Grain colour is one of the stable taxonomic characters used for the description and classification of barley (Hordeum vulgare L., 2n = 2× = 14, HH) accessions. Barley grain at maturity may have white, yellow, purple, blue, or black pigmentation. In white grains, the pigments are absent, whereas two different classes of compounds determine the other types of colours. Yellow, purple, and blue are caused by a diverse subgroup of flavonoid compounds, synthesized in different grain tissues [1, 2]. Unlike the pigmentations caused by flavonoids, studies of black coloration are very scant from biochemical and molecular genetic points of view.

Black coloration is caused by ‘melanin-like’ pigments, which are groups of high molecular weight irregular polymers, arising in the course of oxidation and polymerization of phenolic compounds [3]. Because of their complex structure and resistance to almost all solvents, their chemical nature and the metabolic pathways leading to these black pigments have not yet been determined [4].

In barley, the ‘melanin-like’ pigments accumulate in the lemma and pericarp of grains during the later stages of growth, slightly before maturation of the spike [1, 5]. This trait is controlled by the monogenically inherited Blp1 (black lemma and pericarp 1) locus that has been mapped to chromosome 1HL [6, 7]. Three dominant alleles of the Blp1 locus conferring intensity of grain pigmentation have been reported: Blp1.b (B), Blp1.mb (B ^mb ), and Blp1.g (B ^g ) determine extreme black, medium black, and light black or grey colours, respectively [5, 8, 9].

Barley landraces with black grains were found in southwestern Asia, and sparsely in Tibet and China [10]. In Syria, the black-grained locally adopted landrace ‘Arabi Aswad’ was grown on approximately 75% of the barley growing area, particularly in the driest part of the country, unlike the white-seeded ‘Arabi Abiad’ landrace which was grown in more favourable conditions [11–13]. Compared to white-seeded varieties, black-seeded barley types are considered to be more drought tolerant, more vigorous in early stages of growth, more cold-tolerant, more prostrate, taller and faster-maturing [11, 12].

In addition to better viability under abiotic stress conditions, black-grained barley demonstrate higher resistance to fungal infection. FHB screening showed that among black-grained accessions, 20% were resistant, while among non-coloured accessions only 5% of genotypes were found to be resistant [14, 15]. Barley recombinant inbred lines (RILs) with black grains demonstrated lower Fusarium head blight incidence and lower accumulation of mycotoxin deoxynivalenol than yellow RILs, at least at some disease pressure levels [16]. Increased resistance to Fusarium disease has also been reported in oat genotypes having dark-coloured floral glumes in comparison with light-coloured ones [17].

Some studies of the phytochemical composition of barley grains with black colour have revealed that real ‘melanin-like’ black pigments may be present simultaneously with anthocyanins and other related co-pigments that contribute to the total phenol content [18, 19].

Comparative analysis of phenolic profiles and transcription of flavonoid biosynthesis pathway genes in black- and white-grained barley near-isogenic lines (NILs) demonstrated that none of the key flavonoid biosynthesis genes were upregulated in the black NIL, whereas in the presence of a dominant Blp1 allele, the total amount of phenolic compounds was enhanced in the black-grained line in comparison with the white-grained one [20]. These data suggest that neither coloured (proanthocyanidins, anthocyanins, phlobaphenes) nor uncoloured flavonoids participate in ‘melanin-like’ pigment biosynthesis, and instead, other classes of phenolic compounds are involved.

In the current study, to reveal genetic networks (and hence metabolic pathways) participating in formation of the ‘melanin-like’ black pigments, differentially expressed genes (DEGs) in black- vs white-grained NILs were determined by RNA-seq.

Among all genome fragments, 22,976 were identified as having non-zero expression levels. The fragments, their locations, annotations and respective RPKM values are listed in Additional file 1 (Additional file 1, Table ‘Genes’). These fragments include genes listed in genome assembly annotations as well as genome fragments for which expression has been detected by the Cufflinks pipeline and are further referred to as ‘transcripts’. RPKM values for the transcripts were used for library clustering (Fig. 1). Libraries BLP (2) and BLP (3) are highly similar in terms of gene expression, as are the BW (2) and BW (3) libraries. BLP (1) differs somewhat from the other two libraries belonging to the BLP line. Library BW (1) is further separated from the two other libraries belonging to the Bowman line (Fig. 1). Nevertheless, two clusters can be distinguished on the map, each containing all libraries of one of the two biological lines.

EdgeR detected 648 genes with lower and 1155 genes with higher expression levels in line BLP (Additional file 1, Table ‘EdgeR’), whereas Cufflinks detected 567 and 938 genes with lower and higher levels of expression in line BLP, respectively (Additional file 1, Table ‘Cufflinks’). In the list of genes mutually identified by both tools, 325 genes with decreased and 632 with increased expression levels in BLP were detected (Fig. 2; Additional file 1, Tables ‘UpReg DEGs Cufflinks & EdgeR’, ‘DownReg DEGs Cufflinks & EdgeR’). Thus, 957 transcripts with differential expression were detected in total, comprising 4.1% of all transcripts expressed in either of the biological lines.

