The MLA6 coiled-coil, NBS-LRR protein confers AvrMla6-dependent resistance specificity to Blumeria graminis f. sp. hordei in barley and wheat

Authors

  • Dennis Halterman,

    1. Corn Insects and Crop Genetics Research, USDA-ARS, Iowa State University, Ames, IA 50011–1020 USA,
    2. Department of Plant Pathology, Iowa State University, Ames, IA 50011–1020 USA,
    Search for more papers by this author
  • Fasong Zhou,

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK
    Search for more papers by this author
  • Fusheng Wei,

    1. Department of Plant Pathology, Iowa State University, Ames, IA 50011–1020 USA,
    2. Interdepartmental Genetics Program, Iowa State University, Ames, IA 50011–1020 USA, and
    Search for more papers by this author
  • Roger P. Wise,

    Corresponding author
    1. Corn Insects and Crop Genetics Research, USDA-ARS, Iowa State University, Ames, IA 50011–1020 USA,
    2. Department of Plant Pathology, Iowa State University, Ames, IA 50011–1020 USA,
    3. Interdepartmental Genetics Program, Iowa State University, Ames, IA 50011–1020 USA, and
      For correspondence (fax +515 294 9240; e-mail rpwise@iastate.edu).
    Search for more papers by this author
  • Paul Schulze-Lefert

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK
    Search for more papers by this author
    • Present address: Max-Planck-Institut für Züchtungsforschung, Abteilung Biochemie, Carl-von-Linne-Weg 10, D-50829 Köln, Germany.


For correspondence (fax +515 294 9240; e-mail rpwise@iastate.edu).

Summary

The barley Mla locus confers multiple resistance specificities to the obligate fungal biotroph, Blumeria (= Erysiphe) graminis f. sp. hordei. Interspersed within the 240 kb Mla complex are three families of resistance gene homologs (RGHs). Probes from the Mla-RGH1 family were used to identify three classes of cDNAs. The first class is predicted to encode a full-length CC-NBS-LRR protein and the other two classes contain alternatively spliced, truncated variants. Utilizing a cosmid that contains a gene corresponding to the full-length candidate cDNA, two single-cell expression assays were used to demonstrate complementation of AvrMla6-dependent, resistance specificity to B. graminis in barley and wheat. The first of these assays was also used to substantiate previous genetic data that the Mla6 allele requires the signaling pathway component, Rar1, for function. Computational analysis of MLA6 and the Rar1-independent, MLA1 protein reveals 91.2% identity and shows that the LRR domain is subject to diversifying selection. Our findings demonstrate that highly related CC-NBS-LRR proteins encoded by alleles of the Mla locus can dictate similar powdery mildew resistance phenotypes yet still require distinct downstream signaling components.

Introduction

Resistance (R) gene products play an important role in the recognition of invading pathogens and the activation of responses that impede pathogen growth (Keen, 1990). The R-gene mediated response is dependent on the expression of a complementary pathogen avirulence (Avr) gene. If a host R gene is paired with an appropriate pathogen Avr gene, recognition occurs and an incompatible interaction ensues. This incompatibility results in a rapid signal cascade, leading to an active defense response. In the absence of either the R gene or the corresponding Avr gene, a compatible interaction occurs, and the pathogen is able to proliferate and cause disease. This genetic relationship between hosts and pathogens, termed a gene-for-gene interaction, is involved in resistance to a wide range of pathogen types including fungi, bacteria, viruses and nematodes (Baker et al., 1997; Bent, 1996; Flor, 1971; Van Der Biezen and Jones, 1998).

Resistance genes that function in a gene-for-gene manner generally belong to one of four general classes based on amino-acid motifs that are found within the encoded protein sequence (reviewed by Hammond-Kosack and Jones, 1997). Members of the largest class encode cytoplasmic proteins with a nucleotide-binding site (NBS) and several leucine-rich repeats (LRRs). Sequence diversity within the LRRs is thought to determine recognition specificity for proteins that are otherwise quite similar. Proteins encoded by the NBS-LRR class of resistance genes can be further subdivided into those with either a coiled-coil or Toll-interleukin-1 receptor (TIR) homology domain at the amino terminus, where they may have a function in directing certain protein–protein interactions (Lupas et al., 1991; Whitham et al., 1994).

R genes commonly belong to large, clustered families. These large arrays of similar sequences allow for recombination events that can lead to the evolution of gene products with novel recognition specificities (Michelmore and Meyers, 1998). These recombination events may be accompanied by mutations directed at solvent-exposed residues within the LRR regions to further modify recognition specificity. These events can result in the formation of several closely related alleles that function to recognize different isolates of the same pathogen containing the corresponding Avr genes. Despite selection for divergence, many of these race-specific resistance genes retain the requirement for common downstream signaling components (Aarts et al., 1998; Century et al., 1995; Parker et al., 1996). The NDR1 and EDS1 genes of Arabidopsis encode two signaling components that are required by different subsets of NBS-LRR R genes (Aarts et al., 1998; Century et al., 1995; Parker et al., 1996). Although members of these subsets are not necessarily closely related overall, it has been suggested that NDR1 and EDS1 dependency may be determined by the presence of coiled-coil or TIR domains, respectively (Aarts et al., 1998; McDowell et al., 2000).

