Phenotypes of the wheat tillering mutant lt1
The lt1 mutant was derived from an EMS mutagenesis pool of the elite wheat landrace C6878. This recessive mutant exhibits reduced tillering, typically producing four tillers compared to eighteen of C6878 at the heading stage (Fig. 1A-B). To explain whether the reduced tillers are due to defects in bud initiation or bud elongation, we observed the dynamic development process of tiller buds. At first, we found that the number of AMs remained consistent during the coleoptile and two-leaf stages but started to diverge by the four-leaf stage, with four in the wild type and two in lt1 (Fig. 1C-D). This revealed that the reduced tillering number of lt1 is partially due to the defective AM initiation. Furthermore, we examined the number of ceased lateral buds at the heading stage. This thorough examination revealed a reduced outgrowth ratio of lt1 compared to C6878 (Fig. 1E-F). Taken together, the tillering defect of lt1 appears attributable to both its reduced AM formation and bud outgrowth.
Additional pleiotropic defects in lt1, including decreased stature, short roots, chlorotic leaves, and wrinkled seeds, are concomitant (Fig. S1). These global impacts on lt1 development suggest that LT1 plays significant roles in multiple processes.
Isolation of LT1 by an upgraded bulked segregant method, uni-BSA
The LT1 gene was proven to be a recessive gene via assessing the F2 segregating population. However, the hexaploidy wheat genome is highly complex, which makes traditional map-based cloning more time-consuming. To expedite the cloning of LT1, we utilized the BSA-based method uni-BSA, using the big data from WES combined with our newly developed algorithm, making it cost-friendly and effective. Firstly, the WES data was used to minimize the genome size without the penalty of losing protein-encoding genes while guaranteeing enough SNPs to carry out linkage analysis. Secondly, to address the ambiguous mapping when alignment is performed due to the high duplication proportion of the wheat genome, which may result in aligning one read to multiple loci, we tailor-make a Perl script called Filter.ambi.pl integrated into the uni-BSA protocol (Fig. 2, Fig. S4A). This algorithm potentially leverages reads as more as possible that are uniquely mapped and their mate reads, even if their mate reads are mapped ambiguously to several locations. Accordingly, this filtering method retained 61% of total reads, compared to 48% when discarding all ambiguous reads (Fig. S2B). As a result, the average percentage of each gene coverage was over 81%, with the majority of genes covered at 100% (Fig. S2C). The average coding sequencing depth reached 70X, implying robust sequencing quality for accurate variant calling (Fig. S2D).
Application of uni-BSA narrowed LT1 to a 6 Mb region on the short arm of chromosome 2D (Fig. 3A), compared to 8 Mb without ambiguous read filtering (Fig. S4D). This interval contains 140 genes, of which 65 genes have variations, including SNPs and Indels. As EMS tends to cause SNPs over Indels, 17 Indels were excluded, thus eliminating five genes. Additionally, 41 SNPs of lt1 matching the reference Chinese Spring are unlikely EMS-induced mutations. Ultimately, four genes were identified as candidate genes (Table S2). Interestingly, one gene, TraesCS2D03G0082100, encoding a nucleotide-binding (NB) domain protein (Fig. 3C), harbors an SNP mutation in the 793rd base (C-T), causing a premature of this gene in lt1 (Fig. 3C), while the other four genes had UTR mutations. In addition, the individuals of the F2 population with this homozygous mutation displayed lt1 phenotypes (Fig. 3B). Further, TraesCS2D03G0082100 was not expressed in 2-leaf, 3-leaf, and 4-leaf of lt1, compared to the wildtype (Fig. 3D). We initially considered TraesCS2D03G0082100 the likely causal LT1 gene, given its severe mutation and undetectable expression.
Verification of LT1
To validate TraesCS2D03G0082100 as the LT1 gene regulating tillering in wheat, we used CRISPR/Cas9 to create knock-out mutants in Fielder. The three independent edited lines with different mutations within its coding sequences were obtained (Fig. 4A). The LT1-CR1 and LT1-CR2 show the mutations at gRNA targeted sites, and LT1-CR3 has 239 bp deletion (Fig. 4A). Intriguingly, all three edited homozygous individuals produced fewer tillers than the wildtype (Fig. 4B-D). Moreover, these three lines exhibit other defects of lt1, like yellow leaves (Fig. 4B-D), thus confirming TraesCS2D03G0082100 as the LT1 locus. Therefore, we hereby designate TraesCS2D03G0082100 as LT1.
