The semi-dwarfing gene Rht-dp from dwarf polish wheat (Triticum polonicum L.) is the "Green Revolution” gene Rht-B1b

The wheat dwarfing gene increases lodging resistance, the grain number per spike and harvest index. Dwarf Polish wheat (Triticum polonicum L., 2n = 4x = 28, AABB, DPW), initially collected from Tulufan, Xinjiang, China, carries a semi-dwarfing gene Rht-dp on chromosome 4BS. However, Rht-dp and its dwarfing mechanism are unknown. Homologous cloning and mapping revealed that Rht-dp is the ‘Green Revolution’ gene Rht-B1b. A haplotype analysis in 59 tetraploid wheat accessions showed that Rht-B1b was only present in T. polonicum. Transcriptomic analysis of two pairs of near-isogenic lines (NILs) of DPW × Tall Polish wheat (Triticum polonicum L., 2n = 4x = 28, AABB, TPW) revealed 41 differentially expressed genes (DEGs) as potential dwarfism-related genes. Among them, 28 functionally annotated DEGs were classed into five sub-groups: hormone-related signalling transduction genes, transcription factor genes, cell wall structure-related genes, reactive oxygen-related genes, and nitrogen regulation-related genes. These results indicated that Rht-dp is Rht-B1b, which regulates pathways related to hormones, reactive oxygen species, and nitrogen assimilation to modify the cell wall structure, and then limits cell wall loosening and inhibits cell elongation, thereby causing dwarfism in DPW.


Background
Plant height is an important agronomic trait of crops. The discovery and utilization of semi-dwarfing genes in rice (Oryza sativa) and wheat (Triticum aestivum) triggered the "Green Revolution", as dwarfism not only improves lodging resistance [1], but also increases the grain number per spike and harvest index [2,3]. Increasing numbers of dwarf varieties of crops are being bred for production [4], and the dwarfing mechanisms in many crops are clearly revealed [5][6][7].
As a gibberellin (GA)-insensitive semi-dwarfing gene, Rht-dp was identified from dwarf Polish wheat (DPW, T. polonicum) originally collected from Tulufan, Xinjiang province, China [10,23]. Transcriptomic and proteomic analyses suggested that Rht-dp is probably involved in the phenylpropanoid pathway. It was found to reduce the contents of lignin, cellulose, and S-adenosylmethionine, and increase the contents of flavonoids, which ultimately limits cell expansion and causes dwarfism [24]. Although those results indicated the potential mechanism of Rht-dp, the candidate gene of Rht-dp remained unknown. Genetic analysis of F 2 population derived from the cross of DPW and tall Polish wheat (TPW) indicated that Rht-dp should be a recessive gene [10]. However, the separated threshold of plant height was significant larger than the plant height of DPW [10,23], which implied that the effect of Rht-dp on reducing plant height might be partially covered by one or more non-allelic loci. Further study mapped Rht-dp onto chromosome 4BS between the SSR markers Xgpw3017 and Xwmc511, and suggested that Rht-dp may be an alternative allele at the Rht-B1 locus [10]. However, due to the limited numbers of F 2 plants and molecular markers used in the analysis, a genomic alignment against the genome of Triticum aestivum 'Chinese Spring' (IWGSC RefSeq v1.0) (International Wheat Genome Sequencing Consortium, 2018) indicated that the region between Xgpw3017 and Xwmc511 did not include the Rht-B1 locus. Additionally, Rht-B1b and its alleles are semidominant genes [6,25,26]. Thus, we can't confirm whether Rht-dp is Rht-B1b or its allele, or a new gene.
To test this hypothesis and to understand the dwarfing mechanism of Rht-dp in DPW, we firstly cloned Rht-B1 to investigate sequence differences in Rht-B1 between DPW and TPW. Secondly, we developed and applied a specific molecular marker of Rht-B1 and SSR markers on 4BS to genetically confirm the candidate region using three recombinant inbred lines (RILs). Thirdly, two pairs of near-isogenic line (NIL) obtained from the F 7 population of DPW × TPW were conducted transcript analyses to reveal the molecular mechanism of Rht-dp; meanwhile, F 1 plants and a F 2 population derived from the cross of a pair of NIL were developed for further genetic analysis. Finally, we conducted a haplotype analysis of Rht-dp to reveal the natural distribution among 59 tetraploid wheat accessions.

