Phenotypic analysis of barley rls mutant
The mutant rls was initially identified from screening of 60Co-γ and EMS-treated mutant library in the genetic background of E934. Leaf color of rls was much darker than that of E934 (Fig. 1a), and chlorophyll concentration reflected by SPAD was 29.03 ± 1.11 and 52.53 ± 3.25 in E934 and rls, respectively. Besides, rls was mainly characterized as a mutant with abnormal spike morphology (Fig. 1b). Lateral spikelets of rls were round and sterile. In addition, awn of rls was very short, glume was harder, bilateral empty glumes were also much shorter but a little bit wider than wild type (Fig. 1c-e). Grains of rls were round and shorter than wild type, but there was no significant difference on grain width between rls and E934 (Fig. 1f). Consequently, thousand grain weight of rls was only 32.26g, being much lower than 45.86g of E934. Moreover, outer glume section around the embryo in dried seeds of rls (166.40µm) was much thicker than that in E934 (108.00µm) (Fig. 1g). rls was also characterized as dwarf, plant height of which was about 60% of wild type, and heading date of rls was almost 14 days later than E934.
To find out which stage was crucial for the differentiation of rls spike morphology, 3 mm and 5 mm young panicles of E934 and rls were scanned under electron microscope. It was found that there was no difference between 3 mm spikes of E934 and rls. However, for 5 mm spikes, rls awn was longer than E934, especially the awn at the bottom (Fig. 2a, 2b), though awn of rls was much shorter than E934 at maturity stage. Despite these differences, lateral spikelets clearly increased in size with enlarged lemmas enclosing differentiating floral organs in rls as compared with E934 (Fig. 2c, 2d). What’s more, to determine whether dark green leaf color of rls was associated with development of chloroplast, ultrastructure of chloroplast of E934 and rls leaves was also investigated under scanning electron microscope. Consistently, chloroplast size was increased in rls as compared with that in E934, almost half of the cell being occupied by chloroplasts in rls (Fig. 2e-h).
To further reveal the cellular basis of round grain in rls mutant, cross section of hull in E934 and rls was observed. Significant differences of cell morphology were observed between E934 and rls both longitudinally and horizontally. As shown in Fig. 2, a regular pattern of three cell layers including silicified cells (SC), outer parenchyma cells (OPC) and inner parenchyma cells (IPC) was observed in WT (Fig. 2i, 2j). There was no significant difference between SC of E934 and rls, however, a remarkable difference was observed between OPC and IPC between them. Wider empty region in the OPC layer but a relative bigger and regular IPC was observed both in the lemmas and paleas from rls as compared with that from E934, which may contribute to the thicker glumes and round grains in rls (Fig. 1g).
Genetic dissection of rls mutant underling gene
To decipher genetic feature of the causal gene, rls was crossed with Yan 03174, a typical two-row barley variety with long thin and sterile lateral spikelet, and also long awn, to obtain the F1 seeds. It was found that spikes of the F1 plants showed intermedium type, awns of which were long and same as wild type plants, but were tighter than wild type awns. Lateral spikelets of F1 plants were fatter than wild type and crooked, but bilateral empty glumes of lateral spikelets were the same as rls (m) (Fig. 1b-e). Individuals with the same phenotype as F1 plants were recorded as “mm” in this study (Fig. 1b-e). While the F2 population showed a segregation of the WT, mm and m type spikes, which included 268, 462 and 286 plants, respectively. Chi-test showed that the number of WT, mm and m type plants in the F2 population fitted a ratio of 1:2:1. Moreover, all the F2:3 families derived from WT type F2 individuals had homozygous wide type spikes, all the F2:3 families derived from mm type F2 individuals showed a segregation of three types spikes, while all the F2:3 families derived from m type F2 individuals had homozygous mutant type spikes, which confirmed that this changed spike morphology in rls was controlled by a single semi-dominant gene, and was designed as HvRLS in this study.
