Introduction

MicroRNAs (miRNAs) are a class of abundant endogenous single-stranded small RNAs (~ 21 nt) produced from noncoding, imperfectly complementary (stem-loop) RNA precursors whose biogenesis involves a complex, multistep enzymatic process (Park et al. 2005; Voinnet 2009; Rogers and Chen 2013). miRNAs play important roles in regulating plant growth and development, signal transduction, protein degradation, and responses to environmental stress and pathogen invasion (Shukla et al. 2008; Chen 2009; Voinnet 2009; Song et al. 2019). The main function of miRNAs is to negatively regulate gene expression at the transcriptional and posttranscriptional levels, including mRNA cleavage, translational repression, and epigenetic modification (Voinnet 2009; Shivram et al. 2019). With the development of high-throughput sequencing techniques, an increasing number of plant miRNAs have been identified, most of which are unique and nonconserved (Fahlgren et al. 2007; Cuperus et al. 2011). However, the bulk of studies to date have focused on highly conserved miRNAs whose functions have been broadly maintained throughout plant evolution, with fundamental roles in regulating numerous cellular events (Willmann and Poethig 2007).

The miR164 family (miR164a, miR164b, and miR164c; hereafter referred to as miR164) is a highly conserved miRNA family that functions in various biological processes, including boundary establishment and maintenance, organ formation, lateral root development, age-dependent cell death, and resistance to biotic and abiotic stresses (Laufs et al. 2004; Mallory et al. 2004; Guo et al. 2005; Fang et al. 2014; Feng et al. 2014). miR164 targets a class of NAC (NAM, ATAF, and CUC) transcription factors containing a sequence complementary to miR164 with few mismatches. The miR164-NAC module is highly conserved across plant species. In Arabidopsis thaliana, miR164 constricts the expansion of the boundary domain by inducing the posttranscriptional downregulation of CUC1 and CUC2 at the mRNA level (Laufs et al. 2004). Moreover, miR164-guided cleavage of NAC1 mRNA provides a homeostatic mechanism that attenuates auxin signaling to regulate lateral root development (Guo et al. 2005). Overexpressing miR164-targeted NAC transcription factor genes have negative effects on drought resistance in rice at the reproduction stage (Fang et al. 2014), whereas overexpressing miR164b-resistant OsNAC2 improves plant architecture and increases grain yields (Jiang et al. 2018). NAC21/22 encodes a miR164-target NAC transcription factor that negatively regulates stripe rust resistance in wheat (Feng et al. 2014). Because the biological functions of miRNAs are closely related to the functions of their target genes, identifying additional targets of miR164 should help uncover complex miR164-mediated regulatory mechanisms.

Phytosulfokine (PSK) is a disulfated pentapeptide hormone that regulates cell proliferation and elongation in plants (Matsubayashi and Sakagami 2006). PSK is synthesized from approximately 80-amino-acid precursor polypeptides through posttranslational sulfation and proteolytic processing (Yang et al. 1999; Srivastava et al. 2008). Increasing evidence indicates that PSK plays fundamental roles in plant growth, development, and immunity in both dicot and monocot plants (Sauter 2015). PSK signaling promotes callus growth by inducing cell proliferation (Yang et al. 1999) and enhances root growth and hypocotyl elongation, primarily by promoting cell elongation (Kutschmar et al. 2009; Stuhrwohldt et al. 2011). Overexpression of GhPSK in cotton promotes fiber elongation, leading to the production of longer and finer cotton fibers (Han et al. 2014). PSK signaling also regulates plant immunity by attenuating the pathogen-associated molecular pattern (PAMP) response to bacterial pathogens (Igarashi et al. 2012; Mosher et al. 2013). Therefore, PSK signaling integrates growth and immune responses to meet the competing metabolic challenges that arise during plant development (Sauter 2015).

We previously identified TaPSK5, a wheat gene encoding a PSK precursor, as a potential target of miR164 based on miRNAome and degradome sequencing of developing wheat grains (Li et al. 2015). In the current study, we verified the miR164-mediated cleavage of TaPSK5 mRNA through rapid amplification of 5′ cDNA ends (5′ RACE). Overexpressing TaPSK5-D and miR164-resistant TaPSK5-D (rTaPSK5-D) promoted primary root growth and improved yield traits in rice, including grain number and grain weight. We developed molecular markers based on polymorphisms in TaPSK5-A and TaPSK5-B. Haplotype analysis indicated that favorable haplotypes of TaPSK5-A and TaPSK5-B were associated with higher grain yields in natural populations of Chinese wheat. Our findings provide new insights into the miR164-PSK regulatory network and highlight the potential use of TaPSK5 for crop improvement.

