QTL mapping for benzoxazinoid content, preharvest sprouting, α-amylase activity, and leaf rust resistance in rye (Secale cereale L.)

Mapping population of recombinant inbred lines (RILs) representing 541 × Ot1-3 cross exhibited wide variations of benzoxazinoid (BX) content in leaves and roots, brown rust resistance, α-amylase activity in the grain, and resistance to preharvest sprouting. QTL mapping of major BX species using a DArT-based map revealed a complex genetic architecture underlying the production of these main secondary metabolites engaged in stress and allelopathy responses. The synthesis of BX in leaves and roots was found to be regulated by different QTL. The QTL for the BX content, rust resistance, α-amylase activity, and preharvest sprouting partially overlapped; this points to their common genetic regulation by a definite subset of genes. Only one QTL for BX located on chromosome 7R coincided with the loci of the ScBx genes, which were mapped as two clusters on chromosomes 5RS (Bx3-Bx5) and 7R (Bx1-Bx2). The QTL common for several BX species, rust resistance, preharvest sprouting, and α-amylase activity are interesting objects for further exploration aimed at developing common markers for these important agronomic traits.


Introduction
The most abundant secondary metabolites produced in rye leaves and roots are cyclic hydroxamic acids and their derivatives benzoxazinoids (BXs) playing various physiological roles such as defense against pests, allelopathic agents, inhibition of germination, and detoxification [1][2][3][4]. Rye contains the highest amounts of these compounds in the grain and green tissues, and there is a wide variation of the BX content among cultivars and breeding materials [4][5][6][7]. This points to a possibility for initiating breeding strategies aimed at increasing the BX content in rye cultivars, which then may be treated as a source of allelopathic compounds against many weed species [4]. Although the quantities of BX in rye grain were extensively studied, a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Statistical methods. Observations of the BX content were transformed by log 10 (x+1). The analysis of variance (ANOVA) was performed in the mixed model with years and locations as fixed effects, and lines (RILs) as random effects. Variance components were compared to (3 × std. err.) to assess their magnitude. Broad sense heritability was calculated according to Cullis et al. [15]. Correlation coefficient was calculated on average data over years and locations. Computations were performed using Genstat 17 [16].
QTL mapping. QTL mapping was performed with the software package WinQTL Cartographer 2.5 [17] using the procedure for composite interval mapping (CIM), with the threshold value of LOD set at a score at 3.0. QTL with an LOD value between 2.5 and 3.0 was considered putative.

