–10 genes are located on the long arm of chromosomes 5. Thus the deletion of chromosome 5AL may lower the level of expression for the homeologous group by nature of the deletion of the Esi3–3-A and Esi3–10-A gene copy. However, the high level of the reduction in expression suggests that it is due to the loss of a regulatory gene on chromosome 5AL that affects the expression of Esi3–3 and Esi3–10, as this chromosome has been reported to contain several genes associated with the response to abiotic stress [34].

Fig. 4
figure 4

Long term drought stress. Esi3 response to mild and severe drought stress in (a) the T. turgidum cultivar Creso, (b) the T. aestivum cutivar Chinese Spring (CS) and (c) a Chinese Spring line with a deletion in the 5AL chromosome (CS_5AL-10). Esi3 expression was quantified with Affymetrix microarrays, Esi3–10 is not represented on CS and CS_5AL-10 graphs since values are zero. Plants were grown as described by Aprile et al., 2009 [32]; briefly, soil grown plants were treated at anthesis at different levels of stress: Control field capacity 28% water content, mild stress 18% water content or severe stress at 12.5% water content. Background of 3.3 subtracted from all values, values are in log2 units. Letters signify Duncan’s multiple range values

Full size image

Members of the Esi3 gene family were also shown to have altered levels of expression in wheat leaf tissue in response to cold treatment (Fig. 5 and Additional file 13: Table S10). Esi3–1-A, -B and -D copies showed more than 6, 12 and 30-fold increase in expression levels, respectively, as well as high absolute levels of expression in leaf tissue of plants after they were shifted to growth from 23 °C to 4 °C for two weeks (Fig. 5 and Additional file 13: Table S10) [35]. Esi3–2-A, Esi3–3-B, Esi3–3-D, and Esi3–10-D also showed significant increased levels of expression though their absolute levels of expression were lower than Esi3–1’s. Esi3–4-B and -D as well as Esi3–5-A, -B and -D showed significant reduction in the levels of transcripts in response to cold treatment. In most cases homeologous copies of gene family members responded similarly, they tended to be either all induced or all repressed by cold treatment, though their absolute level of expression and the degree of change in expression within homeologous groups varied several fold; in multiple cases the changes in expression for some members of the homeologous groups were not statistically significant.

Fig. 5
figure 5

Short term cold treatment. Gene expression changes in Esi3 in response to cold treatment at 4 °C. (a) Esi3–1, (b) Esi3–2, −3, −4, -5, and −10 (c) Esi3–6, −7, and −8 expression in response to cold treatment. The Esi3s are separated into three graphs due to large differences in gene expression corresponding to large differences in Y-axis values. Esi3–9 is not graphed since there was no gene expression detected. * indicates significant changes in gene expression determined by a one-way ANOVA. Values are normalized to reads per kilobase per million (RPKPM)

Full size image

Fusarium graminearum infection

Analysis of gene expression profiles by transcriptome sequencing for developing spikes inoculated with Fusarium graminearum [36] showed striking up-regulation for Esi3–4-A and Esi3–4-D at 24 and 48 h after inoculation. The up-regulation observed in both the Fusarium resistant line NIL 38 and the susceptible line NIL 51 was statistically significant, particularly at 48 h post inoculation (Fig. 6 and Additional file 14: Table S11). Esi3–5-A, -B, and -D as well as Esi3–9-A and -B showed higher absolute levels of expression than the Esi3–4s, but levels were not significantly changed in response to pathogen inoculation (Additional file 14: Table S11). However, the D copy for Esi3–3 was up-regulated 6.5 fold 24 h post inoculation in the susceptible line NIL 51, however there was no induction in the resistant line NIL 38 (Additional file 14: Table S11). In fact, the only tissues in which Esi3–9 expression was detected was in the spike and, as mentioned above, in the anthers.