Materials and methods

Prediction and verification of miR164 target genes

Targets of miR164 were predicted using the psRNATarget program (http://plantgrn.noble.org/psRNATarget/) based on a Triticum aestivum cDNA library (EnsemblPlant, release 43). A four-mismatch cutoff was used to filter predicted miR164 targets, with a G–U bond counted as a 0.5 mismatch (Li et al. 2011). To map the cleavage sites in target mRNAs, RNA ligase-mediated 5′ RACE was performed using a FirstChoice RLM-RACE Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Gene-specific primers used for 5′ RACE are listed in Supplementary Table S1.

RNA extraction and quantitative reverse-transcription PCR (qRT-PCR)

Total RNA was extracted from plant materials using Tiangen RNAplant plus Reagent (Tiangen Biotech, Beijing, China). The cDNA was synthesized using a FastQuant RT Kit (Tiangen Biotech, Beijing, China). qRT-PCR analysis was performed on a LightCycler 96 real-time PCR system (Roche Applied Science, Germany) using SYBR Premix Ex Taq (Takara Bio, Beijing, China). The wheat Actin gene was used as the internal reference (Li et al. 2015). miRNA expression was quantified by qRT-PCR using a miRcute miRNA first-strand cDNA synthesis kit (Tiangen Biotech, Beijing, China). Relative gene expression levels (presented as fold-change values) were calculated using the comparative CT method (Livak and Schmittgen 2001). All primers are listed in Supplementary Table S1.

Vector construction and plant transformation

The coding sequences (CDSs) of TaPSK5-A and TaPSK5-D were amplified from Chinese Spring wheat cDNA and cloned into the binary vector pCAMBIA3301 under the control of the CaMV 35S promoter. The recombinant constructs were transformed into Arabidopsis thaliana (ecotype Col-0) by the floral-dip method using Agrobacterium tumefaciens GV3101. Positive transgenic plants were screened by spraying with Basta (0.1% solution). Arabidopsis plants were planted into small pot with soil and grown in a controlled growth chamber at 22 °C, with a 16 h light /8 h dark photoperiod, light intensity of 120 mmol m−2  s−1, and 70% relative humidity.

To construct the TaPSK5 overexpression vector for rice transformation, the CDS of TaPSK5-D was cloned into a modified pCAMBIA2300 vector under the control of the rice ACTIN1 promoter (Liu et al. 2005). To obtain the miR164-resistant TaPSK5-D (rTaPSK5-D), mutations were introduced into the miR164 binding site of the TaPSK5-D CDS by PCR-mediated mutagenesis. To construct the miR164-overexpression vector, a 620-bp stem-loop precursor sequence of miR164 was amplified from wheat genomic DNA and cloned into the pCAMBIA2300 vector in a similar manner. The resulting overexpression constructs were mobilized into Agrobacterium tumefaciens strain EHA105 and transformed into rice (Oryza sativa L. ssp. japonica) cv. Kitaake as described (Hiei et al. 1994). Positive T3 generations of transgenic lines were grown in the Chinese Academy of Agricultural Sciences (CAAS) experimental field (Beijing, China) for phenotypic assessment.

Phenotypic analysis

For Arabidopsis root length assay, seeds of wild-type (Col-0) and transgenic plants were sown on MS agar medium at 4 °C for 2.5 days and then incubated in a growth chamber at 22 °C for 3 days. The resulting seedlings were transferred to new MS plates and grown vertically for 5 days. For phenotypic analysis of rice roots, wild-type (WT) and transgenic plants were grown hydroponically at the seedling stage. In brief, three days after germination, the seedlings were transferred to plastic pots containing 1 L Hoagland nutrient solution and grown for 14 days at 25 °C in a controlled growth chamber under a 16 h/8 h light/dark photoperiod and 70% relative humidity. Hoagland solution was supplied every 3 days to provide nutrients to the seedlings. Root length was measured in at least 30 plants per line. For yield trait analysis, WT and transgenic rice plants were grown in the CAAS experimental field and observed at the maturity stage. Yield traits including panicle length, grain number, thousand-grain weight, and grain size were measured as described (Li et al. 2016).

