Barley (Hordeum vulgare cv. Morex) seeds were treated with 0.03 mol/L EMS for 16 h, and the yellow leaf phenotype mutant was identified in M2 individuals. The mutant was self-pollinated for another five generations to obtain a relatively stably inherited yellow leaf mutant named ylg8. Morex was crossed with the ygl8 mutant, and F₁ plants were self-pollinated to generate F₂ populations for genetic analysis. The greenhouse and field experiments were carried out for hydroponic experiments and BSA and RNA-seq experiments at the Zhejiang Academy of Agricultural Sciences (Hangzhou, Zhejiang Province in China), respectively.

For hydroponic experiments, the barley seeds were sterilized with 10% commercial NaClO, rinsed with tap water, and then germinated in BSM solution (0.5 mM KCl + 0.1 mM CaCl2) for 2 days. Afterward, BSM was changed to 1/5 Hoagland solution with a photoperiod of 14/10 h, light intensity of 200 ± 25 μmol·m−2·s−1, relative humidity of 60%, and constant temperature of 24°C for 8 days (Cai et al., 2022). For different temperature treatments, the seedlings were grown in incubators with 14/10 h light/dark at a constant air temperature of 10°C, 24°C, or 30°C.

Equal fresh weights (approximately 50 mg) of the newly expanded leaves were cut and immersed in 10 ml extract solution (ethanol: acetone: water = 4.5: 4.5: 1) for 16 h in the dark. The homogenates were centrifuged at 3,000 × g for 5 min to remove residual plant debris. The supernatants of these samples were analyzed, and the chlorophyll a and b and carotenoid contents were measured using a spectrophotometer at 663, 645, and 470 nm, and determined using the Lichtenthaler method (Chen et al., 2013).

The newest fully expanded leaves of Morex and the ygl8 mutant at the four-leaf age were collected 1 month after planting in the field and quickly frozen in liquid nitrogen. The total DNA was extracted following the modified hexadecyl trimethylammonium bromide (CTAB) method (Murray and Thompson, 1980). For whole-genome sequencing, the bulked DNA for MutMap analysis was prepared by mixing DNA from 30 ygl8 individuals in an equal ratio, and the parent DNA for MutMap analysis was prepared by extracting DNA from one Morex sample. A paired-end sequencing library was constructed with an insert size of approximately 350 bp and sequenced on the Illumina Hiseq X Ten platform.

For RNA-seq, the newest fully expanded leaves of Morex and the ygl8 mutant were collected at the stem elongation stage in the field. Total RNA was isolated using an RNAprep Pure Plant Kit (TIANGEN). RNA-seq library construction was carried out following a method described previously (Shen et al., 2014) and sequenced on the Illumina Hiseq X Ten platform. All sequencing was carried out at Biozeron Biotech, Shanghai.

For MutMap analysis, more than 30 depth re-sequencing raw data of bulked pool and parent Morex ygl8 were filtered using fastp (Chen et al., 2018b) (Version 0.12.4) with the default parameters. The high-quality cleaned reads were aligned to the barley reference genome (Mascher et al., 2021) with BWA (Li and Durbin, 2009) (Version 0.7.17-r1188) using the default parameters. SAMtools (Li et al., 2009) (Version 1.14) was used to convert the mapping results into the sorted and duplication-marked BAM format. SNP detection was performed by BCFtools (Danecek and McCarthy, 2017) (Version 1.14) using the bcftools mpileup command with the parameters ‘–min-MQ 40 –min-BQ 18 –adjust-MQ 50’ and the bcftools call command with the default parameters. The resulting vcf files were filtered using the bcftools filter command with the parameters ‘INFO/MQ>=40 && %QUAL>20’, and the SNP information was extracted using VCFtools (Danecek et al., 2011) (Version 0.1.17). The average SNP-index of the bulked pool of ygl8 and the parent Morex was estimated via the 2 M sliding window with a 10-kb walking step using Mutplot (Sugihara et al., 2022) (Version 2.3.3) with the parameters ‘–N-bulk 30 –window 2000 –step 10 –min-depth 14 –max-depth 90 –N-rep 10000 –min-SNPindex 0.3’. The genome-wide SNP-index plot was obtained using a custom script written in R (version 4.3.0). The SNP variants were functionally annotated using ANNOVAR (Wang et al., 2010) (Version 2019-10-24).