Preliminary mapping of rls candidate gene via BSA based on RNA-seq
To map the causal gene for rls, BSA based on RNA-seq was initially conducted using mutant pool and wild type pool derived from 60 heterozygous F2:3 families (progenies of mm type F2 individuals). Totally 494 SNPs/Indels were identified as potential linkage with the phenotype, half (246) of which could be anchored to the end of long arm of chromosome 4H (Totally 647.06Mbp), falling into the interval of 619Mbp-646Mbp (Morex V1) (Fig. 3a). However, only 23, 40, 67, 42, 32 and 40 SNPs/Indels were on chromosome 1H, 2H, 3H, 5H, 6H and 7H, respectively, indicating that HvRLS gene was on chromosome 4H.
To validate the BSR analysis, 15 polymorphic markers (Supplementary Table S1) in the target region were developed according to the BSR-SEQ results. These markers were then used to screen the genotypes of rls and Yan 03174 as well as the F3 and F4 population derived from mm type F2 and F3 individuals, respectively. Based on PARMS, totally 505 m, 16 mm and 12 WT type plants from F3 and F4 population were genotyped and analyzed. Finally, HvRLS gene was mapped between 4H-6375 and 4H-60210 (Fig. 3b and 3c), which was located at 601,621,013 bp and 602,104,747 bp on chromosome 4H according to the updated reference genomics of barley cv. Morex (Barley pseudomolecules Morex v3.0) (Jayakodi et al. 2020), spanning about 480 kb. The most interesting was that the genotypic data of 4H-60205, representing a single nucleotide polymorphism (SNP) G/A, co-segregated with rls phenotype in the mapping population (Fig. 3d).
Cloning of rls candidate gene
Further analysis showed that there were 17 high-confident genes (Table 1) in this 480 kb target region. It was very interesting that the G to A substitution represented by 4H-60205 was on a beta TUBULIN gene. Further blast analysis showed that this beta TUBULIN gene was the previously annotated beta tubulin HvTUB8. According to the updated barley genome, genomic sequence and coding region of HvTUB8 was 1589 bps and 1344 bps in length, respectively, which had three exons and encoded a polypeptide composed of 447 amino acid residues harboring the conserved “tubulin beta chain” domain (1-430/447), and the G-A variation was on the third exon (+ 1061) of HvTUB8 (Fig. 4a).
Table 1
High-confident genes in the target region
Gene ID
|
Location on chromosome 4H (Mbp)
|
Gene description
|
|
Start
|
End
|
|
4HG0415300
|
601627654
|
601629075
|
RNA-directed DNA polymerase (reverse transcriptase)-related family
|
4HG0415310
|
601641518
|
601641826
|
Homer protein homolog
|
4HG0415320
|
601652557
|
601652931
|
Non-structural protein NS1
|
4HG0415330
|
601729791
|
601733093
|
Cytochrome P450 family cinnamate 4-hydroxylase
|
4HG0415340.1
|
601760190
|
601765625
|
ATP-binding cassette transporter subfamily A
|
4HG0415350
|
601810049
|
601810648
|
DNA topoisomerase
|
4HG0415360
|
601811485
|
601812995
|
F-box family protein
|
4HG0415370
|
601821300
|
601821821
|
villin-like 1
|
4HG0415380
|
601823536
|
601823943
|
Leucyl/phenylalanyl-tRNA–protein transferase
|
4HG0415390
|
601825582
|
601825959
|
Protein TolB
|
4HG0415400
|
601853564
|
601854058
|
myosin heavy chain, embryonic smooth protein
|
4HG0415410
|
602046060
|
602048127
|
AT hook motif DNA-binding family
|
4HG0415420
|
602053484
|
602055456
|
Tubulin beta
|
4HG0415430
|
602085760
|
602087165
|
MYB-related transcription
|
4HG0415440
|
602095267
|
602095641
|
RNA-directed DNA polymerase (reverse transcriptase)-related family
|
4HG0415450
|
602102008
|
602102721
|
Receptor-like kinase
|
4HG0415460
|
602104336
|
602106756
|
Receptor-like protein kinase
|
To validate the co-segregated G-A substitution, full length coding regions of HvTUB8 were cloned from genomic DNA and cDNA of young panicles of E934 and rls subsequently. Both of the sequences from gDNA and cDNA confirmed the G to A mutation in rls, while no other variation was detected in coding region between E934 and rls. Moreover, there was no difference between the sequence of 1000bp upstream region from E934 and rls either. Whereas, coding regions of the other 16 genes showed no difference between E934 and rls. Further study exhibited that the cysteine (C) at amino acid position 354 of HvTUB8 was replaced by a tyrosine (Y) in rls due to the G-A substitution (Fig. 4a).