Transient assay of miRNA–target interactions in Nicotiana benthamiana leaves

As described previously (Liu and Axtell 2015), a transient assay was performed to quantify miR164-mediated target repression in N. benthamiana leaves using a dual-luciferace reporter system with some modifications. Briefly, reporter and effector plasmids were both derived from a modified pCAMBIA1300 vector containing a CaMV 35S promoter (Chen et al. 2008). The CDSs encoding firefly luciferase (LUC) and Renilla luciferase (REN) were cloned into pCAMBIA1300 vector to construct 35S-LUC and 35S-REN vectors. Similarly, the miR164 precursor was cloned into pCAMBIA1300 to construct 35S-miR164 vector. Two unique restriction sites (BamHI and SalI) embedded in the 3′ UTR of LUC facilitated cloning of potential miR164 target sites. Oligos of target sites from TaNAC92 and TaPSK5 were inserted into the 3′ UTR of LUC to construct corresponding reporter plasmids. Four-week-old N. benthamiana leaves were infiltrated by Agrobacterium tumefaciens carrying 35S-miR164, 35-LUC-reporter, and 35S-REN (ratio 4:1:0.5). The relative LUC activities (LUC/REN) were measured at 72 h after infiltration using the Dual-Luciferase Reporter Assay System on a GloMax-Multi luminescence reader (Promega, Madison, WI, USA).

Transcriptome analysis

Total RNA was isolated from roots of 2-week-old rice seedlings from WT and rTaPSK5-D overexpression plants using Trizol reagent according to manufacturer’s instructions (Invitrogen). For each genotype, 1-g root sample was collected for RNA extraction and pooled for each of the two biological replicates. Library construction and sequencing were carried out commercially on the Illumina Hiseq Novaseq platform at Kaitai Mingjing Genetech Corporation (Beijing, China). After raw sequence processing, clean reads were then mapped to the Oryza sativa reference genome using the HISAT2 program. The expression level of each gene was normalized by fragments per kilobase of transcript per million fragments (FKPM) using StringTie (Pertea et al. 2015). Differential expression analyses between the samples were performed using DESeq2 (Love et al. 2014). Differentially expressed genes between sample libraries were identified based on the threshold with fold-change > 2 or < 0.5, P value < 0.05, and FDR < 0.05.

Haplotype analysis of TaPSK5 homoeologs

Thirty-six wheat accessions (24 modern cultivars and 12 landraces) were initially chosen to detect TaPSK5 polymorphisms (Ma et al. 2016). Genome-specific primers were designed to amplify the genomic regions of three TaPSK5 homoeologs (Table S1). The resulting DNA fragments were purified and sequenced on an ABI 3730XI DNA analyzer (Applied Biosystems). Sequence alignments were performed with DNAMAN (http://www.lynnon.com/), and sequence variations (SNPs and InDels) were identified using DNASTAR (http://www.dnastar.com/). Molecular markers were developed based on the InDels located in the TaPSK5-A and TaPSK5-B promoters. The PCR products were separated by polyacrylamide gel electrophoresis and detected by silver staining to distinguish fragment lengths.

Marker screening and haplotype association analysis were performed using 505 Chinese wheat accessions, including 157 landraces and 348 modern cultivars from core collections of Chinese wheat (Liu et al. 2020). Agronomic trait data for these wheat accessions were collected in three environments: Luoyang in Henan Province in 2002 and 2005, and Shunyi in Beijing in 2010. Phenotypic differences between haplotypes were determined by one-way analysis of variance (ANOVA) and a Tukey test at a significance level of P < 0.05.

Results

TaPSK5 is a less conserved target of miR164

Our previous degradome analysis identified TaPSK5 as a new target of miR164 in wheat (Li et al. 2015). To further explore potential miR164 targets in wheat, we performed a bioinformatics search on the psRNATarget web server based on Triticum aestivum cDNA sequences (EnsemblPlant, release 43). As shown in Supplementary Table S2, five NAC transcription factors were predicted to be miR164 targets, which is consistent with the finding that the miR164-NAC regulatory module is highly conserved. TaNAC92 is a representative NAC family target of miR164 whose homologs in monocots and dicots all contain highly conserved miR164-binding sites (Fig. 1a). TaPSK5 homoeologs were also predicted to be miR164 targets, with a slight increase in the number of mismatches in the miR164 complementary sites (3.0–4.0 mismatches). Unlike highly conserved NAC targets, TaPSK5 homologs in monocots and dicots are divergent in the miR164 binding regions (Fig. 1b). Phylogenetic analysis of PSK precursor polypeptides from Arabidopsis, rice, and wheat showed that potential miR164 targets including TaPSK5, OsPSK5, and OsPSK2 were classified in the same subclade (Fig. S1). Allowing for a four-mismatch cutoff, our results suggest that miR164-mediated regulation of TaPSK5 and its homologs may be restricted to only a few monocot plants including Triticeae, Oryza sativa, and Brachypodium. Therefore, TaPSK5 is a less conserved target of miR164 compared with NAC family targets.

