An autoactive NB-LRR gene causes Rht13 dwarfism in wheat

Significance Conventional dwarfing genes increased wheat yields by disrupting plant hormone (gibberellin) signaling. Alternative wheat-dwarfing genes, suitable for use in additional environmental conditions, have been shown to encode components of gibberellin metabolism. Here, we found that the alternative dwarfing gene Rht13 encodes an autoactive NB-LRR gene rather than a component of gibberellin signaling or metabolism. The autoactive Rht13 allele (Rht-B13b) causes up-regulation of pathogenesis-related genes and affects cell wall properties. Rht-B13b reduces height to a comparable degree as conventional dwarfing genes and offers an additional benefit of increased stem strength. This discovery reveals an unexpected class of reduced height gene in wheat and opens up opportunities to use autoactive NB-LRR genes to reduce height in a range of crop species.

Semidwarfing genes have greatly increased wheat yields globally, yet the widely used gibberellin (GA)-insensitive genes Rht-B1b and Rht-D1b have disadvantages for seedling emergence. Use of the GA-sensitive semidwarfing gene Rht13 avoids this pleiotropic effect. Here, we show that Rht13 encodes a nucleotide-binding site/leucine-rich repeat (NB-LRR) gene. A point mutation in the semidwarf Rht-B13b allele autoactivates the NB-LRR gene and causes a height reduction comparable with Rht-B1b and Rht-D1b in diverse genetic backgrounds. The autoactive Rht-B13b allele leads to transcriptional up-regulation of pathogenesis-related genes including class III peroxidases associated with cell wall remodeling. Rht13 represents a new class of reduced height (Rht) gene, unlike other Rht genes, which encode components of the GA signaling or metabolic pathways. This discovery opens avenues to use autoactive NB-LRR genes as semidwarfing genes in a range of crop species, and to apply Rht13 in wheat breeding programs using a perfect genetic marker.
semidwarfing gene | reduced-height (Rht) gene | autoactive NB-LRR | Triticum aestivum L. (wheat) Dwarfing or reduced height genes have been associated with large increases in the yield of cereals since they were introduced during the Green Revolution (1). Most current wheat cultivars carry Rht-B1b or Rht-D1b which encode negative regulators of gibberellin (GA) signaling (2), resulting in GA insensitivity and reduced height. These GA-insensitive alleles confer benefits to yield by optimizing resource partitioning to the grain and reduced lodging (3). However, they have pleiotropic effects on growth including reductions in coleoptile length and seedling leaf area (4) and impact resistance to diseases such as fusarium head blight (5). The use of alternative dwarfing genes that do not disrupt GA signaling, and which can reduce final plant height without adverse effects on seedling growth, will be particularly relevant in water-limited environments (6).
Several alternative dwarfing loci have been discovered (7) which are GA-sensitive and could therefore overcome the limitations of Rht-B1b and Rht-D1b on early growth. Recently, the causal genes for some of these alternative dwarfing loci have been identified, revealing their functions in the GA metabolic pathway. The first of these to be identified was Rht18, which is on chromosome 6A and causes an increased expression of a GA 2-oxidase gene (GA2oxA9) resulting in the removal of GA 12 precursors from the GA biosynthesis pathway, a reduction of bioactive GA 1 and reduced plant height (8). Map position, allelism tests, and increased expression of the same GA 2-oxidase gene in Rht14 lines suggested that Rht14 and Rht18 are allelic (8)(9)(10). Increased expression of related GA 2-oxidase genes was also found to be responsible for other alternative dwarfing alleles such as Rht12 (GA2oxA13 on chromosome 5A) (11,12) and Rht24 (GA2oxA9 on chromosome 6A, not allelic with Rht18) (13). These alternative dwarfing genes appear to operate through a shared mechanism, i.e., reduction of the flux through the GA biosynthetic pathway and subsequently lower GA content. In addition to GA 2-oxidase genes on chromosomes 5A and 6A, other GA 2-oxidase genes were identified in the wheat genome (14), suggesting that other dwarfing genes at different positions may also cause increased expression of other members of the GA 2-oxidase family.