Plant materials and growth conditions
The DPW and TPW lines were originally collected from Tulufan, Xinjiang province, China, by Prof. Chi Yen and Junliang Yang (Sichuan Agricultural University, China) in the 1980s. The F 1 population of DPW × TPW and the F 2 population (401 plants) derived from DPW × TPW were individually developed for trait investigation. Two RIL populations (F 7 including 330 lines and F 8 including 300 lines) derived from DPW × TPW, and a RIL population (F 6 including 194 lines) derived from DPW × Jianyangailanmai (AABB, 2n = 4x = 28, T. turgidum L., Ailanmai), were developed for gene mapping. Two pairs of NILs (D_60/T_58, and D_33/T_35, D and T represent dwarf and tall phenotype, respectively) derived from two heterozygous F 7 lines were selected for transcript analyses. Meanwhile, F 1 plants and a F 2 population (244 plants) derived from the cross of D_60 and T_58 were developed for trait investigation. The haplotype analysis was conducted using 59 tetraploid wheat accessions (Table S1)
The Rht-B1 Indel marker and 15 polymorphic SSR markers were first used for genetic mapping of Rht-dp in the F 7 RIL population. Then, Rht-B1Indel and its four flanking SSR markers (Xgpw2994.1, Xgpw3128.1, Xgpw3427.1, and Xgpw4800.1) were further used to confirm the candidate region in the F 8 RIL and F 6 RIL populations. The F 7 RIL population was hybridized on the wheat 55 K SNP array by CapitalBio Technology (Beijing, China) (unpublished data).
Linkage analysis was performed using the JoinMap software (version 4.0; Kyazma BV, Wageningen, Netherlands) with a logarithm of odds (LOD) threshold of 3.0. The Kosambi mapping function was used to convert the recombination frequencies into genetic distances (cM) [33].

Haplotype analysis of Rht-B1 in 59 tetraploid wheat accessions
Genomic DNA was extracted from each tetraploid wheat accession using a plant genomic DNA kit (TIANGEN BIOTECH, Beijing, China), and PCR amplification was performed as described in the section "Homologous cloning of Rht-B1". The amino acid sequence was deduced using ExPASy software (http://web.expasy.org/ translate/). All sequences were aligned using Vector NTI software (Invitrogen). A phylogenetic tree was constructed using the neighbour-joining algorithm in MEGA5 (https://www.megasoftware.net/).

Expression analysis of Rht-B1b
Tissues at the three growth stages (jointing, booting, and grain filling stages) were collected, including roots, basal stems, leaf sheaths, leaf blades, young leaves, lower leaf blades, first and second internodes, flag leafs, and spikes. The collected tissues were snap-frozen in liquid nitrogen and stored at − 80°C until RNA extraction. Total RNA was extracted using a Plant RNA Kit (Omega Bio-Tek, American). cDNA was synthesized using the M-MLV First Strand cDNA Synthesis kit (Invitrogen).

Transcript analysis of two pairs of NILs Sample collection
At the booting stage, the first internode was collected individually from two pairs of NILs, and then snap-frozen in liquid nitrogen and stored at − 80°C until RNA extraction.

