TaSRO1 plays a dual role in suppressing TaSIP1 to fine tune mitochondrial retrograde signalling and enhance salinity stress tolerance

Summary

• Initially discovered in yeast, mitochondrial retrograde signalling has long been recognised as an essential in the perception of stress by eukaryotes. However, how to maintain the optimal amplitude and duration of its activation under natural stress conditions remains elusive in plants.

• Here, we show that TaSRO1, a major contributor to the agronomic performance of bread wheat plants exposed to salinity stress, interacted with a transmembrane domain-containing NAC transcription factor TaSIP1, which could translocate from the endoplasmic reticulum (ER) into the nucleus and activate some mitochondrial dysfunction stimulon (MDS) genes. Overexpression of TaSIP1 and TaSIP1-∆C (a form lacking the transmembrane domain) in wheat both compromised the plants' tolerance of salinity stress, highlighting the importance of precise regulation of this signal cascade during salinity stress.

• The interaction of TaSRO1/TaSIP1, in the cytoplasm, arrested more TaSIP1 on the membrane of ER, and in the nucleus, attenuated the trans-activation activity of TaSIP1, therefore reducing the TaSIP1-mediated activation of MDS genes. Moreover, the overexpression of TaSRO1 rescued the inferior phenotype induced by TaSIP1 overexpression.

• Our study provides an orchestrating mechanism executed by the TaSRO1–TaSIP1 module that balances the growth and stress response via fine tuning the level of mitochondria retrograde signalling.

Introduction

Soil salinity represents a potent constraint over the growth and productivity of nonhalophytic plants (Munns & Gilliham, 2015; Yang & Guo, 2018b). Of the number of ways in which this stress can affect a plant's physiology, one of the more important is its effect on the regulation of mitochondrial retrograde signalling (Zhao et al., 2020) that, when triggered, activates a series of nuclear genes referred to mitochondrial dysfunction stimulon (MDS) genes (Crawford et al., 2017). As the abundance of their transcripts reflects the intensity of mitochondrial retrograde regulation (MRR), the genes have been identified as markers to uncover the regulatory factor in the response machinery (Ng et al., 2013; da Cunha et al., 2015). In brewers’ yeast (Saccharomyces cerevisiae), two basic helix–loop–helix leucine zipper transcription factors (TFs), Rtg1p and Rtg3p, have been identified as the key regulators of mitochondrial retrograde signalling; in cells where the mitochondria suffer from stress, these factors are translocated from the cytoplasm to the nucleus, thereby acting as the mediator for activating the nuclear-encoded stress response (Butow & Avadhani, 2004; da Cunha et al., 2015). In animal systems, MRR has proven to be similarly dependent on the translocation of various TFs from the cytoplasm to the nucleus (Butow & Avadhani, 2004). Mitochondrial retrograde regulation in the model plant Arabidopsis thaliana, which is generally assessed by tracking the transcriptional response of AOX1a to mitochondrial dysfunction, relies on several endoplasmic reticulum (ER) membrane-bound NTL (NAC with transmembrane motif 1-like) TFs (notably AtANAC017) that activate MDS genes by binding to the mitochondrial dysfunction motif (MDM) present in their promoters (De Clercq et al., 2013; Ng et al., 2013). In A. thaliana, apart from compromising the tolerance to chemical-triggered mitochondrial stress (Van Aken et al., 2016), the absence of AtANAC017 results in a more pronounced sensitivity to natural abiotic stress agents, including drought (Ng et al., 2013) and flooding (Meng et al., 2020). Therefore, the cytoplasmic-to-nuclear shuttling TF is assumed to be a universal component to regulate mitochondrial retrograde signalling and stress response. However, little information is known about these mechanisms in crop plants.

More recently, increasing evidence, however, indicates that the constitutive MRR leads to tumorigenesis in mammals (Wen et al., 2019), and the overexpression of AtANAC017 results in growth retardation in A. thaliana (Meng et al., 2019). Given mitochondrial retrograde signalling is efficiently elicited by environmental stimuli (Van Aken et al., 2016; Meng et al., 2019), whether MRR in the episode of natural stresses also causes an adverse effect requires a careful re-evaluation. To answer this issue is substantially more essential and practical in crops, because vigorous growth and high yield is the ultimate goal in breeding. Nevertheless, current evidence has highlighted the importance to seek factors that strictly regulate the extent of MRR. In yeast, the shuttling of Rtg1/3p from the cytoplasm to the nucleus is negatively regulated by two 14-3-3 proteins Bmh1/2p and a cytosolic phosphoprotein Mks1p (Liu et al., 2003; Butow & Avadhani, 2004); the subcellular translocation and the activity of Rtg1/3p are both suppressed by the target of rapamycin (TOR) kinase pathway (Komeili et al., 2000; da Cunha et al., 2015). However, how the shuttling TF involved in MRR is fine tuned has barely been addressed in plants.

As one of the most important crops, bread wheat (Triticum aestivum L.) belongs to glycophytes and displays high sensitivity to excess soil salinity (Wang & Xia, 2018). In addition, it is an allopolyploid species (2n = 6x = 42) with a large and complex genome (Wang et al., 2015), which has greatly impeded the understanding of its genetic bases of salinity tolerance. Previously, we generated a salinity-tolerant bread wheat (Triticum aestivum) cultivar (cv) Shanrong No. 3 (SR3) through the asymmetric somatic hybrid made between the wheat cv Jinan 177 (JN177) and the halophytic wild species tall wheatgrass (Thinopyrum ponticum) (Xia, 2009; Wang et al., 2018). Through a combined strategy of quantitative trait locus (QTL) mapping and ’omics analysis, a novel allele at the gene TaSRO1 (SIMILAR TO RCD-ONE) has been shown to explain much of the superior salinity tolerance displayed by cv SR3 over cv JN177 (Liu et al., 2014). The discovery that TaSRO1 affected the expression of multiple genes (including AOX1a) involved in reactive oxygen species (ROS) accumulation or removal therefore maintaining ROS homeostasis, has been summarised as one of the major mechanisms for salinity tolerance in wheat (Munns & Gilliham, 2015; Yang & Guo, 2018a). Recent evolutionary developments of whole genome sequences (Wang et al., 2018) have led functional genomics research in wheat into the post-genomic era, providing an opportunity to elucidate how TaSRO1 regulates its downstream genes and to demonstrate the TaSRO1-mediated signal network.

Given that both AOX1a transcript abundance and AOX activity are suppressed in A. thaliana plants constitutively expressing TaSRO1 (Liu et al., 2014), the natural hypothesis is that TaSRO1 acts as a regulator of mitochondrial retrograde signalling. Noteworthily, the phylogenomic comparative analysis has proved that genes related to abiotic stress tolerance are prone to expansion or contraction in the wheat genome (IWGSC, 2018), which can result in wheat-specific gene networks (Wang et al., 2020a). In this context, an intriguing issue is whether TaSRO1 is involved in a wheat-specific signal or regulatory pathway. Therefore, the objective of the present study was to investigate whether and how wheat TaSRO1 is involved in mitochondrial retrograde signalling, and to elucidate the significance of its involvement during an episode of salinity stress.