Rht13 is another promising alternative dwarfing gene that reduces final plant height without affecting seedling growth (15,16). The dwarfing allele Rht-B13b produced a strong height reduction between 17% and 34% compared with Rht-B13a, which is comparable with reductions typical of Rht-B1b and Rht-D1b, depending on the genetic background and growing conditions (16)(17)(18)(19)(20). Genetic mapping located Rht13 on the long arm of chromosome 7B (21), but the underlying gene has not yet been identified. Here, we describe the causal gene that encodes an autoactive allele of a nucleotide-binding site/ leucine-rich repeat (NB-LRR) gene at the Rht13 locus on chromosome 7BL. Autoactivation of Rht13 leads to the up-regulation of pathogenesis-related (PR) genes, including class III peroxidases, which may catalyze the cross-linking of cell wall compounds to limit cell elongation and hence reduce height.

Results
Characterization of the Rht13 Phenotype in Magnif. The Rht13 semidwarfing gene was originally identified as an induced mutant in the Magnif background (22). We carried out a detailed characterization and found that Rht13 caused a 30 to 35% height reduction in both greenhouse and field conditions (birdcage) (Fig. 1). A comparison of internode lengths showed that most of the height reduction occurred in the peduncle, and this effect was confirmed in field-grown plants that were measured for height from early stem elongation to maturity ( Fig. 1 C and D). Height differences were apparent after Zadoks growth stage 50, with reduced peduncle length accounting for most of the effect.
Fine Genetic Mapping of Rht13 to a Region on Chromosome 7B.
Previously, Rht13 was mapped to the long arm of chromosome 7B and genetically linked to simple sequence repeats (SSR) marker gwm577 (21). An F 2 population from a cross between parental lines ML45-S carrying Rht13 and tall line ML80-T was developed for fine mapping. Approximately 2,400 F 2 gametes were screened with SSR markers gwm577 and wmc276 that were previously shown to flank the locus. The screen identified 21 recombinants that corresponded to less than 1 cM of genetic distance between flanking markers ( Fig. 2A and SI Appendix, Table S2). Additional DNA markers were added to the genetic interval after parental lines were screened with the 9K and 90K wheat single nucleotide polymorphism (SNP) arrays (23,24). In addition, the project was given early access in 2013 to the emerging physical map of chromosome 7B, which was part of the international initiative to generate maps of individual Chinese Spring chromosomes led by the IWGSC and Norwegian University of Life Sciences. Several BAC clones were assigned to the region, and markers that were developed from BAC sequences were added to the interval (SI Appendix, Table S2). In total, 33 DNA markers were added to the genetic interval. BAC sequence-derived markers 7J15.144I10_2_2 and 127M17.134P08_3 flanked the Rht13 locus on the proximal and the distal sides, respectively, and defined a genetic interval of approx. 0.1 cM (Fig. 2).

Next-Generation Sequencing Approaches Revealed a Single Amino Acid Change Between Expressed Genes in the Region on
Chromosome 7B. The Rht13 region defined by flanking markers 7J15.144I10_2_2 and 127M17.134P08_3 corresponded to a 1.93-Mb interval on chromosome 7B in Chinese Spring RefSeqv1.0. To identify candidate SNPs in the interval, we generated an additional population from a Magnif x Magnif M cross and selected four short and two tall F 3 :F 4 lines that were homozygous at Rht13. For each of these lines, we isolated chromosome 7B by flow sorting and then sequenced the chromosome using Illumina short-reads. We attempted to identify SNPs within the mapping interval by mapping this chrom-seq data to the RefSeqv1.0 genome sequence (25), but we found that over half of the 1.93-Mb interval had few reads mapping (1.07/1.93 Mb), which suggested haplotype divergence between Chinese Spring and Magnif. We then examined the alignment of chromosome 7B between Chinese Spring and several cultivars whose genome sequences were available from the 10+ Wheat Genomes Project (26). We found that CDC Stanley had significant haplotype divergence from Chinese Spring in the Rht13 interval on chromosome 7B (SI Appendix, Fig. S3); therefore, we tested whether CDC Stanley would be a more appropriate reference sequence. Using CDC Stanley as the reference, the flanking markers spanned 1.86 Mb on chromosome 7B (Fig. 2B). Within this interval, a 0.49-Mb region had more SNPs between all samples (four short and two tall F 3 :F 4 lines derived from Magnif x Magnif M cross) and the reference sequence, suggesting some divergence between CDC Stanley and Magnif.
