Introduction

Plant height is one of the most important agronomic traits of rice. IR8, a high-yielding rice variety with semi-dwarf gene sd1, bring about great promotion of rice production in southeast Asia, which was called the “green revolution” of rice (Hedden 2003). Therefore, semi-dwarf gene sd1 was known as the “green revolution gene”. So far, more than 80 genes related to dwarf mutants had been cloned in rice (http://www.gramene.org/rice_mutant/). However, most of cloned dwarf genes perform severe dwarf and other adverse phenotypes such as more tillers, small grain, brittle stems, which is difficult to apply in rice breeding and unconducive to expanding the genetic diversity (Hargrove and Cabanilla 1979; Piao et al. 2014).

Most of the dwarf mutants in rice have been identified to be defective in biosynthesis or signal transduction of plant hormones such as gibberellins, brassinosteroids, strigolactone (SL) and auxin (Qi et al. 2008; Sazuka et al. 2009; Song et al. 2009). SL is a kind of plant hormone which can inhibit the growth and development of plant branches (Gomez-Roldan et al. 2008; Umehara et al. 2008). At present, several genes related to high tillers and dwarf phenotype have been reported to be involved in the regulation of the SL pathway, including biosynthesis and signal transduction genes, such as HTD2, D10, D27, MIT3, HTD1 (Liu et al. 2009, 2018; Arite et al. 2007; Lin et al. 2009; Zou et al. 2005) and D14, D53, D3, THIS1 (Kameoka and Kyozuka 2015; Jiang et al. 2013; Zhao et al. 2014; Liu et al. 2013). Plant height and tillers are main factors of architecture (Wang et al. 2018). The uniformity of stem height and tillers in rice directly affects yield (Ma et al. 2009). Studies of IPA1 and OsSHI1 have revealed insight into regulation of balance between plant architecture and yield in rice (Wang et al. 2018; Duan et al. 2019). Therefore, discovering and utilizing new dwarf genes can further elucidate the molecular mechanism of plant height and tillers, which is of great significance to rice production.

In the present study, we identified a multi-tillering semi-dwarf rice line, sde, which is a near-isogenic line (NIL) of ZX5T derived from double dwarf local variety Te’ai (Ma et al. 2003). The phenotypic observation, genetic analysis, cytological observation and gene mapping of sde were conducted. The results indicated that the multi-tillering semi-dwarf line sde was probably a novel weak allelic of D3. Our results proved new understanding of regulatory network of plant height in rice.

Materials and methods

Plant materials and agronomic traits analysis

In our previous study, we selected a variety Xinte’ai, which possessing new semi-dwarf gene sd-e(t) from double dwarf Te’ai, a local variety of Zhejiang (Ma et al. 2003). To create near-isogenic line containing sd-e(t) gene, semi-dwarf plants in F2 population of Xinte’ai/Zhongxuan 5 (ZX5T) were selected to backcrossed with ZX5T, then with ZX5T as recurrent parent, Xinte’ai as donor line, a near-isogenic line with sd-e(t) gene, was constructed by consecutive backcrosses and selections. For each generation, multi-tillering semi-dwarf individuals were selected for further backcrossing until BC6F8. After four times self-cross, one stable multi-tillering semi-dwarf plant was used for research and denoted as NIL-sde (simply for sde).

The wild-type Nanjing 6 (NJ6) and ZX5T were preserved by our lab. All plants were grown in experimental fields of China National Rice Research Institute in Hangzhou, China (120° 12′ E, 30° 30′ N), or Lingshui, China (109° 57′ E, 18° 35′ N).

All plants were grown followed normal agricultural field management. At maturity stage, the agronomic traits of plant height, tilliering number, panicle length, grain width, grain length, grain number per panicle, and 1000-grain weight were investigated for sde and ZX5T, respetively. The statistical significant difference determined by the t-test.

Microscopic observations

Microscopic observations method followed by previous methods with minor modification (Wang et al. 2014). The internodes of ZX5T and sde at the mature stage were fixed in the 70 % FAA solution, dehydrated with graded concentrations of ethanol (70 %, 83 %, 95 % and 100 %) successively, transparented with different concentrations of xylene, embedded in paraffin, and sectioned by Leica RM2235 (5 µm thick). Then the prepared microtome sections were stained with Safranin O-Fast Green and investigated under the fluorescence microscope.