Materials and Methods

Wheat genotypes and growing conditions

The bread wheat cultivars JN17, JN177 and SR3 are maintained in house, and the derivation of the TaSRO1 overexpression lines used has been described elsewhere (Liu et al., 2014). The loss-of-function wheat mutants of SRO1, Kronos 2607 and 2825, generated by Krasileva et al. (2017), were procured via the Chinese distribution site, Shandong Agricultural University. To generate the pUbi::TaSIP1 and pUbi::TaSIP1-∆C constructs used to overexpress TaSIP1, the full-length/deleted versions of TaSIP1 were amplified and inserted into pTCK303. To generate an RNAi construct, the sense and antisense fragments covering the conserved region of TaSIP1-A, -B and -D were amplified and inserted into the vector pTCK303. The constructs were transformed into bread wheat (cv JN17) using a shoot apical meristem method (Liu et al., 2014). The sequence of each of the PCR primers required to construct the various transgenes is given in Supporting Information Table S1.

The field performance of wild-type (WT) wheat (cv JN17) and two selected TaSRO1 overexpressors (TaSRO1-OE) was assessed by growing the materials at a protected site at Dongying (Shandong Province, China; 37°42′14″N, 118°59′59″E), where the soil contains c. 0.4 g soluble salt per 100 g. The plot size (three plots per entry) was 1.5 m × 6 m, into which 200 g grain was sown. The analysis of spatial gene transcription was carried out on plants grown in soil-filled pots, sampled according to Wang et al. (2020b). For plants raised by hydroponics, grains were imbibed on moist filter paper at 20°C for 3 d, and the seedlings removed to half strength Hoagland's liquid medium (pH 6.0), which was replaced every 2 d until the seedlings had reached the two-leaf stage. To check the phenotypic consequences, wheat seedlings were exposed to 150 mM/200 mM NaCl, 50 μM antimycin A (AA) or 10 μM rotenone for 5 d. The long-term phenotypic consequences of the imposition of salinity stress were assessed following the method given by Wang et al. (2020a). To characterise the transcriptional response to abiotic stress, the medium was adjusted to contain one of either 200 mM NaCl, 50 μM AA or 10 μM rotenone, and the seedlings were left to grow for a further 0.5, 1, 6 or 24 h. The environment in the growth chamber was as described by Wang et al. (2020a). To avoid any effect of diurnal variation, the sampling time point was fixed. RNA was extracted from seedling tissue and processed to provide a template for qRT-PCR assays following Wang et al. (2020a). The primers used to quantify transcript abundance in wheat tissue were able to amplify all three homeoalleles. The TaEF1-α (M90077) sequence was used as the reference (Paolacci et al., 2009). The relevant primers are given in Table S1.

Arabidopsis materials and growth conditions

For details of the generation of p35S::TaSIP1 and p35S::TaSIP1∆C transgenic lines, atanac017 mutant, the overexpression line of AtANAC017, and Arabidopsis plant growth, please refer to Methods S1. Primers used for plasmid constructs are listed in Table S1.

Gene identification and phylogeny

The putative c. 2000 bp upstream and coding sequences of TaSIP1, TaAOX1 and various wheat MDS genes were obtained from the WheatOmics database (Ma et al., 2021). Bread wheat members of the NAC-b gene family were identified from the published literature (Borrill et al., 2017; Guérin et al., 2019). Homologues in barley, rice, Brachypodium distachyon and A. thaliana were recovered from the Phytozome (phytozome.jgi.doe.gov/) database. Gene IDs are given in Tables S2 and S3. The McScan program (github.com/tanghaibao/jcvi/wiki/Mcscan-(Python-version)) and the online tool Triticeae-gene Tribe (http://wheat.cau.edu.cn/TGT/) was utilised to assess collinearity (Table S4). Deduced NAC-b and AOX1 polypeptide sequences were aligned using ClustalW software (www.clustal.org) and their phylogeny deduced using Mega 6 software.

Yeast-two-hybrid assay

A ProQuest Two-Hybrid System (Invitrogen) was used to identify proteins able to interact with TaSRO1. For details please refer to Methods S2.

Bimolecular fluorescence complementation (BiFC) assays in wheat protoplasts

Bimolecular fluorescence complementation assays were performed as described elsewhere (Walter et al., 2004). The full-length TaSRO1 sequence and a part of its N terminal sequence were introduced into the pSPYNE(R)173 vector to produce the constructs p35S::nYFP-TaSRO1 and p35S::nYFP-TaSRO1NT. The coding sequences of TaSIP1, TaSIP1-∆C and TaNTL39 were cloned into the pSPYCE(MR) vector to generate p35S::cYFP-TaSIP1, p35S::cYFP-TaSIP1∆C and p35S::cYFP-TaNTL39. Vectors were transfected into the wheat protoplasts using polyethylene glycol (PEG)-mediated transformation (Liu et al., 2014). The ER marker BiP-mCherry and nucleus marker NLS-mRFP-3G were used for co-localisation analysis. After incubation in the dark for 16 h at 23°C, yellow fluorescent protein (YFP) and red fluorescent protein (RFP) activity were detected using confocal laser scanning microscopy, and excitation wavelengths of 488 and 543 nm, respectively.

Subcellular localisation of TaSIP1

For details of plasmid construction of p35S::GFP-TaSIP1 and localisation analysis in response to different stresses, please refer to Methods S3.

In vitro glutathione S-transferase (GST) pull-down assay

The TaSRO1-His construct used was generated by Liu et al. (2014). To obtain GST-tagged TaSIP1 and TaSIP1-∆C sequences, TaSIP1 and TaSIP1∆C were introduced into pGEX-6P-1, and from then into Escherichia coli BL21 (DE3). TaSRO1-His and GST-TaSIP1/TaSIP1∆C recombinant protein were purified from the culture supernatant using a Ni-NTA purification system (Invitrogen) and a Glutathione Sepharose 4B affinity column chromatography (GE Healthcare, Chicago, IL, USA), respectively. Following their purification, the recombinant proteins were quantified using a Bio-Rad protein assay reagent. The protocol of pull-down experiments involving GST-TaSIP1 or GST-TaSIP1∆C and TaSRO1-His is described elsewhere (Li et al., 2016).

Cellular fractionation and immunoblot assay

For details of subcellular fractionation of TaSIP1 in protoplasts, please refer to Methods S4. For details of isolation of the cytoplasm and nuclei fraction of TaSRO1 in wheat plants, please refer to Methods S5.

TaSIP1 processing

For details of TaSIP1 processing, please refer to Methods S6.

Electrophoretic mobility shift assay (EMSA)

Electrophoretic mobility shift assay was performed following the protocol supplied with the Cool-Shift Non-Radioactive EMSA kit (Viagene Biotech, Ningbo, China). The TaAOX1a.1 and a.2 oligonucleotide probes, comprising the MDM elements (Table S7) were biotin-labelled at their 5′ end. Purified GST-TaSIP1, with or without purified TaSRO1-His, was incubated for 20 min with one of the probes. The reaction products were subsequently separated electrophoretically at room temperature through a native 5.5% polyacrylamide gel and the signal was detected using an ImageQuant 400 ECL CCD camera (GE Healthcare).