RNA extraction, library preparation and sequencing
Total RNA was isolated as described above, and RNA degradation and contamination were monitored on 1% agarose gels. A NanoPhotometer® spectrophotometer (Implen GmbH, Munich, Germany) RNA purity was used to check RNA purity. The mRNA was purified from total RNA using poly-T oligo-attached magnetic beads and divided into short fragments using NEBNext First Strand Synthesis Reaction Buffer (5×) (New England Biolabs, Ipswich, MA, USA). The cDNA was synthesized using the fragments as templates and then purified and resolved with EB buffer for the end-repair step and addition of a single adenine (A) nucleotide. To select cDNA fragments 250~300 bp in length, the library fragments were purified with the AMPure XP system (Beckman Coulter, Beverly, CA, USA), and suitable fragments were chosen for a PCR amplification. The PCR products were purified (AMPure XP system) and the library quality was assessed using the Agilent Bioanalyzer 2100 system. The prepared libraries were sequenced on the Illumina Hiseq platform.

RNA-seq data analysis
Raw data (raw reads) of in fastq format were first processed using in-house perl scripts. In this step, clean data (clean reads) were obtained by removing reads containing adapters, reads containing poly-N, and low-quality reads from the raw data. All the downstream analyses were conducted using clean, high-quality data.
The Chinese Spring (IWGSC RefSeq v1.0) reference genome and gene model annotation files were downloaded from the genome website (https://urgi.versailles. inra.fr/download/iwgsc/IWGSC_RefSeq_Assemblies/v1.0). The D genome sequences were excluded from the reference before mapping the processed reads of the tetraploid lines (A and B genomes). An index of the Chinese Spring reference genome was built using Bowtie v2.2.3 and paired-end clean reads were aligned to the reference genome using TopHat v2.0.12. HTSeq v0.6.1 was used to count the number of reads mapped to each gene. The mean fragments per kilobase of transcript per million mapped reads (FPKM) value for each gene was calculated based on the length of the gene and the number of reads mapped to it [35].

Differential expression analysis
Read counts were adjusted by the edgeR program package through one scaling normalized factor. Analysis of differential gene expression between two pairs of NILs (D33/T35 and D60/T58) was performed using the DEG-Seq R package. The P values were adjusted using the Benjamini and Hochberg method. A corrected P-value of 0.005 and log2 (fold change) of 1 were set as the thresholds for significantly different gene expression.

Phenotype of plants harbouring Rht-dp
The average heights of DPW and TPW were 91.52 ± 2.97 cm and 189.88 ± 1.72 cm, respectively. No significant difference in plant heights between F 1 plants (179.12 ± 3.65 cm) and TPW was observed (Fig. S1). The plant heights of F 2 plants ranged from 65 to 185 cm. According to the frequency distribution of plant height, F 2 plants were separated into two groups of dwarf and tall phenotypes at 110 cm (Fig. 1a). The dwarf and tall phenotype groups included 107 and 294 plants, respectively, consistent with the expected Mendelian segregation ratio of 1:3 (X 2 = 0.606, p < 0.05). These results validate that Rht-dp should be a major recessive gene. However, the separated threshold of plant height with 110 cm was significantly larger than the plant height of DPW with 91.52 ± 2.97 cm, which implied that the effect of Rht-dp on reducing plant height might be partially covered by one or more non-allelic loci.
To fine-map Rht-dp, two RIL populations including 330 F 7 and 300 F 8 plants were constructed. The plant heights of F 7 and F 8 plants ranged from 65 to 165 cm (Fig. 1b) and from 65 to 170 cm (Fig. 1c), respectively. For the F 7 population, the average heights of dwarf and tall phenotypes were 84.07 ± 1.97 cm and 133.75 ± 2.01 cm, respectively. Compared with the tall phenotype, the lines harbouring Rht-dp showed a reduction in plant height of up to 37.14%. The reduced plant height was because of the shortened first internode (by 14.83%), second internode (by 7.15%), and basal internode (by 1.46%), but the length of the spike was not affected (Fig.  1d). These results indicate that Rht-dp reduces plant height mainly by restricting elongation of the first and second internodes at the booting stage.
To validate the candidate region of Rht-dp in a different genetic background, an F 6 RIL population including 194 lines derived from DPW × Ailanmai was constructed. The average height of Ailanmai was 100.98 ± 0.37 cm. Ailanmai has a recessive dwarfing gene Rht22, which has an additive effect with Rht-dp. The RIL population was grouped into dwarf and tall phenotypes with heights ranging from 20 to 60 cm and from 120 to 160 cm, respectively (Fig. 1e).
Characterization of Rht-dp in F 1 plants and F 2 population derived from the cross of a pair of NIL Since genetic analysis suggested that the effect of Rht-dp on reducing plant height was probably influenced by one or more non-allelic loci derived from TPW, a QTL analysis was performed on the F 7 RIL population using the wheat 55 K SNP array. Beside of a major-locus on 4BS (Rht-dp) derived from DPW caused dwarfism, a micro-locus on 5A derived from TPW heightened plant was detected (unpublished data). To further confirm the information of Rht-dp, we measured the plant height of F 1 plants and F 2 population derived from the cross of a pair of NIL (D_60 and T_58). The average heights of D_60 and T_58 were 93.52 ± 1.83 cm and 159.67 ± 2.72 cm, respectively; the average plant height of F 1 was 123.23 ± 2.55 cm. Compared with T_58, F 1 plants harbouring Rht-dp showed a reduction in plant height up to 22.82%. The plant heights of F 2 plants ranged from 65 to 155 cm. According to the frequency distribution of plant height, F 2 plants were separated into two groups of dwarf and tall phenotypes at 95 cm (Fig. 2). The dwarf and tall phenotype groups included 62 and 182 lines, respectively, consistent with the expected Mendelian segregation ratio of 1:3 (X 2 = 0.021, p < 0.05). Meanwhile, the separated threshold of plant height with 95 cm was similar to the plant height of D_60 with 93.52 ± 1.83 cm. These results indicate that the dwarfing gene of Rht-dp should be a single semidominant gene, and further imply that the candidate gene is Rht-B1b.

