Research Article

Horizontal gene transfer of Fhb7 from fungus underlies Fusarium head blight resistance in wheat

See allHide authors and affiliations

Science  22 May 2020:
Vol. 368, Issue 6493, eaba5435
DOI: 10.1126/science.aba5435

Fungal disease meets its match

Fusarium head blight (FHB), caused by a fungus, reduces wheat crop yield and introduces toxins into the harvest. From the assembly of the genome of Thinopyrum elongatum, a wild relative of wheat used in breeding programs to improve cultivated wheat, Wang et al. cloned a gene that can address both problems (see the Perspective by Wulff and Jones). The encoded glutathione S-transferase detoxifies the trichothecene toxin and, when expressed in wheat, confers resistance to FHB.

Science, this issue p. eaba5435; see also p. 822

Structured Abstract


Fusarium head blight (FHB) is a fungal disease that devastates global wheat production, with losses of billions of dollars annually. Unlike foliar diseases, FHB occurs directly on wheat spikes (inflorescences). The infection lowers grain yield and also causes the grain to be contaminated by mycotoxins produced by the Fusarium pathogen, thus imposing health threats to humans and livestock. Although plant breeders have improved wheat resistance to FHB, the lack of wheat strains with stable FHB resistance has limited progress.


Many genetic loci in wheat affect FHB resistance but most only have minor effects; only a few exhibit a stable major effect on resistance. Wheat relatives in the Triticeae tribe carry resistant genes to different diseases including FHB and thus can be alternative sources of FHB resistance for wheat breeding. Thinopyrum wheatgrass has been used as a source of beneficial genes transferable to wheat by distant hybridization breeding since the 1930s. Fhb7, a gene transferred from Thinopyrum to wheat, showed a stable large effect on FHB resistance. However, the lack of a Thinopyrum reference genome hampered gene cloning and marker development, delaying the use of Fhb7 in wheat breeding. Here, we cloned Fhb7 using a reference assembly that we generated for Th. elongatum and characterized its resistance mechanisms and evolutionary history.


Using sequence data from Th. elongatum, we assembled the Triticeae E reference genome with 44,474 high-confidence genes annotated. Using this reference, we genetically mapped Fhb7 and located it to a 245-kb genomic region. We determined a gene encoding a glutathione S-transferase (GST) as Fhb7 by virus-induced gene silencing and evaluated mutants and transgenic plants. We discovered that Fhb7 detoxifies pathogen-produced trichothecene toxins by conjugating a glutathione (GSH) unit onto the epoxide moieties of type A and B trichothecenes. Fhb7 GST homologs are absent in the plant kingdom, but one sequence showing ~97% identity with Fhb7 was found in endophytic fungi of an Epichloë species that establishes symbiosis with temperate grasses. This result suggests that Fhb7 might have been transferred from Epichloë to Th. elongatum through horizontal gene transfer. Finally, we demonstrated that Fhb7, when introgressed into diverse wheat backgrounds by distant hybridization, confers broad resistance to both FHB and crown rot without penalizing wheat yield. Our results suggest a source of Fusarium resistance for wheat improvement.


Th. elongatum carries biotic and abiotic resistance genes and is a useful resource for wheat breeding. The assembled Th. elongatum reference genome can aid identification and cloning of such genes for wheat improvement. Cloning of Fhb7 revealed that it encodes a GST that can detoxify trichothecene toxins. Thus, Fhb7 resistance differs from Fhb1 resistance, which depends on a reduction of pathogen growth in spikes, although both confer durable resistance. The ability of Fhb7 to detoxify multiple mycotoxins produced by various Fusarium species demonstrates its potential as a source of resistance to the various diseases for which Fusarium trichothecenes are virulence factors. The deployment of Fhb7 in commercial wheat cultivars could alleviate both the food safety issue for consumers and the yield loss problem for growers. Sequence homologies between fungal and plant Fhb7 suggested that horizontal gene transfer may help to shape plant genomes.

Fhb7 confers FHB resistance by detoxifying trichothecenes.

(A) Fhb7 in Th. elongatum genome likely came from an Epichloë fungus through horizontal gene transfer. Fhb7 drives FHB resistance when introgressed from Thinopyrum into wheat. (B) Fhb7 encodes a GST that detoxifies Fusarium-produced trichothecenes by conjugating GSH (blue) to the epoxy group (red). R1 to R5 refer to the variable groups in trichothecenes.