For RNA-seq analysis, the RNA-seq raw sequencing data of leaf tissues of Morex and the ygl8 mutant were filtered using fastp (Chen et al., 2018b) (Version 0.12.4) with the default parameters. The high-quality cleaned reads were aligned to the reference genome consisting of the barley reference genome (Mascher et al., 2021) and chloroplast reference genome (Saski et al., 2007) using HISAT2 (Kim et al., 2019). Following alignments, the raw counts for each gene were derived using featureCounts implemented in the R package Rsubread (Liao et al., 2019) and normalized into the number of transcripts per kilobase of exon sequence in a gene per million mapped reads (TPM) using the TMM method (Robinson and Oshlack, 2010). The read alignments of the RNA-seq data to candidate genes were shown in a sashimi plot implemented in the IGV genome browser. The differentially expressed genes in leaf tissues between Morex and the ygl8 mutant were detected using the DESeq2 method (Love et al., 2014), and the significantly differentially expressed genes were identified with the standard of |log₂FC| >1 and an adjusted P < 0.05. The GO and KEGG enrichment analysis were performed using the R package clusterProfiler (Yu et al., 2012) (Version 4.8.2), and the enrichment results were shown using the R package ggplot2 (Wickham, 2016) (Version 3.4.2).

The reference barley protein sequences were downloaded from Phytozome (https://phytozome-next.jgi.doe.gov/). Then, Arabidopsis GA20ox (AT4G25420) and CesA (AT4G32410) were used as query proteins to carry out a BLASTP search in the barley protein sequences with an E-value < 1e^-10 as the threshold using DIAMOND (Buchfink et al., 2021) (Version 2.0.11) to identify the barley homologous GA20ox and CesA proteins. Furthermore, the hidden Markov models (HMMs) of the zf-UDP (PF14569.9) and Cellulose_synt (PF03552.17) domains, downloaded from Pfam 35.0 (http://pfam-legacy.xfam.org/), were used to identify the barley homologous CesA proteins using HMMER (http://hmmer.org/) with a “trusted cutoff and E-value < 0.01” as the threshold. The HMMs of the DIOX_N (PF14226.9) and 2OG-FeII_Oxy (PF03171.23) domains were used to identify the barley homologous GA20ox proteins. A total of seven homologous CesA and 24 GA20ox were identified in barley.

Primers were designed according to the open reading frames of the barley YGL8 and OsYSS1 (Zhou et al., 2017) genes. The corresponding sequences were amplified from the cDNA of barley (Morex or ygl8) or rice and cloned into the pMD18T vector (Takara). The clones were selected by PCR and sequenced for confirmation. For the subcellular localization of YGL8 and ygl8 in tobacco leaves, seamless cloning was used to introduce the YGL8 or ygl8 genes, fused to the GFP gene amplified from the pH7WGF2.0 vector, into the pCAMBIA1300 vector, after the 35S promoter amplified from the pH7WGF2.0 vector in advance, to obtain YGL8-GFP or ygl8-GFP. Additionally, seamless cloning was used to introduce the OsYSS1 gene, fused to the mCherry gene amplified from pSAT4a-mCherry-N1, into a modified pCAMBIA1300 vector to obtain OsYSS1-mCherry. The primer information is listed in Supplementary Table 1 .

The subcellular localization vectors of YGL8 were transiently expressed in tobacco (Nicotiana benthamiana) leaves by Agrobacterium-mediated infiltration. OsYSS1-meCherry was used as a chloroplast marker. The Agrobacterium strain C58C1 harboring p19 was used to prevent the onset of post-transcriptional gene silencing (PTGS) in the infiltrated leaves. Infiltrated tobacco plants were grown for another 3 days for GFP and mCherry imaging using a Zeiss LSM710NLO confocal laser-scanning microscope. The excitation/emission wavelengths were 488 nm for GFP and 561 nm for mCherry.

The protein structures of YGL8 and ygl8 were predicted using SWISS-MODEL (https://swissmodel.expasy.org/) with the default parameters. The resulting pdb files of the corresponding proteins were compared and shown using PyMOL (DeLano, 2002).

