Isolation of the TaAGPL1 promoter
Bread wheat genome contains three closely related, yet distinct, subgenomes (AABBDD) and three homoeologs, with over 95% sequence identity in their coding regions for the majority of genes [17]. In bread wheat, rice and maize, AGPL1 gene transcripts showed high levels in endosperm, whereas they were not detected in leaves or other organs [7,8,9, 12], suggesting that the promoter could be endosperm-specific. The cDNA sequence of the TaAGPL1 gene, which was highly expressed in endosperm during the grain filling period in our previous study [10], was searched against the recently published IWGSC databases (RefSeq v1.0) [21], and a high level (100% identities) to a chromosome-located contig (TraesCS1D01G427400.1) was found, suggesting its localization on 1D chromosome.
The sequence of the TaAGPL1-1D gene was used as a query to search its upstream promoter sequence in the IWGSC database, and a 1280-bp fragment (− 1269 bp to + 11 bp) was retrieved and used to design the primer pairs to amplify its promoter sequence. The amplified TaAGPL1-1D promoter demonstrated high similarity (99.8%) to the retrieved IWGSC sequence (Additional file 2: Fig. S1). Using the classic transient transformation system [14], we did not detect histochemical GUS activity of the TaAGPL1-1D promoter in the leaves of transiently transformed-N. benthamiana plants (Additional file 3: Fig. S2), possibly due to its endosperm-specific expression profile. Alternatively, a gene gun-mediated transient transformation protocol for wheat grains was used to detect TaAGPL1-1D promoter activity [16]. Our experiment showed that the TaAGPL1-1D promoter qualitatively and quantitatively drove expression of the GUS gene (Fig. 1), arguing that this promoter contains cis-acting elements that regulate TaAGPL1-1D gene expression in wheat grains.
The binding between TaPDIL1-2 gene and TaAGPL1 promoter
Wheat grains were sampled at 5, 10, 15, 20, 25, 30, and 35 days after anthesis, endosperm was reserved and other organs were removed. The mRNA isolated from the sampled endosperm was used as template to synthesize the full-length double-stranded complementary DNA and together with the linearized pGADT7-Rec vector was co-transformed into yeast strain Y1HGold cells for directional combination through homologous recombination. The combined plasmids were extracted and subsequently transformed into E. coli strain stellar cells. The total capacity was 1.08 ± 0.3 × 106 cfu with more than 77.9% inserted fragments greater than 750 bp (Additional file 4: Fig. S3), which was suitable for yeast screening.
Using the above-amplified TaAGPL1-1D promoter as bait and wheat endosperm cDNA library as prey, we performed Y1H screening (Fig. 2a), and identified 39 positive clones, which were subsequently sequenced and functionally annotated (Additional file 5: Table S2). Of these clones, the TaPDIL1-2 gene was screened out. The complete open reading frame (ORF, 1539 bp) of the TaPDIL1-2 gene was subsequently cloned (Additional file 6: Fig. S4), and its prey vector was constructed and transformed into the bait strain Y1HGold containing three fragments (1280 bp, 881 bp, or 494 bp) of the TaAGPL1-1D promoter (Fig. 2b), respectively, to perform the second Y1H screening with 0 or 100 ng/mL AbA. Our experiment showed that the transformed yeast cells containing Y1HGold/Pro-1 (− 1269 bp to + 11 bp) grew optimally in SD/-Leu, -Ura, +AbA100 medium, followed by Y1HGold/Pro-2 (− 870 bp to + 11 bp) and Y1HGold/Pro-3 (− 483 bp to + 11 bp) (Fig. 2c). These data confirmed the binding between TaPDIL1-2 protein and TaAGPL1-1D promoter, with the potential binding region located in the promoter between − 483 and − 1269 bp.
Characterization and expression of TaPDIL1-2
By searching the IWGSC databases, the isolated TaPDIL1-2 gene was localized to 4BS, and its ORF encoded a 512-amino-acid protein of 56.43 kDa with a predicted pI of 5.03. The deduced TaPDIL1-2 protein contained 4 classical thioredoxin domains, including 2 redox CGHC active sites (a and a′), 2 inactive domains (b and b′), and 1 C-terminal KDEL signal sequence (KDEL), which is a classcial endoplasmic reticulun (ER)-retention signal supporting ER-localization (Fig. 3a). Phylogenetic analysis indicated that the thioredoxin domain of TaPDIL1-2 shared more than 75.7% identity with those of AtPDIL1-2, OsPDIL1-1, OsPDIL1-2 and ZmPDIL1-2, belonging to one sub-branch of group I in the PDIL subfamily (Fig. 3b). Some PDIL proteins have been identified and functionally characterized in the wheat genome [22]. To our knowledge, however, there have been no reports regarding TaPDIL1-2 or its orthologous genes, and the amino acid sequence of TaPDIL1-2 had low similarities (< 24.3%) to other wheat PDIL proteins (Fig. 3b).
At the transcriptional level, the in silico expression data revealed three homoeologs of the TaPDIL1-2 gene that were more highly expressed during the reproductive stage than the vegetative stages, and during the reproductive stage, their expression level were higher in the endosperm/grain and spikes than in stem and leaf (Fig. 3c). The bread wheat endosperm mainly accumulates starch (≥ 70%) and protein (10–14%), forming the bulk of the grain. These findings suggest that TaPDIL1-2 homoeologs can function in starch biosynthesis in wheat grain.