Differences in sequence of Rht-B1 between DPW and TPW
To test the implication that the candidate gene of Rhtdp is Rht-B1b or one of its alleles, the sequences of Rht-B1 were cloned from DPW and TPW. Sequence analysis showed that Rht-B1 of DPW is Rht-B1b, with a single nucleotide change from C to T at the nucleotide position 190 when compared with Rht-B1a (Fig. 3a) that results in a premature termination codon at amino acid position 64 (Fig. 3b). Although Rht-B1 of TPW did not have this single nucleotide change from C to T at nucleotide position 190, it had a three-nucleotide deletion at nucleotide position 386-388 when compared with Rht-B1a (Fig. 3a), resulting in a serine (S) deletion at amino acid position 129 (Fig. 3b). These results imply that the candidate gene of Rht-dp might be Rht-B1b. An Rht-B1 Indel marker was developed from the threenucleotide deletion of Rht-B1 in TPW for further analysis.
Mapping of Rht-dp To confirm that the candidate gene of Rht-dp is Rht-B1b, the Rht-B1Indel marker was first used to determine whether Rht-B1 was tightly linked with Rht-dp. Genetic mapping analyses confirmed that the Rht-B1Indel marker completely co-segregated with Rht-dp in three RIL populations and a F 2 population derived from a pair of NIL (Fig. 4).
To further confirm that Rht-B1b is located in the candidate region of Rht-dp, 190 pairs of SSR markers were exploited according to the genome reference of 4BS (Table S2). Fifteen pairs of SSR markers exhibited polymorphism between DPW and TPW, and were linked with Rht-dp in the F 7 RIL population. Of them, two SSR markers, Xgpw2994.1 and Xgpw3128.1, were tightly linked with Rht-dp with a genetic distance of 0.6 cM ( Fig. 4a; Table S3). Xgpw2994.1 and Xgpw3128.1 were further confirmed as tightly linked markers flanking Rht-dp in the F 2 population derived from NIL (Fig.  4b), and the F 6 ( Fig. 4c) and F 8 (Fig. 4d) RIL populations (Table S3).
Based on the gene annotation of wheat 4BS from 29.94 to 31.29 Mbp, flanked by Xgpw2994.1 and Xgpw3128.1, there were five potential genes: TraesCS4B01G042700 (encodes a teosinte branched 1 protein), TraesCS4B01G042800 (encodes an uncharacterized protein), TraesCS4B01G042900 (a RING finger protein), TraesCS4B01G043000 (EamA-like transporter family), and TraesCS4B01G043100 (Rht-B1 encodes a DELLA protein) (Fig. 4e). Apart from Rht-B1, sequence difference of other four genes (primers listed in Table S4) between DPW and TPW was not found. These results indicate that the candidate gene of Rht-dp should be Rht-B1b.