We identified 12 SNPs and 1 INDEL between the tall and short fixed lines in the mapping interval (Fig. 2C). To identify potential causal genes for Rht13, we carried out RNA-seq on developing peduncle tissues from four fixed short and four fixed tall F 3 :F 4 lines from the same Magnif x Magnif M population that was used for chrom-seq. We found that one transcript within the interval from our de novo annotation was more highly expressed in Magnif M than Magnif samples (2.5-fold change, padj < 0.001; indicated by * in Fig. 2D). However, this transcript did not translate to a protein longer than 76 amino acids in any frame, suggesting that pseudogenization might have occurred. Since there were no obvious changes in expression levels of genes within the interval, except the putative pseudogene, we examined whether the SNPs detected by chrom-seq were contained within any of the de novo assembled transcripts. We found that only 1 SNP (G to A at chr7B:714,391,008) was located within a transcript (Fig. 2E), and this SNP was predicted to cause The Amino Acid Change S240F Reduces Plant Height. The expressed transcript with an amino acid change was predicted to encode an NB-LRR protein (Fig. 2F). The mutation was predicted to cause an amino acid substitution of the serine at position 240 to phenylalanine (S240F) in the RNBS-A motif (27). To test whether this amino acid change caused the reduced height phenotype observed in Magnif M, we searched the Cadenza TILLING population for mutations within closely related genes (28). Line Cadenza0453 was identified as carrying a gene that was 100% identical at the nucleotide level to the mutant NB-LRR gene at the Rht13 locus, resulting in the same amino acid change (S240F) as found in Magnif M. Examination of mega base-scale haplotypes (29) did not indicate a conserved haplotype across this region between Cadenza and CDC Stanley, instead only a small region encompassing Rht13 (10,377 bp) was 100% identical between these cultivars before Ns in the contig interrupted the alignment at both ends. The KASP marker developed for the mutation segregated within progeny derived from Cadenza0453.
Homozygous mutant plants (Rht-B13b) were on average 16.7 cm shorter than homozygous wild-type plants (Rht-B13a) at maturity in the Cadenza0453 background ( Fig. 3 A and B; P < 0.001, Student's t test). This difference in height was reflected in shorter peduncle and internode lengths, except for the first internode (SI Appendix, Table S5).
To confirm that the amino acid change caused the reduction in height, we transformed the mutant allele from Magnif M (Rht-B13b) into Fielder (Fig. 3 C-E). Fielder did not contain an endogenous copy of Rht13 (best BLAST hit < 86% identical to Fielder genome assembly (30)). We found that the expression of the transgene caused a strong reduction in height, compared with null segregants (Fig. 3 C-E and SI Appendix, Table S6), although there was variation in the degree of dwarfism, which did not relate to the copy number or expression levels (SI Appendix, Fig. S5 and Table S6).

Characterization of the Rht13 Reduced Height Phenotype in
Different Genetic Backgrounds. To assess the potential for use of Rht-B13b in breeding programs, we generated sister lines for Rht13 in three Australian elite backgrounds, alongside Rht-B1b (in EGA Gregory) or Rht-D1b (in Espada and Magenta) dwarfing alleles for comparison. We found that Rht-B13b stems elongated earlier than Rht-B1b or Rht-D1b stems, but final lengths were shorter  than the tall sister lines due to an earlier arrest in growth ( Fig. 4 A-C). This lower final length is largely due to the peduncle being shorter in Rht-B13b than in Rht-B1b or Rht-D1b plants (Fig. 4 D-F). No differences in spike length were observed. We found some differences in the effect between cultivars. In Magenta, Rht13 is a stronger dwarfing gene than Rht-D1b (shorter peduncle, no difference in lower internodes; 33.2% and 23.0% stem length reduction, respectively, compared with tall at Zadoks 77.0; P < 0.001, ANOVA with the post hoc Tukey test; Fig. 4 C and F).