Protein–DNA pull-down assay

Mitochondrial retrograde regulation-associated promoter sequences containing MDM were labelled with biotin and PCR amplified using a Platinum Taq DNA polymerase kit (Invitrogen). The resulting amplicons were purified using a DNA Gel Extraction kit (Fermentas, St Leon-Rot, Germany). The biotinylated DNA was attached to 1 μg streptavidin-coated beads (Sigma) by holding a suspension of the DNA, beads and buffer (15 mM HEPES/NaOH (pH 7.9), 0.1 M potassium glutamate, 5 mM MgCl₂, 5% (v/v) glycerol and 0.1% v/v NP40) overnight at 4°C. After rinsing the beads in the same buffer, they were resuspended at 4°C for at least 2 h in the same buffer supplemented with 0.1 g ml⁻¹ bovine serum albumin (BSA) and 0.2 M dithiothreitol (DTT) either with or without TaSRO1-His. The beads were then rinsed five times with buffer (15 mM HEPES/NaOH (pH 7.9), 0.1 M potassium glutamate, 5 mM MgCl₂, 0.1% v/v NP40), and the attached proteins were first eluted by the addition of 50 μl 2× SDS sample buffer, then were electrophoretically separated on an 8% SDS-polyacrylamide gel. Before their incubation with streptavidin beads, an aliquot of each protein sample was set aside to estimate its content of TaSIP1∆C. Western blots were probed with either a 1 : 3000 dilution of anti-GST antibody (Merck, St Louis, MO, USA) or a 1 : 1000 dilution of anti-His antibody (Qiagen). The primers are given in Table S1.

Dual luciferase transient expression assay using wheat protoplasts

The effector constructs p35S::TaSRO1-T7 and p35S::TaSIP1-T7 were generated via introducing the TaSRO1 or TaSIP1 open reading frame into the p326-T7 vector. Reporter constructs were produced by introducing the promoters of MRR-associated genes into the pGreen II 0800-LUC vector (Wang et al., 2020b). Each of the reporter constructs, together with either p35S::TaSRO1-T7 and/or p35S::TaSIP1-T7, were co-transformed into wheat protoplasts. Firefly and Renilla luciferase activity was measured using a dual luciferase reporter assay system (Promega). The primers are given in Table S1.

Transcriptomic analysis

Wheat WT JN17 and three TaSIN∆C overexpression (OE1, OE2, OE3) lines were used for RNAseq analysis. The VC, Col-0, TaSINFL OE (FL3) and TaSIN∆C OE (∆C26, ∆C27) seedlings (T₃ generations) were used for microarray analysis. For details please refer to Methods S7.

Results

TaSRO1 overexpressors gave better yield than WT plants under salinity stress and showed an altered profile of AOX1 transcription

Confirming the outcome of previous trials (Liu et al., 2014), field experiments carried out over three consecutive seasons (2016–2017, 2017–2018 and 2018–2019) showed that plants overexpressing TaSRO1 outyielded WT ones by c. 5% when grown in a saline–alkali soil (Fig. 1a–d). Assessing the effect of the transgene on the abundance of AOX1 transcript, which was suppressed in the TaSRO1 transgenic A. thaliana plants (Liu et al., 2014), first required the identification of the set of AOX1 homologues harboured by the bread wheat genome. A query based on rice and barley AOX1 sequences (Wanniarachchi et al., 2018) uncovered the presence of six homologues in bread wheat, namely TaAOX1a.1 through 4, TaAOX1c and TaAOX1d (Fig. S1a; Table S2). The four TaAOX1a genes (TraesCS2A01G438200, −438300, −439100 and −439400) were closely linked to one another on chromosome 2A, probably a result of multiple duplication events (Fig. 1e; Table S4).

Fig. 1

TaSRO1 overexpression enhances yield of bread wheat, and reduces TaAOX1as expression under salinity stress. (a) Abundance of TaSRO1 transcript in wild-type (WT) bread wheat (cv JN17) and two independent TaSRO1 overexpressors (OE16 and OE18). (b) Grain yield. (c) Thousand kernel weight of grain. (d) Spike number set by plants grown under saline–alkali conditions. Data shown in (a, b, c, d) in the form mean ± SE (n = 6). *,**: means differ significantly from WT at P < 0.05 and < 0.01, respectively, based on Student’s t-test. (e) AOX1a homologues display synteny in the genomes of rice, Brachypodium distachyon, barley and bread wheat. Genes having a ‘+’ orientation are shown as green boxes, and those having a ‘−’ orientation as blue boxes. (f) Transcriptional response to salinity stress of TaAOX1 genes shown by WT bread wheat and two TaSRO1 overexpressors. na, no data available. Data points represent the mean ± SD (n ≥ 3).

In seedlings exposed to salinity stress for 6 h, TaAOX1a.1, a.2 and a.3 were all induced to a similar extent, whilst a.4 appeared to be silent (Fig. 1f). TaAOX1d required a 24 h exposure to salinity to become marginally induced, whilst no induction of TaAOX1c was recorded. The abundance of TaAOX1a.1, a.2 and a.3 transcript was much lower in the TaSRO1 overexpressors than in the WT, whether or not the plants had been exposed to salinity stress, whilst that of both TaAOX1c and TaAOX1d was unaffected by the presence of the transgene (Fig. 1f). In addition, in the wheat mutants of SRO1, the expression levels of TaAOX1a.1, a.2 and a.3 were significantly higher than those in WT (Fig. S1b). In plants challenged with either antimycin A (AA) or rotenone (both of which directly induce mitochondrial stress and then trigger MRR), only TaAOX1a.1, a.2 and a.3 proved to be responsive (Fig. S1c). Intriguingly, TaSRO1 could also be induced by exposure to either AA or rotenone (Fig. S1c). A further analysis revealed that the cis-regulatory element MDM was present within the first 1 kbp upstream of each of TaAOX1a.1, a.2 and a.3, but not in any of TaAOX1a.4, TaAOX1c or TaAOX1d (Table S5). The indication was therefore that TaAOX1a.1, a.2 and a.3 all encode MDS components and are each downregulated by TaSRO1.

TaSRO1 interacts with a membrane-bound NAC TF

To illustrate how TaSRO1 is involved in MRR, a yeast-two-hybrid (Y2H) screen of a cv SR3 seedling cDNA library using the TaSRO1 as the bait (Fig. 2a) yielded several positive clones (Fig. S2a,b; Table S6), some of which encoded the identical NTL TF, harbouring a NAC domain at its N terminus and an α-helical transmembrane domain at its C terminus (Fig. 2b). The protein, from this point forwards is referred to as TaSIP1 (TaSRO1 Interacting Protein 1, accession number in GenBank database: MH165286.1). A further Y2H assay revealed that the RCD-SRO-TAF4 (RST) domain (Fig. 2a) was essential for TaSRO1 to interact with TaSIP1 (Fig. 2c). In addition, a truncated form of TaSIP1 (TaSIP1∆C), which lacked the transmembrane and adjacent sequence segment (Fig. 2b), also showed an ability to interact with TaSRO1 (Figs 2c, S2c). A GST pull-down assay further proved that both GST-TaSIP1 and GST-TaSIP1∆C, but not GST, interacted with TaSRO1 in vitro (Fig. 2d).