Fig. 1
figure 1

miR164 targets TaNAC92 and TaPSK5 in wheat. a, b Sequence alignments of the miR164 binding sites of TaNAC92 homologs (a) and TaPSK5 homologs (b) in monocots and dicots. c, d Mapping of TaNAC92 (c) and TaPSK5 (d) mRNA cleavage sites by 5′ RACE. Arrows indicate cleavage sites, and numbers indicate the frequency of clones sequenced

Full size image

To verify the finding that the predicted NAC genes as well as TaPSK5 are direct targets of miR164 in wheat, we identified the cleavage sites in two genes (TaNAC92 and TaPSK5) by 5′ RACE analysis. The miR164-mediated cleavage products of TaNAC92 and TaPSK5 mRNAs were successfully detected (Fig. 1c, d). TaNAC92 mRNA had a very high frequency (10/10) of cleavage at the predicted position (10th nucleotide) of the miR164 binding site, whereas TaPSK5 mRNA had a low frequency (2/15) of canonical cleavage, suggesting that miR164-mediated cleavage of TaNAC92 mRNA is more efficient than that of TaPSK5 mRNA. To further verify miR164–target interactions in vivo, we constructed reporter plasmids containing miR164 target sites in the 3′ UTR of a constitutively expressed LUC gene (Fig. 2a). Transient assays by co-expressing miR164 and LUC reporters in N. benthamiana leaves confirmed that miR164 could mediate expression repression of TaNAC92 and TaPSK5-D, leading to approximately 48% and 30% decreases in relative LUC activities, respectively (Fig. 2b). In contrast, co-expression of miR164 did not cause significant expression differences of SbPSK5 and AtPSK6, probably due to more mismatches in the miR164 binding regions (Figs. 1b and 2b).

Fig. 2
figure 2

miR164-mediated expression repression of TaNAC92 and TaPSK5 in N. benthamiana leaves. a Schematic diagram of reporter and effector constructs in dual-luciferase transient expression system. ‘Term’ indicates RBS-S terminator. b Transient assay by co-expression of miR164 and LUC reporters fused with potential miR164 target sites in N. benthamiana leaves. Target sites derived from the miR164 binding regions of TaNAC92, TaPSK5, SbPSK5, and AtPSK6. Empty vector was used as negative control (–). The relative LUC activities were measured at 72 h after infiltration. Values presented as mean ± standard deviation (SD) of five independent replicates. **P < 0.01 (ANOVA)

Full size image

Temporal and spatial expression patterns of TaPSK5 homoeologs

To characterize TaPSK5 and explore its potential biological function, we analyzed the spatial and temporal expression patterns of TaPSK5 homoeologs in Chinese Spring wheat by performing qRT-PCR with genome-specific primers (Supplementary Table S1). As shown in Fig. 3, all TaPSK5 homoeologs were constitutively expressed in wheat tissues such as roots, leaves, and stems at the seedling stage or heading stage, but they were expressed at higher levels in developing young spikes. The TaPSK5 homoeologs were expressed at higher levels in younger spikes or grains during spike or grain development, which is consistent with their function in promoting cell division and elongation. Although the TaPSK5 homoeologs exhibited similar expression patterns, TaPSK5-D was expressed at a higher level than TaPSK5-A and TaPSK5-B, which is similar to the results in the wheat expression database (http://www.wheat-expression.com/).

Fig. 3
figure 3

Temporal and spatial expression patterns of TaPSK5 homoeologs in wheat. R root, L leaf, S stem, 1YS–7YS 1–7-cm young spike, SP spike at heading stage, 1D–20D grain at 1–20 days after pollination. Genome-specific primers were used for TaPSK5 homoeologs, and the wheat Actin gene was used as internal reference. Relative gene expression levels were calculated using the comparative CT method. Values presented as mean ± SD of three independent replicates

Full size image

Overexpression of TaPSK5-D and rTaPSK5-D promotes root growth in rice

To determine whether TaPSK5 is a functional PSK gene, we generated transgenic Arabidopsis plants expressing TaPSK5-A or TaPSK5-D under the control of the CaMV 35S promoter. The roots of all TaPSK5-A and TaPSK5-D transgenic lines were significantly longer than those of the Col-0 control (Supplementary Fig. S2), which is consistent with the phenotypes of plants overexpressing functional Arabidopsis PSK genes (Yu et al. 2016).