In Espada and EGA Gregory, the effect of Rht-B13b on height is comparable with Rht-D1b and Rht-B1b (Fig. 4 A, B   Therefore, we hypothesized that the S240F mutation in Rht13 would also result in autoactivation of the NB-LRR protein, upregulating defense responses and reducing plant growth. We first tested this through heterologous expression of the wild-type (Rht-B13a) and mutant Rht13 gene (Rht-B13b) in tobacco leaves. We found that the Rht-B13b allele induced more cell death 5 d post inoculation than the Rht-B13a allele (Fig. 5B), which is a typical defense response to pathogen invasion.

Rht-B13b is Autoactive and Causes a Cell Death Response in
No visible signs of cell death were observed in any of the wheat backgrounds containing Rht-B13b. It is possible that autoactivation of Rht13 in wheat might nevertheless enhance defense responses leading to a reduction in growth, without leading to cell   (E, H). PR gene expression was normalized to actin. For each graph, the expression level is normalized to be 1 in Rht-B13a, error bars represent the SE (n = 3-4). Significant differences were calculated using a t test on log transformed values, *P < 0.05, **P < 0.01, ***P < 0.001. death. We found that the expression level of PR genes PR3 and PR4 were >20-fold up-regulated in the basal peduncle in the Rht-B13b mutant compared with the Rht-B13a wild-type sibling Cadenza0453 (Fig. 5 C-F), suggesting that autoactivation of defense responses occurred in the Rht13 mutant plants in rapidly expanding tissue. PR4 was 15-fold up-regulated in the apical peduncle, but no significant difference was observed in PR3 expression (Fig. 5 D-G). No differences were observed in PR gene expression between Rht-B13b and Rht-B13a in the flag leaf blade (Fig. 5 E-H).

RNA-seq Analysis Reveals that Class III Peroxidases are Up-
Regulated by Autoactive Rht13. To further explore the pathways through which Rht13 reduces height, we used the same RNAseq data from peduncle samples of fixed lines from the Magnif x Magnif M population, which was previously used to identify the causal gene (see Fig. 2). We confirmed that PR genes were upregulated in Magnif M (Rht-B13b) compared with Magnif (Rht-B13a) (SI Appendix, Fig. S6), similar to observations in Cadenza (Fig. 5 C-H). The fold changes observed were higher in the RNA-seq data (SI Appendix, Fig. S6) than the qPCR data (Fig. 5  C-H); however, PR4 up-regulation was only borderline significant (P = 0.05). The up-regulation of PR genes was consistent with up-regulation of defense response-associated genes in the Magnif M plants compared with Magnif, identified by gene ontology (GO) term enrichment (SI Appendix, Fig. S7). Overall, we found that more genes were up-regulated (1,560 genes) than downregulated (726 genes) in Magnif M compared with Magnif (>two-fold, padj < 0.001). Up-regulated genes were enriched for GO terms including defense responses, cell wall organization, regulation of hydrogen peroxide metabolic processes, and salicylic acid biosynthetic processes. We did not detect any enrichment for genes related to GA signaling or biosynthesis. Down-regulated genes were associated with flavonoid biosynthetic processes, responses to cytokinin and photosynthesis (SI Appendix, Fig. S7).
We further hypothesized that the autoactivation of defense responses in the mutant line will cause the production of reactive oxygen species, which can promote cross-linking and cell wall stiffening leading to less growth (32,33). To investigate this, we examined the expression of class III peroxidases that can use hydrogen peroxide in cross-linking reactions during cell wall organization and pathogen defense (34). We identified 218 class III peroxidases that were expressed in Magnif or Magnif M peduncle samples. Of these, 28 were significantly up-regulated in Magnif M (Rht-B13b) compared with Magnif (Rht-B13a) in the peduncle (padj < 0.001, >two-fold, Fig. 6A), which is a significantly greater proportion than would be expected for a set of 218 random genes (12.8% vs. 2.6%, chi-squared test, P < 0.001). Furthermore, many of the class III peroxidase genes were very strongly up-regulated (11/28 are up-regulated >10-fold).