Phylogenetic analysis of β-TUBULIN proteins in plant
BlastP and BlastN of the wide type HvTUB8 protein was then conducted online, and homologue proteins from barley and its relatives, such as Triticum dicoccoides, Triticum aestivum, Aegilops tauschii, Oryza sativa and Zea mays, were downloaded. As shown in Fig. 4b, these proteins were highly conserved among different species, which showed 95–100% similarity to each other. Phylogenetic analysis indicated that HvTUB8 belonged to a mini cluster that only included members from Triticum and its ancestor (Fig. 4b). HvTUB8 protein showed 100% and 96.89% similarity to its counterparts from Triticum dicoccoides (XP_037437345.1) and Oryza sativa (OsTUB2), respectively. Though the Cys to Try substitution mentioned above located in a region that was not quite conserved as compared with other regions, the wide type Cys was highly conserved among barley and its relatives, the substitution could only be detected in rls (Fig. 4c).
Spatial and Temporal expression of HvTUB8 in E934
Then spatial and temporal expression profile of HvTUB8 was analyzed in 10 tissues of E934 at different developmental stages. Result showed that expression pattern of HvTUB8 was highly tissue-specific. It mainly expressed in young panicles and leave, and a relative lower expression was also detected in stamen and internode below the spike at the heading stage (Fig. 4d). However, it was hardly detected in other tissues.
Haplotype analysis of HvTUB8 gene in barley accessions
The co-segregated maker QDD4H-60205, which represented the G to A mutation in HvTUB8, was also applied to screen a panel of more than 400 worldwide barley collections with typical two- or six-row type spikes. As expected, none of these accessions had the “A” type allele as rls. To further explore the natural diversity of HvTUB8, genetic variation of the gene was analyzed in the 20 barley accessions together with E934 using the barley pan-genome data. Coding region from the start codon to stop codon was extracted from 19 accessions except for “Igri”, with nine bps deletion at the very beginning of the gene. It was also found that all these 20 accessions had the same allele (G) as E934 at + 1061. Additionally, a total of 14 other SNPs were detected in coding region of HvTUB8, which resulted in seven haplotypes (Supplementary Table S2) in the 20 accessions, while HvTUB8 in E934 was distinctive and different from all the seven haplotypes. In addition to coding region, 1000bps upstream regions of HvTUB8 were analyzed between these 19 accessions except for HOR3365. Abundant variations were also detected in 1000bps upstream of this gene. However, none of these haplotypes was associated with barley row type.
Functional characterization of HvTUB8 by BSMV-VIGS
To confirm the participation of HvTUB8 in barley spike development, we used BSMV-VIGS to silence the gene in wild-type E934 plants. Control infections with BSMV-γindicated that viral infection did not affect spike morphology of barley, and with BSMV-TaPDS indicated infection and gene silencing efficiency, as shown by bleached lines on infected leaves. Expression analysis of the target gene in leaves of these BSMV:HvTUB8 infected plants suggested that expression of HvTUB8 was reduced about 70% as compared with in non-transformed E934. As shown in Fig. 4e, irregular spikes were observed in plants infected with BSMV:HvTUB8, and extra spikes were produced at the base of the main spike, which was different from the known barley branched spikes. However, no other difference, like grain shape, awn length, leaf color was detected in BSMV:HvTUB8 infected E934 plants as compared with the non-treated E934. The results confirm that the wide type HvTUB8 gene is essential for maintaining the interity of barley spike, but is not for the rls phenotype in the present study.