Expression patterns of Rht-B1b in DPW
To confirm that Rht-B1b reduces plant height via its effects on elongation of the first and second internodes at the booting stage, the transcriptional patterns of Rht-B1b were investigated in different DPW tissues at the jointing, booting, and grain-filling stages. Rht-B1b was mainly expressed in the first and second internodes at the booting stage, and at dramatically higher levels in those tissues than in other tissues at the jointing, booting, and grain-filling stages (Fig. 5).

Dwarfism-related DEGs induced by DELLA mutant Rht-B1b
To understand the molecular networks of Rht-B1b, the DEGs induced by the DELLA mutation Rht-B1b in the first internode of two pairs of NILs were investigated. A total of 41 DEGs was obtained, 30 of which were successfully functionally annotated (Table S5). Twentyeight DEGs were further classed into five sub-groups; hormone-related signalling transduction genes, transcription factor genes, cell wall structure-related genes, reactive oxygen-related genes, and nitrogen regulationrelated genes (Table 1). Among the hormone-related signal transduction genes, two brassinolide (BR) signalrelated genes serine carboxypeptidase II-3 (SCP) and cytochrome P450 710A1 (CYP450) were down-regulated; and genes encoding salicylic acid (SA)-binding protein 2 and ARP were up-regulated in the dwarf phenotype. The only down-regulated transcription factor gene was MybAS2. Fifteen DEGs were grouped into cell wall structure-related genes (seven pectin-related genes and eight xylan acetylation-related genes). In the dwarf phenotype, five pectin-related genes [encoding a pectate lyase 15 (PEL15), three subtilisin-like protease (SBT1.7), and an alpha-galactosidase (α-Gal)] involved in pectin modification were down-regulated; while all eight xylan acetylation-related genes, including three GDSL esterase/ lipase genes, two ESKIMO genes, IRX15-L, ALTERED XYLOGLUCAN 4-like (AXY-L), and an uncharacterized acetyltransferase gene were up-regulated. For the reactive oxygen-related genes, plant cysteine oxidase 2 (PCO2) and ASCO were down-regulated; and genes encoding germin-like protein 5-1 (GLP) and blue copper protein (BCP) were up-regulated in the dwarf phenotype. For nitrogen assimilation-related genes, two phosphoenolpyruvate carboxylase kinase 2 (PPCK2) genes and early nodulin (ENOD) were down-regulated; and asparagine synthetase (APS) was up-regulated in the Fig. 4 Mapping of Rht-dp. a: mapping of Rht-dp in the DPW × TPW RILs F 7 ; b: mapping of Rht-dp in the F 2 population from D_60 × T_58; c: mapping of Rht-dp in DPW × Ailanmai RILs F 6 ; d: mapping of Rht-dp in DPW × TPW RILs F 8 ; e: candidate genes between SSR markers Xgpw2994.1 and Xgpw3128.1 dwarf phenotype. We verified the expression of ARP and ASCO in the first and second internodes at the booting stage (Fig. S2).