Fusarium head blight (FHB), a fungal disease caused by Fusarium species that produce food toxins, currently devastates wheat production worldwide, yet few resistance resources have been discovered in wheat germplasm. Here, we cloned the FHB resistance gene Fhb7 by assembling the genome of Thinopyrum elongatum, a species used in wheat distant hybridization breeding. Fhb7 encodes a glutathione S-transferase (GST) and confers broad resistance to Fusarium species by detoxifying trichothecenes through de-epoxidation. Fhb7 GST homologs are absent in plants, and our evidence supports that Th. elongatum has gained Fhb7 through horizontal gene transfer (HGT) from an endophytic Epichloë species. Fhb7 introgressions in wheat confers resistance to both FHB and crown rot in diverse wheat backgrounds without yield penalty, providing a solution for Fusarium resistance breeding.

Wheat (Triticum aestivum L.) is a leading source of calories for the human population (1). The prevalence and widespread outbreaks of the devastating Fusarium head blight (FHB) disease, exacerbated by recent changes in climate and certain cropping practices, has posed a threat for global wheat production and food safety. Fusarium species cause not only FHB in wheat, barley, and oat, but also crown rot in wheat and ear rot in maize. However, F. graminearum is the prominent pathogen of wheat FHB in China, the United States, Canada, Europe, and many other countries (2). Fusarium produces epoxy-sesquiterpenoid compounds known as trichothecenes, some examples of which are deoxynivalenol (DON), T-2 toxin, HT-2 toxin, and nivalenol (NIV), among others; these compounds are inhibitors of protein synthesis and virulence factors for pathogenicity (2). Trichothecene contamination in cereal grain results in immunotoxicity and cytotoxicity in humans and animals and thus has aroused public safety concerns (3). Despite global screening efforts examining tens of thousands of wheat accessions, a limited number of quantitative trait loci (QTLs) have been verified to confer a stable effect on FHB resistance (4). Fhb1 on chromosome 3B is the only QTL that has been used in breeding programs worldwide. Although it has been cloned from different Chinese wheat sources, its molecular identity and resistance mechanisms remain equivocal (58).

Wheat relatives have proven to be alternative sources for improvement of resistance to both biotic and abiotic stresses in wheat (9). Distant hybridization, the practice of making crosses between two different species, genera, or higher-ranking taxa, makes it possible to transfer alien genes from Triticeae tribe relatives to wheat (911). Tall and intermediate wheatgrasses of the Thinopyrum genus (forage grasses) are sources of resistance to salinity, drought, and disease for wheat. Several disease resistance genes, including stem rust (e.g., Sr24, Sr25, Sr26, Sr43, Sr44, and SrB), leaf rust (Lr19, Lr24, Lr29, and Lr38), powdery mildew (Pm40 and Pm43), barley yellow dwarf virus (Bdv2 and Bdv3), and Fusarium head blight (Fhb7), have been introduced from Thinopyrum into wheat for resistance breeding (10, 1216).

Fhb7 is a QTL introduced from Thinopyrum elongatum and shows a similar effect on FHB resistance as Fhb1. Th. elongatum (syn. Agropyron elongatum or Lophopyrum elongatum), a grass of the Triticeae family with a diploid E genome (2n = 2x = 14), is native to Eurasia and is thought to be a genome donor species for various tetra-, hexa-, and even decaploid species in the Thinopyrum genus (14). The lack of a reference sequence for the E genome has impeded the process of cloning and the development of diagnostic markers for the deployment of Fhb7 and other E genome–derived resistance genes. Here, we report the assembly of a reference genome for Th. elongatum and describe the cloning and biomolecular characterization of Fhb7. Using the newly assembled E genome reference, we identified a GST gene as a candidate for Fhb7 by map-based cloning and confirmed its function in FHB resistance using transgenics. Fhb7 can detoxify trichothecenes by catalyzing the conjugation of a glutathione (GSH) unit onto their toxic epoxide moiety. Fhb7’s coding sequence has no obvious homology to any known sequence in the entire plant kingdom but shares 97% sequence identity with a species of endophytic fungus (Epichloë aotearoae) known to infect temperate grasses, which provides evidence that Fhb7 in the Th. elongatum genome might be derived from the fungus through HGT. We demonstrate here that Fhb7 confers resistance to both FHB and crown rot without yield penalty in wheat.


Th. elongatum genome assembly and evolution

To sequence and assemble the genome of Th. elongatum, 1.1 Tb of high-quality sequence reads were generated from a series of libraries, which is about 236× coverage of the Th. elongatum genome (table S1). We initially assembled the short sequence reads using DeNovoMAGICTM3.0 software (NRGene) and then filled the gaps using ~145 Gb (~31×) PacBio SMRT reads. The initial assembly was finely tuned using high-quality paired-end polymerase chain reaction (PCR)–free reads. Two Bionano optical maps (based on enzymes BspQI and DLE1 data) were further used to extend the scaffolds (tables S2 and S3), which resulted in a 4.63-Gb assembly with a contig N50 size of 2.15 Mb and a scaffold N50 size of 73.24 Mb (Table 1).