We found that Magnif M (Rht-B13b) peduncles had lower hydrogen peroxide content than Magnif (Rht-B13a) (Fig. 6B, P < 0.05, Student's t test), consistent with the up-regulation of class III peroxidases in the mutant (Fig. 6A). Expression levels of other families of enzymes that affect hydrogen peroxide levels in cell walls were not substantially different between Magnif M (Rht-B13b) and Magnif (Rht-B13a) peduncles (1/23 NADPH oxidase genes was up-regulated padj < 0.001, >two-fold, none were down-regulated, 0/4 oxalate oxidases (germin-like proteins), and 0/15 Cu/Zn superoxide dismutase genes were differentially expressed, P > 0.05). To test whether the changes to class III peroxidase gene expression and hydrogen peroxide content influence cell wall mechanical properties, we used a three-point bend test to measure peduncle strength and rigidity. We found that the Magnif M (Rht-B13b) peduncles were stronger and more rigid than Magnif (Rht-B13a) peduncles ( Fig. 6C and D, P = 0.02 and P = 0.003 respectively, Student's t test). The Magnif M (Rht-B13b) peduncles had shorter cell lengths in their epidermis, with cell lengths of approximately 2/3 of wild type, suggesting a lower level of cell expansion (Fig. 6E). To investigate whether these mechanical changes are mediated by changes to lignification, we examined cross-sections of the peduncle taken from the apical part of the peduncle immediately under the ear, the midpoint of the peduncle, and the basal part of the peduncle just above the node. Using toluidine blue, we did not observe any obvious morphological changes (Fig. 6F), and no significant differences in lignification were observed between Magnif and Magnif M in the apical or middle peduncle (Fig. 6G). However, the basal sections of Magnif M (Rht-B13b) peduncles had much lower staining of lignin in and around vascular bundles than the Magnif (Rht-B13a) (Fig. 6G).

Discussion
Novel Mechanism for a Wheat Rht Gene. A striking difference to other reported Rht genes in wheat is that Rht13 is not directly involved in GA signaling or metabolism, as is the case for conventional dwarfing genes Rht-B1b and Rht-D1b (2) and the cloned alternative dwarfing genes Rht12 (11), Rht18 (8), and Rht24 (13). Instead, Rht13 is an NB-LRR gene with a point mutation that induces autoactivation. The amino acid change in Rht13 is the same mutation as previously characterized in the tomato protein I-2, which impeded ATP hydrolysis and promoted an ATP-bound active form of the protein (31). Due to the high conservation between the RNBS-A motif between I-2 and Rht13, we hypothesize that the mutation in Rht13 has the same biochemical function to impede ATP hydrolysis, consistent with the hypersensitive response (HR) we observed upon expressing Rht-B13b in N. benthamiana leaves.
Autoactive NB-LRR genes have been reported to reduce growth in several plant species (35)(36)(37), including causing reduced internode length in flax (38). However, autoactive NB-LRRs are often associated with negative pleiotropic effects including a spontaneous HR resulting in necrotic lesions. We did not observe any spontaneous HR or necrosis in any of the wheat genetic backgrounds tested. Similarly, transgenic flax lines expressing specific autoactive alleles of the L6 NB-LRR gene showed a reduction in height without necrosis (38), suggesting that it may be possible to identify autoactive alleles to alter growth without negative pleiotropic effects in a range of plant species. Rht-B13b behaves differently from known autoactive NB-LRR genes in cereals that reduce height, such as Rp1-D21 in maize which induces a spontaneous HR in a range of genetic backgrounds, although to differing degrees of severity (35). Nevertheless, Rht-B13b induced a HR in tobacco, which could be a result of high transient expression in tobacco, although overexpression of Rht-B13b in wheat did not cause a HR despite severe stunting. One possibility is that the cell death response in wheat is suppressed by the presence of homologous genes, as was observed for the Pm8 resistance gene to powdery mildew, which was suppressed by its homolog Pm3 (39). It is also possible that tissue-specific expression of Rht13 in wheat or differences in signaling pathway thresholds between tobacco and wheat may explain these differences. This is supported by our finding that PR genes were up-regulated only in peduncle tissues, and not in the flag leaves of Cadenza Rht-B13b. The up-regulation of PR genes in Rht-B13b containing wheat raises the question whether Rht-B13b could also enhance resistance response to certain pathogens. Autoactive mutants in flax, potato, and tomato were shown to gain additional specificities to strains of the same pathogen or became effective against other pathogen species (38,40,41), but further research will be required to determine any association between Rht-B13b and enhanced disease resistance.