Total RNA was extracted from the newest fully expanded leaves of Morex and the ygl8 mutant at 2 weeks in the hydroponic experiment or stem elongation stage in the field experiment using an RNAprep Pure Plant Kit (Tiangen Co. Ltd., China). First-strand cDNAs were synthesized using a Fasting RT Kit with gDNase (Tiangen Co. Ltd., China). qRT-PCR was performed using the Talent qPCR PreMix (Tiangen Co. Ltd., China), and the barley actin gene was used as the endogenous control. Reactions containing SYBR premix were carried out in final volumes of 20 ul containing 0.3 umol of the appropriate primers and 2 * PCR master mix. The 2^-△△t method was used to calculate the relative levels of gene expression. The primers used for qRT-PCR are listed in Supplementary Table 1 .

We isolated a yellow leaf mutant, ygl8, from the progeny of barley variety Morex seeds treated with EMS, and the newly emerged leaves of the ygl8 mutant were yellowish throughout the vegetative growth period ( Figures 1A, B ). To investigate whether the pigment contents changed in the yellow leaf of the ygl8 mutant, we detected the pigment contents of the newest fully expanded leaves of Morex and the ylg8 mutant grown in hydroponic solution for 3 weeks and observed significantly decreased pigment contents in the ygl8 mutant compared with Morex ( Figure 1D ). The chlorophyll a and b and carotenoid contents of the ygl8 mutant decreased 15%, 18%, and 29%, respectively, compared with those of Morex ( Figure 1D ). Additionally, we observed that the plant height of the ygl8 mutant decreased to 83% of that of Morex at the mature stage in the field ( Figures 1C, E ). These results suggest that pigment synthesis in the leaves of the ygl8 mutant was dramatically inhibited.

For the rapid identification of the gene responsible for the yellow leaf phenotype in the ygl8 mutant, the bulked DNA pool, consisting of 30 randomly selected ygl8 individuals with a yellowish leaf phenotype, and Morex were subjected to whole-genome sequencing for SNP detection. In total, we obtained 1,397,745 M and 1,002,230 M clean reads for the Morex and ylg8 bulked pools, respectively ( Supplementary Table 2 ). These reads were aligned to the Morex reference genome (Mascher et al., 2021) and 953,254 SNPs were detected.

For each SNP position, the value of the SNP-index was derived, and the ΔSNP-index values (the difference of the SNP-index of Morex and the bulked ygl8 pool) were plotted against each SNP position on the seven chromosomes ( Supplementary Figure 1 ). Furthermore, 4,312 SNP positions with an ΔSNP-index = 1 were identified on seven chromosomes and functionally annotated according to genetic variant types. Then, we observed 130 SNP positions with an ΔSNP-index = 1 in genic regions, including 2 kb upstream and downstream, and the related 129 genes were focused on as candidate genes for the yellow leaf phenotype in the ygl8 mutant ( Supplementary Table 3 ). Moreover, we compared gene expression in the leaf tissues of Morex and the ygl8 mutant at the stem elongation stage in the field using RNA-seq and found that 59 out of the 129 candidate genes were expressed in the leaf tissues ( Supplementary Table 3 ).

To alleviate the effects of large variations of single ΔSNP-index values, the sliding window average values of the ΔSNP-index were calculated, and we further narrowed down the candidate genomic region to the chromosome 3H: 604,680,000-622,110,000, comprehensively considering the genetic variant types, gene functional annotations, and gene expression ( Figure 2A ; Supplementary Table 3 ). There were seven genes with an ΔSNP-index =1 in this region but only four were expressed in our RNA-seq data ( Figure 2B ). Based on the genetic variant types and functional annotations, we finally focused on HORVU.MOREX.r3.3HG0325830 (named YGL8), which encoded one UMP kinase and contained one G to A single-nucleotide transition in the 5th exon/intron junction in the ygl8 mutant; the transition might lead to a splicing site change between exon 5 and 6 of YGL8 in the ygl8 mutant. Notably, the mutations in the Arabidopsis and rice orthologous YGL8 genes were previously reported to affect chloroplast development, and thus resulted in a low chlorophyll content and yellowish leaf phenotype (Hein et al., 2009; Zhu et al., 2016; Chen et al., 2018a; Dong et al., 2019). To confirm the splicing site change between exon 5 and 6 of the YGL8 gene in the ygl8 mutant, we investigated the read alignments around the YGL8 gene using the RNA-seq data of Morex and the ygl8 mutant and observed the existence of the splicing site change between exon 5 and 6 of YGL8 in the ygl8 mutant ( Figure 2C ). Furthermore, Sanger sequencing revealed that the splicing site change between exon 5 and 6 of YGL8 resulted in a 30-bp insertion in the 5th exon in the ygl8 mutant ( Figure 2C ).