Table 1 Summary statistics for Th. elongatum genome assembly.

View this table:

To construct the pseudochromosomes, high-throughput chromosome conformation capture (Hi-C) data were used to categorize and order the assembled scaffolds (table S4). A total of 141 scaffolds were anchored and oriented onto seven pseudochromosomes, which account for 95% of the estimated genome size (4.78 Gb; fig. S1) and 98% of the assembled genome sequences (fig. S2). About 97.6% complete and 1.3% fragmented Embryophyta genes were detected in our assembly according to BUSCO [Benchmarking Universal Single-Copy Orthologs (17)], proportions comparable to other Triticum genomes (table S5). The quality of the E genome assembly was validated by assessment of the long terminal repeat (LTR) completeness using LTR Assembly Index (LAI) software (18) (table S6), by genomic alignment with 61 randomly selected bacterial artificial chromosome (BAC) clones (fig. S3 and table S7), and by the consistency of our assembly with a high-density genetic map from a hexaploid Thinopyrum species (19) (fig. S4).

Repetitive elements are dispersed throughout the E genome, with ~81.29% of the Th. elongatum assembly being annotated as repetitive elements, including retrotransposons (62.39%), DNA transposons (17.83%), and unclassified elements (1.07%) (table S8 and table S9). Analysis of the Cereba and Quinta LTR retrotransposons supported that the centromere regions were appropriately assembled (fig. S5). The composition of different classes of repetitive DNA in the E genome was similar to those of the wheat A, B, or D subgenomes (fig. S6). No recent LTR burst was detected in the E or common wheat genomes (fig. S7), suggesting relatively stable genomes and helping to explain the success of distant hybridization breeding efforts using these materials. A total of 44,474 high-confidence protein-coding genes were predicted on the basis of a combination of methods [ab initio, protein homology based, and RNA-sequencing (RNA-seq) based], and 44,144 (99.3%) of the predicted genes were anchored onto the seven assembled pseudochromosomes (figs. S8 and S9 and tables S10 to S12).

Gene family analysis identified 32,048 orthologous genes between the E genome and the wheat A, B, or D genome or the barley genome (fig. S10). A synonymous substitution rate (Ks) value was calculated using a moving-average model with the ortholog dataset, which revealed similar Ks peak values between the E genome and the wheat subgenomes (E and A: 0.0645, E and B: 0.0645, E and D: 0.062), indicating a branching time for Th. elongatum and Triticum of ~4.77 to 4.96 million years ago when a nucleotide substitution rate of 6.5*10−9 was used (Fig. 1A) (20).

Fig. 1 Genome evolution of Th. elongatum and cloning of Fhb7.

(A) Maximum likelihood phylogenetic tree of the genomes of Triticeae species and the Ks distributions of ortholog genes between the E genome and the wheat Chinese Spring A, B, and D subgenomes. mya, million years ago. (B) Syntenic blocks between the E genome and the three wheat subgenomes. The representative inversion fragment is indicated in green; chromosomal translocations for the wheat A subgenome compared with the E genome are also indicated in blue. (C) Map-based cloning of Fhb7 at the distal region of chromosome 7E. Using the BC6F1 population derived from the cross between two wheat-Thinopyrum substitution lines, 7E1/7D and 7E2/7D, Fhb7 was initially mapped to an interval between the markers XsdauK79 and XsdauK80 (~1.2 Mb on the E reference genome) (second bar from the top). The expressed genes are labeled as follows: gray refers to no expression in the E reference genome; blue refers to E reference genome–specific expression; orange refers to expression in the E reference, 7E1 and 7E2 genomes; red refers to expression in FHB-resistant donor genomes of 7E2 and E reference (third bar from the top). BAC clones containing Tel7E01T1020600.1 were identified from the substitution line 7E2/7D, based on which genetic markers (XsdauK86 and XsdauK87) were developed for recombinant screening of the CS ph1bph1b population. Finally, Fhb7 was genetically confirmed within a 245-kb region between markers XsdauK86 and XsdauK88, with only the candidate gene Tel7E01T1020600.1 encoding a GST [CDS is shown in red; untranslated region is shown in gray] (fourth bar from the top). (D) FHB was evaluated for wild-type (WT, KN199) and transgenic wheat KN199 expressing the native promoter and the 846-bp open reading frame of Fhb7. T3 plants containing Fhb7 from three different lines were evaluated for FHB resistance using single floret inoculation (35). The FHB was scored for at least five spikes per repeat, with at least three repeats for each transgenic line.