GA3 and SL treatments

GA3 and SL treatment was conducted as previous study with minor modification (Li et al. 2010; Umehara et al. 2008). 20 seeds of sde and ZX5T were sterilized in 75 % ethanol for 5 min, disinfected with 25 % NaClO for 30 minutes, then washed with sterile distilled water for three times, finally immersioned into the sterile water in dark circumstance at 28 °C for 3 days. The conformity germinating seeds were selected to grown in the Yoshida solution containing different concentrations of GA3 or GR24 at 30 °C, 16 h light/8 h dark cycle (Yoshida et al. 1976). Seedling height or tiller numbers was measured after 7 days of GA3 treatment or SL treatment.

Assay of α-amylase activity

The assays of α-amylase activities were following the previous methods (Piao et al. 2014; Yamaguchi et al. 1999). The embryoless half seeds were sterilized with 75 % alcohol for 5 min, disinfected with 20% sodium hypochlorite for 20 min, and then rinsed with sterile distilled water for 5 times. Then the embryoless half seeds were placed vertically on 2% agar plates containing 0.2% soluble potato starch, 10 mM sodium acetate and 2 mM CaCl2 at pH 5.3. In addition, 1 µM GA3 was added to the experimental group. After 4 days at 30°C in dark environment incubated, the plates were smoked with iodine vapor until the plate turned blue-purple.

Quantitative real‐time PCR analysis

Total RNA were isolated from leaf, sheath, stem and spike of sde and ZX5T at the mature Stage using AxyPrep total RNA preparation kit. cDNAs were synthesized by ReverTra Ace qPCR RT Master Mix with gDNA Remover (TOYOBO). Real-time PCR were performed with SYBR Green Realtime PCR Master Mix (TOYOBO) on 7900HT Fast Real-time PCR System. Real-time PCR was conducted in a 10 µL volume containing 0.4 µL of each primer, 1 µL of cDNA, 5 µL of SYBR Green Realtime PCR Master Mix (TOYOBO) and 7.7 µL sterile water. The rice OsActin gene was used as the internal control. The primers used for real-time PCR are as follows: 5′-GAGTTGCAACACCGGCTACA-3′ and 5′-AACTAAAGCAGCTTTCAATC-3′ for D3 and 5′-TGGACAGGTTATCACCATTGGT-3′ and 5′-CCGCAGCTTCCATTCCTATG-3′ for OsActin. Four biological replications were implemented and the 2−ΔΔCT method was used to calculate relative expression levels (Livak and Schmittgen 2001).

Gene mapping

To map SDE, a cross was made between sde and NJ6. Plants with multi-tillering and semi-dwarf phenotypes were selected from the F2 populations for mapping. Chi-square (χ2) test was conducted to analyze the hereditary rule. Bulked segregant analysis was used to seek markers linked to SDE gene. The semi-dwarf and wild type DNA pool were constructed with 20 multi-tillering and semi-dwarf plants and 20 wild type plants randomly selected from the F2 population respectively. Total DNA was extracted from young leaves with TPS method following Lei et al. (2014). A total of 622 SSR markers were used to preliminary genetic mapping. The PCR reaction was conducted according to Fu et al. (2016). Then the products were separated on 8 % non-denaturing polyacrylamide gels and visualized with silver staining following (Zhang et al. 2011). To fine map the SDE gene, 3 Indel and 4 dCAPS polymorphic markers were developed according to the available sequence difference between NJ6 and sde.

Candidate gene analysis

Candidate genes were predicted according to the available sequence annotation databases (http://rapdb.dna.affrc.go.jp/). New markers (Table S1) were designed based on the reference sequences of putative genes in the mapping region. The related sequences were amplified from wild-type ZX5T and sde for sequencing analysis.

Results

Phenotype analysis of the sde

The multi-tillering semi-dwarf line sde was the NIL of ZX5T and selected from double dwarf local variety Te’ai in our previous study (Ma et al. 2003). Compared to ZX5T, the sde was no difference at seedling stage, while showed significantly dwarfism and multi-tillering at mature stage, with an average plant height of 88.23 cm, 62 % of that in ZX5T (Fig. 1; Table 1). Additionally, panicle and internode of sde were significantly shortened in relative to ZX5T (Fig. 1). Each internode length of sde was reduced in almost equal proportion to that of the wild type ZX5T (Fig. 1). Therefore, it could be inferred that sde was a dwarf mutant of “dn” type, kind of with normal internode proportion feature. In addition to dwarfism, compared with the wild-type ZX5T, the sde also showed increased tiller number, and the decreased grain number per panicle (Table 1).

