Introduction

Fusarium head blight (FHB) or scab is a devastating wheat disease worldwide, leading to yield reduction and poor grain quality. FHB resistance is a typical quantitative trait and controlled by multiple genes. Even though chemical and biological measures could be employed to control this disease, the deployment of disease-resistant varieties is the most effective strategy. There are two major FHB resistance types in wheat, with type I against initial penetration and type II against fungal spread within spikes (Schroeder and Christensen 1963). They were governed by different genetic factors (Lin et al. 2004, 2006). Usually, the single floret or point inoculation method is used for evaluation of type II resistance. To evaluate type I resistance, inoculation is usually carried out by spraying Fusarium graminearium conidiospores at anthesis or by planting in disease nurseries and percentage of infected spikes (PIS) and percentage of diseased spikelets (PDS) were used as the resistance indexes. Although PDS could be affected by both type I and type II resistances, it represents mainly type I resistance if the data are collected at the suitable time after the infection (Xue et al. 2010b). In breeding practice, much attention has been paid to type II resistance due to lack of germplasm with excellent type I resistance and a more tedious task in measuring type I resistance.

The availability of molecular markers and genetic maps makes quantitative trait loci (QTLs) mapping a routine strategy for the discovery of genes involved in complex quantitative traits in wheat (Liu et al. 2005; Quarrie et al. 2005). So far, more than 100 QTLs for FHB resistance have been identified in Sumai No. 3 and its derivatives as well as in a few other resistant germplasms using recombinant inbred lines (RILs) and doubled haploid lines (Buerstmayr et al. 2009; Löffler et al. 2009). Most of these QTLs are type II resistance related. In the indigenous germplasm Wangshuibai originating in Jiangsu, China, we have identified two chromosomal intervals, one on chromosome 4B and one on chromosome 5A, showing major effects on type I resistance (Lin et al. 2006).

With primary mapping populations, QTL is usually placed to a large confidence interval due to the limitation in population size and marker density of the genetic maps. Thus, secondary populations created with near-isogenic lines (NIL) or substitution lines of the target QTLs that have reduced genetic background noise are often used in QTL fine mapping and isolation (Zhang et al. 2006; Shan et al. 2009). A number of QTLs of agronomical importance have accordingly been isolated in several plant species, for instance, Gpc-B1 of wheat (Triticum aestivum L.) (Uauy et al. 2006), Gw2 of rice (Oryza sativa L.) (Song et al. 2007), and fw2.2 of tomato (Frary et al. 2000). QTL fine mapping is not only the prerequisite for QTL isolation but also beneficial for marker-assisted breeding. In marker-assisted selection (MAS), selection for large intervals often means cotransfer of deleterious genes or unfavorable traits together with the target QTLs due to linkage drag (Gervais et al. 2003; Somers et al. 2003; Draeger et al. 2007; McCartney et al. 2007; Xue et al. 2010a). Currently, the FHB resistance QTLs that have been fine mapped using a similar strategy included Fhb1, Fhb2, and Fhb4 (Cuthbert et al. 2006, 2007; Liu et al. 2006; Xue et al. 2010b). Fhb4 is for type I resistance.

A previous study using multiple-year trials has revealed that the type I resistance QTL on chromosome 5A of Wangshuibai, Qfhi.nau-5A, consistently explained 16.6–27.0% phenotypic variation in the RIL population (Lin et al. 2006). This QTL interval has also been associated with type I resistance in a few other germplasms, such as CM-82036 (Buerstmayr et al. 2003), DH181 (Yang et al. 2005), and W14 (Chen et al. 2006). Evaluation of the Qfhi.nau-5A NIL in susceptible background demonstrated that Qfhi.nau-5A only conditioned type I resistance and its presence reduced PIS by over 60% (Xue et al. 2010a). A recent meta-analysis for FHB resistance using 30 mapping populations again detected this QTL interval (Löffler et al. 2009).

In this study, we report the precise mapping of Qfhi.nau-5A by screening and evaluation of recombinants identified in two segregating populations created with the Qfhi.nau-5A NILs and in the Nanda2419 × Wangshuibai (NW) RIL population.