Fig. 2

TaSIP1 interacts with TaSRO1 in vitro and in vivo. (a) Schematic representation of TaSRO1, which includes a PARP signature and the two protein–protein interaction domains WWE and RST. (b) Predicted structure PARP signature of TaSIP1. NAM, no apical meristem protein domain; TM, transmembrane motif. (c) Interaction of TaSRO1 with TaSIP1 demonstrated by a yeast-two-hybrid assay. ++, strong positive control; +, weak positive control; −, negative control; VC, empty vector control. (d) Pull-down assays of TaSRO1-His involving GST-TaSIP1and GST-TaSIP1∆C. GST was used as a negative control. The middle TaSRO1-His panel shows the input (20%) of the TaSRO1-His protein from Escherichia coli. Expressed products. (e) Bimolecular fluorescence complementation (BiFC) analysis used to detect the TaSRO1/TaSIP1 and TaSRO1/TaSIP1∆C interactions. The YFP N terminus was fused to the full length of TaSRO1 to produce p35S::nYFP-TaSRO1, whilst the full length or ∆C fragment of TaSIP1 was fused to the YFP C terminus to produce p35S::cYFP-TaSIP1 or p35S::cYFP-TaSIP1∆C. BIP-mCherry, subcellular marker of ER; NLS-mRFP-3G, subcellular marker of the nucleus. Combinations of these plasmids were transformed into wheat protoplasts. The right column presents the ratio of wheat protoplasts co-expressing p35S::nYFP-TaSRO1 and p35S::cYFP-TaSIP1 (n = 54) that exhibit an ER-localised or ER + nucleus-localised fluorescence signal, and the ratio of wheat protoplasts co-expressing p35S::nYFP-TaSRO1 and p35S::cYFP-TaSIP1∆C (n = 51) that exhibit a nucleus-localised fluorescence signal. Bar, 10 μm. (f) Phylogeny of the NAC-b family of proteins encoded by Arabidopsis thaliana (green circles), rice (yellow squares), Brachypodium distachyon (light blue triangles), barley (dark blue triangles) and wheat (diamonds). TaSIP1 is highlighted by the red arrow.

To test whether the TaSRO1–TaSIP1 interaction occurred in planta, a BiFC assay was applied to bread wheat protoplasts. When p35S::nYFP-TaSRO1 and p35S::cYFP-TaSIP1 were co-transfected, a YFP signal was detectable (Figs 2e, S2d). Intriguingly, cells with the YFP signal could be divided into two types; c. 90% of cells exhibited the ER-localised signal, and the rest exhibited the signal in both the ER and the nucleus (Fig. 2e). By contrast, when p35S::nYFP-TaSRO1 and p35S::cYFP-TaSIP1∆C were co-transfected, the signal was restricted to the nucleus (Figs 2e, S2e). No signal was detected in protoplasts co-expressing p35S::nYFP-TaSRO1 and p35S::cYFP-TaNTL39 (TraesCS7A02G000300) (Fig. S3b), another wheat NTL TF that showed no interaction with TaSRO1 in the Y2H assay (Fig. S3a). The conclusion was that TaSRO1 has the capacity to interact with TaSIP1 both in vitro and in planta.

Evolutionary analysis of TaSIP1

We compared TaSIP1 with the reference genome v.1.1 of bread wheat model cv Chinese Spring (IWGSC, 2018), and found it shared 100% identity with TraesCS5D02G279100, whose homoeo-alleles were TraesCS5A02G271500 and TraesCS5B02G271800. Through genomic collinearity analysis (Fig. S4a), the orthologues of TaSIP1 were HvHNAC046 (HORVU5Hr1G074230) in barley, BdBNAC079 (BRADI_4g34022) in B. distachyon, and OsONAC042 (LOC_Os09g32040) in rice, which were all assigned to the NAC-b phylogenetic group, a subfamily mainly consisting of transmembrane domain-containing NACs (Murozuka et al., 2018). Intriguingly, zooming in on the previously generated phylogenetic trees (Murozuka et al., 2018) revealed that HvHNAC046, BdBNAC079 and OsONAC042 were the unique members in barley, B. distachyon, and rice that shared the same clade with Arabidopsis AtANAC013/016/017. We next isolated all the NAC-b type genes in bread wheat (Borrill et al., 2017; Guérin et al., 2019), and re-generated the phylogenetic tree based on amino acid sequences (Fig. 2f). TaSIP1 uniquely located in the clade with the orthologues above as well as AtANAC013/016/017.

TaSRO1 influences the subcellular localisation of TaSIP1

When the p35S::GFP-TaSIP1 transgene was transiently expressed in wheat protoplasts, c. 70% of transgenic cells displayed a network pattern of GFP signal in the cytoplasm (Fig. 3a), which was co-localising with that of BiP-mCherry, an ER marker (Fig. S5a). By contrast, c. 20% of cells showed a GFP signal in both the cytoplasm and the nucleus, whilst a small proportion of cells (<10%) show a concentrated nuclear signal (Figs 3a, S5a). When the sequence was altered to remove the predicted transmembrane domain, the GFP signal became confined to the nucleus (Figs 3a, S5b). Analysis of the sequence of TaSIP1 revealed that it contained a putative rhomboid protease cleavage site (Ng et al., 2013) at the C-terminal TM domain (Fig. S5c), suggesting that TaSIP1 could be proteolytically processed. Following the transient expression of p35S::GFP-TaSIP1 in A. thaliana protoplasts and the separation of expressed proteins into the soluble and membrane fractions, western blotting using anti-GFP antibody showed that bands at c. 140 and c. 116 kDa were only detected in the membrane fraction (Fig. S5d). The upper one, larger than the estimated molecular mass of the full-size TaSIP1 (c. 110 kDa), was predicted to be a post-translationally modified form, which was previously also found with other membrane-associated TFs, such as AtANAC062 (Seo et al., 2010; Yang et al., 2014), MfNACsa (Duan et al., 2017) and AtbZIP28 (Liu et al., 2007). By contrast, when p35S::TaSIP1∆C-GFP was transiently expressed, its product was mainly detected in the soluble fraction (Fig. S5e). The implication was that TaSIP1 is an ER membrane protein, able to proteolytically release a form that translocates to the nucleus.

Fig. 3

Subcellular localisation of TaSIP1, which is influenced by TaSRO1. (a) Site of GFP activity in wheat protoplasts expressing p35S::GFP, p35S::GFP-TaSIP1 or p35S::TaSIP1∆C-GFP transgene. Bar, 10 μm. (b) Fluorescence signal in wheat protoplasts co-expressing p35S::RFP-TaSIP1 and p35S::TaSRO1-GFP. Bar, 10 μm. (c) Proportion of TaSIP1 located in the nucleus (N), on the endoplasmic reticulum (ER), or at both sites (ER + N) of wheat protoplasts (n = 100) co-expressing p35S::RFP-TaSIP1 and p35S::GFP (left column) or p35S::RFP-TaSIP1 and p35S::TaSRO1-GFP (right column). (d) Proportion of TaSIP1 located in the N, on the ER, or at both sites (ER + N) of wheat protoplasts (n = 100) expressing a p35S::GFP-TaSIP1 transgene. Protoplasts were extracted from the wild-type (WT; left column) or TaSRO1 overexpressing wheat (middle and right columns) lines. (e) Western blot analysis using anti-GFP antibody illustrates TaSIP1 abundance in the cytoplasmic (α-GFP-C) and nuclear (α-GFP-N) fractions in the absence or presence of TaSRO1. Total protein was extracted from protoplasts expressing the p35S::GFP-TaSIP1 without or with the p35S::TaSRO1-T7. Molecular masses in kDa are indicated on the left side. Bands based on anti-T7 antibody were used to show the abundance of TaSRO1. The large subunit of RuBisCo (RbcL) and actin were used as the loading controls. Experiments were repeated three times independently with similar results. (f) Relative processing efficiency of TaSIP1 in (e). Intensities of blots were quantified using ImageJ software. The ratios of the processed TaSIP1 form (α-GFP-N) relative to the full-length TaSIP1 form at each concentration were compared with the ratio at 0. Data are shown in the form mean ± SD (n = 3). *, Means differ significantly from one another at P < 0.05, based on Student’s t-test.