We also compared the E genome with other Triticeae genomes that have been used for distant hybridization based on a maximum likelihood tree built using single-copy genes from available Triticeae genome assemblies; the tree also incorporated transcript data for several diploid species, including the Triticeae R, Q, V, F, and Ns genomes (table S13). The three wheat subgenomes are more closely related to the E genome of Th. elongatum than they are to the R genome of rye, another species frequently used in wheat distant hybridization (Fig. 1A). A syntenic block analysis indicated genome-wide colinearity between the E genome and the A, B, or D genomes, which helps to explain the success of E-genome–based distant hybridization breeding in wheat (Fig. 1B and data S1). Substantial colinearity notwithstanding, we did identify 18 fragmental inversions between the E genome and the wheat subgenomes, with sizes ranging from 1.5 to 18 Mb, which is supported by both the Bionano maps and Hi-C data (fig. S11 and table S14).

Map-based cloning of the Fusarium resistance gene Fhb7

A total of 1897 resistance gene analogs (RGA) were annotated in the E genome (fig. S12 and table S15). An apparent RGA expansion, especially for CC-NBS-LRR (CNL), on the distal end of the long arm of chromosome 7E (7EL) is accompanied with the expansion of this genomic region (fig. S13 and table S16). Some of the alien resistance gene introgressions into wheat are located in this region, including Lr19, Sr25, Bdv3, and Fhb7 (10, 13, 14).

Previously, we mapped the Fhb7 to the distal end of the 7EL (based on recombination between 7E1 and 7E2 in a common wheat background) using a recombinant inbred line (RIL) population from a cross between an FHB-susceptible substitution line (7E1/7D) and an FHB-resistant substitution line (7E2/7D) (13, 21). For further mapping of this gene, we developed a segregation population derived from BC6F1 with the same cross, in which FHB resistance was tracked as monogenic trait for validation of phenotypes. We also developed a population to promote 7E recombination by introducing the CS ph1bph1b locus (fig. S14). Because of the semidominant nature of Fhb7, the homozygous offspring of the recombinants were evaluated for FHB resistance. With analysis of 258 recombinants (between the XBE45653 and XsdauK67 markers) screened from 19,200 progeny of BC6F1 population, we confirmed that Fhb7 is positioned between the XSdauK79 and XSdauK80 markers within an ~1.2-Mb region based on the E reference genome (Fig. 1C and fig. S15).

Analysis of the RNA-seq data of E reference genome from Th. elongatum spikes identified eight expressed genes in the Fhb7 region (Fig. 1C and table S17). However, when conducting transcriptomics analysis of the parental lines of 7E1/7D (S) and 7E2/7D (R), we found that only two candidate genes (Tel7E01T1020600.1 and Tel7E01T1021800.1) were expressed in a manner specific to the 7E2 genome (the resistant parent) and E reference genome [which also confers FHB resistance (12, 22)] (Fig. 1C and tables S18 and S19). BAC clones containing Tel7E01T1020600.1 were identified from the resistant donor line and new markers (XsdauK86 and XsdauK87) derived from the BAC ends were developed to screen recombinants among 5760 progeny of the segregation population harboring the CS ph1bph1b locus (Fig. 1C, fig. S14, and table S20). Analysis of phenotypic data of the three key recombinants verified that Fhb7 is located between the XsdauK86 and XsdauK88 markers, thereby delineating this locus to a 245-kb region containing a single expressed gene: Tel7E01T1020600.1 (Fig. 1C). This gene is present in the E reference genome and 7E2 genome but absent in the susceptible 7E1 genome based on analysis of genomics and transcriptomics data (table S19 and table S21).

Gene expression analysis using quantitative PCR indicated that Tel7E01T1020600.1 was constitutively expressed in all tissues examined, including root, leaf, shoot, and spike (fig. S16). Moreover, barley stripe mosaic virus (BSMV)–induced gene silencing of Tel7E01T1020600.1 in wheat leaves revealed that it conferred resistance to F. graminearum, supporting that this gene represents Fhb7 (fig. S17). Sequence analysis of 22 ethyl methanesulfonate (EMS)–induced mutants identified five amino acids that were implicated in Fhb7’s FHB resistance–related function: S34F, T48I, A98V, A9V, and P106L (fig. S18 and data S2). Moreover, two stop-gain mutations at position 209 or 243 led to reduced resistance to F. graminearum (fig. S18 and data S2). To confirm Tel7E01T1020600.1 as Fhb7, we transgenically introduced a construct with the native promoter and the 846–base pair (bp) coding sequence of this gene into the FHB-disease–susceptible wheat cultivar KN199 and assessed three independent T3-transgenic plants. The Fusarium-inoculated transgenic plants exhibited lower FHB symptom with substantially fewer diseased spikelets per spike than the control (Fig. 1D).