Materials and methods

Plant materials

A BC3F2 population of 94 plants and a BC4F2 population of 71 plants were created for recombinant identification by selfing two different Qfhi.nau-5A NILs heterozygous at Qfhi.nau-5A. These two NILs were developed, respectively, by using susceptible cultivars Mianyang 99-323 and PH691 as the recurrent parent and confirmed with marker genotypes. They had over 97% recipient genome composition (Xue et al. 2010a; unpublished data). In addition, a RIL population of 530 lines derived from Nanda2419 × Wangshuibai (Lin et al. 2004) was also used for recombinant screening. A F2 population of 491 plants developed by crossing a homologous Qfhi.nau-5A NIL with its recurrent parent Mianyang 99-323 was used for genetic map construction of the Qfhi.nau-5A region.

Wheat cultivar “Chinese Spring” (CS), its ditelosomic lines, as well as a set of deletion lines of chromosome 5A (Endo and Gill 1996) were used for chromosomal arm and bin assignments of molecular markers.

Genotyping, mapping, and recombinant screening

Genomic DNA was extracted from young leaves according to Ma and Sorrells (1995). Polymerase chain reaction (PCR) was performed following the procedure of Ma et al. (1996). The PCR products were separated in 8% non-denaturing polyacrylamide gels (Acr:Bis = 19:1) at room temperature with 1 × TBE buffer and visualized by silver staining (Bassam et al. 1991).

For map construction, all DNA markers mapping to the Qfhi.nau-5A interval flanked by Xbarc56 and Xbarc100 on chromosome 5A (Röder et al. 1998; Somers et al. 2004; Song et al. 2005; Lin et al. 2006; Torada et al. 2006; Xue et al. 2008) were employed in genotyping. Linkage map was constructed with MAPMAKER/EXP version 3.0 (Lincoln et al. 1992) using the Kosambi mapping function (Kosambi 1944) and drawn with MapDraw version 2.1 (Liu and Meng 2003).

Plants with recombination occurring within the Qfhi.nau-5A interval were identified in the above-mentioned BC3F2 population with the Mianyang 99-323 background using flanking markers Xbarc303 and Xbarc180, and in the BC4F2 population with the PH691 background using flanking markers Xbarc56 and Xbarc100. The selfed progenies of the identified plants were then genotyped for recombinant screening with all the markers mapped in the QTL interval and polymorphic in the respective populations. The identified homozygous recombinants were selfed for seed production for resistance evaluation.

In the RIL population, lines with recombination occurring within the QTL interval and without the Wangshuibai chromatin corresponding to Fhb4 were selected based on the marker genotypes within the respective regions. Markers closely linked to Fhb4 were used in selection against Fhb4 (Xue et al. 2010b).

FHB resistance evaluation

Evaluation of the BC3F2 population and the BC4F2 population was carried out in 2006 and 2008 seasons, respectively, in a field of Nanjing Agricultural University (NAU) Jiangpu (JP) experimental station. Single-space planted plants were evaluated in these experiments. At planting, fifteen seeds were evenly distributed in each 1.5-m row. The row spacing was 50 cm. Inoculation was carried out 1 week before anthesis through scattering on the soil surface wheat grains fully infected with four locally isolated highly virulent strains F4, F15, F34, and F7136, courtesy of Plant Protection Institute of Jiangsu Academy of Agricultural Sciences. Fifteen days after the anthesis, the total number of spikelets and the number of diseased spikelets (with visible disease symptom) of each spike were counted and the percentage of diseased spikelets (PDS) was then calculated. The PDS mean (from 3 to 5 spikes) of each F2 plant was used here as the index for type I resistance. In our experiments, the PDS data collected at 15 days after anthesis mainly reflected type I resistance in our experimental conditions (Xue et al. 2010b).