Next, we generated TaPSK5- and rTaPSK5-overexpressing rice plants expressing TaPSK5-D or miR164-resistant TaPSK5-D (rTaPSK5-D) under the control of the rice ACTIN1 promoter. For rTaPSK5-D, two point mutations were introduced into the miR164 binding site without altering the amino acid sequence (Fig. 4a). Transient assays in N. benthamiana leaves showed that the point mutations of rTaPSK5-D could perturb the miR164-mediated expression repression (Fig. S3). Consistently, TaPSK5 was expressed at significantly higher levels in rTaPSK5-D-OE versus TaPSK5-D-OE plants (Fig. 4b and Supplementary Fig. S3), indicating that the miR164-resistant form of TaPSK5-D could result in enhanced accumulation of TaPSK5 transcripts. Transgenic rice seedlings harboring both TaPSK5 and rTaPSK5 showed accelerated root growth, but the effects caused by overexpression of rTaPSK5 were stronger (Fig. 4c). The average primary root length of TaPSK5-OE plants was significantly greater than that of the WT control and lower than that of rTaPSK5-OE plants (Fig. 4d). Therefore, TaPSK5 encodes a positive regulator of root growth, and its regulatory effect is restricted by miR164.

Fig. 4
figure 4

Overexpression of TaPSK5-D and rTaPSK5-D promotes root growth in rice seedlings. a Schematic representation of the TaPSK5-D and rTaPSK5-D constructs. b Relative expression levels of OsPSK5 and TaPSK5 in 2-week-old wild-type (WT), TaPSK5-D-overexpression (OE), and rTaPSK5-D-overexpression (rOE) seedlings. Rice Actin gene was used as internal reference. Values presented as mean ± SD. **P < 0.01 (ANOVA). N.D., not detectable. c Comparison of root growth among WT, OE, and rOE transgenic rice seedlings. Bar = 5 cm. d Primary root length of WT, OE, and rOE transgenic rice seedlings. *P < 0.05 and **P < 0.01 (ANOVA)

Full size image

We also generated miR164-overexpressing rice plants to clarify the effects of this miRNA on root growth. Compared with the WT plants, all three independent miR164 overexpression lines showed a more than one-fold increase in miR164 abundance and a corresponding decrease in OsPSK5 transcript levels (Fig. S4), confirming the notion that both TaPSK5 and OsPSK5 are targets of miR164. The root lengths of miR164-overexpression plants were reduced by more than 25% compared with the WT (Fig. S4), in contrast to the phenotype observed for the TaPSK5-overexpression lines. However, miR164 overexpression also led to reduced shoot length (Fig. S4), whereas TaPSK5 overexpression did not lead to altered shoot growth (Fig. 4b). These results indicate that miR164 plays a widespread role in plant development, possibly by regulating the expression of multiple targets including NACs and PSK5.

Overexpression of TaPSK5-D and rTaPSK5-D improves yield traits in rice

The finding that TaPSK5 was expressed at high levels in developing young spikes suggests that it might also be involved in regulating yield traits. To test this hypothesis, we measured multiple yield-related traits in WT, TaPSK5-D-OE, and rTaPSK5-D-OE rice plants, including plant height, tiller number, panicle length, grain number per panicle (GN), thousand-grain weight (TGW), grain length, and grain width. Although no significant differences in plant height or tiller number were observed between the WT and transgenic plants (Figs. 5a and Supplementary Fig. S5), TaPSK5-D and rTaPSK5-D plants showed better yield traits than the WT, such as longer main panicles, higher GN, larger grain size, and higher TGW (Fig. 5b–g). Notably, the increased TGW in the transgenic plants mainly resulted from an increase in grain length rather than grain width (Fig. 5g and Supplementary Fig. S5). Although the difference was not significant, the rTaPSK5-D plants showed slight increases in yield traits compared with TaPSK5-D plants (Fig. 5b–g). These results confirm the notion that overexpressing TaPSK5-D and rTaPSK5-D improves yield traits in rice.

Fig. 5
figure 5

Overexpression of TaPSK5-D and rTaPSK5-D improves yield traits in transgenic rice. a–c Phenotypic comparison of whole plants (a), panicles (b), and grains (c) at adult stage among wild-type (WT), TaPSK5-D overexpression (OE), and rTaPSK5-D overexpression (rOE) plants. Bar = 5 cm in a and bar = 1 cm in (b). d–g Panicle length (d), grain number per panicle (e), thousand-grain weight (f), and grain length (g) in WT, OE, and rOE plants. Values presented as mean ± SD. *P < 0.05 and **P < 0.01 (ANOVA) indicate significant difference from WT

Full size image

Identification of differentially expressed genes between WT and rTaPSK5-D-OE by RNA-seq