Fig. 6
figure 6

Fusarium graminearum inoculation. Esi3–4-A and Esi3–4-D changes in gene expression in NIL 38, a disease resistant line and NIL 51, a disease susceptible line, in response to F. graminearum inoculation after 24 and 48 h. Letters correspond to Duncan’s multiple range values to highlight significant differences in gene expression. Values are normalized to reads per kilobase per million (RPKPM)

Full size image

Discussion

The members of the Esi3 gene family identified in Triticum aestivum in this study likely represent all members of the family. Whole genome sequencing for T. aestivum [22] and the extensive transcriptome databases for T. aestivum and other Triticum species were thoroughly searched for Esi3-like sequences. The resulting gene sequence set was verified in the recently released diploid genome for the D genome donor, Ae. tauschii, for which the full pseudomolecule sequences for the seven chromosomes are available [24]. The identification of homeologous genes in the A, B and D genomes for each family member provides additional confidence for the completeness of the searches. The set was compiled and refined through a series of iterative searches, and expanded when it was apparent that there may be members of the family missing from the dataset, such as an absent homeolog or a missing ortholog to a gene family member identified in closely relates species. Orthologs for all members of the Esi3/RCI2/PMP3 gene family were also identified in closely related diploid species belonging to the Triticeae tribe, Secale cereale, H. vulgare and Ae. tauschii. The high degree of similarity among homeologous gene copies is similar to that described for other gene families in Triticum aestivum, alpha tubulins and caleosins, i.e. approximately 97% nucleotide sequence identity and 99–100% amino acid sequence identity for many homeologous sets [21, 37]. The degree of sequence similarity among homeologous gene copies is also similar to that among orthologs to the Esi3 genes in other species of the Triticeae, H. vulgare and Secale cereale.

The comparison of the structure of the proteins predicted for the gene family members in Triticum indicate that species in the Triticeae contain a novel class of Esi3-like proteins in addition to the two groups of Esi3/RCI2/PMP3 gene family members previously identified in other species [5]. Esi3–9 constitutes a new class of proteins, Group III, characterized by long extensions of coding sequence on their N terminal ends. Esi3–9-like genes were also identified in the closely related species H. vulgare, Secale cereale, Ae. tauschii as well as T. turgidum. Esi3-like genes with the Group II structure were not identified in Arabidopsis nor in other monocot species surveyed which include Z. mays, S. bicolor, O. sativa, and B. dystachion. Other genes in the Esi3 gene family can be classified into groups previously described in other species [5]. The Esi3 genes were dispersed in the genome, however two pairs of genes were found to be localized in tandem. Esi3–1 was localized near Esi3–2, and Esi3–3 was localized near Esi3–10 in the A, B and D genomes of T. aestivum as well as in Ae. tauschii. In H. vulgare the Esi3–1 and Esi3–2 genes are approximately 200 kb apart, but the species did not contain a copy of Esi3–3 [23]. Other paralogous Esi3 genes located on the same chromosomes are quite far from each other. The size of the Esi3/RCI2/PMP3 gene family is similar in the other species studied; there were ten Esi3/RCI2/PMP3 genes in Ae. Tauschii and Secale cereale, nine in H. vulgare, eight in B. dictachyon, and A. thaliana, seven in Sorghum bicolor, 11 in Z. mays, and 12 in O. sativa. The gene families for A. thaliana, B. distachyon, O. sativa, and Z. mays have been previously described [5].

The phylogenetic tree created with the nine species suggests that there have been independent radiations of the gene family among the monocot species (Additional file 15: Figure S2). Among the Triticeae most gene family members are common, with the exception of Esi3–2 which is not present in the B genome of T. aestivum and T. turgidum, which suggests that it was deleted in these lineages (Additional file 16: Figure S3). Overall this pattern indicates that the ten-member gene family evolved to its current number before the divergence of the members of the Triticeae tribe. Though the dicot Arabidopsis has eight members of the Esi3/RCI2/PMP3 gene family they do not have a close relationship with the gene family members in the Triticeae and other monocot species. The phylogenetic tree suggests that there have been a number of gene duplication events since the divergence of monocots and dicots and there have been additional gene duplication and gene losses since the divergence of O. sativa, Z. mays, B. distachyon and the Triticeae.

