HSP90s are highly conserved within triplets
As a result of sequence searches and domain confirmation, 18 HSP90s were identified in the wheat genome. The identified HSP90s were only present in homeologous groups 2, 5 and 7, and members in each subfamily were evenly distributed among the A, B and D subgenomes (regarded as a triplet), exhibiting a conserved copy number during wheat polyploidization. The 18 HSP90s were classified into HSP90AAs (three, triplet AA), HSP90ABs (three, triplet AB-5 and three, triplet AB-7), HSP90C1s (three, triplet C1), HSP90C2s (three, triplet C2) and HSP90Bs (three, triplet B) based on the phylogenetic tree (Fig. 1, Table 1). The characterized cytosolic HSP90s were identical to the previously reported cytosolic HSP90 sequences [21], suggesting the high reliability of our characterization procedure.
Table 1
Summary information of wheat HSP90s.
Gene | Subfamily/ Triplet | Exon number | Isoform number | FPKM value range |
Grains | Flag leaves |
TraesCS2A01G033700 | AA | 3 | 5 | 22–363 | 2–1896 |
TraesCS2B01G047400 | AA | 3 | 6 | 9–275 | 2–1486 |
TraesCS2D01G033200 | AA | 3 | 8 | 50–541 | 18–2496 |
TraesCS7A01G242200 | AB | 3 | 10 | 61–445 | 148–2249 |
TraesCS7B01G149200 | AB | 3 | 13 | 163–555 | 160–2045 |
TraesCS7D01G241100 | AB | 3 | 6 | 83–378 | 94–1361 |
TraesCS5A01G260600 | AB | 3 | 6 | 159–241 | 141–350 |
TraesCS5B01G258900 | AB | 3 | 9 | 79–181 | 88–589 |
TraesCS5D01G268000 | AB | 3 | 7 | 91–222 | 86–632 |
TraesCS7A01G529900 | B | 15 | 7 | 19–59 | 67–232 |
TraesCS7B01G446900 | B | 15 | 13 | 24–66 | 65–203 |
TraesCS7D01G517800 | B | 15 | 7 | 20–78 | 65–235 |
TraesCS5A01G251000 | C1 | 19 | 2 | 32–89 | 29–190 |
TraesCS5B01G249000 | C1 | 19 | 4 | 22–119 | 77–422 |
TraesCS5D01G258900 | C1 | 19 | 2 | 15–99 | 27–220 |
TraesCS5A01G101900 | C2 | 20 | 5 | 5–12 | 4–24 |
TraesCS5B01G106300 | C2 | 20 | 9 | 6–12 | 19–104 |
TraesCS5D01G113700 | C2 | 20 | 7 | 7–13 | 11–56 |
(A) Expression abundance of the major isoforms of HSP90s in triplet AA. Expression is represented as the log2-transformed value (FPKM + 1). |
(B) Splicing patterns of the major isoforms of HSP90s in triplet AA. Solid boxes represent exons, lines represent introns, arrows indicate the splicing sites that differ from the longest intact coding sequence. |
(A) Regulation of HSP90s at different levels under different conditions. TR, transcriptional regulation. |
(B) The most abundant isoform in each HSP90s under different conditions. Major 0 indicates that the corresponding HSP90 gene only transcribed one major isoform. Major 1 indicates the major isoform with the complete reading frame. Major 2 and Major 3 (for TraesCS5B01G258900 only) represent the other major isoforms. Minor represents the minor isoform. |
Additional files |
Additional file 1: Table S1. Identities between each HSP90 proteins. Table S2. Detail information for MEME motifs. Table S3. Isoform type classification based on FPKM value. |
Additional file 2: Figure S1. (A) Protein sequence motifs of HSP90s, motif analysis was performed by MEME (http://meme-suite.org/). (B) Gene structure analysis of HSP90s was carried out at Gene Structure Display Server 2.0 (http://gsds.cbi.pku.edu.cn/). |
Additional file 3: Figure S2. Expression profiles each HSP90 homeolog in different triplet under various conditions. Log2-transformed fold-change values were used to construct a heat map. The samples are indicated by a character that refers to the tissue, followed by the heat stress duration. G, grain. L, flag leaf. “0” is regarded as a control. The left panel shows the phylogenetic tree. |
Additional file 4: Figure S3. Expression abundance and splicing modes of major isoforms among HSP90 homologs in triplets AB-5 and − 7. Expression abundance of the major isoforms of HSP90s in triplet AB-5 (A) and AB-7 (C). Expression is represented as the log2-transformed value (FPKM + 1). Splicing patterns of the major isoforms of HSP90s in triplet AB-5 (B) and AB-7 (D). Solid boxes represent exons, lines represent introns, arrows indicate the splicing sites that differ from the longest intact coding sequence. |