There are approximately 30 Mla variants in barley that confer ‘gene-for-gene’ specificity to the powdery mildew fungus, Blumeria (=Erysiphe) graminis f. sp. hordei DC. Merat Em. Marchal (Bgh) (reviewed by Jørgensen, 1994). Many of these, including Mla1, Mla6 and Mla13, confer a rapid defense response phenotype, while others, such as Mla7, Mla14 and Ml-Ru3 confer a delayed and more intermediate response (Wei et al., 1999; Wise and Ellingboe, 1983). Moreover, there are different requirements among these variants for the downstream signaling components, Rar1 and Rar2 (Freialdenhoven et al., 1994; Jørgensen, 1988; Jørgensen, 1996; Peterhänsel et al., 1997; Torp and Jørgensen, 1986). In contrast to other resistance clusters that contain only homologous genes, the Mla locus contains three distinct families of NBS-LRR resistance gene homologs (RGHs) (Wei et al., 1999). Hence, one might expect signaling specificity to be due to structural differences within the R-gene sequence, much like the differential requirements for NDR1 and EDS1 in Arabidopsis.

In this report, we utilized the information derived from our physical analysis of the Mla locus in cultivar (cv.) Morex (Wei et al., 1999) to rapidly identify functional cDNA and genomic copies of the Mla6 allele. Two single-cell expression assays were employed to demonstrate functionality of the cloned Mla6 allele in barley and wheat. The first of these assays was used to substantiate that Mla6 function is dependent on the Mla-signaling gene, Rar1. Despite their differences in downstream signaling requirements, sequence comparison of Mla6 and the recently cloned Mla1 allele (Zhou et al., 2001) has revealed that the two predicted proteins are highly similar, thus providing a unique framework from which to explore the subtle features of recognition and signaling specificity.

Results

Isolation of transcribed RGHs from the Mla6 locus

Previous research resulted in the development of a physical contig of YAC and BAC clones co-segregating with and spanning the Mla locus. Sequence analysis of Mla-spanning BACs from cv. Morex revealed the presence of three families of NBS-LRR resistance gene homologs (RGHs). These families were designated RGH1, RGH2 and RGH3 based on their sequence divergence (Wei et al., 1999; F. Wei, R. Wing and R. Wise, unpublished results). Although Morex does not contain a characterized Mla resistance specificity, we utilized the information derived from our physical mapping efforts to identify candidates for the Mla6 allele present in C. I. (Cereal Introduction) 16151, a Franger-derived, near-isogenic line (Moseman, 1972).

Genomic DNA of C.I. 16151 was used as substrate for PCR amplification of the LRR regions from the three Mla-RGH families (Figure 1). These low-copy genomic DNAs were used individually to hybridize to 400 000 Pfu of an unamplified Lambda-Zap cDNA library constructed from C.I. 16151 seedlings inoculated with an AvrMla6-containing isolate of Bgh (see Experimental procedures). No plaques were identified using the Mla-RGH2a or Mla-RGH3a probes, however, 29 cDNAs with homology to the NBS-LRR class of plant disease resistance genes hybridized to a mixed Mla-RGH1a/RGH1e probe. Thirteen of the 29 cDNAs contained 5′ untranslated regions (UTRs) up to 400 nt in length. The largest of the cDNAs was used as a probe to re-screen the same library, which resulted in the isolation of nine previously unidentified cDNAs, including two truncated classes with no NBS- or LRR-encoding domain. In total, this screen revealed the presence of three classes of transcripts with 5′ UTRs, containing 13, 2 and 1 members, respectively.

Figure 1.

Selection scheme for complementation of Mla6 specificity.

(a) Genomic DNA of the Mla6-containing line, C.I. 16151, was used as a template to amplify the LRR encoding regions of RGH1a, RGH1e, RGH2a and RGH3a (see Experimental procedures). (b) RGH family specific probes were used individually to hybridize 400 000 Pfu of a C.I. 16151 lambda-ZapII cDNA library. Twenty-nine NBS-LRR encoding cDNAs were identified with the RGH1a/RGH1e probe. (c) The Mla6 co-segregating, C.I. 16151 cDNA sequence was used to design PCR primers to screen super pools of a 3-genome equivalent C.I. 16151 cosmid library. Individual cosmid clones were purified from the identified pools, fingerprinted by restriction digestion, and confirmed via hybridization to the candidate cDNA. (d) Cosmid 9589–5a was used to complement AvrMla6-dependent resistance specificity via the 3-component single-cell assay.

