Introduction

The annual production of wheat exceeds 700 million tons and provides approximately 20% of the calories consumed around the world (Ma et al. 2005; Buerstmayr et al. 2009, 2020; Yan et al. 2021). Among these QTLs, the formally named QTLs are Fhb1-Fhb7. Fhb1, Fhb2, Fhb4, and Fhb5 are located on chromosomes 3BS, 6BS, 4B, and 5AS, respectively, with Fhb1 and Fhb2 being associated with resistance to expansibility, while Fhb4 and Fhb5 are associated with resistance to invasion (Bai et al. 1999; Waldron et al. 1999; Anderson et al. 2001; Buerstmayr et al. 2003; Zhang et al. 2004; Lin et al. 2006). Fhb3, Fhb6, and Fhb7 originated in the wheat relatives Leymus racemosus, Elymus tsukushiensis, and Thinopyrum pontium and were transferred to chromosomes 7AS, 1AS, and 7DL, respectively (Qi et al. 2008; Cainong et al. 2015; Guo et al. 2015; Bai et al. 2018). However, only Fhb1 consistently confers high resistance (HR), explaining up to approximately 50% of the phenotypic variation in different genetic backgrounds and testing environments (Zhang et al. 2021b). Furthermore, Fhb1 was cloned from the Chinese wheat variety Sumai 3, and the results showed that a pore-forming toxin-like (PFT) gene confers FHB resistance (Rawat et al. 2016). Subsequently, Fhb1 was also isolated as a histidine-rich calcium binding protein (HRC), and diagnostic markers were developed (Li et al. 2019; Su et al. 2019).

Fhb1 has been widely used in breeding programs, resulting in the production of numerous new wheat varieties, such as Ningmai 13, Ningmai 15, Ningmai 16, Ningmai 18, Ningmai 23, Shengxuan 6, Nannong 0686, 25R18, 25R51, 25R42, Jaceo, and MS INTA 416, which show significantly improved FHB resistance (Buerstmayr et al. 2020; Ma et al. 2022b; Zheng et al. 2022). Nevertheless, Fhb1 alone is not sufficient for impeding the damage of FHB under severe FHB epidemics because of its partial resistance (Bai et al. 2018; Brar et al. 2019; Su et al. 2019). Pyramiding Fhb1 with other FHB resistance QTLs/genes from elite genetic backgrounds is effective to rapidly arrive both acceptable resistant levels of FHB and desired agronomic traits (Kang et al. 2011; Tamburic-Ilincic et al. 2019). Some Chinese landraces with HR or moderate resistance (MR) to FHB were identified, but it is difficult to use these landraces directly as parents in modern breeding programs due to their inferior agronomic performance (Li et al. 2011; Zhang et al. 2012; Cai et al. 2014; Li et al. 2016; Zhu et al. 2019). The objective of the present study was to perform QTL analysis for FHB resistance using TL-RILs in different environments and then to obtain candidate genes from stable QTLs.

Materials and methods

Plant materials

QTL analysis was used a RIL population derived from the cross of ‘TN18 × LM6’ (TL-RILs, 184 lines) (Zhang et al. 2019; Han et al. 2023). TN18 shows MR to FHB which was identified by CVAC of Shandong Province. LM6 is high susceptible (HS) to FHB. The two parents show obvious different for FHB resistance, with TN18 having higher resistance than LM6 overall.

Experimental design and trait measurement

The TL-RIL population and its parents were planted with a single row for each line, 60 seeds per row, 1.5 m row length, 25 cm row spacing, and two replications. FHB resistance was evaluated in four environments: a field at the Nan**g Agricultural University in 2017–2018 (E1), a field at the Jiangsu Academy of Agricultural Sciences in 2019–2020 (E2), a greenhouse at the Jiangsu Academy of Agricultural Sciences in 2019–2020 (E3), and a field of the Nan**g Agricultural University in 2019–2020 (E4).