To investigate whether the yellow leaf phenotype of the ygl8 mutant was controlled by a single gene, we crossed Morex and the ygl8 mutant to obtain F1 progeny, and the F1 plants were self-pollinated to generate F2 progeny. All F1 progeny exhibited the normal green leaf phenotype, and the leaf phenotype of the F2 plants segregated in a 3:1 ratio (150 green: 40 yellow), which indicated that the yellow leaf phenotype of the ygl8 mutant was conferred by a single recessive mutation. These results reveal that the yellow leaf phenotype of the ygl8 mutant might be the result of the splicing site change between exon 5 and 6 of YGL8 due to a G to A single-nucleotide transition in the 5th exon/intron junction.

To comprehensively understand the transcriptional expression patterns of YGL8, the RNA-seq data from different developmental tissues of Morex were derived (Mascher et al., 2017), and the analysis demonstrated that YGL8 was constitutively expressed, with a relatively higher expression level in photosynthetic tissues ( Figure 3A ). To determine the influences of the splicing site change between exon 5 and 6 on the transcriptional level of YGL8, we examined the transcripts of YGL8 in the 3-week-old leaf tissues of Morex and the ygl8 mutant grown in hydroponic solution, and no significant transcriptional changes were observed ( Figure 3B ).

To investigate whether the splicing site change between exon 5 and 6 of YGL8 affected the subcellular localization, the YGL8 or ygl8 fused with GFP (YGL8-GFP or ygl8-GFP) were transiently expressed in tobacco leaves. We observed strong GFP signals of YGL8 and ygl8 in the chloroplast, which coincided with the chloroplast marker signals of OsYSS1-mCherry (Zhou et al., 2017) ( Figure 3C ). These results demonstrate that the splicing site change between exon 5 and 6 of YGL8 has no effect on the gene transcript and chloroplast subcellular localization.

We compared the YGL8 orthologous proteins of different species and observed high similarity among them, especially in the UMP kinase domain ( Figure 4A ). Given that the splicing site change between exon 5 and 6 of YGL8 was located in the UMP kinase domain ( Figure 4A ), we assumed that the mutation might affect the protein structure of the conserved UMP kinase domain. To verify this hypothesis, we predicted and compared the protein structures of YGL8 with ygl8 and found that the splicing site change between exon 5 and 6 of YGL8 resulted in an additional loop in the UMP kinase domain structure in the ygl8 protein ( Figure 4B ). Furthermore, the ten key amino acids, previously reported to be important for interactions with the substrates UMP or UTP (Hein et al., 2009), were examined, and we only observed slight conformational changes for five of the ten amino acids in the predicted ylg8 protein structure ( Figure 4C ; Supplementary Figure 3 ). These results suggest that the newly generated loop in ygl8 protein did not significantly affect the UPM domain structure, whereas the additional loop might disturb the access of the substrates to the kinase domain, thus resulting in a full or partial malfunction of the YGL8 gene.

To clarify which genes are affected in terms of transcription levels due to the splicing site change between exon 5 and 6 of YGL8, we investigated the transcriptomes of the leaf tissues of Morex and the ygl8 mutant. Compared with Morex, we observed 2,401 significantly differentially expressed nuclear-encoded genes, including 817 upregulated genes and 1,584 downregulated genes ( Figure 5A ; Supplementary Table 4 ). The most significantly enriched GO terms included response to karrikin, chlorophyll biosynthetic process and saccharide biosynthetic process, and response to abiotic stresses such as cold stress and light intensity ( Figure 5D ; Supplementary Table 6 ). The most significantly enriched KEGG pathways included photosynthesis, carbon fixation and metabolism, plant hormone signal transduction and MARK signaling pathway, and linoleic acid metabolism ( Figure 5E ; Supplementary Table 7 ). These results suggest that photosynthesis and plant signaling transduction are affected by the splicing site change between exon 5 and 6 of the YGL8 gene in the ygl8 mutant.