When p35S::TaSRO1-GFP and p35S::RFP-TaSIP1 were co-expressed in wheat protoplasts (Fig. 3b), c. 90% of transgenic cells displayed the exclusively ER-localised RFP signal (Fig. 3c). By contrast, when p35S::TaSRO1-GFP was replaced with the empty vector, this proportion became c. 70% (Fig. 3c). When p35S::GFP-TaSIP1 was transformed into protoplasts extracted from TaSRO1-overexpressing wheat lines, cells exhibiting ER-localised signal accounted for c. 85% of transgenic cells, whilst this ratio was c. 65% when the transformation was performed in WT wheat protoplasts (Fig. 3d). Moreover, a western blot analysis provided further evidence that increasing the titre of p35S::TaSRO1-T7 co-transformed into protoplasts alongside a constant titre of p35S::GFP-TaSIP1, could significantly enhance the TaSIP1 protein abundance in the membrane fraction (Fig. 3e) and reduce the proteolytic processing efficiency (Fig. 3f). The above results suggested that TaSRO1 was able to retain TaSIP1 on the ER membrane and therefore regulate the ER-to-nucleus shuttling of TaSIP1. Consistent with the subcellular result that a few of wheat protoplasts solely expressing p35S::TaSRO1-GFP exhibited the cytoplasmic signal (Fig. S6a), western blotting using anti-TaSRO1 antibody showed that the TaSRO1 protein was detectable in the cytosolic fraction extracted from cv SR3 seedlings, and the salinity treatment strongly increased the cytosolic TaSRO1 protein abundance (Fig. S6b).

The response of TaSIP1 and its encoded protein to stress agents

Both the developmental expression atlas of bread wheat (Ramírez-González et al., 2018) and a quantitative real time PCR (qRT-PCR) assay (Fig. S7) confirmed that TaSIP1 was active in the root and leaf of plants at the vegetative stage, and particularly so in senescent tissue of plants at the reproductive stage. When exposed to either AA or rotenone, TaSIP1 was significantly induced within 0.5 h, but after 6 h the level of transcript fell to the basal expression level observed in nontreated plants. When exposed to salinity, the induction of TaSIP1 was slower (at 1 h) and less intense (Figs 4a, S8).

Fig. 4

TaSIP1 reduces salinity stress tolerance of bread wheat. (a) Transcription profiling of TaSIP1 in response to either salinity stress or chemical-triggered mitochondrial stress. Mock condition is normal liquid medium (for salinity), medium plus ethanol (for antimycin A (AA)) or medium plus dimethyl sulphoxide (DMSO) (for rotenone). CT, nonstressed conditions. Data points represent the mean ± SD (n ≥ 3). Different letters above the columns indicate significant (P < 0.05) differences between means, based on a one-way Waller–Duncan test. (b–d) Phenotypic effect of manipulating the expression of TaSIP1 following a limited-term episode of salinity stress: (b) seedling phenotype, (c) root elongation, and (d) shoot elongation. WT, wild-type wheat (cv JN17); FL-OE, TaSIP1 overexpressors; ∆C-OE, TaSIP1-∆C overexpressors; Ri, TaSIP1-RNAi knockdown plants; CT, nonstressed conditions. Bar, 5 cm. (e–g) Phenotypic effect of manipulating the expression of TaSIP1 following a longer term episode of salinity stress: (e) plant phenotype, (f, g) accumulation of fresh weight. CT, nonstressed conditions. Bar, 10 cm. Data in (c, d, f, g) are shown in the form mean ± SE (n = 20). Different letters above the columns indicate significant (P < 0.05) differences between means, based on a one-way Waller–Duncan test.

A further investigation was performed to answer whether stress agents affected the subcellular localisation of TaSIP1. When wheat protoplasts expressing p35S::GFP-TaSIP1 were treated with AA, the proportion of protoplasts, in which GFP signal was restricted to the nucleus, was increased (Fig. S9a). After the exposure of A. thaliana transgenics harbouring p35S::GFP-TaSIP1 to AA, the titre of TaSIP1 in the nuclear form increased greatly compared with that in the ER-localised form (Fig. S9b,c). By contrast, exposure to mild and moderate salinity stress had no effect on TaSIP1 subcellular distribution, whilst high salinity stress tended to suppress the processing efficiency, as shown by both the pattern of GFP activity and a western blot analysis (Fig. S9d–f).

TaSIP1 improves the tolerance to mitochondrial stress but reduces the tolerance to salinity stress

A series of genotypes in which the expression of TaSIP1 was altered (Fig. S10) was investigated in an attempt to understand the phenotypic significance of this gene. Under nonstressed conditions, there was no visible phenotypic effect of either overexpressing (FL-OE) or knocking down TaSIP1 (Ri) in wheat (Fig. 4b–g), nor were A. thaliana plants phenotypically affected by the constitutive expression of TaSIP1 (FL-CE) (Fig. S11); however, overexpressing TaSIP1-∆C (∆C-OE) in wheat resulted in a marked retardation of growth (Fig. 4b–f), whilst its constitutive expression in A. thaliana (∆C-CE) reduced leaf size (Fig. S11a).

FL-OE, and particularly ∆C-OE wheat plants challenged by exposure to NaCl suffered a greater reduction in both shoot and root growth than did WT seedlings. By contrast, knocking down of TaSIP1 boosted shoot and root elongation under salinity stress conditions (Fig. 4b–d). A longer exposure to a lower concentration of NaCl inhibited the growth of FL-OE and ∆C-OE wheat seedlings more strongly than that of WT seedlings, whilst the Ri plants were the least affected (Fig. 4e–g). Similarly, the A. thaliana transgenics FL-CE and ∆C-CE responded less well to salinity stress than did WT, with respect to both their ability to germinate and their seedling growth (Fig. S11d–i). Intriguingly, compared with WT, AtANAC017 overexpressor in A. thaliana (Hu et al., 2016) also showed a more sensitive phenotype under salt stress, whilst atanac017-1 loss-of-function mutant (SALK_022174) was less sensitive (Fig. S12). By contrast, the extent of the inhibition over both stem and root growth imposed by exposure to either AA or rotenone was less serious for ∆C-OE than for WT wheat seedlings, whilst Ri seedlings were more strongly compromised (Fig. S13). Similarly, the constitutive expression in A. thaliana of either TaSIP1-FL or TaSIP1-∆C enhanced the plants' level of tolerance to both AA and rotenone (Fig. S11b,c). Therefore, the effect of TaSIP1 on mitochondrial stress tolerance was opposite to its effect on salinity tolerance.