Evaluation of the RIL and NIL homozygous recombinant lines was made in 2009 in a field of the Weigang (WG) campus of NAU, in 2010 in a field of the Pailou (PL) experimental station of NAU, and in 2009 and 2010 in a field of the JP experimental station, all using a randomized complete block design with two replicates. Each line was planted in one 1.4-m row in the WG and PL trials and two 1.5-m rows in the JP trials. The row spacing was 25 cm, and 20 seeds were planted per row. Inoculation was carried out at anthesis through spraying mixed F. graminearum conidial suspension prepared with the four virulent strains with a concentration of 5 × 104 conidia/ml in the WG and PL trials as described in Lin et al. (2006) and through scattering the scabby wheat grains on the soil surface in the JP trial as described above. Percentage of infected spikes (PIS) in a sample size of about 100 spikes was investigated in all but the 2010 JP trial. PIS indicates the number of spikes with at least one floret with visible FHB symptom 15 days after anthesis divided by the total number of spikes of each line. In the 2010 JP trial, PDS was investigated in the NIL recombinants due to heavy infection, for which about 30 randomly chosen spikes of each line were scored. Under the condition of heavy infection or small spike sample size as for the F2 plants, PDS is more suitable as the type I resistance index than PIS.

In these trials, the Qfhi.nau-5A NILs and RIL NW127 that has Wangshuibai chromatin in the Xbarc56–Xbarc100 interval and has Nanda2419 chromatin in the Fhb4 region were used as the resistant control; Nanda2419, Mianyang 99-323, and PH691 were used as the susceptible control.

Statistical analysis

Statistical analysis was performed using software Data Desk v. 5.0 (Data Description, Inc., Ithaca, NY, USA). The effects of genotypes and environments on PIS were determined according to the fixed-effects model of two-factor analysis of variance (ANOVA). In this study, each site-year combination was treated as an environment. Dunnett’s test was used for recombinant-versus-control comparisons, and Tukey’s HSD test was applied for multiple mean comparisons of the BC3F2 and BC4F2 genotypes.

Results

Inheritance of Qfhi.nau-5A

The mode of inheritance of Qfhi.nau-5A was analyzed using the PDS data of 94 BC3F2 plants derived from the Qfhi.nau-5A NIL heterozygous at Qfhi.nau-5A with the Mianyang 99-323 background and of 71 BC4F2 plants derived from the Qfhi.nau-5A NIL heterozygous at Qfhi.nau-5A with the PH691 background. PDS in these two populations all displayed a two-peak frequency distribution with the valley at 25% (Fig. 1). Divided at the valley, the resistant group or group with smaller PDS and the group with larger PDS fit the 3:1 ratio in each of these two populations (χ2 = 0.13 and 0.38, respectively), suggesting that Qfhi.nau-5A was inherited like a single dominant gene. The means of the resistant and susceptible groups of the BC3F2 plants were 17.7 ± 0.5% and 37.7 ± 1.0%, and those of the BC4F2 plants were 13.8 ± 0.5% and 32.1 ± 0.8%, respectively. The differences between the group means were all significant at P ≤ 0.0001.

Fig. 1
figure 1

Frequency distribution of percentage of diseased spikelets in the BC3F2 population with Mianyang 99-323 background (a) and the BC4F2 population with PH691 background (b)

Full size image

To prove that Qfhi.nau-5A represents a single dominant gene, the PDS data were classified into three classes based on the marker genotype at Xgwm415 that was tightly linked to Qfhi.nau-5A (Lin et al. 2006; Xue et al. 2010a), including Wangshuibai genotype (WW), Mianyang 99-323 (MM) or PH691 genotype (PP), and heterozygous genotypes of Wangshuibai with Mianyang 99-323 (WM) or PH691 (WP). As shown in Table 1, the PDS means of the MM or PP classes were all significantly larger than those of the WW class and the WM or WP classes, while the difference between the PDS means of the WW class and the WM or WP classes did not reach the significant threshold. These results showed that plants with Qfhi.nau-5A at either homozygous or heterozygous state contributed equally to type I resistance.