Distances are in centimorgans across the top horizontal line and YACs/BACs in the 400 kb contig are drawn to scale in kilobases below. The Franka YAC is designated by a ‘Fr’ prefix, whereas the Morex BACs are designated by a ‘Mo’ prefix. An orange filled-in circle designates that YAC/BAC was amplified by the respective end-clone primer set or it hybridized to the amplified product. Horizontal arrowheads designate the T7 side of the BAC vector. Locations of Morex RGH sequences are designated by vertical rectangles. RGH1 sequences are designated as red, RGH2 as white, and RGH3 as diagonal hash marks within the rectangle. The number of RGH1 family members shown in this figure has been reduced from what is shown in Figure 5 of Wei et al. (1999). The recent availability of the finished sequence of BACs 711N16 and 80H14 (F. Wei, R. Wing and R.P. Wise, unpublished results) has revealed the presence of a BARE-1 retroelement in the intron between the NBS and LRR encoding domains and also a 29 nt deletion in the LRR domain of RGH1bcd. This complicated the interpretation of RGH1 bcd, originally thought to be three separate RGHs from the low-pass sequence data.

Architecture of Mla-RGH1 cDNAs

As shown in Figure 2, members of cDNA classes B and C are severely truncated and contain only 663 nucleotides (nt) after the start AUG, compared to the 2871 nt open reading frame of class A. The first 584 nt of the ORFs contain four nucleotide differences between class A and classes B and C. One of these mutations, an insertion at base 250 in the open reading frame of classes B and C, causes a frame shift leading to termination of the protein sequence after only 87 amino acids. Another striking difference between these classes occurs 584 nt downstream of the start AUG, where 79 nt of classes B and C have no significant similarity to class A cDNAs.

Figure 2.

Three classes of candidate Mla6 transcripts.

(a) Representation of the 5′ untranslated regions of the 3 Mla6 cDNA classes. Black arrowheads indicate the position of the 17 nt repeat in class A (Mla6). Classes B and C (Mla6–2) differ only by the presence or absence of introns 1 and 2 and are divergent from Mla6 but identical to Mla1 near the 5′ end (colored green). All cDNAs encode a small 9 amino acid peptide (uORF) located before the first putative 5′ UTR intron (designated by red arrowheads). An identical peptide is encoded within the 5′ UTR of Mla1 (Zhou et al., 2001). The 3′ end of this uORF spans the first intron-splicing site. The presence of this intron, as in classes A and B, results in the addition of only one amino acid to the uORF because of a stop codon existing very early in the intron. The genomic sequence of Mla6 was obtained from cosmid 9589–5a that was shown to be functional in the three-component transient assay.

(b) Representation of the open reading frames (including introns) encoded by Mla6 and Mla6–2. The open reading frame of Mla6 contains two introns, 992 nt and 115 nt in length. The open reading frame of Mla6–2 is nearly identical to Mla6 µp to 584 nt downstream of the AUG start codon. The four divergent bases within this region are designated with arrows. An insertion at base 250 causes a frame-shift leading to an early stop codon in Mla6–2. The remaining 79 bases have no significant similarity to Mla6.

Significant differences between the three classes of RGH1 cDNAs were also found within the 5′ UTRs. Aside from different intron splicing events, the 5′ UTRs of classes B and C contain identical nucleotide sequences, but are different from class A cDNAs in a small region near the 5′ end (see Figure 2). This divergent region in the first cDNA class is 68 nt in length and contains two 17 nt repeated sequences separated by 10 bases. In contrast, in classes B and C, this region is 28 nt shorter and is identical to the corresponding section of the 5′ UTR of Mla1 (Zhou et al., 2001) but shares no similarity to class A cDNAs. In summary, these data suggest the presence of separate genes encoding class A and class B/C cDNAs. The presence of at least two genes is corroborated by the observation of three or more hybridizing restriction fragments with multiple enzymes on genomic DNA gel blots (data not shown). The fact that class B and C cDNAs were isolated implies that the gene encoding them contains a functional promoter, although premature termination within the open reading frame and the absence of any NBS or LRR encoding sequence suggests that the function of these proteins could be compromised. Therefore, we focused on determining whether the gene encoding class A alone is capable of conferring Mla6 specificity.