Discussion
The GA-insensitive dwarfing gene Rht-B1b is the predominant source of the semi-dwarf growth habit of wheat plants grown in parts of Northern Europe [36], the Mid and Lower Yangtze Valley Autumn-sown Spring Wheat Region in China [37], and the Great Plains Hard Winter Wheat Region in the USA [38]. Because Rht-B1b significantly decreases plant height to reduce plant lodging and increase wheat yield [37,39], it has been introduced into tetraploid wheat T. durum for dwarf breeding [28]. However, it is well known that the progenitor of T. polonicum is not Norin 10 or T. durum. Additionally, DPW was originally collected from Tulufan, Xingjiang, China [23]. Thus, the dwarfing gene Rhtdp of T. polonicum cannot be derived from Norin 10 or T. durum. However, our results show that the candidate gene Rht-dp of DPW is Rht-B1b. This conclusion is supported by the following evidences: (1) Rht-dp is a single semi-dominant dwarfing gene, as is Rht-B1b [6].
(2) Rhtdp and Rht-B1b reduce plant height mainly via reducing the length of the first and second internodes (Fig. 1d), and their effects on reducing plant height are similar with 22% [18,39]. (3) The sequence of Rht-B1 of DPW is the same as that of Rht-B1b (Fig. 3). (4) Mapping work revealed that the candidate region of Rht-dp was between SSR markers Xgpw2994.1 and Xgpw3128.1 (Fig.  4b-d). This region contains five potential genes including Rht-B1 (Fig. 4e); except of Rht-B1, other four genes have no sequence difference between DPW and TPW. (5) The Rht-B1 Indel marker developed based on the sequence difference of Rht-B1 between DPW and TPW is completely co-segregated with Rht-dp in a F 2 population derived from NIL and three RIL populations (Fig. 4). In the haplotype analysis, Rht-B1b was only obtained from T. polonicum (Fig. 6a), implying that it might originate from this species, or might be introduced into T. polonicum from other unknown species but not T. aethiopicum and T. durum.
In wheat, Rht-B1b encodes a DELLA mutant protein resembling the SLRL1 protein. Its accumulation represses GA-regulated growth and developmental responses and causes the typical semi-dwarf phenotype [6,40]. DELLA not only regulates the expression of downstream genes but also interacts with DNA-binding transcription factors. Our transcript analysis identified 28 DEGs regulated by the DELLA mutant Rht-B1b involved in the processes of nitrogen assimilation, oxidationreduction, modification of the cell wall components and structures, and hormone-related signal transduction (Table 1). However, this list of DEGs only slightly overlaps with those identified in previous studies, suggesting that the effects of DELLA on transcription depend on the species, organ, and developmental context [41][42][43][44].
Since Rht-B1b is mainly expressed in the first and second internodes (Fig. 5) to dramatically reduce their lengths at the booting stage in DPW (Fig. 1d), we explored the molecular network of Rht-dp by conducting a transcript analysis of the first and second internodes at the booting stage. The control of plant growth and development by DELLA is dependent on GA-regulated growth and developmental responses [44][45][46]. However, we did not find genes involved in GA metabolism among the DEGs in this study. Instead, the DEGs identified in this study included auxin-, SA-and BR-related genes (Table 1). These results suggested that GA interacts with these hormones [46]. DELLA can directly trigger the expression of auxin-and BR-related genes to affect plant growth [47,48]. For example, the expressions of SCP and CYP450 (both grouped into BR-related genes) were dramatically down-regulated by the DELLA mutation Rht-B1b to potentially cause dwarfism in DPW (Table 1), because the expression of SCP positively affects plant growth [49]. Auxin represses the expression of ARP genes [50,51]. In a previous study, overexpression of an ARP of Brassica rapa caused a reduction in vegetative growth [50]. Auxin also modulates the expression of ASCO, which encodes a crucial enzyme that produces oxidative molecules, including H 2 O 2 [52]. Overexpression of an ASCO in cotton enhanced the accumulation of H 2 O 2 and promoted cell elongation, whereas suppression of an ASCO in tobacco and Arabidopsis inhibited stem cell growth [53].
Our results show that the DELLA mutation Rht-B1b resulted in dramatically up-regulated ARP and downregulated ASCO in DPW (Table 1). Auxin-induced growth inhibition is accompanied by decreased levels