Additional file 5: Figure S4. Expression abundance and splicing modes of major isoforms among HSP90 homologs in triplets B, C1 and C2. Expression abundance of the major isoforms of HSP90s in triplet B (A), C1 (C), and C2 (E). Expression is represented as the log2-transformed value (FPKM + 1). Splicing patterns of the major isoforms of HSP90s in triplet B (B), C1 (D), and C2 (F). Solid boxes represent exons, lines represent introns, arrows indicate the splicing sites that differ from the longest intact coding sequence. |
Sequence analysis showed that the protein sequence identities of three HSP90 homologs in a triplet were above 96% (Additional file 1: Table S1), indicating that their sequences were highly conserved. Accordingly, the protein sequence motifs (Additional file: 2, Figure S1A, Additional file 1: Table S2) and gene structures (Additional file 2: Figure S1B) were also highly conserved within a triplet. In detail, the HSP90AAs and HSP90ABs contained three exons and the HSP90C1s, HSP90C2s and HSP90Bs contained 19, 20 and 15 exons, respectively.
Conserved heat response pattern among HSP90 homeologs within a triplet
Using the previously determined dynamic and intensive heat response transcriptomes of filling grain and flag leaves of wheat [32], we investigated the heat response patterns of wheat HSP90s (Fig. 2). Consistent with reports that HSP90AA is highly heat inducible in Arabidopsis and rice, the wheat HSP90AAs were remarkably induced (fold change > 2 and FDR-adjusted P-value < 0.01) after 10 min and 30 min heat treatment in leaves and grains, respectively. However, the wheat HSP90ABs were also highly heat inducible, whereas they are regarded as constitutively expressed genes in Arabidopsis and rice [17, 18]. It is also worth noting that HSP90C2s and HSP90AB-5 did not respond to heat stress in grains, but did respond to heat stress in leaves. Our results demonstrated that the heat response patterns of HSP90s were highly dynamic and more intensive studies should be considered to uncover the array of heat responses of HSP90s.
We next compared the expression profiles of wheat HSP90s and surprisingly found a similar trend in the response of the three HSP90 homologs within a triplet (Fig. 2, Additional file 3: Figure S2). For example, all three of the HSP90 homologs in triplet AA were lowly expressed under normal conditions and 5 min heat treatment, but were sharply upregulated at later heat treatment time points. In conclusion, consistent with the high level of sequence conservation, the expression and heat stress response profiles of HSP90 homeologs were also conserved within a triplet, implying that no significant functional divergence of HSP90 homologs occurred during wheat polyploidization.
Large number of novel isoforms produced by HSP90s under heat stress
Recent findings have suggested the importance of AS regulation in abiotic stress responses [24, 26, 33, 34]. Using the qualitative and quantitative heat response transcriptomes produced by combining second- and third-generation sequencing in our previous study [32], we investigated the roles of AS in the heat stress response of HSP90s.
Firstly, 126 isoforms of HSP90s were identified from our data, including the 36 isoforms that have been annotated in IWGSC RefSeq v1.0 and 90 newly identified isoforms. The isoform number per HSP90 gene ranged from 2 to 13, with an average of 9, 8.5, 7, 6.3 and 2.7 for the subfamilies B, AB, C2, AA and C1, respectively (Table 1). The isoform number did not correlate with the exon number for each gene in this gene family. Secondly, we detected the expression abundances of these isoforms and found them to vary remarkably among the isoforms generated by the same gene. Intriguingly, some isoforms responded to heat stress with transcriptional regulation according to the following criteria: fold change > 2 and FDR-adjusted P-value < 0.01 (Fig. 3), while the related genes that produced these isoforms were not significant heat response genes under these criteria, for example, the 5B- and 5D-homeologs in triplet AB and the 5D-homeolog in triplet C2. Thus, these transcriptionally heat-responsive isoforms extended our understanding of the transcriptional regulation of HSP90s and further revealed the complexity of the heat stress response for this gene family.