To explore the possible functional mechanism of TaPSK5 in promoting root growth, we analyzed the transcriptome of 2-week-old seedling roots of WT and rTaPSK5-D-OE1 using RNA-SEq. Based on a log2 fold-change threshold of < −0.5 or > 0.5, with a P value and FDR < 0.05, a total of 348 genes were identified as differentially expressed genes between WT and rTaPSK5-D-OE1 (Fig. S6). Compared with the WT, 240 and 148 genes were determined to be up- and downregulated, respectively, in the rTaPSK5-D-OE1 line (Fig. S6). Interestingly, some hormone-related genes and transcription regulators were significantly upregulated in the rTaPSK5-D-OE1 line. These genes were involved in brassinosteroid (BR), auxin, and jasmonate (JA) signaling and transcription regulation network, such as OsBRI1 (BR), OsIAA8 and OsIAA14 (auxin), OsJAZ9 (JA), and several NAC and HOX family transcription factors (Fig. 6a). Moreover, multiple genes encoding peroxidases were differentially expressed between WT and rTaPSK5-D-OE1, and most of them were upregulated in the rTaPSK5-D-OE1 line (Fig. 6b), indicating that homeostasis of ROS might play a role in accelerating root growth in transgenic plants. We performed qRT-PCR analysis to further confirm the expression patterns of five representative genes (OsBRI1, OsIAA14, OsJAZ9, OsNAC12, and Os07g0676900) among WT, TaPSK5-D-OE, and rTaPSK5-D-OE plants. As shown in Fig. 6c, all five genes were upregulated in TaPSK5-D-OE and rTaPSK5-D-OE lines compared with the WT, consistent with our transcriptome data. Notably, only OsBRI1 and Os07g0676900 exhibited TaPSK5 dose-dependent effects, showing higher expression levels in rTaPSK5-D-OE than TaPSK5-D-OE lines (Fig. 6c).

Fig. 6
figure 6

Transcriptome and qPCR analysis of differentially expressed genes between WT and rTaPSK5-D-OE1 (rOE1) line. a Heatmap showing that representative genes involved in hormone signaling and transcription regulation were upregulated in the rOE1. Values of log2 fold-change (rOE1/WT) are indicated in different colors (red to green). b Heatmap showing peroxidase-encoding genes differentially expressed between WT and rOE1. c Validation of five representative upregulated genes by qRT-PCR. Rice Actin gene was used as internal reference. Values presented as mean ± SD

Full size image

Haplotype identification of TaPSK5 homoeologs in Chinese wheat accessions

To investigate genetic effects of TaPSK5 homoeologs and identify their favorable alleles in wheat, we sequenced the coding and promoter regions of TaPSK5 homoeologs (TaPSK5-A, -B, and -D) in 36 Chinese wheat accessions and detected polymorphic sites only in the promoter regions of TaPSK5-A and TaPSK5-B (Fig. 7a, b). No variations were observed in the coding regions of TaPSK5-A, -B, or -D. Among promoter regions, only one simple sequence repeat (SSR) was detected in the 2.1-kb region upstream of the ATG start codon of TaPSK5-A (Fig. 7a), and 17 SNPs and 2 InDels were present in the 3.0-kb region upstream of the ATG start codon of TaPSK5-B (Fig. 7b). These variations resulted in three TaPSK5-A haplotypes (TaPSK5-A-Hap1/2/3) and two TaPSK5-B haplotypes (TaPSK5-B-Hap1/2) among the common wheat accessions examined. We developed two molecular markers based on the SSR (–249 bp) in TaPSK5-A and the 13-bp InDel (–2760 bp) in TaPSK5-B to distinguish the three TaPSK5-A haplotypes and two TaPSK5-B haplotypes, respectively (Fig. 7a, b).

Fig. 7
figure 7

Haplotype identification in common Chinese wheat accessions and development of molecular markers for TaPSK5-A and TaPSK5-B. a An SSR (GT repeat) located in the TaPSK5-A promoter region (forming three haplotypes) was used to develop a molecular marker. b Two InDels and 17 SNPs located in the TaPSK5-B promoter region (forming two haplotypes) were used to develop a molecular marker based on the 13-bp InDel

Full size image

TaPSK5-A and TaPSK5-B haplotypes were associated with yield traits in natural populations of Chinese wheat

