Identification of QTL for barley grain size

Introduction

Barley (Hordeum vulgare L.) is considered to be the first crop domesticated by human ancestors back to 10,000 years ago (Zohary, Hopf & Weiss, 2012). It is not only an important crop model for studies on genetics, biochemistry and biology development (Giraldo et al., 2019), but is also widely used for animal feeding, malting and brewing, as well as human food (Zhou, 2010; Walker & Panozzo, 2016; Wendt et al., 2016). Barley grain yield is determined by various yield components (Benbelkacem, Mekni & Rasmusson, 1984; Hadjichristodoulou, 1990; Peltonen-Sainio et al., 2007). The sink capacity, a key to grain yield, is a function of the number of grains per unit land area and their potential size (Bingham et al., 2007). Barley grain size correlates with not only barley yield (Xu et al., 2018) but also grain texture, such as malting quality, starch and protein content (Walker et al., 2013; Yu et al., 2017), is thus an important end-use quality parameter (Holopainen et al., 2005; Psota et al., 2007; Mayolle et al., 2012). Enhancing grain size is, therefore, one of efficient ways of increasing yield and improving end-use efficiency of barley.

Grain size in rice has attracted more attention than that in barley, with many genes/QTL determining grain sizes having been identified. These include grain size 5, GS5, which encoding a putative serine carboxypeptidase (Li et al., 2011), qGL3 and 1GW2a (Sun et al., 2013), GL3.1, a protein from phosphatase kelch (PPKL) family which influences protein phosphorylation in the spikelet and accelerates cell division (Qi et al., 2012), and GS3 which are composed of four domains that function differently in determining grain sizes (Fan et al., 2006; Mao et al., 2010). In comparison to rice, there are only limited reports on QTL for barley grain size. From a DH population of a cross between six-rowed and two-rowed barleys, eight QTL were identified for grain length (GL) and nine for grain width (GW) (Wang et al., 2019). Among these QTL, the vrs1(HORVU2Hr1G092290) was annotated to be the candidate gene for the major QTL on 2H as a homeobox-Leucine Zipper Protein which controls row type and nud (HORVU7Hr1G089930) that determines the hulled/naked caryopsis morphology is more likely the candidate gene for chromosome 7H QTL hotspot region underlying grain size, though this candidate gene was annotated an ethylene-responsive transcription factor (Wang et al., 2019). Five QTL were identified from crosses between a wild barley and two cultivated barley cultivars with two major ones on 3H (20.7–25.6 cM) and 4H (4H 62.9–69.9 cM), respectively (Zhou et al., 2016). Eleven grain length QTL were identified from a cross between two Australian malting varieties, Vlamingh and Buloke (Walker et al., 2013) but only two major QTL for grain length (qGL2H and qGL5H) were identified from the same cross in a different study (Watt et al., 2019). qGL5H has been fine mapped to a 1.7 Mb interval comprising of 23 candidate genes related to grain sizes (Watt et al., 2019). By employing the same material, qGL2H was fine mapped to a 140.9 Kb interval and a gene that encodes a MYB transcription factor is likely the candidate for this QTL (Watt et al., 2020). Genome-wide association studies (GWAS) have also been successfully used in mapping grain size related QTL in sorghum (Tao et al., 2020), rice (Ma et al., 2019) and Wild Wheat Aegilops tauschii (Arora et al., 2017).

In the current study, 188 doubled-haploid (DH) lines derived from the cross between a Japanese malting barley (Naso Nijo) and a Chinese feed barley (TX9425) were grown in eight different environments (three different sites × multiple years). Seven significant QTL for grain size were identified on 1H, 2H, 3H and 5H, respectively. One of the QTL (QGl.NaTx-2H) was located at a similar position to a previously reported malt extract QTL (Wang et al., 2015). Given that malt extract QTL were once reported to corelate with grain size (Walker et al., 2013), we analysed the QTL for grain size using malt extract as covariates. Our results suggested a close linkage between malt extract and grain size, but these two traits are controlled by different genes, which was confirmed by various near iso-genic lines, adding the novelty to our findings. Overall, this study identified new QTL for grain sizes and investigated the relationship between grain size and major quality traits.