Fig. 1
figure 1

Morphological comparison between sde and ZX5T. a The plant architecture at mature stage (left: ZX5T; right: sde) (Bar = 10 cm); b Different internodal length at mature stage (left: ZX5T; right: sde) (Bar = 5 cm); c Panicles (left: ZX5T; right: sde) (Bar = 5 cm); d The comparisons of plant height between ZX5T and sde; e, f Length of the panicles and internodes of ZX5T and sde at the mature stage, P indicates the panicle, and I, II, III, IV, V indicate five internodes, respectively (n = 10). **Significantly difference at P = 0.01

Full size image
Table 1 Comparison of agronomic traits between sde and ZX5T
Full size table

Microscopic observation of sde

The internode length was proved to be connected with cell division and cell elongation in the interstitial meristem in rice (Kitano and Futsuhara 1981). To identify the factors in responsible for shorten internode in sde, paraffin sections of the first internode were observed at maturity stages in the wild type and sde. The results showed that in longitudinal sections of the stem, cell in wild type and sde were similar, both regular and oblong (Fig. 2). Fewer cells were observed in sde than wide type ZX5T in the longitudinal sections of the first internode, while there was no significantly change in cell length compared with the wild type (Fig. 2). Therefore, these results indicated that the decreased number of cells was the interpretation of the shortened internode of sde.

Fig. 2
figure 2

Longitunal section of the 1st internode in the wild type and sde. a, b Longitunal section of the first internode of the wild type and sde, (a ZX5T; b sde) Bar 50 µm; c Cell numbers of the first internode; d Cell length of the first internode. **Significantly difference at P = 0.01

Full size image

sde was insensitive to GA3

To clarify the effects of SDE mutation on GA3 sensitivity, both of seedlings height and α-amylase activity were detected in seedlings treated with different concentrations of exogenous GA3 (Fig. 3). Compared to ZX5T, sde showed significantly decreased plant heights and amylase activities within the investigated concentrations of exogenous (Fig. 3). However, no significant difference was detected in seedling height of sde for varied GA3 concentrations, whereas sde showed relatively weaker activities of α-amylase than ZX5T (Fig. 3). The results indicated that sde was less sensitive to exogenous GA3 than ZX5T (Fig. 3).

Fig. 3
figure 3

Resposes of ZX5T and sde to gibberellic acid (GA3) and GR24 treatment. a Plant height of ZX5T and sde with varying concentrations of GA3 trestment; Data are mean ± SD (n = 10); b α-amylase activity induced by gibberellic acid (GA3) in ZX5T and sde. c The tiller phenotype of WT (ZX5T) and sde after four weeks treatment with 1 µmol/L GR24; CK, Without GR24; Arrows indicate the tiller buds. Bars 5 cm. d Tiller numbers after 4 weeks treatment with GR24. Tillers (> 2 mm) were counted (values are mean ± SD, n = 8)

Full size image

sde was insensitive to GR24

Most high tillering and dwarf mutants are associated with defects in SLs biosynthesis or signal transduction (Liu et al. 2009, 2018; Arite et al. 2007; Lin et al. 2009; Zou et al. 2005). Previous studies have revealed that the tiller number of the SL-deficient mutant d10 or d27 was significantly decreased under GR24 treatment while d3 mutants show no significant change of phenotype (Arite et al. 2007; Lin et al. 2009; Zhao et al. 2014). Thus, to identify whether the sde was related to SLs, seedlings of wild-type ZX5T and sde were treated with GR24, a synthetic SL analogue. The results showed that the tiller number of sde was not significantly decreased under GR24 treatment (Fig. 3), suggesting that sde was insensitive to GR24.