Table 1 Average percentage of diseased spikelets (PDS) of plants with the same marker genotypes in the BC3F2 and BC4F2 populations
Full size table

Marker saturation of the Qfhi.nau-5A interval

Qfhi.nau-5A was previously mapped to a 12.6-cM confidence interval flanked by Xbarc56 and Xbarc100 (Lin et al. 2006). To construct a fine map of this region, a F2 population of 491 plants developed from the cross of a Qfhi.nau-5A NIL × Mianyang 99-323 was used. A total of 25 SSR and STS markers mapping to the Qfhi.nau-5A interval were surveyed for polymorphism between the Mianyang 99-323-derived Qfhi.nau-5A NIL and Mianyang 99-323. The flanking markers Xbarc56 and Xbarc100 were monomorphic between the two mapping parents. Ten markers detected polymorphism and were used in genotyping of the F2 plants. With these marker data, a 4.2-cM map bordered by Xbarc303 and Xmag3794 was constructed (Fig. 2), which had an average marker density of one marker per 0.4 cM.

Fig. 2
figure 2

Molecular marker map of the Qfhi.nau-5A region (left) constructed using the 491 F2 plants derived from a Qfhi.nau-5A NIL × Mianyang 99-323, with the markers aligned to the deletion bin map (right) of chromosome 5A. The short arm is at the top. The black oval indicates the approximate position of centromere. Numbers to the left of the genetic map are interval distances (cM). The highlighted markers Xbarc56 and Xbarc100 were placed on the map according to Xue et al. (2008). The bin assignments delineated by neighboring deletion breakpoints are shown to the right of deletion bin map

Full size image

Recombinant identification of the Qfhi.nau-5A interval

Recombinants were identified from the mentioned BC3F2 and BC4F2 populations and from the Nanda2419 × Wangshuibai RIL population.

Genotyping the 94 BC3F2 plants with the Mianyang 99-323 background at Xbarc303 and Xbarc180 led to identification of four plants with recombination occurring between these two markers. Twelve homozygous recombinants were then obtained by surveying 10–15 BC3F3 plants derived from each of these plants with the ten mapped polymorphic markers. These recombinants represented four kinds of genotypes at the Qfhi.nau-5A interval (Fig. 3a). No recombinants were found between Xbarc303 and Xgwm304 and between Xbarc180 and Xmag3794.

Fig. 3
figure 3

Graphical illustrations of the recombinant genotype types in the Qfhi.nau-5A interval and their phenotypes in multiple trials. a Mianyang 99-323 background recombinants; b PH691 background recombinants; c Nanda2419 × Wangshuibai RIL recombinants. The black rectangles indicate Wangshuibai chromatin, the open rectangles indicate Mianyang 99-323, PH691, or Nanda2419 chromatin. The broken rectangle defines the interval harboring Qfhi.nau-5A. R, resistant and S, susceptible. Asterisk significantly different from Mianyang 99-323, PH691, or Nanda2419 at P ≤ 0.05 according to Dunnett’s test

Full size image

Based on the genotypes at Xbarc56 and Xbarc100, three heterozygous recombinants were identified among the 71 BC4F2 plants with the PH691 background. By genotyping their BC4F3 plants with the 12 markers shown in Fig. 2, which all detected polymorphism between the PH691-derived Qfhi.nau-5A NIL and PH691 as well as between Wangshuibai and Nanda2419, seven homozygous recombinants, representing three kinds of genotypes at the Qfhi.nau-5A interval, were obtained (Fig. 3b). Among the 530 RILs derived from Nanda2419 × Wangshuibai, nine lines without Fhb4 were identified with recombination occurring within the Xbarc56–Xbarc100 interval. These RIL recombinants represented four kinds of genotypes at the Qfhi.nau-5A interval (Fig. 3c). No recombinants were obtained in the Xbarc56–Xbarc180 interval in these two populations (Fig. 3b, c). The marker order determined using the identified recombinants fits well with the genetic map (Fig. 2).