Next, to characterize the predominant isoforms of each gene, we introduced the isoform expression percentage (IEP), which was calculated as the expression abundance ratio of one isoform to all isoforms produced by the same gene. An isoform with an average IEP of more than 30% across all of the time points in a tissue was regarded as a major isoform, an isoform with an IEP less than 5% in all time points was regarded as a rare isoform and all other isoforms were classified as minor isoforms. This analysis led to the classification of isoforms into major (30), minor (44) and rare (52) isoforms (Additional file 1: Table S3). For 18 HSP90s, HSP90AB (TraesCS5B01G258900) produced three major isoforms and 12 HSP90s generated two major isoforms. Interestingly, among the two or three major isoforms, one was already annotated in IWGSC RefSeq v1.0 and contained the longest complete coding region comprising the HATPase domain and HSP90 domain. However, another major isoform was newly discovered from our hybrid sequencing data and this produced truncated peptides that possessed only the HSP90 domain. Furthermore, expression analysis showed that the major isoforms generated by the same genes had comparable expression levels and response patterns (Fig. 4, Additional file 4: Figure S3, Additional file 5: Figure S4), making it intriguing as to what roles these truncated peptides play in the heat stress response.
Varied number and splicing modes of major isoforms generated by HSP90 homologs within triplets
The above transcription analysis showed that the expression abundance and response patterns of HSP90 homeologs within triplets were conserved. However, regarding the number of major isoforms produced by each HSP90 homolog within a triplet, only one or two homologs produced two major isoforms, with the exception of the homeologs of triplet AA and triplet AB-5, which produced two major isoforms for each homolog (Fig. 4, Additional file 4: Figure S3, Additional file 5: Figure S4). Furthermore, the newly identified major isoforms (NIMIs) from our hybrid sequencing data exhibited different exon and intron compositions among the homeologs in the triplet, suggesting that the homeologs exploited different transcripts and proteins to respond to heat stress. For example, among the three homologs of triplet AA, the NIMI of the 2A-homeolog underwent AS at both the 5ʹ (Alt5ʹSS) and the 3ʹ ends of the first intron (Alt3ʹSS), producing a truncated peptide of the HSP90 domain. The NIMI of the 2B-homeolog underwent intron retention, producing a polypeptide with an incomplete HSP90 domain as a result of using an alternative open reading frame. The NIMI of the 2D-homeolog underwent exon skipping, producing a truncated peptide of the HSP90 domain (Fig. 4). In conclusion, the number and transcript structure produced by the three homologs in a triplet are significantly different, despite the homologs being conserved at the sequence level and at the transcriptional response level. These results suggested a new evolutionary direction for HSP90 homologs in wheat polyploidization.
Differential AS responses of HSP90 homologs within a triplet
By comparing the highly expressed isoform sets and their expression changes between the control and heat stress treatment samples, we identified 11 differentially spliced HSP90s under heat stress (i.e. HSP90s that respond to heat stress with AS regulation) for at least one time point sample (Fig. 5A) using the criterion defined in our previous study [32]. Interestingly, some HSP90s that did not respond to heat stress with transcriptional regulation, did respond to heat stress with AS regulation by generating new isoforms or changing the expression level of highly expressed isoforms, extending our understanding of the heat stress response of HSP90s and their functional conservation. Significantly, for the three HSP90 homeologs within a triplet, only one or two responded to heat stress with AS regulation, demonstrating the differential responses at the AS regulation level and suggesting diverged evolution in the heat stress response among the three HSP90 homologs.
Furthermore, using the qualitative and quantitative isoforms, we investigated the isoform with the highest abundance in each sample (Fig. 5B). The results showed the highest abundance isoform of HSP90s that generated more than one major isoform and produced intact HSP90 proteins or truncated peptides that altered among heat stress treatment time points. Interestingly, some minor isoforms defined by our criterion switched to become the highest abundance isoforms, highlighting the important role of AS regulation. It was also worth noting that the highest abundance isoforms generated by three HSP90 homeologs at specific time points were also different, providing further evidence of the differential responses and diverged evolution. In conclusion, inconsistent with the conserved sequences and transcriptional regulation, AS regulation and the most highly abundant isoforms differed among the three HSP90 homeologs within a triplet, diversifying their heat stress response.