TraesCS6A01G315800
Blue copper protein 13 12 Nitrogen regulation-related genes

TraesCS3B01G385400
Asparagine synthetase 11 12 of reactive oxygen species [54]. Thus, the accumulation of the DELLA mutant protein regulated via auxin-mediated signal transduction may reduce the contents of reactive oxygen species such as H 2 O 2 [41], thereby limiting cell expansion to cause dwarfism in DPW.
In rice, over-expression of an early nodulin gene resulted in improved nitrogen-use efficiency and increased nitrogen assimilation [55]. In C 3 plants, nitrogen assimilation is positively correlated with phosphoenolpyruvate carboxylase (PEPC) phosphorylation [56,57], which is catalysed by phosphoenolpyruvate carboxylase kinase (PPCK). The extent of phosphorylation is largely determined by PPCK activity, which is controlled by the level of PPCK transcripts [56,58,59]. A reduction in PEPC activity leads to serious stunting of growth [60]. Our results showed that the DELLA mutation Rht-B1b led to significant down-regulation of early nodulin and two PPCKs in DPW (Table 1). Thus, decreased nitrogen assimilation and PPCK activity may decrease the activity of PEPC [43,59] to cause dwarfism in DPW.
The hemicellulose xylan and pectins are two abundant polysaccharides in the plant cell wall [61]. Their modifications, such as methylesterification and acetylation, have been proposed to influence cell wall architecture and function, causing various plant growth phenotypes [61][62][63][64]. Our results showed that the DELLA mutation Rht-B1b led to significant down-regulation of the expression of several pectin-related genes, including PEL, three SBTs, and α-Gal (Table 1). Decreases in the transcript levels of these genes may lead to the repression of pectin degradation and the accumulation of deesterified pectin [63], enhanced pectin methylesterase activity to stiffen the cell wall [65], and reduced adherence of pectin to the cell wall [66]. Thus, the DELLA mutation Rht-B1b may result in modifications of pectin that limit cell wall loosening and inhibit cell elongation, thereby causing dwarfism in DPW.
Many studies have reported that either excess or inadequate acetylation of xylan disrupts the cell wall structure, thereby causing dwarfism in plants [67,68]. Our results show that the DELLA mutation Rht-B1b upregulated eight xylan acetylation-related genes, including three GDSL esterase/lipase genes, two ESKIMO genes, IRX15-L, AXY-L, and an uncharacterized acetyltransferase gene (Table 1). ESKIMO and AXY-L are xylan acetyltransferases, and IRX-L is involved in synthesis of the xylan backbone [61,[67][68][69][70]. A specific interaction between acetyltransferases and xylan backbone biosynthetic enzymes may repress acetylation of adjacent residues [68,70]. Therefore, even though the transcript levels of ESKIMO, IRX15-L, AXY-L and IRX-L were upregulated (Table 1), the acetylation of xylan might be decreased. GDSL esterase/lipase is a xylan deacetylation enzyme [64]. The DELLA mutation Rht-B1b resulted in up-regulated expression of GDSL esterase/lipase, leading to enhance xylan deacetylation. Therefore, the DELLA mutation Rht-B1b may reduce acetylation of xylan to limit cell wall loosening and inhibit cell elongation, causing dwarfism in DPW. A model summarizing how the DELLA mutation Rhtdp causes dwarfism in DPW is proposed (Fig. 7). Whether the DELLA mutation Rht-B1b regulates the pathway of hormones, reactive oxygen species, and nitrogen assimilation, it ultimately affects the cell wall structure to limit cell wall loosening and inhibit cell elongation, thereby causing dwarfism in DPW.

Conclusion
In summary, our results indicated that the semidwarfing gene Rht-dp is the"Green Revolution" gene Rht-B1b. It regulates pathways related to hormones, reactive oxygen species, and nitrogen assimilation to modify the cell wall structure, and then limits cell wall loosening and inhibits cell elongation, thereby causing dwarfism in DPW.