FHB resistance was estimated via single-floret inoculation as described by Zhang et al. (2004). For all the four environments, inoculation was performed at anthesis via the single-floret inoculation of approximately 1,000 conidiospores from a mixture of four local virulent strains of F. graminearum (Li et al. 2019). The spikes were covered with a plastic bag to retain high humidity and the plastic bags were removed 72 h later. Then, the spikes were sprayed using water to keep them moist. For each line, ten spikes were inoculated. The scabbed spikelets were recorded at 21 days postinoculation, and the number of diseased spikelets (NDS) was determined using the average of 10 inoculated spikes. The relative disease index (RDI) was calculated by dividing NDS by the average of all NDS value.

Data analysis, QTL detection and candidate gene identification

SPSS 17.0 software was used for analysis of variance (ANOVA) (SPSS Inc., Chicago, IL, USA). Genotypes and environments were considered as two factors using the data of 184 lines under four environments. All factors involved were considered sources of random effects.

QTL analysis was performed using the genetic map of unigenes (UG-Map) based on their physical positions, which including 31,445 sub-unigenes (Zhang 2019; Han et al. 2023). QTL map** was performed using Windows QTL Cartographer 2.5 software (http://statgen.ncsu.edu/qtlcart/WQTLCart.htm) with the parameters of composite-interval map** (CIM), model 6 standard analysis, walk speed of 0.5 cM, forward and backward regression, up to five control markers, and blocked window size of 10 cM. A significant QTL for each environment was declared to be present with LOD more than 3.0, and the QTL interval was decided by drop** 1 unit in both directions from the peak LOD value. We defined a QTL as a stable QTL that was found over the AV + 1 environments.

The unigenes covered by the interval of QTLs were defined as the candidate genes of the corresponding QTLs. The RNA-Seq data were previously reported to be consistent with qRT-PCR results (Li et al. 2021; Zhang et al. 2021a). So, we used FPKM (Trapnell et al. 2010) to quantify the expression level of the candidate genes.

Results

Phenotypic variation

TN18 and LM6, the parents of the TL-RILs, were obvious differences for the number of diseased spikelets (NDS) and relative disease index (RDI) (Table 1). For the RILs, ANOVA indicated that the variance of NDS, RDI and the interactions of environment × genotype were significant at p ≤ 0.01 (Table S1). The RIL population exhibited wide variations, with the CV (coefficient of variation, %) ranging from 23.35 (AV, average value) to 50.84 (E3) (Table 1). The transgressive segregation was appeared in all ten trait-environment combinations (including AV). The two investigated traits under each treatment showed a continuous distribution (Figure S1), which indicates their quantitative trait nature.

Table 1 Phenotypic parameters of the TL-RILs and their parents
Full size table

Major characteristics of stable QTLs

QTL analysis was performed using the UG-Map of the TL-RILs, which included 31,445 sub-unigenes (Zhang 2019; Han et al. 2023). Using the software of Windows QTL Mapper 2.5, a total of 19 stable QTLs for NDS and RDI were located distributed across 13 chromosomes,1A, 1B, 1D, 2A, 2B, 3A, 3B, 4B, 5A, 5B, 5D, 6B, and 7A (Table S2, Table 2, Fig. 1). The additive effects of ten QTLs, QNds-1AS-2225, QNds/Rdi-1BS-1759, QNds/Rdi-1BL-16514, QNds-2BL-16606, QNds-3AS-3130, QNds/Rdi-4BL-7422, QNds/Rdi-5AL-7560, QNds-5BL-3900, QNds/Rdi-5BL-9509, and QNds/Rdi-6BS-6922, were positive, indicating that the female parent TN18 was increased the QTL effects. Contrarily, the additive effects of the other nine QTLs, QRdi-1BS-4752, QNds-1DS-527, QRdi-1DS-575, QRdi-2AL-11212, QNds/Rdi-3AL-7033, QNds/Rdi-3BS-6183, QNds/Rdi-4BL-4553, QRdi-5DL-7988 and QNds/Rdi-7AL-14499, were negative, indicating that the male parent LM6 was increased the QTL effects. The maximum LOD value for a single QTL under the different trait-environment combinations was 12.45 (QNds/Rdi-5BL-9509 in E2 of RDI). The contribution of a single QTL ranged from 5.61% (QNds-5BL-3900 in E2) to 51.14% (QNds-1AS-2225 in E3).