Evolutionary history and molecular function of Fhb7

Protein domain–based functional annotation predicted that Fhb7 likely encodes a GST enzyme. A BLAST search of the Fhb7 sequence against the National Center for Biotechnology Information (NCBI) GenBank database (23) did not find any homolog of Fhb7 in the Triticum genus or in the entire plant kingdom. However, there is a homolog sharing 97% identity in the genome of E. aotearoae (Fig. 2A and fig. S19). A phylogenetic analysis of the Fhb7 sequence revealed its distribution among Epichloë species, endophytic fungi of temperate grasses (Fig. 2A). Thus, the occurrence of the Fhb7 gene in the Th. elongatum genome might be caused by fungus-to-plant HGT (FP-HGT) event. Because the Fhb7 locus is present both in the diploid E genome of Th. elongatum and in 7E2 from decaploid Th. ponticum, this FP-HGT event apparently occurred after the divergence of the E genome from Triticum sp. but before the formation of the decaploid Th. ponticum (Fig. 2A).

Fig. 2 Fhb7 confers FHB resistance by detoxifying DON.

(A) Maximum likelihood phylogenetic tree of the closest homologs of Fhb7 from plants and fungi. The DNA sequence similarity with Fhb7 is marked in red. (B) Horizontal gene transfer of Fhb7. The transcripts CDS (purple), and possible untranslated regions (yellow) of Fhb7 are shown along chromosome 7E, and the sequence sharing high similarity with the E. aotearoae genome is presented as a gray block. The genomic fragment (897 bp) containing full CDS and partial untranslated region of Fhb7 showed 97% identity between the two genomes. (C) DON tolerance of Fhb7-transgenic wheat. Seedlings (4 days old) were moved to a petri dish containing 25 mg L−1 DON and seedling length was evaluated 7 d after the DON treatment at room temperature. (D) Extracted ion chromatograms (EICs) at m/z 604.2173 revealing the presence of two DON-glutathione adducts. The Fhb7 NIL, Fhb7-transgenic wheat, and Fhb7-transgenic yeast (P. pastoris) cultures were treated with 25 mg L−1 DON for 24 hours. A product that elutes at 1.68 min accumulated in Fhb7(+) samples and a known, nonenzymatically produced DON-glutathione adduct product that elutes at 2.4 min accumulated in the corresponding Fhb7(–) control samples. (E) Relative abundances of the de-epoxidated Fhb7-catalyzed DON-glutathione (green) adduct and the known nonenzymatic DON-glutathione adduct (blue) in spikes of Fusarium-challenged NIL plants contrasting in Fhb7. After inoculation of F. graminearum on spike glumes, the Fhb7(+) NIL accumulated a copious amount of de-epoxidated DON-glutathione adduct. By contrast, the DON substrate reduced the accumulation in Fhb7(+) NIL compared with that in Fhb7(–) NIL, as shown in the bottom bar chart. (F) Molecular structure of the de-epoxidated DON-glutathione adduct catalyzed by Fhb7.

The horizontal transfer of the Fhb7 sequence did not occur as a part of a gene cluster (presuming that it is from E. aotearoae as the donor genome; this is the species harboring the closest identified homolog of Fhb7) (fig. S20). On the basis of sequence similarity, the sequence was transferred into the diploid E genome as a short fragment, including the 846-bp coding sequence for Fhb7, a 32-bp sequence before the start codon, and a 19-bp sequence after the stop codon (Fig. 2B). At the position 535 bp upstream of Fhb7’s start codon in the E genome, another 90-bp sequence shows high identity to a sequence in E. aotearoae (Fig. 2B), suggesting the possibility that a larger sequence was initially transferred to Th. elongatum but late mutations occurred in the transferred sequence. The insertion of the Epichloë genome fragment in the E genome was also identified in a BAC clone harboring Fhb7 (Fig. 1C and data S3), confirming that the sequence is not an artifact from the genome assembly process.

Phylogenetic analysis of the GST superfamily showed that Fhb7 belongs to the fungal GTE (glutathione transferase etherase–related) subfamily (fig. S21 and tables S22 and S23), wherein all members contain a LigE domain, but none of which has been functionally characterized to date (24). The Fhb7 gene is conserved in Epichloë species and in multiple Thinopyrum species, emphasizing its role in protecting organisms from the cytotoxic damage caused by Fusarium species (Fig. 2A and fig. S20). Gene expression analysis in a time course of Fusarium infection in Th. elongatum and the 7E2/7D substitution line (table S18) showed that the transcription levels of Fhb7 were induced at 48 hours after infection (fig. S22).