Materials and Methods

Results

QTL analysis for grain length (GL)

Four significant grain length QTL (QGl.NaTx-1H, QGl.NaTx-2H, QGl.NaTx-3H and QGl.NaTx-5H) were identified using MET-BLUP data from all different environments (Fig. 1, Table 3). These QTL explained 6.8–29.8% of the phenotypic variation. The total phenotypic variation explained by these six QTL was about 70%. Naso Nijo alleles of QGl.NaTx-1H and QGl.NaTx-3H contributed to longer grain while for QGl.NaTx-2H and QGl.NaTx-5H, TX9425 alleles contributed to longer grain (Table 3). The most significant QTL, QGl.NaTx-2H, was identified on 2H with the closest marker of 3256205S2. This QTL explained 29.8% of the phenotypic variation with a LOD value of 21.95 (Fig. 1, Table 3). Another major QTL QGl.NaTx-3H.1 with the closest marker of 6283018S3 explained 21.9% of the phenotypic variation. These two major QTL were less affected by growing conditions and were detected in most of the trials (Table S2). The other QTL showed a significant interaction with growing conditions and only detected in some environments, four of eight for the QTL on 1H and two of eight for the QTL on 5H (Table S2).

Table 3:

QTL for barley grain size traits in the DH population of Naso Nijo × TX9425.

Trait	Linkage group	QTL name	Nearest marker	Position (cM)	Two LOD support intervals	LOD	R2 (%)	Source of positive effect	Additive effect	Malt extract as covariate	Uzu Gene as covariate
GL	1H	QGl.NaTx-1H	3255878S1	66.29	54.98–72.38	10.64	11.9	NN	0.073	NC	NC
	2H	QGl.NaTx-2H	3256205S2	10.02	9.56–15.44	21.95	29.8	TX	−0.111	23.0	NC
	3H	QGl.NaTx-3H	6283018S3	58.43	57.51–60.84	17.46	21.9	NN	0.259	NC	11.6
	5H	QGl.NaTx-5H	3264393S5	47.19	40.34–47.01	6.25	6.2	TX	0.044	NC	NC
GW	1H	QGw.NaTx-1H	4170979D1	65.7	54.98–66.00	4.13	9.5	NN	0.02	NC	NC
	2H	QGw.NaTx-2H	5258068D2	15.44	9.24–18.91	7.69	18.5	TX	−0.025	15.7	NC
	5H	QGw.NaTx-5H	3430425D5	44.24	29.10–59.58	3.73	8.5	NN	0.018	NC	NC

DOI: 10.7717/peerj.11287/table-3

Notes:

The position is that of the nearest marker; R² means percentage genetic variance explained by the nearest marker; Two LOD support intervals were used to indicate the 95% confidence intervals (van Ooijen, 1992); NC means no significant changes.

Discussion

Barley grain sizes positively associated with starch contents which contributes to process performance in human food, animal feed and brewing (Yu et al., 2017). Improving grain size is one of the objectives in breeding programs for not only improved quality but also high yielding (Serrago et al., 2013). Grain size is a quantitative trait controlled by multiple genes (Zhou et al., 2016) and can also be affected by the environment (Walker et al., 2013). The identification of QTL and molecular markers linked to grain size is essential for breeders to pyramid different QTL through marker assisted selection. Only a limited number of QTL for grain size have been identified with some larger effect ones on 2H, 3H, 4H and 5H. In this study, we have identified four QTL for GL, and three QTL for GW. Among the four identified significant QTL for GL, QGl.NaTX-2H and QGl.NaTx-3H were less affected by environments and determined a large percentage (29.8% and 21.9%, respectively) of phenotypic variation (Table 3). A major QTL for GL have been reported on 2H from the cross of Vlaminh and Buloke which is located at 70-80 cM in one report (Walker et al., 2013) but at 159–179 cM in another report (Watt et al., 2019) which is further fine mapped to a 140.9 Kb interval (Watt et al., 2020). The QTL are apparently different from our major QTL on 2H for GL which was located at 10.02 cM (Table 3). The 2H QTL identified in this study determined nearly 30% of the phenotypic variation with the closest marker of 32562045S2 at the position of 10.02 cM. At a similar position (16.3–17.5 cM), a QTL was also found from a cross between a long-kernel wild barley and cultivated barley cultivars but only determined a small proportion (10.4%) of the phenotypic variation (Zhou et al., 2016).