Accuracy of gene annotation

The importance of manual annotation of gene families is apparent from the miss-annotation of the Esi3/RCI2/PMP3 gene family members in several full genome annotation databases. This underscores the challenge of perfecting automated annotation of extensive DNA sequence sets. Automated gene annotations are particularly troublesome with genes with short coding sequences such as the Esi3s. Some miss-annotations detected in the whole genome database appear to be based on a preference for long open reading frames (ORFs). Several of the miss-annotated loci were identified with incorrect long ORFs that overlapped those of Esi3/RCI2/PMP3 genes that would encode hypothetical proteins that have no similarity to those in other species. The genome for S. cereal is not annotated at Ensembl Plants [23]; these discrepancies are summarized in Additional file 3: Table S3. The annotations reported here relied heavily on the availability of transcript sequences from EST and TSA databases for gene identification and the demarcation of intron/exon junctions. Additional considerations included the detection of open reading frames, sequence and gene structural conservation among family members. We expected sequence similarity among homeologous groups of genes to be higher than 95% in most cases [21, 37]. When they are available, the transcript sequence is an important consideration in gene annotation. The transcripts in the TSA database were extremely useful because of the exceptional depth of the second generation sequences, as well as the accuracy of their assembly. However, they have a tendency to have excessively long 5′ and 3′ untranslated regions (UTR) sequences that do not appear to realistically represent the UTRs. These may be due to rare extended transcripts that are detected in the high depth of sequences available from second generation sequencing and that are included in the automated assemblies of transcripts. These extended UTR sequences are generally not seen in the EST databases which have traditionally contained sequences obtained from the Sanger sequencing methods on individual cDNA clones. It is advisable that these two types of transcript sequences remain separated in the public databases.

Expression

Nearly all members of the Esi3 gene family showed differential expression in response to abiotic or biotic stresses; this differential gene expression is taken as an indicator of the role these genes play in stress tolerance. The constitutive overexpression of select members of the Esi3/RCI2/PMP3 gene family in Arabidopsis [10, 13,14,15] and Nicotiana tobacum [16] have demonstrated that these genes can increase abiotic stress tolerance and reduce Na+ accumulation in transgenic plants. Similarly, the expression of a large number of plant Esi3/RCI2/PMP3 gene family members in Saccharomyces demonstrated that they can improve salt stress tolerance and complement the deletion of the yeast PMP3 gene. The studies of Aprile et al., 2009 and 2013 [31, 32] compared wheat cultivars that differed in stress tolerance and water use efficiency and found expression patterns in which higher levels of expression or higher levels of induction paralleled the level of tolerance for the species or genotype. Long term drought treatment showed that Esi3–1 was more strongly induced in the drought tolerant species T. aestivum than in the more sensitive T. turgidum. However, the opposite trend was seen for Esi3–1 when comparing the high water use efficiency T. turgidum cultivar, Cappelli, with the low water use efficiency cultivar Ofanto. Esi3–1 showed decreased expression in Cappelli in response to stress and increased expression in Ofanto [31].

The gene expression datasets explored to characterize Esi3 gene family members in response to abiotic stresses included a wide range of stress treatments, such as short term and long term osmotic stress, heat, cold treatments and Fusarium graminearum inoculation. Plants that were used for these studies were grown under different conditions and analyzed at different stages of growth, including soil-grown plants at flowering stage, seedlings grown in sterile petri dishes and in the case of the Fusarium inoculation, during grain development. Thus the patterns of gene expression were quite diverse; nevertheless, nearly all members of the gene family were shown to be up-regulated in at least one stress condition.

Experiments that were assayed by transcriptome sequencing facilitated the comparison of expression of individual homeologs. In most cases the members of the same homeologous set showed similar gene expression changes, though in some cases these changes were not statistically significant. We point out that most gene expression assays for stress treatments were measured on leaf tissue, whereas studies of tissue-specific expression showed that the Esi3 gene family members were differentially expressed over a broad range of tissues and developmental stages. If these tissues were evaluated separately a more complex pattern of gene expression in response to stress may be recognized. Indeed, the initial description of Esi3 differential gene expression was described in the relatively strong induction by salt treatment in the roots of the tolerant species Lophopyrum elongatum for the Esi3–1 gene family member [1].