TaSIP1 directly participates in MRR

To gain a comprehensive insight into the role of TaSIP1, particularly in the regulation of MRR, a transcriptomic analysis was performed using A. thaliana genetic materials. More genes were up/downregulated in plants constitutively expressing TaSIP1-∆C (∆C-CE) than those in FL-CE (Fig. S14a). Notably, almost all (23/24) of reported MDS genes (De Clercq et al., 2013) – including AtAOX1a – were upregulated in both ∆C-CE and FL-CE plants (Fig. S14b; Tables S7, S8), but the upregulation in the former was more pronounced (Fig. S14b,c). In addition the WT and three independent TaSIP1-∆C overexpressing wheat lines were used to perform RNA sequencing, and both transcriptomic analysis (Fig. S15a) and qRT-PCR validation (Fig. S15b) showed that MDS homologues (Table S5) also tended to be significantly upregulated in the ∆C-OE wheat transgenics. By contrast, qRT-PCR assay revealed that in Ri seedlings these genes were typically transcribed less abundantly than in WT seedlings (Fig. S16). It should be noted that the homologous genes of At2g32020, At3g50930 and At5g43450, similar to the homologues of AtAOX1a (Fig. 1d), were tandemly duplicated in bread wheat genome (Figs S4b,c, S15a; Table S5). A query to the reference genome of bread wheat showed that MDM was frequently present in the putative promoter region of these MDS genes (Fig. 5a; Table S5).

Fig. 5

TaSIP1 participates in the activation of mitochondrial retrograde regulation (MRR)-associated genes, which are repressed by TaSRO1. (a) Schematic diagram of MDM element (highlighted by underlining) present in the promoter of selected MRR-associated genes. The core part of the MDM element is highlighted in red. (b) Binding affinity of TaSIP1, in the absence or presence of TaSRO1, to the MDM element harboured by the TaAOX1a.2 promoter (MDM2), as shown by electrophoretic mobility shift assay (EMSA). Competitor-M, the competitor-mutated MDM element. (c) Protein–DNA pull-down assay of biotinylated DNA sequences harbouring an MDM element present in the promoters of TaAOX1a.2 (MDM2), TaAOX1a.3 (MDM3) and TraesCS7D01G484500 (MDM4) with purified TaSIP1∆C-His protein, in the absence or presence of TaSRO1-His. (d) Schematic representation of various constructs used in the dual luciferase transient expression assay. Arrows after the 35S promoter indicate the transcriptional start sites. (e) TaSIP1-mediated activation of mitochondrial dysfunction stimulon genes and self-activation of TaSIP1, in the absence (dark grey) or presence (light grey) of TaSRO1, as shown by a dual luciferase transient expression assay. FL, TaSIP1; ∆C, TaSIP1-∆C. Data are shown in the form mean ± SD (n = 4). Different letters above the columns indicate significant (P < 0.05) differences between means, based on a one-way Waller–Duncan test.

An EMSA experiment revealed that TaSIP1 could bind with probes containing the MDM domain presented in the promoters of TaAOX1a genes (Figs 5b, S17); by contrast, when MDM in the competitor was mutated, the competition against the interaction between TaSIP1 and probes was much suppressed (Figs 5b, S17). The outcome of a pull-down experiment further supported the notion that TaSIP1 can bind to the MDM presented in MRR-associated genes (Fig. 5c). When a dual luciferase assay was applied to wheat protoplasts (Fig. 5d), the introduction of either TaSIP1-FL or -∆C was found to activate MRR-associated genes; the presence of the latter transgene had the larger effect (Fig. 5e). Note that the TaSIP1 promoter also harbours an MDM element (Fig. 5a; Table S5), which may explain why TaSIP1 is able to self-regulate (Fig. 5e). The overall conclusion was that TaSIP1 acts as a positive regulator of MRR in bread wheat.

The interaction between TaSRO1 and TaSIP1 also suppresses the TaSIP1-mediated transcriptional activation of MRR-associated genes in the nucleus

Both transcriptomic profiling (Zhang et al., 2016) and qRT-PCR analyses confirmed that most TaSIP1-targeted MDS genes in WT plants were strongly upregulated by salinity stress (Figs 1e, 6d), whilst this induction was suppressed in TaSIP1-Ri plants (Fig. S16), suggesting that TaSIP1 mediated the salinity stress response of these genes. However, in TaSRO1-OE plants, not only were TaAOX1a genes suppressed (Fig. 1f), but also the transcriptional abundance of other MRR-associated genes was lower than in WT plants, whether or not the plants were subjected to salinity stress (Fig. 6d). In the cv SR3, as anticipated given its enhanced level of TaSRO1 activity (Liu et al., 2014), most of the genes targeted by TaSIP1 were less abundantly transcribed than in its parental cv JN177 (Fig. S18). The implication was that TaSRO1 negatively modulated TaSIP1 in a way that affected the expression of its downstream targets.

Fig. 6

Counteracting consequence of overexpressing TaSRO1 in a TaSIP1-∆C overexpressor. (a) Seedling phenotype of wild-type (WT) wheat (cv JN17), a TaSIP1-∆C overexpressor (TaSIP1∆C-OE), a TaSRO1 overexpressor (TaSRO1-OE18) and a plant overexpressing both TaSIP1∆C and TaSRO1 (double-OE) in the presence or absence of salinity stress. Bar, 5 cm. (b) Root and shoot length of WT, TaSIP1∆C-OE, TaSRO1-OE18 and double-OE plants grown in the absence of salinity stress or when exposed to a short-term episode of strong salinity stress. (c) Shoot fresh weight and percentage of yellow leaves of plants grown under control conditions or when exposed to a long-term episode of moderate strength salinity stress. Data in (b, c) are shown in the form mean ± SE (n = 20). (d) Abundance of transcript generated by mitochondrial retrograde regulation (MRR)-associated genes in plants grown in the absence or presence of salinity stress. CT in (b–d), nonstressed conditions. Data points represent the mean ± SD (n ≥ 3). Different letters appearing above the columns indicate significant (P < 0.05) differences between means, based on a one-way Waller–Duncan test.

One observation above was that the interaction between TaSRO1 and TaSIP1 in the cytoplasm (Fig. 2e) suppressed the ER-to-nuclear translocation of TaSIP1 (Fig. 3b–e), providing a possible explanation for how TaSRO1 inhibited TaSIP1-triggered MRR. Given that TaSRO1 was also able to interact with TaSIP1 in the nucleus (Fig. 2e), we then sought to investigate the biochemical significance of this kind of interaction. A pull-down experiment revealed that, intriguingly, the addition of TaSRO1 could enhance the ability of TaSIP1 to bind to sequences harbouring an MDM element, which was positively associated with the titre of TaSRO1 (Fig. 5c). An EMSA experiment further established that TaSRO1 strengthened the DNA binding activity of TaSIP1 (Fig. 5b). A follow-up dual luciferase assay using transgenic wheat protoplasts sought to check whether TaSRO1 affected TaSIP1’s transcriptional activation ability (Fig. 5d). When p35S::TaSRO1-T7 was additionally introduced, the level of LUC expression representing the activations of either five MDS genes or TaSIP1 by TaSIP1 or TaSIP1∆C, was significantly reduced (Fig. 5e). Therefore, the nuclear interaction between TaSRO1 and TaSIP1 could further suppress the TaSIP1-mediated activation of MDS genes, as well as the self-activation of TaSIP1.