QTL for benzoxazinoid (BX) content
Differences in BX quantities between the parental lines were low (although statistically significant) and unstable in consecutive years (Table 1). Variation ranges found among RILs were much larger than those found in the parental generation in both years (from about 1.5 to 47 times), which suggests a substantial genetic transgression component in the observed variation. The prevailing BX species in rye leaves were DIBOA and GDIBOA. Lower amounts of GDIMBOA and HBOA were found in this tissue. Roots contained higher amounts of GDIM-BOA, MBOA, and GDIBOA, and lower amounts of DIBOA, DIMBOA, and HBOA. Broad sense heritability of the BX species ranged between c. 30% and 40% with two exceptions  Table 2). But, the substantial lines × years interaction (Table 3) and apparent deviations from the normal distribution lowered the chances of finding stable QTL across different environments (Fig 1). A number of significant QTL for particular BX species were found (Tables 4 and 5, Fig 2). Depending on a year of study, 2 (2014) to 3 (2013) QTL were detected for HBOA in leaves, with one common on chromosome 6RL and three separate on chromosomes 1R, 4RS, and 7RS. QTL for HBOA in roots were found on chromosome 5R in 2013 and on chromosomes 5R (1) and 6R (2) in 2014. The QTL mapping of the GDIBOA amount in leaves revealed 1 QTL in 2014 and 4 QTL in 2013, located on the distal part of chromosome 1RS (1); distal part of chromosomes 2RS (1), 5R (1), and 7RL (2) coinciding with the locus for the ScBx1-2 genes. No QTL for GDIBOA in roots were found in 2013, but in 2014, there were 3 detected QTL on chromosomes 3RS, 6RS, and 6RL. QTL for DIBOA in leaves were dispersed on chromosomes 3R and 7RL (adjacent to the ScBx1 locus) in 2013, whereas in 2014, they were detected on chromosomes 1R, 6R, and 7R. The QTL for DIBOA in roots was found on chromosomes 3R, 6RS, and 7RL in 2013 and on chromosomes 1RL, 2RL, 4RL, and 6R (2) in 2014. Closely linked but not overlapping QTL on chromosomes 1R, 3R, and 6R were detected for DIBOA from both leaves and roots. The QTL for GDIMBOA from leaves were mapped on chromosomes 1R, 2RS, 5RS, and 7R in 2013 and only on 6R in 2014. The distribution of GDIMBOA QTL in leaves partially overlapped with that in roots where the QTL were located on chromosomes 1R, 2RS (coinciding QTL found in leaves and roots), and 6RS in 2014 and on chromosome 7R (2013) near the ScBx1 locus. DIM-BOA QTL were detected only in roots. They were located on chromosomes 3R and 6R in 2013 and on chromosomes 1RL and 6RS in 2014. MBOA QTL were not detected in leaves. MBOA QTL in roots were represented in 2013 by two loci on chromosomes 1R and 4RL and by one locus on chromosome 2RL in 2014. The R 2 value for most of the QTL for BX was close to 10%. The greatest impact on the BX content in leaves was exerted by GDIBOA_L/13_1 (R 2 = 34%) on chromosome 2R (2013), and in roots, by GDIMBOA_R/14_3 (R 2 = 39.3%) on chromosome 6R (2014). QTL were significant only in one year of the study (with one exception on 6RL), which illustrates their instability due to the significant environmental component of variation including lines versus years relationship.
Except for the QTL on the distal part of chromosome 2RS, there were no coinciding QTL for the same BX species in leaves and roots, which strongly supports the hypothesis of the differences in the regulation of BX production in these two plant tissues. In contrast, there are several coinciding QTL for different BX species like those for leaf GDIBOA, GDIMBOA, and HBOA on 5RS; leaf GDIBOA and GDIMBOA on 2RS; and leaf GDIBOA, GDIMBOA, and DIBOA on 7R. Similarly, coinciding QTL for different types of BX in roots were found on chromosomes 2RL (DIBOA and MBOA), 3R and 6RS (DIMBOA and DIBOA), 4RL (MBOA and DIBOA), and 6RS (GDIBOA and DIBOA). They may represent pleiotropic genes regulating the production of more than one BX species or clusters of tightly linked genes with specificity for a given species of BX. Coinciding QTL for BX in both leaves and roots may constitute the genetic background for the significant correlation of their amounts detected in this study (Table 6).
Structural loci encoding enzymes of the BX biosynthesis pathway were mapped on the distal part of chromosome 5RS (a cluster of the ScBx3, ScBx4, and ScBx5 genes) and on the central part of the 7R chromosome (a cluster of the ScBx1 and ScBx2 genes) (Fig 2). The QTL for BX in leaves and roots did not coincide with these structural genes except for the QTL for GDIM-BOA in roots, which was mapped near the ScBx1 and ScBx2 gene cluster on chromosome 7R.

QTL for resistance to preharvest sprouting (PHS)
In each experiment with one exception (Choryń 2014), differences between parental lines were substantial, showing that line Ot1-3 represents resistant and line 541 susceptible phenotypes ( Table 2). The RIL variation ranges were large in each experiment overcoming those of parental lines, which led to high coefficients of variation. But, because of a very high line × location and line × year interactions, the heritability of the trait calculated for the whole experiment was very low (2.18%) ( Table 7). Only four QTL for PHS were detected in 2013 and three others in 2014 (Tables 4 and 5, Fig 2). The highest determination coefficient has PHS_WPUT/ 13_2 QTL from the distal part of chromosome 6RS. The QTL for PHS were distributed on 1RS (1), 1RL (1), 2R (2), 5R (2), and 6RS (1).

QTL for α-amylase activity (AMY)
The grains of parental lines exhibited large differences in the α-amylase activity, and a large variation range was detected among RILs in each experiment ( Table 2). The environmental component of variation due to the line × location interaction was significant (Table 7). In contrast to the other traits, the heritability of the α-amylase activity proved to be very high (71.85%), and its distribution was close to normal (Fig 1). Among the seven QTL detected in 2013, the one on chromosome 6R was the most efficient (R 2 = 38.8%) ( Table 4, Fig 2). Five QTL detected in 2014 had similar low determination coefficients (c. 10%) ( Table 5, Fig 2). QTL were found on distal parts of chromosome 1RL, on the proximal part of

QTL for leaf rust resistance (RR)
Parental lines did not differ significantly with respect to rust resistance (Table 7). But, the variation ranges among RILs were substantial because of transgression effects, which assured that the genetic variation in a mapping population although not high (CV close to 10%) was sufficient for successful QTL mapping. In addition, the results of ANOVA showed that the environmental component of variation was low and the heritability was high (Table 4).
Chromosome 1R carried four QTL for leaf rust resistance. They were distributed on the distal part of the 1RS arm (2) and on the 1RL arm (2) (Tables 4 and 5, Fig 2). Three QTL for rust resistance were detected on chromosome 3R: the first on the short arm, the second in the centromeric region 3Rc, and the third on the long arm. Chromosome 4R contained one Lr QTL on the 4Rc. Two other QTL for rust resistance were in the 5Rc region. A majority of the RR QTL determined the trait variation in c. 10%, while the RR QTL on chromosome 5R (RR_WPUT/13_2) had a much higher impact (25.5%). Each of the RR QTL was detected only in one year and in one location, which shows that their effectiveness was highly dependent on environmental changes.