Table 2 Summary for the stable QTLs and their candidate genes in TL-RILs (see Table S2 for details). “TraesTLxx02G……” are the genes annotated in the TL-RILs; “STRG……” are ncRNAs
Full size table
Fig. 1
figure 1

Locations of the 19 stable QTLs and their 36 candidate genes using the TL-RILs. CS and TL are abbreviations of gene names of “TraesCSxx02G” and “TraesTLxx02G”. Highlighted in red are the seven candidate genes which should be closely associated with FHB resistance in wheat

Full size image

Five stable QTLs, QNds-1AS-2225, QNds-1DS-527, QNds-2BL-16606, QNds-3AS-3130, and QNds-5BL-3900, were associated solely with NDS. These QTLs were distributed across chromosomes 1A, 1D, 2B, 3A, and 5B. Except for QNds-5BL-3900, the average contributions of the QTLs were all over 10.00%, showing that these QTLs should be major stable QTLs.

Four stable QTLs, QRdi-1BS-4752, QRdi-1DS-575, QRdi-2AL-11212, and QRdi-5DL-7988, were associated solely with RDI. These QTLs were located on chromosomes 1B, 1D, 2A, and 5D. The average contributions were all over 10.00%, indicating that they were major stable QTLs.

Ten stable QTLs were associated with both NDS and RDI and should thus be particularly important. These QTLs included QNds/Rdi-1BS-1759, QNds/Rdi-1BL-16514, QNds/Rdi-3AL-7033, QNds/Rdi-3BS-6183, QNds/Rdi-4BL-4553, QNds/Rdi-4BL-7422, QNds/Rdi-5AL-7560, QNds/Rdi-5BL-9509, QNds/Rdi-6BS-6922, and QNds/Rdi-7AL-14499, distributing on eight chromosomes, 1B, 3A, 3B, 4B, 5A, 5B, 6B and 7A. The average contributions of all the QTLs were over 10.00%, showing that these QTLs should be major stable QTLs.

Candidate genes from stable QTLs

A total of 36 candidate genes (including noncoding RNAs, ncRNAs) from the 19 corresponding stable QTLs were identified (Table S2, Table 2, Fig. 1), which were covered by the intervals of the QTLs. Of these candidate genes, 24 were annotated in the reference genome Chinese Spring (CS) RefSeq v1.1 (IWGSC 2018), four were annotated in TL-RILs, and eight were ncRNAs. The average number of candidate genes per QTL was 1.89 (36/19), with 14 (73.7%), two (10.5%), and three (15.8%) QTLs including one, two, and 3–10 candidate genes, respectively.

Seven candidate genes were obtained for NDS, including three candidate genes annotated in CS RefSeq v1.1, two candidate genes annotated in TL-RILs, and three ncRNAs. Among these genes, TraesCS1D02G017800 and TraesCS1D02G017900 for QNds-1DS-527 were high-confidence (HC) genes, and TraesCS1A02G037600LC for QNds-1AS-2225 were low-confidence (LC) genes.

Four candidate genes were found for RDI, including three candidate genes annotated in CS RefSeq v1.1 and one ncRNA. Among these genes, TraesCS1D02G018000 for QRdi-1DS-575 was an HC gene, and TraesCS1B02G179200LC for QRdi-1BS-4752 and TraesCS5D02G555900LC for QRdi-5DL-7988 were LC genes.