Negative regulation of the TaPDIL1-2 gene in wheat starch biosynthesis
Gene function in higher plants is often explored through the use transgenic and mutational assays. In bread wheat, however, the multiple copy insertions, low transformation efficiency, cultivar specificity, time consumption, and high cost of transgenic approaches have greatly limited gene function studies in this species [23]. The functional redundancy among homoeologs in this species also causes some difficulties in terms of generating null mutants [24]. BSMV-VIGS method facilitates the rapid generation of gene knockdown phenotypes in polyploid species because plant transformation is not required, accelerating the characterization of target genes [25]. In this study, this method was used to evaluate the function of the TaPDIL1-2 gene. We constructed BSMV-TaPDIL1-2 and BSMV-GFP vectors, and the latter was used as the control. To prevent functional complementation and allow complete silencing, we selected a conserved cDNA fragment (190 bp) of the TaPDIL1-2 gene, three homoeologs of which shared 99.1% similarity (Additional file 7: Fig. S5), potentially enabling the simultaneous silencing of its three homoeologs using the BSMV-VIGS method.
In previous studies, the BSMV-VIGS method has mostly been applied to explore the function of candidate genes of cereal crops in controlled conditions considering responses to single or several environmental factors [26,27,28]. However, there are some differences (e.g. stronger plants, more yields) in growth and phenotypes of most higher plants including important crops grown between field conditions with multiple and variable factors (light, temperature, etc.) and under the controlled conditions [26]. Thus, field experiments can more efficiently explore the function of the target genes. By using arch plastic sheds combined with water-sprayed spikes, a few efficient and convenient approaches were designed in our previous study for application of the BSMV-VIGS method to wheat plants under field conditions [13]. In the present study, BSMV-TaPDIL1-2 and BSMV-GFP vectors were separately used to inoculate 142 and 167 wheat spikes at anthesis under field conditions. We observed that BSMV-VIGS-induced chlorosis occurred on inoculated-spikes at 19 days after anthesis, gradually extended throughout the entire spike, and persisted until the mature stage (Fig. 4a), suggesting that these two virus vectors were successfully inoculated into the wheat spikes.
At 19, 24, 29, and 34 days after anthesis, the transcription levels of the TaPDIL1-2 gene were determined by using qPCR, and the β-actin and GAPDH genes were used as two internal controls. Our results showed that transcription levels of the TaPDIL1-2 gene in grains of BSMV-TaPDIL1-2-inoculated spikes were markedly decreased by 42.2–79.2% (Fig. 4b, c), similar to previous studies [13, 29]. The starch contents and 1000-kernel weight in grains of BSMV-TaPDIL1-2-inoculated wheat spikes were increased by 16.6% and 6.8%, respectively (Fig. 4d, e), demonstrating that TaPDIL1-2 silencing caused enhanced starch biosynthesis in the wheat endosperm. Thus, TaPDIL1-2 might act as a negative regulator of starch biosynthesis. Similarly, the transcription levels of the TaAGPL1-1D gene in the endosperm of BSMV-TaPDIL1-2-infected wheat spikes at the four sampling time points increased remarkably (Fig. 4f, g). Because our previous studies have confirmed the important role of the TaAGPL1 gene in starch biosynthesis in wheat grain [11], we inferred that the TaPDIL1-2 gene could play a crucial role in wheat starch biosynthesis by negatively regulating TaAGPL1-1D gene expression.
Indirect regulation of TaPDIL1-2 gene on expression of TaAGPL1-1D gene
Most of PDIL proteins contain a KDEL at the C-terminus, and are localized in ER, where they combine with other protein substrates to form enzymatic complexes, in which their classical thioredoxin domains are responsible for interacting with substrates [30]. In the enzymatic complexes, PDIL proteins function as chaperones to bind to the unfolded or partially folded protein substrates by ascertaining their correct folding and preventing their aggregation [31]. Cereal endosperm storage proteins (e.g., gliadins and glutenins) have many disulfide bonds, which are bound to some PDIL proteins [32].
Transcription factors (TFs) function by binding to cis-acting elements preferentially located in the promoter region of the target genes [33], some TFs are also localized in ER to act as homo- or hetero-dimers with some chaperones (e.g. DnaK), where the cytoplasm-facing N-terminal regions are released by chaperones to be active TFs, and then, active TFs have been transferred from the ER into the nucleus, in which they activate the transcription of downstream genes [34]. For example, DRBEIII-1, a plant-specific TF in Brassica napus, was fused to a human PDIL protein to form a fusion protein complex (HDP), in which DREBIII-1 exhibited a highly soluble and biologically active form and HDP was confirmed to have the biological function of DRBEIII-1 [35]. Our sequence analysis indicated that the potential binding region was located from − 483 to − 1269 bp in the TaAGPL1-1D promoter containing the Auxin-responsive element and GCC-box-binding cis-acting elements (Additional file 8: Fig. S6). Previous studies have shown that a few PHD-, or AP2/ERF, bZIP-type TFs negatively regulate the expression of some starch biosynthesis genes [10, 36,37,38,39,40], and some members (e.g. bZIP28 and bZIP60) are localized in ER and possess the capacity to preferentially on the promoters of downstream target genes via binding to the above-mentioned cis-acting elements [41, 42]. Therefore, we speculated that functional TaPDIL1-2 protein could combine with TFs in ER to ascertain their correct folding to activate TFs, and then, active TFs are transferred into nucleus, in which they negatively regulate the expression of the TaAGPL1-1D gene. Based on these results, we propose one pathway for the TaPDIL1-2 chaperone to indirectly and negatively regulate the expression of the TaAGPL1 gene in starch biosynthesis (Fig. 5).