TaSRO1 partially counteracts the TaSIP1-mediated stress response

The phenotypic consequence of the interaction between TaSRO1 and TaSIP1 was explored by generating bread wheat plants in which both TaSRO1 and TaSIP1-∆C were overexpressed (double-OE) (Fig. S19). Under nonstressed growing conditions, TaSRO1-OE seedlings produced larger than normal leaves and ∆C-OE smaller ones, whilst the double-OE plants developed leaves of normal size (Fig. 6a–c). ∆C-OE seedlings were sensitive to salinity stress, but the introduction of TaSRO1 was able to partially complement this sensitivity (Fig. 6a–c). A similar mitigation effect on salinity tolerance was achieved by crossing transgenic A. thaliana plants harbouring either FL-CE or ∆C-CE lines with those constitutively expressing TaSRO1 (Fig. S20a). With respect to the tolerance to mitochondrial dysfunction, double-OE wheat plants were able to tolerate exposure to rotenone rather better than WT plants, but their tolerance was less pronounced than that shown by ∆C-OE plants (Fig. S21).

A comparison of the transcript abundance of TaSIP1-targeted MDS genes showed that, under nonstressed conditions, the upregulation of the MDS genes in the double-OE was less pronounced compared with that in the single ΔC-OE, whilst under salt stress condition, the upregulation was similar to that in the WT plants (Fig. 6d). In addition, the abundance of transcript of these MRR-associated genes appeared to be negatively correlated with the level of salinity tolerance. Similar results were obtained from a comparison between the transgenic A. thaliana plants (Fig. S20b).

Discussion

A wheat-specific TaSIP1-MRR signalling pathway

Here, the bread wheat protein TaSIP1 was shown to be an ER membrane-bound NAC domain-containing TF able to interact with TaSRO1 (Figs 2, S2). Among the family of bread wheat proteins harbouring a NAC domain (Borrill et al., 2017; Guérin et al., 2019), this is the only one that maps to the same phylogenetic clade as AtANAC017 (Figs 2f, S4a). As is also the case for AtANAC017 (Ng et al., 2013), mitochondrial stress appears to facilitate the proteolytic release of TaSIP1 from the ER membrane, allowing it to migrate to the nucleus (Fig. S9), where it binds to the MDM domains present in the promoters of MDS genes such as TaAOX1a (Figs 5a–c, S17), thereby activating them (Fig. 5d,e). The constitutive expression of TaSIP1 in both wheat and A. thaliana had the effect of enhancing the level of tolerance to either AA or rotenone treatment, whilst the gene's knockdown had the opposite effect (Figs S11b,c, S13). The conclusion is therefore that TaSIP1 is shuttled from the cytoplasm to the nucleus and that it participates in mitochondrial retrograde signalling (Fig. 7a).

Fig. 7

Working model for the TaSRO1–TaSIP1 module. (a) Under optimal conditions, TaSIP1 is expressed at a relatively low level and its product mainly remains anchored to the endoplasmic reticulum (ER) in a latent form. Therefore, mitochondrial retrograde regulation (MRR) is not triggered. (b) Upon antimycin A (AA)/rotenone-triggered mitochondrial stress, TaSIP1 is induced and its product is translocated from the ER to the nucleus. Salinity stress also induces TaSIP1, but the translocation of TaSIP1 to the nucleus is less efficient. In the nucleus, TaSIP1 binds to the mitochondrial dysfunction motif (MDM) element present in the promoter of MRR-associated genes, thereby triggering MRR. As the TaSIP1 promoter includes an MDM element, the gene is self-activated, thereby amplifying the TaSIP1–MRR cascade. As a result, the plant's mitochondrial stress tolerance is enhanced, but the high level of MRR activity requires the expenditure of energy, resulting in growth retardation and an attenuated salinity stress tolerance. (c) As is the case for TaSIP1, TaSRO1 is also induced by either an AA/rotenone-triggered mitochondrial stress or salinity stress. Notably, the cytoplasmic TaSRO1 protein abundance is increased by exposure to salinity stress. In the cytoplasm, the interaction between TaSRO1 and TaSIP1 buffers the nuclear import of TaSIP1 to avoid the nuclear overaccumulation of TaSIP1. In addition, in the nucleus, TaSRO1 also interacts with TaSIP1, but reduces the trans-activation activity of TaSIP1, allowing for some mitigation in the transcriptional activation of TaSIP1 itself and MRR-associated genes. As a result, the plant's mitochondrial stress tolerance is retained, whilst at the same time the tolerance of salinity stress is enhanced.

The behaviour of TaSIP1 showed clear differences from that of AtANAC017. The latter is constitutively expressed when the plant experiences either mitochondrial stress or certain other stresses (Ng et al., 2013; Meng et al., 2019), but the translocation of its product to the nucleus is enhanced by AA treatment (Ng et al., 2013). In the nucleus, AtANAC017 could activate AtANAC013 via binding to the MDM element in the promoter of AtANAC013, and therefore AtANAC013 is induced by stress (Meng et al., 2019), resulting in an AtANAC017–AtANAC013 signalling cascade (De Clercq et al., 2013; Ng et al., 2013). AtANAC017 and AtANAC013 are phylogenetically related to one another, but in bread wheat and other grasses, only one homologue (i.e. TaSIP1) is present (Figs 2f, S4a). Under conditions of mitochondrial stress, TaSIP1 moves from the ER to the nucleus (Fig. S9a–c), but unlike AtANAC017, its encoding gene's promoter harbours an MDM domain (Fig. 5a), allowing it to self-activate and therefore amplify the signalling directly (Fig. 5e). These differences give an explanation for the phenomenon that TaSIP1 and its downstream MDS genes respond earlier to stresses, compared with what happens in A. thaliana (Van Aken et al., 2016). More importantly, although as the sole NTL-type regulator (Fig. 2f), TaSIP1 can target more MDS genes, such as TaAOX1a, which experience gene expansion due to the tandem or block duplication in bread wheat (Figs 1e, S4a, S15a), rendering a more straight and extensive activation of MRR. The clear inference is that TaSIP1 is an indispensable and efficient regulator of MRR in bread wheat.