PlantCyc database (BarleyCyc division) annotations include 196 differentially expressed genes, identified by Cufflinks and EdgeR, as participating in known BarleyCyc pathways (Additional file 2). The five pathways with the highest DEG PPS scores are ‘superpathway of pyrimidine deoxyribonucleotides de novo biosynthesis’, ‘UTP and CTP dephosphorylation II’, ‘flavonoid biosynthesis (in equisetum)’, ‘flavonol biosynthesis’ and ‘UTP and CTP dephosphorylation I’. Pathways with high numbers of DEGs include ‘suberin monomer biosynthesis’, ‘cutin biosynthesis’, ‘phenylpropanoid biosynthesis, initial reactions’, ‘superpathway of scopolin and esculin biosynthesis’ and ‘eugenol and isoeugenol biosynthesis’. Among additional pathways potentially related to the differing phenotypes of the NILs were ‘cuticular wax biosynthesis’ and ‘diterpene phytoalexins precursors biosynthesis’.

Using Blastp, homologies among Affymetrix Barley Genome Array sequences were found for 508 upregulated and 270 downregulated transcripts in BLP. Gene Ontology enrichment analysis showed that the GO term ‘fatty acid biosynthesis process’ is significantly (FDR = 7,5·10⁻⁴) enriched in transcripts with higher expression in BLP (Additional file 3), while cellular component GO terms ‘nucleoplasm’ (FDR = 0,03) and ‘chloroplast thylakoid’ (FDR = 0,029) and molecular function GO terms ‘transporter activity’ (FDR = 3,4·10⁻³) and ‘transcription factor activity’ (FDR = 5,5·10⁻³) were significantly overrepresented among transcripts with lower expression in line BLP (Additional files 4-5).

qPCR differential expression verification

qRT-PCR was used to evaluate the transcription of eight genes in seeds of the BLP and Bowman NILs, and results are summarized in Fig. 3. Six genes with higher expression levels in BLP and two genes with higher expression levels in the Bowman line were tested to verify the RNA-seq results.

Among the genes from the RNA-seq data showing increased expression levels in BLP are MLOC_62322 encoding phenylalanine ammonia-lyase (PAL), MLOC_76018 encoding Caffeic acid O-methyltransferase, MLOC_55029 encoding lipoxygenase B, MLOC_51393 encoding isoflavone reductase, MLOC_3802 encoding universal stress protein and MLOC_80301 encoding polyphenol oxidase. We also experimentally verified two genes from the RNA-seq data showing increased expression in the Bowman line: MLOC_54913 encoding lys-63-specific deubiquitinase and MLOC_16300 encoding apyrase 3. The log₂(FC) values, p-values from RNA-seq data corrected with the Benjamini-Hochberg procedure and the qRT-PCR experimental fold changes and associated p-values for verified genes are shown in Table 2. The obtained experimental data are consistent with the RNA-seq predictions.

Gene	Product	log2(FC)	p-value (RNA-seq)	Experimental expression fold change	p-value (experimental)
MLOC_62322	Phenylalanine ammonia-lyase	-3,97	1,1·10−7	5,05	0,0095
MLOC_76018	Caffeic acid O-methyltransferase	−8,81	2·10−40	27,31	0,0059
MLOC_55029	Lipoxygenase B	−4,14	2,4·10−21	5,06	0,0031
MLOC_51393	Isoflavone reductase	−10	1,9·10−4	5,15	0,0578
MLOC_3802	Universal Stress Protein	−6,46	6,8·10−39	45,08	0,0068
MLOC_80301	Polyphenol Oxydase	−4,61	8,2·10−31	26,07	0,0018
MLOC_54913	Lys-63-specific deubiquitinase	6,45	4,5·10−2	2,39	0,0423
MLOC_16300	Apyrase 3	11,95	2,4·10−14	43,33	0,0064

The diversity of metabolic pathways up- and down-regulated in the presence of Blp (Blp1) suggests a pleiotropic nature of this gene. This finding is in agreement with previous observations at the phenotypic and physiological levels, suggesting that in addition to conferring black lemma and pericarp colour, the presence of the dominant Blp allele may contribute to total antioxidant level in barley grains (Additional file 6), abiotic stress tolerance [11–13] and disease resistance [14–16]. Investigation of DEG functions allowed us to identify genes and metabolic pathways presumably related to known phenotypic and physiological effects of the Blp locus.

Black ‘melanin-like’ pigments in plants have complex structures and can withstand almost all solvents of different chemical natures [4]. It has been proposed that these pigments are high molecular weight products of oxidation and polymerization of phenolic compounds [3, 21]. Indeed, an increase of total phenolic compounds in BLP in comparison to BW has been observed [20]. Here, we report significantly increased levels of PAL mRNA in the BLP line (Table 2, Fig. 3). PAL encodes phenylalanine ammonia-lyase, the key enzyme of the phenylpropanoid pathway, underlying biosynthesis of phenolic compounds in plants. Nevertheless, these quantitative changes in phenolic metabolites and PAL transcripts do not explain the qualitative differences between genotypes carrying dominant and recessive Blp alleles. Investigating additional DEGs, we propose that black pigmentation may be associated with high expression level of polyphenol oxidase (PPO) in BLP (Table 2, Fig. 3). PPO catalyses the oxidation of phenols to quinones. The products of the polymerization of quinones cause the darkening of some vegetables and fruits upon contact with air [22]. In humans, a similar enzyme, tyrosinase, is involved in the synthesis of melanin [23].