Precise mapping of Qfhi.nau-5A

To precisely map Qfhi.nau-5A, the recombinant lines as well as the controls were surveyed for resistance with two to four field trials. Analysis of variance for PIS of the Mianyang 99-323 NIL recombinants and RIL recombinants across the environments revealed highly significant effects for both genotypes and environments, but not for the replications and genotypes × environments (Table 2). Moreover, the mean squares for genotypes were much larger than those for the genotype-by-environment interactions. The between-environment correlation coefficients of the genotype variable for the NIL and RIL recombinants ranged from 0.78 to 0.87 and from 0.91 to 0.98, respectively, all significant at P ≤ 0.0001. The 2-year data of the PH691 NIL recombinants, one in PIS and one in PDS, were highly correlated (r = 0.96, P ≤ 0.0001). Similarly, the PDS data of the Mianyang 99-323 NIL recombinants were also highly correlated with the three sets of PIS data (r = 0.89–0.94, P ≤ 0.0001). A discontinuous bimodal distribution showed up for the PIS data of all three kinds of recombinants (Fig. 4), with the two areas of the distribution spaced by 13.8–23.7%, which were significant at P ≤ 0.001. This made it possible to classify the recombinant lines into resistant group (R, falling in the area with lower PIS) and susceptible group (S, falling in the area with higher PIS). The classification of the recombinants did not change in trials. For the PH691 NIL recombinants, the PIS means of the R and S groups ranged from 3.2 to 7.5% and from 21.4 to 22.9%. For the Mianyang 99-323 NIL recombinants, the PIS means of the R and S groups over all the experiments ranged from 9.7 ± 0.4 to 17.2 ± 5.1% and from 31.5 ± 3.7 to 34.6 ± 6.4%. Those for the RIL recombinants ranged from 8.9 ± 1.1 to 16.3 ± 0.5% and from 40.1 ± 2.0 to 49.3 ± 5.7%. On the whole, the R group had over 55% less infection than the S group.

Table 2 Results of two-way analysis of variance for percentage of infected spikes (%) of NIL and RIL recombinants
Full size table
Fig. 4
figure 4

Distribution of percentage of infected spikes (PIS) of the twelve Mianyang 99-323-derived NIL recombinants (a), seven PH691-derived NIL recombinants (b), and nine RIL recombinants (c). R-resistant control and S-susceptible control

Full size image

Of the four Mianyang 99-323 NIL recombinant genotypes, three were resistant and one was susceptible (Fig. 3a). Common to these resistant genotypes, the Xgwm304–Xgwm415 interval or one of these two loci was derived from Wangshuibai. Both Xgwm304 and Xgwm415 loci in the susceptible genotype were derived from Mianyang 99-323. Among the PH691-derived NIL and RIL recombinants, the resistant genotypes had the Xgwm304–Xgwm415 interval derived from Wangshuibai and the susceptible genotypes had this interval derived from PH691 or Nanda2419 (Fig. 3b, c). These results indicated that Qfhi.nau-5A fallen in the 0.3-cM interval flanked by Xgwm304 and Xgwm415.

Physical mapping of Qfhi.nau-5A

To determine the physical position of Qfhi.nau-5A, all markers mapping to the Xbarc56–Xbarc100 interval were mapped using the CS chromosome 5A ditelosomic and deletion lines. As shown in Fig. 2, the Xbarc56–Xbarc100 interval covered nearly the whole short arm and about one-half of the long arm of chromosome 5A. However, Xgwm304, as well as the cosegregating Xbarc358 and Xgwm293, was placed to the 5AS1-0.40–0.75 bin, and Xgwm415 was placed to the C–5AS1-0.40 bin, all on the short arm. Thus, Qfhi.nau-5A resides in the C–5AS3-0.75 bin (Fig. 2). Since Xbarc117 was the locus closest to the distal boundary of the 5AS1-0.40–0.75 bin and Xgwm415 was the locus closest to the centromere, the 0.5-cM interval of Xbarc117–Xgwm415 was corresponding to 219 Mb at the most, estimated based on the physical lengths of chromosome 5AS and the entire genome (Smith and Flavell 1975; Gill et al. 1991).

Discussion

A major type I FHB resistance QTL on chromosome 5A has been reported in several germplasms (Buerstmayr et al. 2003; Yang et al. 2005; Chen et al. 2006). Qfhi.nau-5A in scab resistance germplasm Wangshuibai explained over 16% phenotypic variation of type I resistance in a RIL population (Lin et al. 2006). In this study, we conducted precise mapping of this QTL. Using the genotype and phenotype data of the identified lines with recombination occurring within the Xbarc56–Xbarc100 interval, Qfhi.nau-5A was placed in the 0.3-cM region flanked by Xgwm304 and Xgwm415 (Fig. 2).