Research in plant pathology about the progression of F. graminearum infection in wheat has established that the fungus starts to produce its DON mycotoxin, an inhibitor of protein synthesis that targets ribosomal machinery, by the 48-hour infection time point (25). We therefore conducted DON assays on wheat seedlings of the 7E2/7D substitution line. The results showed that the expression of Fhb7 can be induced within 6 hours after DON treatment (fig. S22), suggesting that this putative GST enzyme may have a role in xenobiotic detoxification. To test this hypothesis, we conducted a growth inhibition assay by growing Fhb7 near-isogenic lines (NILs) and Fhb7-transgenic wheat seedlings in media containing DON and found that the plants with Fhb7 grew better (assessed as seedling length) than the plants without Fhb7 (Fig. 2C and fig. S23). We also expressed Fhb7 in yeast to test its growth on DON-containing media and found that both the Fhb7(+) and Fhb7(–) yeasts grew well in the absence of DON; however, only the Fhb7(+) yeast grew normally on the media containing 400 mg L−1 DON (fig. S24).

Further evidence for the involvement of Fhb7 in detoxification was demonstrated by its direct use of DON as a substrate. We treated the seedlings of NILs, Fhb7-transgenic wheat, and Fhb7-expressing yeast cultures with DON, and found that the presence of Fhb7 in wheat and yeast caused accumulation of a chromatographic peak at 1.68 min, but the accumulation was not detected in the corresponding control samples without Fhb7 (Fig. 2D). This peak had a mass/charge (m/z) value of 604.2173 (± 3 ppm) under positive ion mode, which is equal to the value for the molecule comprising DON (296.1259), a glutathione group (307.0838), and a hydrogen atom (1.0078), therefore suggesting that Fhb7 confers GST activity to form a glutathione adduct of DON (DON-GSH) (Fig. 2D and fig. S25).

Previous studies on FHB- and DON-associated chemistry (26, 27) using nuclear magnetic resonance spectroscopy confirmed the nonenzymatic formation of a DON-GSH adduct that was formed through a reaction with the double bond at C10 on DON’s first planar ring. This product was mainly detected in the DON-treated Fhb7(–) yeast cultures and Fhb7(–) wheat samples with the peak at 2.4 min (Fig. 2D and fig. S25). Although the two detected DON-GSH isomers had identical m/z values, tandem mass spectrometry with collision-induced dissociation experiments unequivocally supported that the Fhb7(+) samples produce a de-epoxidated DON-GSH adduct (figs. S25 to S28); that is, the GSH group added by Fhb7 is attached to the C13 carbon, which disrupts the epoxy group known to be critical in DON’s toxicity (Fig. 2F) (28). Further, we used liquid chromatography–high-resolution mass spectrometry (LC-HRMS) to profile DON-treated spikes from 37 diverse wheat germplasm accessions and cultivars without Fhb7. We detected the DON-GSH (C10) peak at 2.4 min in all of these plants but did not detect the 1.68-min de-epoxidated DON-GSH (C13) adduct in any of them (fig. S29).

Fusarium species produce a series of trichothecene mycotoxins, including DON, 3-ADON, 15-ADON, T-2, HT-2, fusarenon-X, NIV, diacetoxyscirpenol, and others, the distribution of which varies among Fusarium chemotypes (24, 26). Considering the common occurrence of epoxy groups in these trichothecene compounds, we hypothesized that Fhb7 may be able to detoxify trichothecenes other than DON. Indeed, LC-HRMS analysis of trichothecene-treated wheat samples revealed the presence of GSH adducts for all the trichothecenes that we tested in this study (figs. S30 to S37). In light of Fhb7’s wide catalytic spectrum for these mycotoxins, we investigated whether it can confer resistance to other Fusarium chemotypes, including F. pseudograminearum for crown rot and F. asiaticum, a predominant FHB-causing strain in south China. Assays using detached wheat leaves showed that the Fhb7-transgenic plants exhibited smaller lesions than wild-type plants for all the tested Fusarium species (fig. S38). F. pseudograminearum was also inoculated on the base of wheat seedlings, and the results confirmed that the transgenic plants also exhibit improved crown rot resistance compared with the nontransgenic controls (fig. S39). These results further demonstrate how Th. elongatum benefits from Fhb7 through the FP-HGT event, which protects plants from Fusarium-caused cytotoxic damage by detoxifying trichothecene through de-epoxidation (fig. S20).