Coinciding QTLs for different traits
The most interesting was the coinciding of QTL for rust resistance affecting leaves and the αamylase activity in rye grain. Such cases were found on chromosomes 1RL and 3R (Fig 2), but they did not result in a significant correlation between these traits ( Table 6). The two coinciding QTL for PHS and rust resistance were detected on chromosomes 1RS and 5R (Tables 4  and 5, Fig 2). Except for the distal region on chromosome 1RL, the proximal region of 2RL, and the distal region of 6RS, there were no genomic locations with coinciding PHS and AMY QTL, which may partially confirm the lack of correlation between these two traits ( Table 6).
The QTL for BX in leaves or roots were found in several locations accompanied by AMY QTL as on chromosomes 1RL, 3R, 5R, 6RS, and 6RL. The most apparent coincidence for PHS and GDIMBOA QTL from roots was on the distal part of chromosome 6RS. A common position of QTL for the HBOA content in roots and the PHS QTL was also detected on chromosome 5R. The detected similarities between BX and AMY or PHS QTL distribution did not result in a significant correlation between these traits (Table 6). Coinciding QTL for different traits raise a question on the pleiotropic effects of the underlying genes. They also provide strong support for the validity of the QTL detection.

Discussion
Results of mapping the ScBx genes were in agreement with the earlier reports on their distribution in the contigs of the rye BAC library [13]. The ScBx3, ScBx4, and ScBx5 genes were mapped as a one gene cluster on the distal part of chromosome 5RS and ScBx1 with the ScBx2 gene as another cluster on chromosome 7R. This result clearly shows that both groups of genes are not linked to other BX genes: ScGT encoding UDP-glucosyltransferase and ScGlu encoding beta-glucosidase which were shown to be located on chromosomes 4R and 2R, respectively [18]. The map positions of the ScBx1-5 cluster on 5RS and the ScBx1-2 cluster on 7R, determined for the first time in this study, confirm the earlier results attributing the Bx genes to rye chromosomes [18,19]. The map position of Bx genes in the rye genome is thus not syntenious to that found in corn where they are all clustered on the distal part of the short arm of chromosome 4 [20]. The two clusters of the BX genes were also assigned to chromosomes in wheat: Bx1 and Bx2 to chromosome group 4 and Bx3, Bx4, and Bx5 to the short arm of the chromosomes of homoeologous group 5 [18]. Since a segment of chromosome 4L is located on rye chromosome 7RS because of ancient translocation [21], it may be inferred that the Bx1-2 cluster from wheat chromosome 4DL and that from rye chromosome 7R are located synteniously. The studied complex QTL system for BX in the rye genome constitutes a network of regulatory genes affecting the expression of the structural Bx genes encoding the enzymes of the BX biosynthesis pathway. The Bx genes located in two clusters on chromosomes 5RS and 7R have no or a slight effect on the variation of BX production. Only one QTL for GDIBOA_L with a determination coefficient of 16.5% and additive value of 34.5 coincided in one year with the Bx1-2 genes on chromosome 7R. A similar situation was found for the α-amylase activity in rye where none of the QTL detected coincided with the structural genes encoding the α-amylase isozymes playing a major role in the catabolism of the endosperm starch [22,23]. Allelic forms of the Bx structural genes have little effect on their expression. Instead, a high level of genetic variation in regulatory genes enables a fast and specific adjustment of enzyme synthesis in response to environmental conditions. Another observation supporting this hypothesis is that of the completely different sets of QTL regulating the production of a particular BX species in roots and leaves. These organ-specific regulatory networks probably reflect different defensive mechanisms against biotic and abiotic stresses acting upon roots and leaves or may result from the activation of different members of the multigene families in each tissue. A second possible explanation of the apparently small impact of polymorphisms in the Bx structural loci on the BX content, found in this study, is the lack of functional polymorphism in the Bx genes within the mapping population. This hypothesis is supported by earlier results of association mapping, which showed the presence of SNPs in the Bx genes associated with the BX content within a wide range of rye genotypes [7]. Similarly, the results reported by [8] and [9] support the significant role of polymorphisms within the Bx1 sequences in controlling the BX content in corn.
Results of the QTL mapping revealed that although the heritability of the BX content in rye is promising for breeding purposes, individual QTL are not stable across environments and years. This is a common conclusion from many QTL studies based on the whole mapping populations where phenotypic differences between lines are most often small and prone to environmental factors. This problem is overcome in bidirectional selective genotyping (BSG) studies where only extreme phenotypes, derived through divergent selection and much more independent of environmental changes, are analyzed [22][23][24]. These studies show that the individual effects of QTL are usually low, but their impact on the trait due to their interactions with other QTL may be substantial.