Isolation of cosmids containing a candidate Mla6 genomic sequence

The class A RGH1 cDNA co-segregated with Mla6- mediated resistance using a previously described high-resolution mapping population (Mahadevappa et al., 1994; Wei et al., 1999). In order to isolate a genomic clone for functional testing, we designed a series of unique PCR primers unique to the class A cDNA sequence and screened a three-genome-equivalent cosmid library constructed from genomic DNA of C.I. 16151. In contrast to conventional cosmid library screening via colony hybridization, this procedure targeted cosmid clones specifically containing the class A cDNA sequence as opposed to potentially cross-hybridizing copies. Seven out of 347 pools containing 3.4 × 106 cosmid clones (10 000 clones/pool) yielded PCR products that co-migrated with products amplified from C.I. 16151 genomic DNA. Individual cosmids purified from these pools ranged between 27 and 39 kb in length. DNA gel-blot analysis of EcoRI, HindIII, EcoRV and BclI digested cosmids and subsequent hybridization with the RGH1 class A cDNA probe revealed that five cosmids contained identical restriction site patterns that are found in the class A cDNA sequence (Figure 1), whereas the other two cosmids contained related, but not identical, cross-hybridizing members.

The smallest cosmid (9589–5a; 27 kb) containing only one hybridizing candidate gene was selected for sequencing. Sequence analysis of this cosmid revealed a single NBS-LRR gene with an open reading frame identical to the class A RGH1 cDNA. Comparison of the genomic and cDNA sequences predicted the presence of two introns (992 and 115 nt) within the open reading frame and two introns (110 and 102 nt) within the 5′ UTR. However, the first predicted intron (110 nt) was not spliced out of any of the 5′ UTRs of class A cDNAs (Figure 2). Cosmid 9589–5a, containing the genomic copy plus at least 3 kb of upstream sequence of the Mla6 candidate gene, was then used in a single-cell transient expression assay for powdery mildew resistance (Shirasu et al., 1999a; Zhou et al., 2001).

Functional complementation of the Mla6 specificity in a three-component transient expression assay

The Mla6-containing line, C.I. 16151, is known to possess an additional Mla resistance specificity, designated Mla14 (Giese, 1981; Jørgensen, 1994). While Mla6 confers rapid and complete resistance to Bgh, Mla14 is expressed much later and only moderately suppresses sporulation of the fungus. Since Mla6 is epistatic to Mla14 and the two specificities co-segregate in coupling (Wei et al., 1999), Mla14 can only be detected if the infecting Bgh isolate possesses AvrMla14, but lacks AvrMla6. The powdery mildew isolate that we have used does indeed contain AvrMla6 and, hence, the results described below focus on the complementation of Mla6 specificity.

The three-component system adapted for our experiments is based on the simultaneous single-cell transient expression of the reporter GFP, wild-type barley Mlo, and the Mla6 candidate gene in mlo resistant plants. In this assay, GFP fluorescing, leaf-epidermal cells are rendered susceptible to Bgh, due to the presence of wild-type Mlo, whereas neighboring non-transformed cells retain broad-spectrum mlo resistance (Shirasu et al., 1999a). The mlo resistance of non-transformed cells makes it possible to score the infection phenotypes on the few single-cell transformation events which otherwise would become masked by spreading fungal hyphae that originates from neighboring susceptible cells. Particle bombardment is used to deliver a reporter plasmid (pUGLUM) into leaf epidermal cells permitting expression of Mlo and GFP from individual ubiquitin promoters (Zhou et al., 2001).

One set of bombarded leaves was inoculated at high density with Bgh conidia of isolate A6, which contains AvrMla6 but not AvrMla1, and is therefore avirulent on cells expressing a functional Mla6, but virulent on cells that express Mla1. As a control, a duplicate set of bombarded leaves was inoculated with Bgh isolate K1, which does not possess AvrMla6, but contains AvrMla1. Five days post-inoculation, epidermal cells co-expressing GFP and Mlo were scored for Bgh infection. Only GFP fluorescing cells that had attached fungal conidia were counted in these experiments because resistance can only occur after direct contact between conidia and the host cell (Ellingboe, 1972). Green fluorescent cells that supported growth of fungal hyphae were considered susceptible. Results of previous experiments using this system have suggested that fungal growth in approximately 45% of GFP fluorescing cells should be considered complete susceptibility (Shirasu et al., 1999b).

The results of the above-described experiments are presented in Table 1. In leaves that were bombarded with pUGLUM DNA alone, there was no significant difference in the percentage of infected, GFP-fluorescing cells after inoculation with the A6 or K1 conidia (50.0% and 52.3%, respectively). When cosmid 9589–5a DNA was included in the bombardment, the percentage of GFP fluorescing cells that supported growth of isolate A6 was significantly reduced to 9.4%. Cells inoculated with conidia of isolate K1 supported fungal growth 46.5% of the time, which was not significantly different from that of the control. In another control experiment, we bombarded the reporter plasmid pUGLUM together with another cosmid, p6–49–2-15, which contains Mla1 (Zhou et al., 2001). This resulted in significantly reduced infection of GFP-marked cells to the K1 isolate expressing AvrMla1, but did not reduce susceptibility to Bgh isolate A6 (9.3 and 47.1%, respectively). The observed Avr gene-dependent, single-cell resistance indicates that the single candidate gene encoded within cosmid 9589–5a is Mla6. The gene encoding cDNA classes B and C, which is not present in cosmid 9589–5a, has been designated as Mla6-2.