Genetic analysis and mapping of the SDE gene

To reveal the genetic characteristic of the sde, one F2 mapping populations was developed from the cross of sde and NJ6. The plant heights of F1 generation performed the similar wild-type phenotype to that of NJ6. Furthermore, there were two segregated phenotypes showed in F2 population, with one type was in accordance with sde, and the other was similar to NJ6. In the F2 population of 324 plants, the separation ratio of phenotype of wild-type to dwarf was 250:74, corresponding to the separation ratio of 3:1 (χ20.05 = 0.70 < χ20.05 = 3.84), indicating that the traits of the multi-tillering semi-dwarf sde was controlled by a one pair of recessive genes.

A total of 622 SSR markers distributed evenly on 12 chromosomes were employed to map the SDE gene using the aforementioned F2 population. The SDE gene was initially located between RM469 and RM3805 on the short arm of chromosomes 6 by bulked sergeant analysis (BSA) with 74 F2 recessive individuals (Fig. S1). Based on the coarse mapping, 7 additional polymorphic markers were developed and used to narrow the interval with 358 more recessive individuals identified from F2 population (Table. S1). And finally, the multi-tillering semi-dwarf gene SDE was located in a 58Kb region between Indel marker DE2 and dCAPS marker ME2 (Fig. 4).

Fig. 4
figure 4

Molecular mapping of SDE gene in rice. a Fine mapping of the D3 gene. Location of the SDE locus was narrowed down to a 58 kb region between DE2 and ME2 on chromosome 6. The number below the corresponding markers indicates the numbers of recombinants between the markers and sde. b Structure of the D3 gene and its mutation sites in the three alleles. Nucleotide substitutions and inserts in the three mutants are indicated. sde has one nucleotide C2104 substitution in the exon. Black boxes indicate exons, white boxes indicate UTRs and lines indicate introns. c Predicted structures of three-dimensional model of D3 protein in WT, sde and other three mutations

Full size image

There were 9 annotated genes located in the target region (http://rapdb.dna.affrc.go.jp/) (Table 2). To identify the candidate gene of SDE, the entire sequence of located region were sequenced. Sequence comparison revealed that only one bp substitution (C-T) was found in the first exon of Os06g0154200 between wild type ZX5T and sde, which causing amino acid arginine (R) mutates into tryptophan (W) at the 702th amino acids (Fig. 4). Besides, not any other differences were observed in the remaining eight genes sequences between ZX5T and sde were observed. Interestingly, SDE shared the common locus of the previously reported DWARF3 (D3) gene, encoding an F-box protein with leucine-rich repeats and essential for SL signal transduction (Zhao et al. 2014). Dwarf and multi-tillering characters of the sde were similar to that of d3. These results indicate that the sde is probably a novel allelic to d3.

Table 2 Annotated genes and their putative functions in the candidate region
Full size table

Expression patterns of SDE

The real-time RT-PCR was employed to detect the expression patterns of the candidate gene Os06g0154200 (D3) in different tissues of ZX5T and sde. The results showed that D3 was expressed in all tissues (Fig. 5). Nonetheless, compared with ZX5T, the expression of D3 gene in the culm and panicle of the sde was significantly increased (Fig. 5).

Discussion

SL is a kind of plant hormone, which can inhibit the growth and development of plant branches (Gomez-Roldan et al. 2008; Umehara et al. 2008). At present, D3, D14, and D53 has been proved to play various roles in SL signaling pathway (Zhao et al. 2014; Kameoka and Kyozuka 2015; Jiang et al. 2013). D3 encodes an F-box leucine-rich repeat protein, which is assembled into complex SCFD3 with SCF (Zhao et al. 2014). SL signaling is positive regulated by D3, which can inhibit mesocotyl elongation through degrade the OsGSK2-phosphorylated CYC U2 (Sun et al. 2018). D14 encodes α/β-fold hydrolase protein, which inhibits rice branch elongation together with SCFD3 (Kameoka and Kyozuka 2015). D53 is the substrate of SCFD3 ubiquitination complex, forming complexes with D14 and D3 to inhibit the SL signaling pathway (Jiang et al. 2013). OsMADS57, together with OsTB1, target D14 to control tillers in SL signaling pathway (Guo et al. 2013). Here, the treatments of exogenous SL (GR24) indicated that sde is insensitive to GR24, indicating that sde may have a defect in SL pathway and SDE may control tiller development through SL signaling pathway.