The Esi3–9s had the most unique structural features among the gene family members; they also have the most unique expression pattern. Transcripts for these genes were detected in very few datasets, namely in an RNA-seq datasets from anthers in Triticale and the anthers from the Azhurnaya Spring Wheat [21, 28] and in developing seed heads in the experiments that characterized the response to Fusarium inoculation of developing spikes.

The widespread differential gene expression of the Esi3 gene family members in Triticum species in response to abiotic stress suggests that nearly all members of the family may contribute to abiotic stress tolerance. This would merit further investigation through transgenic studies or by comparative expression analysis in genotypes with differing degrees of stress tolerance.

Conclusions

There are twenty-nine Esi3/RCI2/PMP3 gene family members with ten paralogous groups, each with three homeologous copies in T. aestivum except for Esi3–2 for which no B copy was identified. Esi3–9s have an extended N terminal therefore placing it in a group on its own, which has been designated Group III. The other Esi3s fall within Group I or II based on amino acid length and properties [5]. The gene family members were manually curated and compared to sequences that were annotated automatically and the manually curated sequences were more accurate. This study highlights the importance of manually curating sequences to elucidate gene families and their interrelationships, and for the improvement of automated annotation methods. The T. aestivum Esi3s share homology with Esi3/RCI2/PMP3 genes from A. thaliana, Z. mays, O. sativa, B. distachyon, Sorghum bicolor, Ae. tauschii, H. vulgare, and Secale cereale. The ten paralogous Esi3s have differential expression in response to a variety of abiotic stresses and in response to infection with F. graminearum. The homeologous copies also displayed varied responses to abiotic and biotic stress, suggesting that the Esi3 family members play a role in stress tolerance.

Methods

Esi3 gene sequence retrieval

The Esi3 cDNA sequence from L. elongatum (GB Accession U00966.1) and the Lti6A and Lti6B genes from Oryza sativa [8] were used to search the National Center for Biological Information (NCBI) Transcriptome Shotgun Assembly (TSA) databases for T. aestivum and T. monococcum by Blastn and tBlastn; duplicate hits were eliminated. A total of ten paralogous gene sequences were found, and these were used to search the NCBI EST database and the International Wheat Genome Sequencing Consortium (IWGSC) [22] wheat survey sequences (WSS) versions 2 and 3 of individual chromosome arms. The sequences were also re-confirmed using the IWGSC whole genome assembly RefSeq v1.0 [38, 39]. The latter were used to make chromosomal arm assignments to individual homeologous copies of the gene family members from the A, B and D genomes of T. aestivum. Novel sequences identified in either of these databases were iteratively used to query the TSA and EST databases to verify the sequence and delimit exon/intron junctions in the gene sequence. In cases where there was discrepancy between sequences from different databases, contigs were re-assembled with T. aestivum EST sequences that shared a minimum of 99% identity using the CAP3 assembly program at prabi [40]. The paralogous set of Esi3 gene family members were used to identify homologs in other monocotyledonous species, including Secale cereale, H. vulgare, and Sorghum bicolor in the GenBank nucleotide collection, TSA and EST databases. Ae. tauschii sequences were obtained from the NCBI TSA and nucleotide collection databases and from the Ae. tauschii Genome release of unannotated pseudomolecule sequences [25]. Z. mays sequences for Esi3/RCI2/PMP3 genes were taken from those described by Fu et al., 2012 and Zhao et al., 2014 [10, 18]; O. sativa genes were taken from those reported by Morsy et al., 2005 [8] and Medina et al., 2007 [4]. Sequences of Esi3-like genes for B. distachyon were those reported by Rocha [5]; those for Arabidopsis were taken from Capel et al., 1997 [6], and Medina et al., 2007 [4].

Phylogenetic analysis

Phylogenetic trees were constructed using Molecular Evolutionary Genetics Analysis (MEGA) version 7 [41]. The Esi3 homeolog phylogenetic tree was constructed using the nucleotide sequences of the coding regions from all gene family members from T. aestivum. The Maximum Likelihood method was used based on the Jukes-Cantor model [42] with the same parameters used by Khalil et al., 2014 [21]. The analysis involved 29 nucleotide sequences. Codon positions included were 1st + 2nd + 3rd + Noncoding. All positions with less than 95% site coverage were eliminated. That is, fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position. The percentage of trees in which the associated taxa clustered together is shown next to the branches. There were a total of 165 positions in the final dataset.