Table 1.  Results of the 3-component transient assay in mlo-5 barley leaves inoculated with isolates A6 or K1 of B. graminis f. sp. hordei
Test DNAA6 (AvrMla6, virMla1)K1 (virMla6, AvrMla1)
no. GFP
cells with
conidia
no. GFP
cells with
hyphae
% GFP
cells with
hyphae
P valuea
(test vs.
pUGLUM)
no. GFP
cells with
conidia
no. GFP
cells with
hyphae
% GFP
cells with
hyphae
P valuea
(test vs.
pUGLUM)
P valueb
(A6 vs. K1)
  • a

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with hyphal colonies after bombardment with test DNA in comparison to bombardment with the pUGLUM control. A P value <0.05 indicates that the percentages are significantly different.

  • b

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with hyphal colonies after inoculation with isolate A6 in comparison to inoculation with isolate K1. A P value <0.05 indicates that the percentages are significantly different.

  • *

    Raw numbers indicate the combined results of three independent experiments.

pUGLUM522650.0 442352.3 0.9926
pUGLUM + 9589-5a127129.4<0.00011296046.50.4102<0.0001
pUGLUM + p-49-2-15512447.10.61795459.3<0.0001<0.0001

Mla6 signals through the zinc-binding protein, RAR1

Previously, the function of Mla6-mediated resistance was shown to be dependent on Rar1, a powdery mildew resistance-signaling gene (Jørgensen, 1996; Shirasu et al., 1999b). To further substantiate that the full-length NBS-LRR sequence encoded on cosmid 9589–5a encodes Mla6 specificity, we took advantage of a recently isolated, rar1–2/mlo-31 double mutant (J. Orme and P. Schulze-Lefert, unpublished results) to see if Mla6-mediated resistance was compromised in a rar1 mutant background.

The rar1/mlo double-mutant leaves were co-bombarded with pUGLUM and cosmid 9589–5a (Mla6). In this experiment, summarized in Table 2, 41% of the GFP-marked epidermal cells supported growth of Bgh isolate A6 (AvrMla6). In a control experiment, Rar1/mlo leaves were resistant when bombarded with cosmid 9589–5a. In a further control we bombarded pUGLUM together with the Mla1-containing cosmid p6–49–2-15 (Zhou et al., 2001). On both rar1/mlo double mutant leaves and Rar1/mlo leaves, we observed the same percentage of GFP expressing cells supporting the formation of A6 (AvrMla6) powdery mildew colonies (44%). Thus, we conclude that the AvrMla6-dependent activity of the NBS-LRR gene in cosmid 9589–5a is fully compromised in a rar1 mutant background. This finding is consistent with our claim that the single NBS-LRR gene in cosmid 9589–5a is Mla6.

Table 2.  Results of the 3-component transient assay in mlo-5 and rar1-2/mlo-31 barley leaves inoculated with isolate A6 of B. graminis f. sp. hordei (AvrMla6, virMla1)
Test DNAmlo-5 leavesrar1-2/mlo-31 leaves
no. GFP
cells with
conidia
no. GFP
cells with
hyphae
% GFP
cells with
hyphae
P valuea
(9589-5a vs.
p-49-2-15)
no. GFP
cells with
conidia
no. GFP
cells with
hyphae
% GFP
cells with
hyphae
P valuea
(9589-5a vs.
p-49-2-15)
P valueb
(Rar1 vs.
rar1)
  • a

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with hyphal colonies after bombardment with Mla6 cosmid 9589-5a in comparison to bombardment with the Mla1 cosmid p-49-2-15. A P value <0.05 indicates that the percentages are significantly different.

  • b

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with hyphal colonies in mlo-5 leaves in comparison to mlo-31/rar1-2 leaves. A P value <0.05 indicates that the percentages are significantly different.

  • *

    Raw numbers indicate the combined results of two independent experiments.

pUGLUM + 9589-5a7200.0 783241.0 0.0
pUGLUM + p-49-2-151024544.10.0873843.70.72870.8976