In this study, we characterized a new allelic mutant of D3, sde, which showed typical multi-tillering and semi-dwarf phenotypes. This was consistent with three reported allelic D3 mutants, i.e. Id3, gsor300097 and ext.-M1B. The Id3 mutant contained a 448 bp insertion in the D3 gene, which included a hypothesized transposon at the 154th amino acid (Fig. 4), resulting in alteration of amino acid sequences and a premature stop mutation (Ishikawa et al. 2005). This mutation in Id3 caused suppression of tiller bud growth, thus more than 45 tillers were noticed in Id3 mutant (Ishikawa et al. 2005). Another high-tillering dwarf D3 mutant, gsor300097, had a single-base mutation from G to A at the 1583th position of D3, converted amino acid from lysine to a premature stop codon (Fig. 4) (Zhao et al. 2014). The mutation from A to CC at the 1000th position of D3 (Fig. 4), caused frameshift mutation and a premature stop codon, is responsible for the 42.21 cm height and nearly 121 tillers phenotype of ext.-M1B allelic mutant (Liang et al. 2017).

However, although exhibiting similar phenotypes, sde was a novel and never-reported mutant. One nucleotide substitution (C-T) at the 1583th position of D3 first exon was responsible for sde. This mutation only result in a substitution of amino acid (R-W) at 702th amino acids (Fig. 4). Compared to the dwarf mutant Id3, gsor300097 and ext.-M1B, the semi-dwarf sde could still remain higher height (88.23 cm) and fewer tillers (34.42) (Table 1). This may be attributed to the maintenance of full length of 720 amino acids in sde (Fig. 4), whereas only 528, 564, 369 amino acids existed in Id3, gsor300097 and ext.-M1B respectively. Obviously, the mutation in Id3, gsor300097 and ext.-M1B caused defect changes in D3 protein (Fig. 4), which would lead to severe phenotypes. Nonetheless, the mutation of SDE showed a relatively little impact on the D3 function, is a weakest mutation of D3. The results of qRT-PCR showed that D3 was expressed in various tissues in wild type (Fig. 5), which was similar with previous research (Zhao et al. 2014; Liang et al. 2017). The premature stop codon would hindered the D3 expression in Id3 mutant (Ishikawa et al. 2005), while the increased expression of D3 in sde probably due to the negative feedback regulation. The weak allelic mutant of sde probably explained the less-tillering and higher height performance in sde than Id3, gsor300097 and ext.-M1B. Moreover, appropriate height and tiller number in rice play an important role in yield composition (Ma et al. 2009; Wang et al. 2018; Duan et al. 2019). Hence, less-tillering and higher height may provide more possibility application for sde in rice breeding.

Fig. 5
figure 5

Expression pattern of the D3 gene in wild type ZX5T and sde

Full size image

Similar situation were also reported about multiple allele mutations in previous research. Three d11 mutants, d11-1, d11-4, and d11-2, have typical dwarf phenotype. The mutations in d11-1, d11-4, and d11-2 generated premature stop codon and caused truncates protein. While another d11 weak mutant d11-3 only has a single-base mutation from C to T at exon 4, which cause an amino acid substitution (Thr to Ile). Thus, d11-3 shows a relatively mild dwarfing phenotype compared with d11-1, d11-4, or d11-2 (Tanabe et al. 2005). cyp2-1 contained a nucleotide mutation that resulted in a premature stop codon and a short protein, while cyp2-2 had a nucleotide mutation that changed an amino acid from Gly (72) to Ala (72) in the OsCYP2 gene. In contrast to cyp2-1, cyp2-2 showed a weaker phenotype with the ability to inhibit lateral root development (Kang et al. 2013). Two eg1 alleles mutant (eg1-1 and eg1-2) exhibited a wide variety of spikelet developmental defects. The strong allelic mutant eg1-1 produces a premature stop codon and causes a short protein, resulting in severe effect on phenotype. While the weak allelic mutant eg1-2 only cause amino acid exchange in EG1 gene, and result in weak effect on phenotype (Li et al. 2009).

Generally speaking, the variation of strong allele may lead to early termination of protein translation or affect the important structural and functional domains, and the variation of weak allele may only leads the substitution of single amino acid. The identification of weak allelic mutants of important functional genes would provide new insight for their application in rice breeding. The sde in this study, as a new weak allelic of D3, have moderate plant height and tiller number and may play an important role in new dwarf varieties creation and application in the future.