The phylogenetic tree and sequence alignment of the Esi3 homologs in nine species was done using the amino acid sequences. The sequences were aligned using MUSCLE [43] and the phylogenetic tree was constructed using ten paralogous Esi3 genes from T. aestivum, one representative from each homeologous group and the full set of protein sequences from the diploid species Ae. tauschii, H. vulgare, B. distachyon, Secale cereale, O. sativa, Z. mays, Sorghum bicolor and A. thaliana. The evolutionary history was inferred using the Maximum Likelihood method based on the Whelan And Goldman model (WAG) [44]. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+G, parameter =1.5141)). The analysis involved 85 amino acid sequences. All positions containing gaps and missing data were eliminated. There were a total of 54 positions in the final dataset.

Esi3 expression analysis

The Esi3 gene expression levels in response to abiotic and biotic stresses including drought, heat, combined heat and drought, low temperature and Fusarium graminearum infection were determined using Illumina RNA-sequencing libraries listed in the NCBI Sequence Read Archive (SRA) database and then retrieved from the European Nucleotide Archive at EMBL-EBI [45] Datasets used for gene expression analysis are listed in Additional file 17: Table S12. The datasets in FASTQ format were converted to FASTA using FASTX Toolkit 0.0.13.2 [46]. Due to the high sequence similarity found within the coding region of the T. aestivum Esi3 sequences, the 3’ UTRs from the EST sequences were used to search the datasets using CD-HIT-EST-2D biological sequence clustering algorithm [47]. The CD-HIT-EST-2D parameters were set to default with the exception of a word size of 5 (n = 5), similarity cut-off of 99% (−c 0.99) and a memory limit of 32G (−M 32000 Mbytes). The expression was normalized to reads per kilobase per million (RPKPM) to normalize for differing library sizes and different 3’ UTR lengths among the gene transcripts. The RPKPM was calculated as the number of reads in a library with a minimum of 99% identity to the query sequence/the length of the query sequence/ number of reads in a library/106.

Affymetrix microarray 61 k RMA normalized datasets (log2 units) [29] for different wheat tissues at different times of development and to assay gene expression in response to drought, heat, and combined drought and heat stress were retrieved from the PLEXdb database [48]. The differences in fold expression were deduced by comparing the levels of treated samples relative to the levels found within the control measured by fluorescence intensity. The levels of Esi3 gene expression in response to drought was analyzed using the dataset from Aprile et al., 2009 [32]. The effects of drought, heat and combined stress was analyzed using the dataset from Aprile et al., 2013 [31]. The Esi3 expression levels in different tissue types were analyzed using a dataset from Schreiber et al., 2009 [29]. Homeologous gene copies in T. aestivum cannot be distinguished by the microarray analysis; results were analyzed for the wheat Esi3 paralogous genes, which represent the combined set of three homeologous genes.

Expression data analysis of the Esi3 gene family across seventy-one tissue types reported in [28] was retrieved from the wheat eFP browser at the University of Toronto BAR [49]. The identifiers for all Esi3 gene family members were retrieved from the Ensembl Plants and listed in Additional file 1: Table S1. Data was not available for Esi3–1-D as well as for Esi3–10-A since the latter has not been identified on Ensembl Plants [23].

Statistical analysis

The significance of differences in Esi3 gene expression, in the RNA-seq datasets, was analyzed using a two-way ANOVA to test for significant differences in expression between genotypes in response to treatments and for genotype x treatment interaction effects. Subsequently, the data was analyzed using a one-way ANOVA test with a Duncan’s multiple range post hoc test to determine the differences in Esi3 induction between the A, B, and D copies in response to stress or in differing tissue types. One-way ANOVA was used to test the significance of the differences in Esi3 gene expression from microarray data. Duncan’s multiple range post hoc test was used to determine the significance of differences in gene expression in response to different stress conditions.