Application of Fhb7 in Fusarium resistance breeding

Considering Fhb7’s functionality, specifically in the enzymatic conversion of trichothecenes, we speculated that incorporating the Fhb7 locus into wheat may confer resistance in different genetic backgrounds without affecting yield traits. Indeed, the translocation of a short fragment [with ~16% of the 7E long arm (13)] on wheat 7D resulted in wheat lines with broad resistance to both FHB and crown rot (Fig. 3, A to C). Detailed characterization of NILs (LX99 background) in field conditions showed no significant difference in agronomic yield traits (e.g., thousand grain weight, flag leaf length, etc.; Fig. 3, D and E). Obvious yield penalty caused by Fhb7 resistance was also not detected when it was transferred into seven additional genetic backgrounds (Fig. 3F and fig. S40). These results demonstrated the advantages of Fhb7-mediated resistance over other QTLs, including high resistance to both FHB and crown rot and detoxifying DON without yield penalty, and thus highlighted the potential utility of the Fhb7 locus in future wheat breeding for improved FHB resistance and good yield traits.

Fig. 3 Application prospects for Fhb7 in wheat resistance breeding.

(A) Genomic in situ hybridization analysis (left panel) showing a translocation of the distal region of 7E (containing Fhb7) from an E genome donor into wheat. Scale bar, 20 μm. Also shown are images of Fusarium-infected spikes (middle panel) and crown rot (right panel) of LX99 NILs contrasting in Fhb7. (B) FHB resistance of Fhb7 in eight different wheat genetic backgrounds evaluated at 21 d after inoculation in field conditions. (C) Crown rot phenotypes were recorded as the death ratio after growth in soil containing F. pseudograminearum at 30 days postinfection. (D) Field plant photographs of two Fhb7(+) NILs and one Fhb7(–) NIL in the LX99 background. (E) Comparison of the yield traits among the two Fhb7(+) NILs and one Fhb7(–) NIL in the LX99 background evaluated in the 2017 field experiment. FLL, flag leaf length (cm); FLW, flag leaf width (cm); SL, spike length (cm); KPS, kernels per spike; IS, infertile spikelets; GL, grain length (mm); GW, grain width (mm); TGW, thousand grain weight (g). (F) Comparison of the grain yield among eight Fhb7 translocation lines in different wheat genetic backgrounds. The grain yield was measured from a 1-m2 plot in the 2017 and 2018 field experiments.


Fusarium diseases are economically impactful because of their effects on the production of cereal crops. In this study, the successful cloning of Fhb7 from the Triticeae E genome and characterization of its molecular mechanism advances the knowledge on the essential role of trichothecenes in the pathogenesis of Fusarium. We have demonstrated that Fhb7 confers FHB resistance in diverse wheat genetic backgrounds without yield penalty and Fhb7 is able to biochemically detoxify trichothecene mycotoxins produced by multiple Fusarium species, which highlights the value of Fhb7 in combating FHB and reducing DON contamination in wheat and other cereal crops through breeding.

The epoxides at the C12/13 of trichothecene mycotoxins are the key contributors to their toxicity. However, to date, genes or proteins with de-epoxidation function have not been identified (3). Fusarium species can reduce DON toxicity by adding an acetyl group on the hydroxyl group at C3 and C15, forming 3-ADON and 15-ADON, respectively; however, the reduction of cytotoxicity for these DON derivatives is modest in plant cells (3). In planta, glucosylation at C3 has been documented to detoxify DON by forming DON-3-glucoside (D3G), which is reversible in animals, causing release of DON during digestion (29). Here, beyond our identification of an FHB resistance gene, the broad detoxification spectrum of Fhb7 through de-epoxidation of trichothecenes suggests the potential utility of the GST enzyme in the biomedicine, feed, and food industries in addition to reducing DON content in wheat grain.

HGT, the transfer of genes between nonmating species, is thought to occur frequently in prokaryotes, but much less so in eukaryotes (30). There is accumulating evidence illustrating instances of HGT events involving bacteria or the organellar genomes of another plant as donor (31). For instance, two Agrobacterium genes were found to be inserted in the genome (with transfer DNA borders) of a cultivated sweet potato [Ipomoea batatas (L.) Lam.], revealing a naturally occurring transgenic food crop (32). However, there is little evidence for HGT events involving nuclear DNA transmission from fungi or other eukaryotes, and such transmission has been thought to be insignificant (33). Fundamentally, our results highlight the roles that FP-HGT has had in shaping plant genomes, which advances the knowledge on disease resistance gene evolution and opens a new avenue for the identification of plant resistance genes.