A lower threshold for the extent of TaSIP1-activated MRR causing the adverse effect on salinity stress tolerance

It has been reported that the overexpression of AtANAC017 results in growth retardation, whilst the growth of knock-out mutants is somewhat enhanced (Meng et al., 2019). Apart from the negative effect on growth, our results further provided a side-effect of excessive/prolonged MRR, via overexpressing TaSIP1 (Figs 4, S11) or AtANAC017 (Fig. S12), on salinity stress tolerance. This seems contradictory to previous findings that some AtANAC017-targeted MDS genes are positive regulators of abiotic stress tolerance (Skirycz et al., 2010; Tognetti et al., 2010; Zhang et al., 2014). However, increasing evidence suggests that some other AtANAC017/TaSIP1-targeted MDS genes play negative roles in abiotic stress tolerance (Wang et al., 2010, 2016; Sakuraba et al., 2015). More importantly, the targeted MDS genes of AtANAC017/TaSIP1 are directly involved in a variety of cellular processes, such as hormone metabolism/transport (auxin, UGT74E2 and ABCB4; salicylic acid, SOT12; Cho et al., 2007; Baek et al., 2010; Tognetti et al., 2010) and signalling (ethylene, HRE2/ERF71; cytokinin, CRF6; Licausi et al., 2010; Zwack et al., 2016), and ROS burst and homeostasis (Rentel et al., 2004; Gong et al., 2021). Therefore, one plausible scenario to account for the adverse effects of AtANAC017/TaSIP1 overexpression is that the overactivation of MRR may cause energy to be dissipated and respiration efficiency to be compromised; although an antagonistic relationship between MRR and auxin has been proposed as an alternative (Ivanova et al., 2014; Kerchev et al., 2014).

Noteworthily, compared with TaSIP1▵C-OE lines that exhibited growth arrest under both nonstressed and salt-stressed conditions (Fig. 4), TaSIP1FL-OE lines displayed a lower MRR activation (Fig. S15b), which was not enough to affect the normal growth under nonstressed condition, but could still cause an inferior performance in exposure to salinity (Fig. 4). The implication is that, under salt stress, the threshold for the level of TaSIP1-triggered MRR that leads to the adverse phenotype is lower, and therefore a more robust regulation on this signalling cascade is essential. From this view, the observation that the cytoplasmic-to-nuclear shuttling of TaSIP1 was essentially unaffected by salinity stress (Fig. S9d–f) appears to be an adaptive mechanism existing in planta to avoid a constitutive or excessive episode of TaSIP1-mediated MRR. However, TaSIP1’s transcription, unlike that of AtANAC017 that is constitutive at a level unaffected by the presence of abiotic stress, is induced by salinity (Fig. 4a) and is self-activated (Fig. 5), therefore more easily reaching an undesirable level. In this scenario, TaSIP1-mediated MRR during an episode of salinity stress in particular needs to be fine tuned.

A dual role of TaSRO1 in fine tuning TaSIP1-mediated MRR under salinity stress

The cytoplasmic-to-nuclear shuttling feature of the key TF involved in mitochondrial retrograde signalling provides two regulatory strategies to avoid excessive/prolonged MRR; one is to control the nucleo-cytoplasmic trafficking process of TF (Liu et al., 2003; Butow & Avadhani, 2004), and another one is to affect its function as the transcription regulator of MRR-associated genes (Komeili et al., 2000; Ruiz-Roig et al., 2012). A recent study showed that AtRCD1, a close homologue to TaSRO1 in Arabidopsis, could interact with AtANAC017 (Shapiguzov et al., 2019). However, many biochemical details (for example, where the interaction happens) and the biological significance (particularly for the tolerance to the natural stress) of this interaction remain to be clarified. In this study, we demonstrated that TaSRO1 could interact with TaSIP1 in both the ER and the nucleus (Fig. 2e). The interaction between TaSRO1 and TaSIP1 in the cytoplasm arrested more TaSIP1 on the membrane of ER, therefore avoiding the nuclear overaccumulation of TaSIP1 (Fig. 3b–e). In addition, the result of nuclear interaction was an enhancement in the TaSIP1 binding to MDS genes (Fig. 5b,c), along with a decrease in its trans-activation activity (Fig. 5e); the net result was a reduction in the expression of MDS genes (Figs 1f, 6d, S18, S20b). Whilst this observation appeared at first to be somewhat unusual, it has recently been reported that the interaction between AtPYL8/9 and AtPIF4 has the effect of enhancing the binding of AtPIF4 to the AtABI5 promoter, thereby negatively regulating AtPIF4-mediated AtABI5 activation (Qi et al., 2020). Furthermore, the overexpression of TaSRO1 was able to rescue the salinity sensitivity and growth retardation of TaSIP1∆C overexpressors, whilst also reducing the extent to which MRR-associated genes were activated (Figs 6, S20). These observations indicate that TaSRO1 plays a dual role in regulating both the nuclear import and the transcriptional activation of TaSIP1, to efficiently refine the level of TaSIP1-mediated MRR (Fig. 7).

An intriguing issue is how TaSRO1 affected the proteolytic cleavage and the trans-activation activity of TaSIP1. One hypothesis is that the three-dimensional structure of TaSIP1 would change when it interacts with TaSRO1, given that TaSIP1/AtANAC017 harbours intrinsically disordered regions (IDRs) (Christensen et al., 2019), which possess structural flexibility depending on cell internal and external conditions (Covarrubias et al., 2017). Moreover, very recent discoveries have highlighted the new role of IDR-based protein interactions in the formation of self-assembled, membrane-less organelles through liquid–liquid phase separation (LLPS) (Emenecker et al., 2020; Huang et al., 2021). Our results revealed that the cytoplasmic interaction between TaSRO1 and TaSIP1 yielded a punctate pattern (Fig. 2e), which has been discovered in LLPS (Emenecker et al., 2020). In the meantime, we found that TaSRO1 could interact with several kinases and phosphatases (Table S6) (Ma et al., 2021), and AtRCD1 was also proved to interact with photoregulatory protein kinases (PPKs) (Wirthmueller et al., 2018). Intriguingly, it has been reported that the phosphorylation status of NTL protein affected its nuclear import (Kim Mi et al., 2012). Therefore, another possibility is that TaSRO1 is a scaffold to recruit other proteins and therefore affect the phosphorylation modification of TaSIP1. Nevertheless, both hypotheses require a further investigation in the future.

Taken together, our study presented a novel concept of fine tuning MRR to avoid the detrimental effects of a prolonged or excessive episode of mitochondria retrograde signalling under natural stress conditions. Furthermore, we proved that TaSRO1 is an important regulator of this process, in particular providing efficiently control over the TaSIP1–MRR cascade (Fig. 7). Together with our previous discovery that TaSRO1 is a major contributor to the maintenance of ROS homeostasis under salt stress (Liu et al., 2014) and recent findings that AtRCD1 coordinates chloroplast and mitochondrial functions via integrating signalling from both organelles (Shapiguzov et al., 2019), this gene family seems to be a coordinator to balance the level of stress-related signals, which is promising in crop breeding.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (31720103910, 31771353, 32072064, 31471158 and U1906202), the Natural Science Foundation of Jiangsu Province, China (BK20200110), and Youth Innovation Promotion Association of Chinese Academy of Sciences (2022314).

Author contributions

GX devised and supervised the project. MW performed most of the experiments. MW wrote the manuscript, performed most of the experiments in wheat, and conducted the evolutionary and bioinformatics analyses. MZ conducted some physiological assays. MW performed the field experiments. SL performed the EMSA. YT conducted western blotting in transgenic plants. BM conducted western blotting and protein fractionation assays in protoplasts. CL and CL characterised the overexpression and RNAi lines of wheat. WS, MB, SL, WZ and IH discussed the manuscript. MW and MW contributed equally to this work.

Open Research

Data availability

The raw RNA-Seq data have been deposited in the Gene Expression Omnibus (GEO) database under accession no. GSE156679 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE156679). Microarray data from this article can be found at GEO (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE124871) under accession no. GSE124871.