To elucidate the possible reasons for pleiotropic phenotypes of the lt1 mutant, we assessed the expression levels of LT1 in various tissues. qPCR analysis revealed ubiquitous expression of LT1, with exceptionally high levels in leaves (Fig. 4E). Given its high level in leaves, it is not strange that lt1 has yellow leaves once LT1 is disrupted. LT1 was detectable in tiller buds, albeit at relatively lower levels (Fig. 4E). The broad expression pattern of LT1 suggests its multiple roles in wheat development. Overall, these data indicate that LT1 likely influences tillering and other developmental processes indirectly or directly.
To determine the sublocation of LT1, we carried out a transient expression experiment of LT1 in wheat protoplasts. In contrast with the control, which is ubiquitous in protoplast cells, the LT1-GFP fusion protein was predominantly localized in chloroplasts (Fig. 4F). The chloroplast location of LT1 implies that LT1 may operate nutrition production, like sucrose, to control tillering.
The regulatory pathways of LT1 in tillering development
LT1 controls lateral bud formation by targeting TaROX/TaLAX1 directly or indirectly
To investigate modular relationships involving LT1 further, we conducted a co-expression analysis using TPM values from shoot base tissues at three developmental stages: the two-leaf, three-leaf, and four-leaf. An initial survey of these RNA-seq datasets revealed that samples belonging to each group clustered well (Fig. S3). The transcripts were grouped into eight clusters representing distinct gene expression trends (Fig. 5A). LT1 expression, which belongs to cluster five, was highest at the two-leaf stage and then decreased at the three- and four-leaf stages. We considered the two-leaf stage to be essential for AM initiation since genes active in this stage showed a pulse expression and then decreased in the following stages. Thus, we performed GO analysis on genes with significant changes between lt1 and C6878, revealing perturbation of various pathways in lt1. We then specially examined genes belonging to the overlap between cluster five and the two-leaf stage to determine which pathways were affected (Fig. 5B). Notably, in this stage, various pathways (Fig. 5C) related to AM formation shared the locus TraesCS3B02G383000, an ortholog of Arabidopsis ROX and LAX1 in rice that regulate AM formation. These pathways include “morphogenesis of a branching structure”, “secondary shoot formation”, and “shoot axis formation”. Moreover, TraesCS3B02G383000, namely Ta3BLAX1, is undetectable in lt1 (Fig. 5D). This is consistent with our previous observation of significant differences in tiller numbers at the four-leaf stage in lt1 mutants (Fig. 1C). Taken together, LT1 might regulate AM initiation by affecting TaROX/TaLAX1 directly or indirectly.
Auxin and cytokinin are involved in tiller development in lt1
Auxin and CK play antagonistic roles in regulating tillering (Yuan et al. 2023). We performed GO analysis on the genes in the intersection between the three developmental stages and cluster 5, respectively. The results revealed perturbation in several phytohormone-related pathways, including auxin, CK, salicylic acid, and jasmonic acid (Fig. 6A). Among these pathways, the indole-containing compound biosynthesis process, in which auxin is biosynthesized, was enriched at all three developmental stages. For example, TrpA family genes Ta5BTrpA and Ta5DTrpA exhibited significant upregulation in lt1. This suggests that higher auxin levels may inhibit tillering in lt1 (Fig. 6B). In addition to auxin, CK levels were suggestively decreased, as indicated by the upregulation of TaCKX5 (cytokinin dehydrogenase 5) genes (Ta3ACKX5, Ta3BCKX5, and Ta3DCKX5) mediating CK degradation (Fig. 6B). These CKX5 genes were also enriched in pathways related to secondary shoot formation (Bartrina et al. 2011), implying CK metabolism may play an important role in tillering in wheat controlled by LT1. Taken together, LT1 may regulate tillering through the involvement of auxin and cytokinin-related pathways.
LT1 may function through the sucrose biosynthesis pathway
As with all organisms, plants require energy for growth. They achieve this by intercepting light and fixing it into usable chemical forms via photosynthesis. The resulting carbohydrate (sugar) energy is then utilized as substrates for growth or stored as reserves (Eveland and Jackson 2012), thus influencing various aspects of plant development, such as tillering (Rabot et al. 2012). Our co-expression analysis revealed perturbations in the fructose 1,6-bisphosphate (FBP) pathway at the four-leaf stage, which is involved in sucrose biosynthesis (Fig. 6C). Coincidentally, RNA-seq analysis using whole seedlings with two leaves also showed perturbations of the FBP pathway genes (Fig. 6D). Within this pathway, three closely related TaFBPase genes involved in sucrose biosynthesis were down-regulated in lt1 mutants (Fig. 6E), implying lower sucrose levels. To determine if the sucrose levels have changed in the lt1 mutant, we collected the shoot base at the two-, three-, and four-leaf stages and measured the sucrose level. Indeed, it decreased significantly in lt1 mutants compared to wildtype (Fig. 6F). Together, these datasets suggest LT1 may exert its influence on tillering and other phenotypes by targeting FBPases, thereby impacting sucrose levels.