Bioactive GAs are vital endogenous plant growth regulators, which affect stem and internode elongation partially by influencing the cell wall characteristics via regulating CesA genes (Gupta and Chakrabarty, 2013; Huang et al., 2015). Among the differentially expressed genes in the RNA-seq data from Morex and the ygl8 mutant, we observed that five homologous GA20ox genes and all seven homologous CesA genes were significantly downregulated in the ygl8 mutant ( Figures 5B, C ; Supplementary Table 8 ). Furthermore, we examined the gene expression in all 24 homologous GA20ox genes in the leaf and internode tissues from a previous published paper (Mascher et al., 2017) and found that the differentially expressed GA20ox genes in our RNA-seq data had the highest expression levels of all the homologous members in the internode tissues ( Supplementary Figure 4 ; Supplementary Table 9 ). These results demonstrate that the splicing site change between exon 5 and 6 of YGL8 downregulates the expression of the GA20ox and CesA genes in the ygl8 mutant and thus leads to the dwarf phenotype.

Many genes responsible for leaf color variations have been reported to be temperature sensitive (Chen et al., 2013; Shi et al., 2015; Liu et al., 2020). To address whether the yellow leaf phenotype of the ygl8 mutant was dependent on the temperature conditions, we grew Morex and the ylg8 mutant in a hydroponic solution under different temperature conditions. Although a more yellowish leaf of the ygl8 mutant was observed at a low temperature (10°C) than at a normal culture temperature (24°C), the newly emerged leaf of the ygl8 mutant was green at a high temperature (30°C) ( Figure 6A ), which was consistent with the pigment content of the ygl8 mutant ( Figure 6B ). Furthermore, we examined the transcript levels of the GA20ox and CesA genes, which showed differential expression in our RNA-seq data, in the leaf tissues of Morex and the ygl8 mutant under different temperature conditions. The downregulation of CesA genes in the ygl8 mutant was only observed at a low temperature and no significant transcript changes were observed at normal or high temperatures ( Figure 6D ). Interestingly, we did not observe obvious transcript changes of GA20ox genes under different temperature conditions in the ygl8 mutant compared with Morex ( Figure 6C ), inconsistent with our RNA-Seq data ( Figure 5B ), which might be caused by different developmental sampling stages. Additionally, different conclusions about affected genes have been obtained in the rice ygl8 mutant for different sampling timings (Zhu et al., 2016; Chen et al., 2018a; Dong et al., 2019).

Given that the disruption of chloroplast-localized UMP kinase affected chloroplast-encoded gene expression and resulted in abnormal chloroplast development, subsequently resulting in a chlorophyll-deficient leaf phenotype in Arabidopsis and rice (Hein et al., 2009; Zhu et al., 2016; Chen et al., 2018a; Dong et al., 2019), we examined the chloroplast-encoded and chlorophyll synthetic genes in the ygl8 mutant under different temperature conditions. We found that the chloroplast thylakoid membrane major component encoding genes, including photosystem complex subunit encoding genes (psaA, psaB, psaI, psaJ, psbE, psbF, psbI, and psbL) (Nelson and Junge, 2015), ATP synthase subunit encoding genes (atpB and atpE) (Dobrogojski et al., 2020), and the cytochrome b₆/f subunit encoding gene petA (Dobrogojski et al., 2020), showed similar expression patterns in the ygl8 mutant to those in Morex, with downregulation at low temperatures and no significant changes at normal or high temperatures ( Figure 6E ). Similarly, several chlorophyll synthetic genes were only downregulated in the ygl8 mutant compared with Morex at low temperatures ( Figure 6F ). These results suggest that the yellow leaf phenotype of the ygl8 mutant is temperature dependent.