For both NDS and RDI, a total of 25 candidate genes were obtained, including 18 genes annotated in CS RefSeq v1.1, two genes annotated in TL-RILs, and five ncRNAs. Among these genes, two LC genes (TraesCS1B02G784400LC and TraesCS1B02G786600LC) for QNds/Rdi-1BL-16514 and one LC gene (TraesCS3B02G203000LC) for QNds/Rdi-3BS-6183 were identified. For QNds/Rdi-4BL-4553, five candidate genes were identified, including four HC genes, TraesCS4B02G227300, TraesCS4B02G227400, TraesCS4B02G229100, and TraesCS4B02G229500. For QNds/Rdi-4BL-7422, one LC gene (TraesCS4B02G528400LC) was identified. For QNds/Rdi-5AL-7560 and QNds/Rdi-7AL-14499, the HC genes TraesCS5A02G321800 and TraesCS7A02G568400 were identified, respectively. Ten candidate genes of QNds/Rdi-5BL-9509 were identified, including eight HC genes: TraesCS5B02G303100, TraesCS5B02G303200, TraesCS5B02G303300, TraesCS5B02G303400, TraesCS5B02G303500, TraesCS5B02G303700, TraesCS5B02G303800, and TraesCS5B02G304000.

Variant types and excellent sites for the candidate genes

For the 24 candidate genes annotated in RefSeq v1.1, 16 and four were non-synonymous and synonymous mutations in exons, respectively; and four were mutant in introns (Table S3). For all 36 candidate genes, the reads of 27 candidate genes were significantly different between the TN18 and LM6 genotypes in TL-RILs, indicating that their mRNA expression levels were changed (Table S3). According to the DNA sequences of the parents and a part of RILs, the promoter region (− 2000 bp from the start site of the 5’UTR) of 14 candidate genes in CS RefSeq v1.1 and three candidate genes in TL-RILs were mutated (Table S4).

The QTLs with negative effects indicated that the decreased of NDS and/or RDI values came from the TN18 parent. Therefore, the excellent mutants of the corresponding candidate genes are TN18 type. Whereas the excellent mutants of the candidate genes are LM6 type for the QTLs with positive effects. For all the 36 candidate genes, the excellent mutants of 15 candidate genes were TN18 type, and the excellent mutants of 21 candidate genes were LM6 type (Table S3).

Discussion

QTLs identified in this study differ from previously named QTLs/genes

TN18, the female parent of TL-RIL, was derived from a cross of ‘Laizhou 137 × Yannong 19’. LM6, the male parent of TL-RIL, was derived from a cross of ‘Lmai 13 × 924,402’. Sumai 3, a famous FHB-resistant variety, is derived from a cross between the Italian variety ‘Funo’ and ‘Taiwanxiaomai’, a landrace from Fujian Province, China. FHB resistance came from ‘Taiwanxiaomai’ (Ma et al. 1999; Anderson et al. 2001; Yang et al. 2005; Lin et al. 2006; Buerstmayr et al. 2003, 2009, 2020; Yan et al. 2021; Zhang et al. 2021b). It is difficult to accurately compare the QTL locations identified in the present study with those in previously studies because of the nature of map** populations and their genetic maps with different marker systems. Among the seven named QTLs/genes, Fhb1, Fhb2, Fhb4 and Fhb5 were identified and located on chromosomes in common wheat, Sumai 3 and/or Wangshuibai, and should not correspond to the QTLs identified in this study according to the pedigree analysis. For example, Fhb1 is located on chromosome 3BS (TraesCS3B02G019900, ~ 8.5 Mb in RefSeq v1.1) (Ma et al. 2015; Guo et al. 2015). Moreover, some reports predicated candidate genes of FHB resistance on the same chromosomes with this study (Sari et al. 2019; Ma et al. 2022a; Song et al. 2022; Serajazari et al. 2023; Wang et al. 2023), but no one were coincident each other.

In addition, using a high-density genetic map of DNA markers, a QTL can cover dozens and even more of genes (Choulet et al. 2014), which is difficult to distinguished highly reliable candidate genes for gene cloning. Using UG-Map according to the physical positions of the TL-RILs, we obtained 36 candidate genes for 19 QTLs for FHB resistance, with an average of 1.89 candidate genes per QTL. This result showed that we can directly found the candidate genes from QTLs, which should facilitate the gene cloning and genetic improvement in wheat breeding programs.