In wheat, black-spiked genotypes carry dominant allele Bg (Rg-A1c) on chromosome 1A, however on the short arm [24, 25], so barley Blp and wheat Bg are unlikely to be orthologues. The origin of black glume colour in wheat remains unknown, although it has been suggested to be related to phlobaphenes, a class of flavonoid compounds [26]. In barley, however, we did not observe phlobaphene biosynthesis structural genes among the identified DEGs.

Higher resistance of black grains to oxidative stress may also associated with increased phenolic content, since many phenolic compounds are known to be antioxidants [27]. Among the DEGs was a gene related to phenolic metabolism, which is of particular interest in relation to mechanisms increasing the antioxidant capacity of grains in the BLP line. This gene, with increased mRNA level in BLP, encodes Caffeic acid O-methyltransferase (Table 2, Fig. 3), which converts caffeic acid into ferulic acid [28]. Ferulic acid is a ubiquitous plant constituent that arises from the metabolism of phenylalanine and tyrosine. Ferulic acid, which accumulates primarily in seeds and leaves (both in its free form as well as covalently linked to lignin or other biopolymers), is known for its ability to terminate free radical chain reactions and hence serves an important antioxidant function to preserve the physiological integrity of cells in unfavourable conditions [29].

Higher resistance of black grains to oxidative stress may partially explain earlier findings that black seeded types are more drought tolerant, more vigorous in early stages of growth and more cold-tolerant than white seeded types [11]. Higher drought tolerance may also be related to cuticular wax biosynthesis, as revealed in our study as one of the metabolic pathways induced in the presence of Blp.

Antioxidant capacity can contribute to higher biotic stress resistance.

Black grain colour is associated with higher resistance to Fusarium in barley [14–16] and oat [17]. Fusarium head blight is a fungal disease favoured by humid conditions during flowering and early stages of kernel development. Due to better heat accumulation and accelerated drying, black-grained cereals may have an advantage. In addition to this possible effect of black pigmentation, the elevated resistance of dominant Blp allele carriers to Fusarium graminearum may also be explained by intensive production of phenolic compounds. Gunnaiah et al. applied an integrated metabolo-proteomic approach to decipher the mechanisms by which a wheat QTL (Fhb1) contributes to resistance against F. graminearum [30]. Fhb1 is not related to Blp or black grain pigmentation. However, it confers resistance to F. graminearum by inducing the phenylpropanoid pathway, leading to secondary cell wall thickening in rachises of wheat NILs with a resistant Fhb1 allele. Finally, increased transcriptional activity of the PAL gene can affect the biosynthesis of suberin and phytoalexins. Suberin, a waterproof waxy substance found in higher plants, can benefit black-grained cereals in humid conditions by preventing F. graminearum development, while phytoalexins (a wide class of plant molecules known for their antimicrobial properties) can act as toxins for the fungi. In addition to the PAL gene, we identified another DEG important for phenolic phytoalexin synthesis, encoding Isoflavone reductase (IFR) (Table 2, Fig. 3). Furthermore, DEGs necessary for the synthesis of terpenoid phytoalexins were identified (Additional file 2; ‘diterpene phytoalexins precursors biosynthesis’).

Improved resistance to F. graminearum in Blp carriers can also be explained by cuticular wax formation. Cuticular wax biosynthesis was revealed in our study as a metabolic pathway induced in the presence of Blp. Another mechanism of resistance can putatively be related to increased expression of the gene encoding universal stress protein (USP), which is usually expressed under environmental stress and stimulates further stress tolerance [31]. The BLP line has an increased USP level in optimal conditions, so it may potentially have increased tolerance to environmental stresses even in the absence of preliminary stimulation.

Metabolic pathways and genes presumably associated with black lemma and pericarp colour as well as with the pleiotropic effects of the Blp locus (i.e., resistance to oxidative stress and pathogens) were revealed. We suggest that black pigmentation of lemmas and pericarps may be related to increased level of phenolic compounds and their oxidation. The effect of Blp presence on the synthesis of ferulic acid and other phenolic compounds may be one of the explanations for the increased antioxidant capacity and biotic and abiotic stress tolerance of black-grained cereals. Their drought tolerance and resistance to diseases affecting the spike may also be related to cuticular wax biosynthesis. In addition, upregulated synthesis of phytoalexins, suberin and universal stress protein (USP) in lemmas and pericarps of the Blp carriers may also contribute to their increased disease resistance. Further description of the DEGs haplotypes and study of their association with physiological characteristics may be useful for future application in barley pre-breeding.