The complex genetic architecture of BX in rye makes it difficult to develop an efficient strategy for increasing the level of these important metabolites in both leaves and roots. From the QTLs for GDIBOA in leaves, the most promising for selection purposes might be those on chromosomes 1RS, 2RS, and 7R (coinciding with the ScBx1 gene). Both QTL for DIBOA in leaves on chromosome 7R seem to also be valuable for such selection. Roots can be enriched in BX by applying the QTL for GDIMBOA on chromosomes 1R, 2R, 6RS, and 7R.
The set of QTL for PHS revealed in this study constitutes only a part of the genetic architecture detected in the 541 × Ot1-3 intercross in earlier studies [22,25,26]. The QTL on chromosomes 1RL, 2R, 3R, 5R (3), and 6RS are located on the dense DArT map of the RIL population applied in this study in similar positions as those found on the F 2 map consisting mainly of the RAPD and SSR markers [27] and agree with the results of association mapping performed within a wide range of rye genotypes [10]. The present experiment however did not identify the QTL for PHS on chromosomes 3RS, 6RL, and 7RS detected earlier by using both QTL mapping and BSG analysis. A proteomic analysis of lines representing the resistant and susceptible lines of the 541 × Ot1-3 cross showed that many identified proteins represented defensive functions against microbial and insect pests [28,29]. This should not be surprising as only the intact grain surface without any destructive effects of pets' activities can resist water infiltration, which directly induces sprouting and the α-amylase production. This observation leads to the hypothesis about the possible role of BX in the resistance to PHS, which can be additionally supported by data about the inhibitory effect of BX on the germination ability [4]. The results of this study suggest that the coinciding positions of QTL for PHS and BX may reflect the common regulation of BX production and PHS in rye grain.
One of the first processes connected with the water uptake by the grain is the increase in the α-amylase activity in the endosperm. Therefore, all factors reducing the water permeability of the grain surface, including resistance against microbial infections and pest infestation should hinder the α-amylase production in aleurone cells. BXs are among metabolites that play a defensive role, and their relationship with the α-amylase activity is an important question. The negative effect of BX on a plant's α-amylase activity was reported by Kato-Noguchi et al. [30]. As shown in this paper, several QTL for AMY coincide with those for BX. This observation opens up an opportunity for their common influence on both traits in the rye grain. Future studies should verify this hypothesis, which may play a positive role in finding valuable genes for rye breeding.
The α-amylase activity in the rye grain is a very variable trait, sensitive to environmental conditions, mainly the precipitation and temperature profiles during the period of grain maturation. Levels of this enzyme in the maturing grain were found to be not correlated with its predisposition to PHS [10,31]. Further, results reported in this paper are in agreement with the QTL mapping performed earlier within the 541×Ot1-3 mapping population [23,26,32]. The nine positions of QTL for AMY found here represent regions similar to those detected in previous reports, including an association mapping study [10]. The AMY and PHS QTL occupy similar positions in three chromosomal regions. This finding to some extent confirms earlier results showing that the genetic architecture of these traits is partially overlapping on chromosomes 1RL, 2R, 3R, and 6RL [23,26,33].
Despite several genetic, physiological, and breeding-directed studies, leaf rust caused by Puccinia recondita f. sp. secalis still constitutes a big challenge for rye breeders [34][35][36]. The SNPs detected for leaf rust resistance using association mapping on chromosome 1R [10] are in a region of RR QTL found in this study. But, no QTL on chromosome 2RS and 5RL, corresponding to the SNPs detected through association mapping, were found. RR QTL were also not found within regions of chromosomes 6RL and 7RL, where the Pr1 and Pr2 genes for leaf rust resistance were mapped [36]. QTL for rust resistance coincide in a majority of cases with the BX QTL and the QTL for the α-amylase activity, along with those for PHS on chromosome 5R. This finding is very important for the possibility of selecting molecular markers aimed at improving AMY, PHS, BX content, and rust resistance in rye. Such markers should be searched for in chromosome regions with coinciding QTL.
The physiological aspect of the common positions of QTL for defensive traits (BX content and rust resistance) and water stress-dependent traits such as PHS and AMY is very interesting. Further, the synthesis of the BX species as defensive metabolites is probably a part of a wider mechanism of mobilizing all necessary plant resources to eliminate the negative aspects of biotic and/or abiotic stresses.
Supporting information S1 Protocol. Sequences of primers, PCR conditions, and restriction enzymes used for revealing the allelic polymorphisms in SNPs of ScBx1-5. (DOCX)