By using the newly developed molecular markers, we genotyped 505 Chinese wheat accessions, including 157 landraces and 348 modern cultivars, from core collections of Chinese wheat. We then performed association analysis between the TaPSK5-A/B haplotypes and multiple yield traits from plants collected in three environments (2002, 2005, and 2010). Because the frequency of TaPSK5-A-Hap3 was very low (less than 4%) in both the landraces and modern cultivars, we did not use this haplotype in the association analysis. Significant differences in GN were detected between TaPSK5-A-Hap1 and TaPSK5-A-Hap2 in both populations. The mean GN of TaPSK5-A-Hap1 was 5.1–5.9 higher in the landraces and 2.4–6.3 higher in the modern cultivars compared with TaPSK5-A-Hap2 in all three environments (Fig. 8a). Moreover, the TaPSK5-B haplotypes were significantly correlated with TGW in both populations. Compared with TaPSK5-B-Hap1, the mean TGW of TaPSK5-B-Hap2 was 5.1–5.9 g higher in the landraces and 2.4–6.3 g higher in the modern cultivars in all three environments (Fig. 8b). These results indicate that both TaPSK5-A-Hap1 and TaPSK5-B-Hap2 are favorable haplotypes associated with superior yield traits, such as higher GN and higher TGW.

Fig. 8
figure 8

Association of TaPSK5-A and TaPSK5-B haplotypes with yield traits in natural populations of Chinese wheat. a, b Association of TaPSK5-A haplotypes (A-Hap) with GN in landraces (a) and modern cultivars (b) in three environments (2002, 2005, and 2010). There were 42 and 108 A-Hap1 and A-Hap2 accessions among landraces (a) and 89 and 257 A-Hap1 and A-Hap2 accessions among modern cultivars (b), respectively. c, d Association of TaPSK5-B haplotypes (B-Hap) with TGW in both landraces (c) and modern cultivars (d) in three environments (2002, 2005. and 2010). There were 136 and 20 B-Hap1 and B-Hap2 accessions among landraces (c) and 131 and 209 B-Hap1 and B-Hap2 accessions among modern cultivars (d), respectively. Values presented as mean ± SE. *P < 0.05 and **P < 0.01 (ANOVA)

Full size image

TaPSK5-A-Hap1 and TaPSK5-B-Hap2 underwent positive selection during Chinese wheat breeding

Favorable haplotypes have always been subjected to positive selection during wheat breeding (Ma et al. 2016; Liu et al. 2020). We investigated the geographic distribution of TaPSK5-A/B haplotypes in both landraces and modern cultivars from China’s ten ecological zones. The favorable TaPSK5-A-Hap1 associated with high GN was subjected to weak positive selection mainly in zones I, II, and III, which are major wheat production zones in China (Fig. S7). By contrast, TaPSK5-B-Hap1, the high-TGW haplotype, was subjected to strong selection in almost all production zones, with an increase in frequency from 13% in landraces to 60% in modern cultivars (Fig. S7). These results indicate that both TaPSK5-A-Hap1 and TaPSK5-A-Hap2 were subjected to positive selection during wheat breeding, but the degree of selection pressure on these two favorable haplotypes varied in different production zones.

Discussion

In plants, most conserved miRNAs regulate highly conserved target genes involved in development and hormone signaling. Many of these target genes encode transcription factors and generally retained conserved miRNA-binding sites across plant evolution. Therefore, many miRNA-target regulatory modules comprising conserved miRNAs and their targets have been identified, such as miR156-SPLs, miR164-NACs, and miR396-GRFs (Willmann and Poethig 2007). Genome-wide analyses of miRNA targets have revealed that conserved miRNAs can also regulate targets that are not broadly conserved (Debernardi et al. 2012; Zhao et al. 2016). In Arabidopsis, bHLH74 is a verified target of miR396, but bHLH74 homologs with miR396 binding sites are only present in species within the sister families Cleomaceae and Brassicaceae (Debernardi et al. 2012). In the current study, by performing 5′ RACE, transient assay, and transgenic analysis, we determined that TaPSK5 is a less conserved target of miR164 (Figs. 1d, 2b, and 4b–d). Indeed, TaPSK5 homologs with miR164 binding sites are restricted to closely related monocot plants including Triticeae, Oryza sativa, and Brachypodium (Figs. 1b, 2b). Thus, our findings provide new evidence supporting the notion that conserved miRNAs have acquired new targets during evolution in various plant species. The identification of additional miRNA targets could help fully elucidate the specific functions of conserved miRNAs.

PSK signaling regulates cell proliferation and elongation in plants (Matsubayashi and Sakagami 2006). In the current study, both transgenic Arabidopsis and rice plants constitutively overexpressing TaPSK5 had significantly longer roots than the WT controls (Fig. 4c and Supplementary Fig. S2), which is consistent with the finding that PSK signaling plays a conserved role in promoting root growth (Kutschmar et al. 2009). More interestingly, overexpressing TaPSK5-D also improved yield traits in transgenic rice, including larger panicles and grain size, higher GN, and higher TGW (Fig. 5a–g). Overexpressing miR164-resistant TaPSK5-D in transgenic rice plants caused more obvious increases in root growth (Fig. 4a–d), indicating that the hormone PSK may have a dose-dependent effect on plants. We believe that the yield improvement obtained by overexpressing TaPSK5-D mainly resulted from the role of PSK in directly regulating reproductive organ growth based on its high expression levels in young spikes and grains (Fig. 3), which can be supported by the fact that the difference in yield traits between TaPSK5-OE and rTaPSK5-OE is not as significant as that in root traits. Moreover, overexpressing GhPSK in cotton promoted fiber elongation and enhanced fiber quality (Han et al. 2014). These findings indicate that PSK signaling is a versatile regulator of plant growth and development and that manipulating its expression has potential applications for crop improvement.