MLA6 belongs to the coiled-coil, NBS-LRR class of resistance proteins

The deduced protein sequence encoded by the Mla6 open reading frame contains 956 amino acids with an estimated molecular mass of 107.8 kDa. An in-frame stop codon 33 nt upstream of the putative start methionine supports our prediction that the ORF depicted in Figure 2 is the entire coding region of Mla6. A COILS analysis (Lupas et al., 1991) of the MLA6 protein sequence revealed with greater than 95% probability that a coiled-coil region is located between amino acids 24 and 50, indicating that MLA6 belongs to the coiled-coil subset of NBS-LRR resistance proteins. As illustrated in Figure 3, the MLA6 protein contains the five conserved motifs indicative of a nucleotide-binding site. The kinase-1a (P-loop), kinase-2a, kinase-3a, and conserved domain 2 motifs are all highly similar to those of other NBS-LRR resistance proteins (Grant et al., 1995; Van Der Biezen and Jones, 1998). However, the conserved NBS domain 3 of MLA6 lacks the conserved phenylalanine found in other NBS-containing resistance proteins. The C-terminal region of the protein contains 11 imperfect leucine-rich repeats with an average size of 26 amino acids. These LRRs conform to the consensus motif LxxLxxLxxLxLxx(N/C/T)x(x)L observed in other cytoplasmic R gene products (Jones and Jones, 1997).

Figure 3.

Amino acid sequence alignment of Mla6 from C.I. 16151, Mla1 and an Mla1 homologue from C.I. 16137, and four Mla-RGH1 family members from the cv. Morex.

Shaded boxes indicate similar residues. Conserved motifs within the NBS region are indicated above the sequence. The stars denote the putative solvent exposed residues of the LRR region. The carets indicate residues conserved between MLA6 and MLA1 but not with any other protein. RGH1e and RGH1f gene sequences differ by only one nucleotide, which does not cause and amino acid change. Note the presence of a premature stop codon at position 151 of these two classes. A large deletion starting at position 714 of RGH1bcd causes a frame-shift mutation. The homologous frame is shown in the alignment after this deletion.

To deduce the conserved amino acids necessary for function of Mla alleles, the protein sequence of MLA6 was compared to MLA1, an MLA1 homologue (MLA1–2; Zhou et al., 2001), and four MLA-RGH1 family members from the barley cultivar Morex (Figure 1). Although there is a high level of conservation between all these sequences, it is apparent that the two proteins with known function, MLA6 and MLA1, are more similar to each other than to any of the proteins with unknown function. MLA6 and MLA1 are 92.2% similar (91.2% identical) at the amino acid level. The MLA-RGH1 protein with the highest overall similarity to these two proteins is MLA-RGH1bcd, which is 87.3% similar (83.6% identical) to MLA1 and 84.2% similar (79.9% identical) to MLA6. Fifty-seven amino acids are conserved between MLA6 and MLA1 that are not conserved in any of the MLA homologs without known function. Interestingly, the majority (38) of these residues are located within the first 160 amino acids.

A comparison of the leucine-rich repeats of these proteins revealed a number of regions of non-conserved amino acids centered primarily around the putative solvent exposed residues of the repeats (Figure 3). The predicted solvent exposed residues in LRR regions of many R gene products are known to be hypervariable (Botella et al., 1998; Dixon et al., 1998; Ellis et al., 1999; McDowell et al., 1998; Meyers et al., 1998; Noël and Moores, 1999; Parniske et al., 1997; Thomas et al., 1997). Amino-acid variations within these exposed residues of NBS-LRR proteins are thought to be one of the R-gene components that determine recognition specificity (Ellis et al., 1999). Our results indicate that residues within these regions are highly variable not only between functional and non-functional proteins but also between the two functional proteins, MLA6 and MLA1. Further analysis indicated that the variability within the LRRs, and more specifically within the solvent exposed residues, is subject to diversifying selection. In any given region of a gene, a greater number of non-synonymous (Ka) than synonymous (Ks) nucleotide mutations indicates selective divergence of the region (Ka/Ks > 1; Hughes and Yeager, 1998; Meyers et al., 1998; Parniske et al., 1997). The Ka/Ks ratio between the solvent exposed residues of MLA6 and MLA1 is 3.75 (15/4) suggesting selection for divergence at these residues. Comparatively, the entire LRR region has a Ka/Ks ratio of 1.64 (36/22) and the region upstream of the LRR has a ratio of exactly 1.0 (26/26).

Barley MLA6 recognizes Bgh AvrMla6 in wheat

The cloning and characterization of resistance genes is important in understanding the fundamental basis of disease resistance, yet ultimately the ability to transfer race-specific resistance genes between plant species will be important to improve agriculturally important crop varieties. We used a recently described system, similar to the three-component assay described above, to test whether the Mla6-containing cosmid, 9589–5a, could confer resistance to Bgh in wheat (Schweizer et al., 1999). In this assay, a GUS reporter plasmid under the control of a ubiquitin promoter (pUGUS) was used in place of pUGLUM, enabling easy visualization of fungal haustoria in GUS-positive, epidermal cells. Mla6 is known to predominantly terminate fungal growth by preventing differentiation of haustoria (Boyd et al., 1995; Ellingboe, 1972). Therefore, lack of these structures would indicate Mla6 activity.