Subcellular localization of HvTUB8 and Hvrls proteins
To demonstrate the subcellular localization of HvTUB8 protein, both the full length HvTUB8 protein and the mutated Hvrls form were fused with GFP reporter and expressed in wheat protoplasts. As shown in Fig. 5, in the positive control cells expressing 35S::GFP protein only, GFP signal was observed throughout the cells, but the single amino acid change of HvTUB8 caused change of subcellular localization of the protein in wheat protoplasts. In the cells expressing 35S::HvTUB8:GFP protein (Fig. 5e-h), a weak signal was detected in cytoplasm specifically. However, in the cells expressing the mutant isoform of HvTUB8 fused GFP protein, a strong green florescence signal was detected both in the nucleus and cell membrane.
Transcriptome profiling reveals the downstream genes regulated by HvTUB8
To interrogate the downstream signaling pathways regulated by HvTUB8 and their contributions to barley development, RNA-seq was performed on 0.5-1 cm young panicles of rls and E934 with three biological replicates. Totally, 40,892 expressed genes were detected with 441 genes being differentially expressed in rls relative to E934, and more than three quarters (340) of these differentially expressed genes (DEGs) were up-regulated by the mutation of HvTUB8 (Supplementary Table S3).
Furthermore, expression of HvTUB8 was analyzed based on the transcriptomic data. It was shown that the gene expression levels were similar between the mutant and the wild type though HvTUB8 could be detected in 0.5-1 cm spikes of both rls and E934. Then expressions of known barley lateral spikelet development regulating genes, including VRS1, VRS2, VRS3, VRS4 and VRS5 were further analyzed, but none of them were altered in rls as compared with in wild type. Two LOB domain-containing protein genes on 3H (HORVU.MOREX.r3.3HG0233680) and 4H (HORVU.MOREX.r3.4HG0414380) were up-regulated by 3 and 7.6 folds, respectively, and a Homeobox-leucine zipper family protein (HORVU.MOREX.r3.2HG0189890) was also up-regulated by 4 folds in rls. However, the two LOB domain-containing genes only showed 30.35% and 33.16% similarity to VRS4/HvRA2 in the coding region, while the Homeobox-leucine zipper family protein showed low level (40.55%) of similarity to VRS1, respectively.
Further GO enrichment analysis of the 441 DEGs showed that about half of the DEGs (213) could be anchored to known GO terms (Supplementary Table S4). Totally 34, 5 and 1 GO terms were enriched for Biological Process, Molecular Function and Cellular Component, respectively. It was found that phytohormones homeostasis was extensively disturbed in rls. Notably, a 1-aminocyclopropane-1-carboxylate oxidase gene and a 1-aminocyclopropane-1-carboxylate synthase gene in “Ethylene biosynthetic process”, together with 12 of the 14 Ethylene-responsive transcription factors in “ethylene-activated signaling pathway” were significantly up-regulated in rls (Fig. 6). Additionally, genes responsive to other phytohormones were also enriched in DEGs, including salicylic acid, brassinosteroid and gibberellic acid, and most of them were transcription factors, such as WRKY and zinc finger proteins. Moreover, the RNA-Seq data also witnessed a large amount of other transcription factors being regulated by mutation of HvTUB8, such as NAC, F-box, MYB, CRT-binding factors, bHLH, MADS-box and so on. The majority of them were up-regulated in rls. On the other hand, expressions of 32 genes encoding various protein kinases (LRR, WAK etc) were changed in rls, with 27 and 5 being up-regulated and down-regulated, respectively.
Considering the difference of leaf color and ultrastructure of chloroplast of E934 and rls, chlorophyll biosynthesis and metabolism related genes were analyzed subsequently. It was found that expressions of 11 genes participating in chloroplast synthesis and metabolism were changed by the mutation of HvTUB8 in rls, nine of which were also up-regulated (Fig. 6)