So far, plant PSK signaling through PSKR receptors has been well established, and some crucial regulators of PSKs have been identified (Matsubayashi et al. 2002; Heyman et al. 2013). However, PSK signaling pathway is far from being fully understood downstream of perception. Our transcriptome data identified hundreds of differentially expressed genes between WT and rTaPSK5-OE plants (Fig. S6), which contributed to elucidate PSK signaling in promoting root growth. Among them, a few hormone-related genes were significantly upregulated in the rTaPSK5-OE plants (Fig. 6a). For example, the brassinosteroid receptor OsBRI1, a key member in BR signaling pathway, demonstrated a TaPSK5 dose-dependent expression pattern, with higher expression level in rTaPSK5-OE plants than WT and TaPSK5-OE plants (Fig. 6c). Previous studies have shown that plant BRI1 has structural similarity with PSKR receptors and BR mediates PSK-dependent elongation growth in roots (Hartmann et al. 2013; Ladwig et al. 2015). Our study indicated that PSK accumulation would in turn induce the expression of BRI1 and enhance the BR signaling in root growth. In addition to BR, the plant hormone auxin (primarily indole-3-acetic acid, IAA) plays an essential role in root growth (Pacifici et al. 2015). Consistently, PSK promoted the expression of auxin-related genes including auxin efflux carrier OsPIN5B and auxin-responsive transcription factors OsIAA8, OsIAA14, and OsLBD16 (Fig. 6a). Transcription factors are also important regulators of root growth (Chen et al. 2018), and PSK-induced NAC and HOX family transcription factors have potential functions in accelerating root growth downstream of PSK signaling. Moreover, ROS act as important signal molecules in controlling root growth and development (Tsukagoshi 2016). We identified multiple differentially expressed peroxidase-encoding genes, and found most of them to be upregulated in the rTaPSK5-OE plants (Fig. 6b), indicating a molecular link between PSK and ROS. Together, our transcriptome data reveal a complex network of interactions between PSK and other plant hormones and signal molecules.

Yield improvement is a major objective of wheat breeding programs due to the enormous demand for wheat by an increasing worldwide population. Marker-assisted selection (MAS) is a highly efficient approach for genetic yield improvement in wheat (Ma et al. 2019). In the current study, we identified three TaPSK5-A haplotypes and two TaPSK5-B haplotypes among Chinese wheat accessions and developed molecular markers to distinguish between the distinct haplotypes of TaPSK5-A and TaPSK5-B (Fig. 7a, b). Our association analysis showed that the TaPSK5-A and TaPSK5-B haplotypes were significantly correlated with GN and TGW, respectively, in both Chinese landraces and modern cultivars (Fig. 8a–d). The different genetic effects of TaPSK5-A and TaPSK5-B haplotypes point to the possible subfunctionalization of homoeologous genes, which has been reported for gene sets such as TaCPK2 and TaGW8 (Geng et al. 2013; Ma et al. 2019). TaPSK5-A-Hap1 is the favorable TaPSK5-A haplotype associated with higher GN, and TaPSK5-B-Hap2 is the favorable TaPSK5-B haplotype associated with higher TGW (Fig. 8a–d). We propose that this functional divergence resulted in differential selection pressure on TaPSK5-A-Hap1 and TaPSK5-B-Hap2 during Chinese wheat breeding. TaPSK5-B-Hap2 was subjected to stronger positive selection than TaPSK5-A-Hap1 (Fig. S7), which is consistent with the finding that TGW was the prioritized target among the three yield components during a century of wheat breeding in China (Hou et al. 2014). As combinations of favorable alleles or haplotypes usually show additive effects, pyramiding favorable haplotypes of TaPSK5-A/B and other yield-related genes could be highly effective for improving wheat yields in the future.

In conclusion, we identified TaPSK5, a less conserved target of miR164, as a novel positive regulator of root growth and yield traits in wheat. Transgenic and haplotype analyses highlighted the potential of targeting TaPSK5 for genetic improvement in crops.