We co-bombarded pUGUS and cosmid 9589–5a (Mla6) into wheat leaves from cv. Cerco followed by inoculation with the B. graminis f. sp. tritici (Bgt) isolates JIW2 and JIW48. Both of these isolates possess most of the currently known wheat powdery mildew Avr genes, yet are virulent on Cerco wheat. Co-bombardment of cosmid 9589–5a with pUGUS did not significantly alter the percentage of infected cells when compared to bombardment with pUGUS alone (data not shown). This suggests that either wheat does not contain the machinery necessary for proper function of Mla6 or that JIW2 and JIW48 do not contain a recognized AvrMla6 gene product.

We therefore repeated the experiment using the barley powdery mildew isolate A6, which possesses a functional AvrMla6, to inoculate the bombarded wheat leaves. Although inoculation of Cerco wheat with Bgh isolate A6 represents a non-host interaction, we observed that up to 30% of the infecting conidia were able to form differentiated haustoria (Figure 4). Conidia from Bgh isolate K1 were not able to form haustoria at a significant level and were not suitable for use as a negative control. Therefore, the virulent Bgt isolate JIW48 was used instead.

Figure 4.

Compatible and incompatible wheat/powdery mildew interactions revealed in the three-component transient assay.

(a,b) CERCO wheat cells after inoculation with Bgh isolate A6 (AvrMla6). (c,d) CERCO wheat cells after inoculation with isolate Bgt JIW48 (virMla6, virMla1). The GUS expressing cells (g) in (a) and (c) have been co-bombarded with the Mla1 cosmid p6–49–2-15. (b) and (d) show GUS expressing cells after bombardment with the Mla6 cosmid 9589–5a. A statistically significant reduction in haustorial development after bombard ment with cosmid 9589–5a suggests that MLA6 is able to recognize Bgh AvrMla6 in wheat. Leaves were stained with X-gluc overnight at 37°C to highlight GUS staining cells (g) and haustorium (h; see Experimental procedures). The leaves were then incubated in Coomassie blue to stain the attached conidia (c).

There was no significant difference between the percentage of wheat epidermal cells with Bgt JIW48 haustoria after co-bombardment of pUGUS with cosmids 9589–5a or p6–49–2-15 (31 and 30%, respectively; Table 3). Similarly, the incidence of wheat epidermal cells with Bgh A6 haustoria following delivery of pUGUS alone or together with cosmid p6–49–2-15 was comparable (20.5% and 22.8%, respectively). In contrast, Bgh isolate A6 consistently established less than half as many haustoria when wheat cells were transformed with pUGUS and cosmid 9589–5a. Thus, Mla6 is able to function in wheat to confer specificity to Bgh expressing AvrMla6.

Table 3.  Results of the 3-component transient assay in CERCO wheat leaves inoculated with isolates A6 of B. graminis f. sp. hordei and isolate JIW48 of B. graminis f. sp. tritici
Test DNABgh isolate A6 (AvrMla6, virMla1)Bgt isolate JIW48 (virMla6, virMla1)
no. GUS
cells with
conidia
no. GUS
cells with
haustoria
% GUS
cells with
haustoria
P valuea
(test vs.
pUGUS)
no. GUS
cells with
conidia
no. GUS
cells with
haustoria
% GUS
cells with
haustoria
P valuea
(test vs.
pUGUS)
P valueb
(A6 vs.
JIW48)
  • a

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with haustoria after bombardment with cosmid 9589-5a in comparison to bombardment with the Mla1 cosmid p-49-2-15 control. A P value <0.05 indicates that the percentages are significantly different.

  • b

    P values were obtained using a random effect model to test for a significant difference between the percentage of cells with haustoria after inoculation with isolate A6 in comparison to inoculation with isolate JIW48. A P value <0.05 indicates that the percentages are significantly different.

  • *

    Raw numbers indicate the combined results of three independent experiments.

pUGUS1954020.5      
pUGUS + p-49-2-152806422.80.54331344130.6 0.0903
pUGUS + 9589-5a322319.60.0005922830.40.9792<0.0001

Discussion

Powdery mildew of barley is becoming a tractable model for investigating the molecular basis of host–pathogen interactions among monocots and obligate biotrophic fungi. Recently, positional cloning efforts have resulted in the isolation of the Mlo gene modulating broad- spectrum resistance (Büschges et al., 1997), the Rar1 gene involved in resistance signaling (Shirasu et al., 1999b), and the Mla1 resistance allele (Zhou et al., 2001). In this report, we present the identification of the Mla6 (CC-NBS-LRR-encoding) allele and show that it is able to confer specificity to Bgh expressing the AvrMla6 gene in both barley and wheat.