The other major GL QTL identified on 3H (QGl.NaTx-3H) in this study was in a similar position to uzu gene which controls plant dwarfness and has a pleiotropic effect on spike morphology (Chen et al., 2016). QTL analysis using uzu gene as a covariate indicated that this QTL was not the same but closely linked to uzu (Table 3). This QTL has also been reported in previous studies (Ayoub et al., 2002; Zhou et al., 2016) from populations with no uzu gene, confirming that the QTL was not due to a pleiotropic effect of uzu. The major QTL for grain size on 2H was not reported before, and the analysis using malt extract as covariate only suggested the linkage of malt extract and grain size, belonging to separate genes/traits that can be selected independently.

QTL alleles determining seed size also tend to determine malt quality. QTL alleles leading to increased variability of kernel size were associated with poor malt quality (Ayoub et al., 2002). In our study, the QTL on 2H for GL (QGl.NaTx-2H) is located at a similar position to a previously reported QTL for malt extract (QMe.NaTx-2H) (Wang et al., 2015). To investigate whether these two QTL are the same, we further applied QTL analysis for GL using malt extract as a covariate. Results suggested that these two QTL are independent, instead of a single one gene with pleotropic effect. To further confirm this, we checked several pairs of near isogenic lines (NILs) differing in malt extract QTL. No significant differences were found between lines with high malt extract and those with low malt extract (Fig. 2).

Three barley ortholog genes were found within the identified QTL regions in this study through protein sequence alignment to the cloned grain size genes in rice. Within QGl.NaTx-1H, HRVU.MOREX.r2.1HG0042890 is an ortholog to OsGS5, encoding a serine carboxypeptidase and functions as a positive regulator of grain size (Li et al., 2011). Another ortholog gene (HORVU.MOREX.r2.1HG0040860) of OsMAPK6 was also identified within this QTL region, encoding mitogen-activated protein kinase 6, determining rice grain size (Liu et al., 2015). SMALL GRAIN 1 is another mitogen-activated protein kinases identified in rice, involving regulating rice grain sizes, its homolog gene HORVU.MOREX.r2.5HG0381450 was identified within QGl.NaTx-5H and encode cytochrome, a protein involving cell wall elongation in barley (Table S1). Several candidate genes linked to cell growth and phytohormones also exist in the GL QTL region. ABC transporters play critical roles in plant growth and development, especially for the development of specialized plant cells (Do, Martinoia & Lee, 2018) and regulation of root cell growth (Larsen et al., 2007). MYB transcription factor is also proposed to be the candidate for GL (Watt et al., 2020). It is linked with cell growth and seed production through interacting with plant hormones, playing roles in sperm-cell, stamen development, cotton fibre and even stomatal cell divisions (Lai et al., 2005; Rotman et al., 2005; Pu et al., 2008; Zhang et al., 2010). Cytochrome P450 gene, which belongs to CYP78A subfamily, was reported to have a function in seed and fruit size development (Tian et al., 2016). Zinc finger proteins could also be good candidates for seed growth as they not only play a strong role in regulating cell growth, but also are critical for chloroplast and palisade cell development, thus affect seed filling and change seed size (Næsted et al., 2004). Manipulating ethylene signalling also indicates evidence to improve yield-related traits in crops. Overexpression of an ethylene response element MHZJ, a membrane protein, promoted grain sizes in rice (Ma et al., 2013). Similar findings were also observed in a wheat study, where overexpression of the transcriptional repressor (TaERF3, ethylene response factor), vice versa, decreased grain size and affected 1000-grain weight (Wang et al., 2020). IAA and gibberellin play critical roles in regulating seeds size, such that IAA-glucose hydrolase gene TWG5 determines grain length and yield (Ishimaru et al., 2013) and the identified quantitative locus GW6 controls rice grain size and yield through the gibberellin pathway (Shi et al., 2020). Based on these reported genes and their functions in determining grain sizes, we highlighted 66 genes involving the discussed functions in accordance with the identified QTL from this population. Most of those candidate genes are located outside the area for malt extract (Fig. 1).

Conclusions

In this study, seven major QTL for grain size were identified. The major one on 2 H (QGl.NaTx-2H) is closely linked to the reported QTL for malt extract (QMe.NaTx-2H, (Wang et al., 2015). The other major QTL on 3H for GL (QGl.NaTx-3H) shares a similar position with a reported dwarf gene, uzu (Chen et al., 2016), but they are two independent genes and control different phenotypes. Therefore, these major QTL can be used in breeding program to improve grain size, independent of malting quality and plant height.

Supplemental Information