Seven candidate genes were related to FHB resistance in wheat, barely or Brachypodium distachyon

Because the cloned genes and their functions in FHB resistance have been less frequently reported, the FHB resistance of candidate genes in this study needs further confirmation. Among the 24 candidate genes annotated in RefSeq v1.1 in this study (Table 2, Fig. 1), the homologous genes of seven candidate genes, including TraesCS4B02G227300 for QNds/Rdi-4BL-4553, TraesCS5B02G303200, TraesCS5B02G303300, TraesCS5B02G303700, TraesCS5B02G303800 and TraesCS5B02G304000 for QNds/Rdi-5BL-9509, and TraesCS7A02G568400 for QNds/Rdi-7AL-14499, were previously reported to be related to FHB resistance in wheat, barely or Brachypodium distachyon (Edwards et al. 2000; Geddes et al. 2008; Tian et al. 2009; Gardiner et al. 2010; Kugler et al. 2013; Pasquet et al. 2014; Schweiger et al. 2016; Wang, et al. 2020; Hu et al. 2022; Wu et al. 2022). These genes should be closely associated with FHB resistance in wheat.

TraesCS4B02G227300 of QNds/Rdi-4BL-4553 is annotated as an actin-depolymerizing factor (ADF) with an SNP in the intron and one SNP in the promoter. The expression levels were significantly different between the TN18 and LM6 genotypes of the TL-RILs (Table S3, S4). In barley, ADF plays an important role in the resistance of cells to infection by external FHB pathogens (Geddes et al. 2008). Tian et al. (2009) confirmed that AtADF4 participates in the process of plant defense in Arabidopsis and considered it to be a component of the plant defense signaling pathway opposing pathogen infection.

TraesCS5B02G303200, TraesCS5B02G303300, TraesCS5B02G303700, TraesCS5B02G303800 and TraesCS5B02G304000 were five of eight annotated candidate genes in RefSeq v1.1 of QNds/Rdi-5BL-9509. TraesCS5B02G303200 and TraesCS5B02G303300 are annotated as glutathione S-transferases (GSTs). TraesCS5B02G303200 has four SNPs and two InDels in the intron, and five SNPs in the promoter. TraesCS5B02G303300 has two synonymous variants in the exon, three SNPs in the intron, and 16 SNPs and one InDel in the promoter. The expression levels of the two genes were significantly different between the genotypes of TN18 and LM6 (Table S3, S4). GSTs are abundant proteins encoded by a highly divergent and ancient gene family (Edwards et al. 2000). GST plays a role in resistance to FHB by reducing the toxicity of F. graminearum, for example, by modulating the activity of DON (Wang et al. 2020; Hu et al. 2022; Wu et al. 2022). TraesCS5B02G303700 and TraesCS5B02G303800 are annotated as NADP-dependent alkenal double bond reductase. TraesCS5B02G303700 has five non-synonymous substitutions (Gln25His Lys72Asn, Gln155His, Ile342Thr and Arg407Stop) in exons, one SNP in 3'UTR and one SNP in the intron (Table S3, S4). TraesCS5B02G303800 has two non-synonymous substitutions (Arg210Gln and Pro339Ala) and one synonymous variant in exons. The expression levels were significantly different between the genotypes of TN18 and LM6 (Table S3, S4). Bradi4g39950 gene (encoding an NADP-dependent alkenal double bond reductase P1) of Brachypodium distachyon has exhibited differential accumulation in spikes inoculated with the DON− and the DON+ strains. Bradi4g39950 was chosen as one of the representative samples of all differential regulation patterns for qRT-PCR analysis (Pasquet et al. 2014). TraesCS5B02G304000 is annotated as an F-box protein with one non-synonymous substitution (Lyr177His) in the exon and one InDel in the 5’UTR. The expression levels were significantly different between the genotypes of TN18 and LM6 (Table S3, S4). F-box genes are one of the largest multigene superfamilies in plants, where they control many crucial processes, such as embryogenesis, senescence, and pathogen resistance (Lechner 2006; Xu et al. 2009). An F-box protein could be involved in reducing the protein levels of a susceptible factor for FHB in wheat (Schweiger et al. 2016). RNA profiling of barley spikes revealed that an F-box domain containing protein (HD08H17r_at) showed significant differential expression between DON and water-treated samples, and the increased transcript accumulation were validated by qRT-PCR (Gardiner et al. 2010).