Among the PR genes up-regulated by Rht-B13b are class III peroxidases which are known to act in a wide range of physiological processes, including cross-linking of cell wall components, formation of lignin, and metabolism of reactive oxygen species such as hydrogen peroxide (34). The up-regulation of class III peroxidases is associated with a decrease in hydrogen peroxide in Rht-B13b, which may be due to its use in cell wall cross-linking. Increased cross-linking could explain the reduced cell lengths observed in Rht-B13b and the increase in peduncle strength and rigidity. Surprisingly, we did not observe an increase in lignin in Rht-B13b compared with the wild type, suggesting that these changes in cell size and tissue strength may be mediated by cross-linking polysaccharides and extensins other than lignin. Alternatively, subtle differences in lignin content may not be detectable by histochemical staining in the middle section of the peduncle, where differences in bending strength were observed. Taken together, we present a possible model through which Rht-B13b operates (Fig. 7). In this model, the up-regulation of class III peroxidases promotes cross-linking of cell walls in the tissues of Rht-B13b carriers, constraining cell elongation and ultimately reducing height. Further work will be needed to validate the pathway through which Rht-B13b acts.
Applications in Agriculture. Rht13 is effective in multiple genetic backgrounds and provides a height reduction similar to conventional dwarfing genes Rht-B1b and Rht-D1b. Rht13 dwarfism is not associated with reduced seedling growth or coleoptile length, and most of the height-reducing effect occurs later in development (after Zadoks stage 50), which is mainly associated with reduction in peduncle growth. Therefore, the gene is well suited to water-limiting environments that require deeper planting to access available moisture and rapid leaf area development to lower evaporative losses from the soil surface. We found that Rht-B13b increased bending strength, which may further decrease lodging and reduce yield losses compared with One representative image from five independent biological replicates is shown. Asterisks indicate statistical differences between genotypes: *P < 0.05, **P < 0.01, ***P < 0.001.
conventional Rht genes. In previous work, Rht-B13b was reported to increase yield by 18 to 21% in recombinant inbred and nearisogenic lines grown in Australia (16,17). In contrast, trials in northwest China reported a decrease in the grain yield per plant of 26 to 29% (18,19). However, similar decreases in the grain yield per plant occurred for Rht-D1b containing lines grown under these conditions, and yields were not assessed at the plot level (18). Further trials in diverse environments with advanced-generation material will be required to establish the effect of Rht-B13b on yields. Testing and deployment of Rht-B13b will be facilitated by the use of a perfect KASP marker for the selection of the allele in breeding programs. It is possible that Rht-B13b mutation is already circulating in some breeding materials, for example, in the WM-800 eight-way MAGIC population of European winter wheat cultivars, a significant quantitative trait locus (QTL) was identified on chromosome 7B, for which the peak SNP marker maps only 10 Mb away from the location of Rht13 (42). However, no height QTL was identified on chromosome 7B in other MAGIC populations including a diverse UK 16 founder MAGIC population (43) and an Australian four-way MAGIC population (44).
In conclusion, the identification of an NB-LRR gene underlying an alternative dwarfing gene in wheat has provided insight into an alternative pathway, where GA biosynthesis or signaling is not directly affected. This discovery will open up opportunities to alter height, potentially through engineering of autoactive NB-LRR genes and cell wall enzymes. More knowledge will be needed to establish whether the activation of defense responses by Rht-B13b could influence disease resistance.

Methods
All methodological information is available in SI Appendix, Materials and Methods. This includes details about plant materials, genetic mapping, chromosome-seq, RNA-seq, candidate gene identification and validation (transgenics and TILLING), heterologous N. benthamiana expression, qPCR, hydrogen peroxide, cell size, and stem property measurements.
Data, Materials, and Software Availability. The data that support the findings of this study are available in the SI Appendix of this article, and raw reads for the chromosome-seq and RNA-seq are deposited as PRJEB51492 (45) Fig. 7. Model of pathway through which Rht-B13b causes semidwarfism. In a wild-type plant (Rht-B13a, Left) the NB-LRR protein is inactive resulting in normal cell wall cross-linking, cell expansion, and growth. In the autoactive mutant (Rht-B13b, Right), PR genes including class III peroxidases are up-regulated in expanding tissues. Class III peroxidases may use H 2 O 2 to increase cell wall cross-linking, which results in reduced cell expansion and growth.