The endophytic Neotyphodium and Epichloë fungi often form mutualistic symbiotic associations with forage grasses and offer hosts bioprotective benefits against pathogens and abiotic stresses, presumably owing to the fungus-mediated anabolism and catabolism of various natural product compounds (34). Here, we showed that the GST encoded by Fhb7 is conserved in Epichloë species and is able to detoxify the trichothecene mycotoxins secreted by Fusarium species. Thus, transfer of this fungal gene into a plant genome could be beneficial to plants, perhaps even eliminating the need for the symbiotic association per se. The finding of Fhb7-mediated resistance to both FHB and crown rot diseases further emphasizes the importance of this HGT in benefiting the perennial Th. elongatum, which is perhaps reflected by constitutive expression of Fhb7 in all examined tissues. However, the molecular machinery that enabled the FP-HGT of Fhb7 and the nature of the promoter evolution underlying the expression of Fhb7 remain to be elucidated.

Methods summary

The Th. elongatum genome was first sequenced by Illumina short-read sequencing and was de novo assembled using the software package DeNovoMAGICTM3.0. PacBio SMRT long reads were used to fill the gaps in the assembly and Bionano optical maps were then used to correct and extend the scaffold sequences. The assembly was anchored into seven pseudochromosomes using Hi-C data. The assembly was validated using independent BAC sequences, genetic maps of related species, and commonly used software programs. Genes, repetitive DNA, and other genomic features in the assembly were annotated to reveal the landscape of the species and to examine their relationship with wheat and other related species by in-depth comparative analyses. Genetic markers in the Fhb7 region were developed by means of the reference genome sequence and used to screen recombinants for fine mapping to identify the Fhb7 candidate gene. The candidate gene was functionally validated by virus-induced gene silencing, EMS-induced mutation, and transgenic approaches. FHB resistance was evaluated by inoculation of Fusarium conidial suspensions on wheat spikes, leaves, or crowns. LC-HRMS(/MS) analysis was used to infer the biochemical structure of trichothecene-glutathione adducts catalyzed by Fhb7. Fhb7 was introgressed into diverse wheat backgrounds using distant hybridization and conventional breeding, and the presence of alien chromatin in wheat was validated by genomic in situ hybridization.

Supplementary Materials

Materials and Methods

Figs. S1 to S40

Tables S1 to S23

Captions for Data S1 to S3

References (3693)

MDAR Reproducibility Checklist

Data S1 to S3

References and Notes

Acknowledgments: We thank Q. Song (Lanzhou University) and X. Zhang (University of Louisville) for advice on entophytic fungi and biochemical analysis of trichothecenes, X. L. Zhang (Northeast Forestry University) for providing some marker sequences of the Triticeae E genome, and Y. Liang and J. Yu (Shandong Agricultural University) for providing the Fusarium stains. Funding: This work was supported by the National Natural Science Foundation of China (31520103911, 31871610, and 31901492), the National Key Research and Development Program (2016YFD0100102-2), the National Key Program on Transgenic Research from the Ministry of Agriculture of China (2016ZX08002003-002 and 2016ZX08009-003), and the Agricultural Variety Improvement Project of Shandong Province (2019LZGC016). Author contributions: H.W. and L.K. designed the project. S.S.X., G.B., E.N., C.G., and H.O. supervised the project. S.S., K.W., L.C., X.F., and F.N. performed bioinformatics analysis. W.G., L.Z., B.H., Z.L., S.S.X., J.G., M.L., P.S., X.F.L., G.W., C.B., W.Z., X.C., J.W., L.D., W.C., W.L., G.X., J.Z., Y.H., Y.X., Y.G., W.J.L., Y.L., H.Y., J.L., X.L., Y.Z., and X.W. conducted experiments. X.M. and A.L. performed the field work. H.W., S.S., G.B., and L.K. wrote the paper with input from all authors. Competing interests: The authors declare no competing interests. Patents with application nos. 2020101464009 and 202010146399X are pending. Data and materials availability: All data are available in the manuscript, the supplementary materials, or at publicly accessible repositories. These data in the public repositories include all raw reads and assembled sequence data for Th. elongatum in NCBI under BioProjectID PRJNA540081, the assembly and annotation data of Th. elongatum, and the draft genomes of 7E/7D substitution lines in the Genome Warehouse in BIG Data Center under accession numbers GWHABKY00000000, GWHABLF00000000, and GWHABLE00000000, which are accessible at All materials are available from L. Kong upon request.
View Abstract

Stay Connected to Science

Navigate This Article