TraesCS7A02G568400 of QNds/Rdi-7AL-14499 is annotated as a disease resistance family protein/LRR family protein with one non-synonymous substitution (Met641Thr) in the exon and no SNP/InDel in the promoter. The expression levels were not significantly different between the genotypes of TN18 and LM6 (Table S3, S4). Many plant disease resistance proteins contain a nucleotide-binding site (NBS), a series of leucine-rich repeats (LRRs), and a putative amino-terminal signaling domain. They are referred to as NBS-LRR proteins (Belkhadir et al. 2004). NBS-LRR resistance genes play crucial roles in pathogen reception and signal transduction. NBS-LRR proteins are relevant factors in the interaction between F. graminearum and wheat and are involved in a certain stage in the host defense response (Kugler et al. 2013).

Five candidate genes were involved in plant defense responses against pathogens

The homologous genes of five genes, including TraesCS1A02G037600LC for QNds-1AS-2225, TraesCS1D02G017800 and TraesCS1D02G017900 for QNds-1DS-527, TraesCS1D02G018000 for QRdi-1DS-575 and TraesCS4B02G227400 for QNds/Rdi-4BL-4553, were involved in plant defense responses against pathogens. These genes should be likely associated with FHB resistance in wheat.

TraesCS1A02G037600LC of QNds-1AS-2225 is annotated as a suppressor of npr1-1 constitutive4 (SNC4) with one non-synonymous substitution (Met156Thr) in the exon and one SNP in the promoter. The expression levels were significantly different between the genotypes of TN18 and LM6 (Table S3, S4). The gene of Arabidopsis NPR1 is a positive regulator of systemic acquired resistance (SAR), which is essential for transducing the SAR signal salicylic acid (SA) (Li et al. 1999). AtSNC1 plays important roles in defense responses (Yeon et al. 2023). The snc1 mutation results in resistance against both bacterial and fungal pathogens (Zhang et al. 2003).

TraesCS1D02G017800 and TraesCS1D02G017900 genes were the two candidate genes of QNds-1DS-527. TraesCS1D02G018000 gene was the only candidate gene of QRdi-1DS-575. These three genes are all annotated as receptor-like kinases (RLKs). TraesCS1D02G017800 has three non-synonymous substitutions (Ile231Leu, Pro265Gln and Arg587Ser) in the exon, four SNPs and one InDel in the intron, and eight SNPs in the promoter. TraesCS1D02G017900 has four non-synonymous substitutions (His670Leu, Val646Phe, Leu475Arg and Leu118Ile) in the exon, and two SNPs in the promoter. TraesCS1D02G018000 has three non-synonymous substitutions (His17Asn, Ser63Ala and Ile501Asn) and one synonymous variant in the exon, and one SNP in the intron, and one SNP in the promoter. The expression levels of the three genes were all significantly different between the genotypes of TN18 and LM6 (Table S3, S4). RLKs play substantial roles in many aspects of plant biology, and some of them are implicated in plant defense responses (Morris et al. 2003). Some RLKs recognize specific pathogens (Gomez-Gomez et al. 2002), and other RLKs are associated with pathogen invasion (Silva et al. 2002).

TraesCS4B02G227400 of QNds/Rdi-4BL-4553 is annotated as a subtilisin-like protease with one synonymous variant in the exon, and no SNP/InDel in the promoter. The expression levels were not significantly different between the genotypes of TN18 and LM6 (Table S3, S4). Subtilisin-like proteases or subtilases are a very diverse family of serine peptidases exist in many organisms, but mostly in plants. Subtilases have been gaining increasing attention regarding their involvement in plant defense responses against highly diverse pathogens (Figueiredo et al. 2018).