The Qfhi.nau-5A NILs used in this study showed a stable and significantly decreased PIS and PDS compared with their susceptible recurrent parents Mianyang 99-323 and PH691 (Fig. 3). Resistance in the BCF2 populations derived from these NILs segregated in the 3:1 ratio, suggesting that Qfhi.nau-5A inherits like a single Mendelian factor. It was thus designated as Fhb5. Although FHB resistance is a quantitative trait, the use of NILs has significantly reduced the background noises and makes it possible to precisely map it. Since there was no statistical difference in resistance performance between plants with homozygous Wangshuibai alleles of this QTL and plants in its heterozygous state (Table 1), Fhb5 functions dominantly.

Fine mapping and cloning QTLs using recombinants identified from the NIL-derived secondary segregating populations or RIL populations have been effective in plant species (Uauy et al. 2006; Andaya and Tai 2007; Chen et al. 2008). Using this strategy, QTLs for wheat FHB resistance, including Fhb1, Fhb2 and Fhb4, have been fine mapped (Cuthbert et al. 2006, 2007; Liu et al. 2006; Xue et al. 2010b). In this study, the 28 homozygous recombinants for the Fhb5 interval were clearly classified into two groups based on their resistance levels (Fig. 4). The categorization of the individual lines did not change in the trials conducted in difference locations or years. This implies that the genetic variation caused by Fhb5 is significantly larger than the variation caused by genotype × environment interactions and experimental errors, thus allowing us precisely determine the location of Fhb5 relative to the molecular markers.

The C–5AS3-0.75 bin harboring Fhb5 has an approximate physical distance versus genetic distance ratio of 438 Mb/cM. Assuming a wheat genome size of 5,000 cM and 16,800 Mb, the recombination rate of this interval (0.002 cM/Mb) is about 130 times smaller than the genome average. However, this could be a severe underestimate, since both Xbarc117 and Xgwm415 could be far from the bin boundaries. In wheat the pericentromeric regions usually have lower recombination rate (Akhunov et al. 2003). This low recombination makes it difficult to conduct map-based cloning of Fhb5. Qi and Gill (2001) obtained similar findings in their studies of the male-sterile gene Ms3 that maps to the C–5AS1-0.40 bin. Most of the wheat genes isolated by map-based cloning so far, such as Q (Faris et al. 2003), VRN2 (Yan et al. 2004), Pm3b (Yahiaoui et al. 2004), and Gpc-B1 (Uauy et al. 2006), reside in regions with a recombination rate of more than 0.5 cM/Mb. However, since the two flanking markers Xgwm304 and Xgwm415 map to two adjacent bins (Fig. 1) that cover 75% of the 5AS physical length, more markers mapping between Xgwm304 and Xgwm415 must be developed and more recombinants of this interval must be identified to get accurate estimate of the recombination rate near Fhb5. This is possible. It was shown that in this study, by using a population of 491 F2 plants derived from the Fhb5 NIL × Mianyang 99-323, two recombinants were identified for the Xbarc117–Xgwm304 interval and three were identified for Xgwm304–Xgwm415 interval, while these three loci cosegregated in the NW map constructed with 136 RILs (Xue et al. 2008) and in the CSSQ1 map constructed with 96 doubled haploid lines (Quarrie et al. 2005). Moreover, the recombination rate near a particular locus could differ in different crosses (Bentolila and Hanson 2001; Imai et al. 2003; Dadkhodaie et al. 2011). Thus, development of multiple populations using different crossing parents could increase the chance to obtain the desired recombinants.

Fhb5 in Wangshuibai is one of the most effective type I resistance QTLs (Lin et al. 2006; Xue et al. 2010a). However, its 12.6-cM Xbarc56–Xbarc100 interval has been associated with flag leaf width, thousand-grain weight, and spike compactness, with the Wangshuibai allele negatively affecting these agronomical traits (Ma et al. 2007, 2008; Xue et al. 2010a). Through precise mapping in this study, Fhb5 is now confined to a 0.3-cM interval. This will facilitate investigation into the genetic association of Fhb5 with these agronomical traits and reduce linkage drag in marker-assisted selection of Fhb5 in breeding programs.