Little is known about the molecular recognition process involved in plant fungal interactions. However, it has been established that, in the case of powdery mildew, Mla-mediated resistance is expressed only after host and pathogen have made intimate membrane-to-membrane contact. Among Mla alleles, resistance specificities conferred by Mla6 (and Mla1) are two of the earliest and most effective at aborting fungal infection attempts prior to haustorium differentiation (Boyd et al., 1995; Wise and Ellingboe, 1983). It is for this reason that we were able to demonstrate that the barley Mla6 gene also functions in wheat. Introduction of Mla6 into wheat epidermal cells reduced the incidence of Bgh haustoria by 50%, indicating that Mla6 retains function and specificity in this heterologous system.

Historically, intergeneric gene transfer among the Triticeae has been accomplished through cytogenetic translocation of chromosome segments. This breeding technology has been exploited to transfer useful traits into polyploid wheat, including several resistance genes (Friebe et al., 1989; Heun and Friebe, 1990; Heun et al., 1990). The chromosome segment that has been most widely utilized for this purpose is rye 1RS. This chromosome segment contains genes for resistance to leaf, stem and stripe rust, as well as for powdery mildew (Singh et al., 1990; Villareal et al., 1991). It is interesting to note that 1RS from rye is homologous to barley 1HS, the chromosome arm that contains the majority of Ml loci, including Mla.

The Lr26 leaf-rust resistance gene, carried on translocated rye 1RS, is fully functional in a wheat background (Hanusováet al., 1996). However, the activity of translocated genes can sometimes be suppressed by interacting factors within the wheat genome (Bai and Knott, 1992; Hanusováet al., 1996; Kema et al., 1995; Kerber and Green, 1980; Ma et al., 1995). For example, a dominant suppressor in wheat inhibits the phenotypic expression of the Pm8 powdery mildew resistance gene, which was also translocated into wheat via rye 1RS. These gene-specific suppressors may represent downstream signaling factors required for the ultimate expression of the resistance phenotype, similar to Rar1 from barley (Freialdenhoven et al., 1994; Jørgensen, 1996; Shirasu et al., 1999b). Our initial data, utilizing isolate A6 of Bgh, indicate that the ability of Mla6 to prevent the formation of haustoria in wheat was dependent upon recognition of Bgh AvrMla6 (Table 3) and that specificity was not suppressed by wheat host factors. The question remains, however, whether barley Mla6 also confers resistance to Bgt isolates that express AvrMla6. In the Solanacae, four R genes have been shown to function in heterologous systems: (1) the Pto gene of tomato (Rommens et al., 1995; Thilmony et al., 1995); (2) the N gene of tobacco (Whitham et al., 1996); (3) the Cf-9 gene of tomato (Hammond-Kosack et al., 1998); and (4) the Bs2 gene of pepper (Tai et al., 1999). Although it still remains to be seen whether R genes can provide resistance outside of their own family, our results demonstrate that CC-NBS-LRR-encoded resistance specificity can be transferred between two different members of the Triticeae, and that heterologous transfer is not limited solely to Solanaceous plants.

Expression of Mla-encoded NBS-LRR genes

Only probes specific for the RGH1 family were useful in the identification of expressed transcripts. The cDNAs that we have isolated appear to correspond to two different genes based on multiple differences within the 5′ untranslated regions, the open reading frames, and the 3′ end (Figure 2). The fact that one of the base substitutions within the truncated coding region of Mla6–2 cDNAs leads to premature termination of the protein suggests that MLA6–2 may be non-functional. The predicted MLA6–2 protein shares some similarity to the truncated products of the N, L6 and RPP5tr genes in that it lacks an LRR domain (Dinesh-Kumar et al., 1995; Lawrence et al., 1995; Parker et al., 1997; Whitham et al., 1994). While the truncated N and L6 transcripts appear to arise from alternative intron splicing events, the truncated RPP5 transcript (RPP5tr) originates from an RPP5 gene family member and not from the gene that determines specificity (Parker et al., 1997). Interestingly, the RPP5tr transcript contains a small region near the 3′ end that is divergent from the functional RPP5 transcript, which is similar to what we have detected in the Mla6–2 cDNAs. The role of these truncated proteins in conferring disease resistance is unclear. While Dinesh-Kumar and Baker (2000) have suggested a dominant negative role for the truncated N gene product, alternative transcripts of the L6 gene appear to have no function in resistance to rust in flax (Ayliffe et al., 1999).

Small upstream open reading frames within the 5′ UTR (uORFs) are present in 5–10% of eukaryotic mRNAs and are thought to control translation or tissue-specific expression of the parent protein within the cell (for reviews see Van Der Velden and Thomas, 1999; Willis, 1999). A number of eukaryotic regulatory genes, such as BCL-2 and c-mos, encode proteins whose over-production could be deleterious to cells. Translation of these proteins appears to be constitutively down-regulated by the presence of small upstream open reading frames (Harigai et al., 1996; Steel et al., 1996). Removal of these uORFs from the 5′ UTRs led to a five- to sevenfold increase in the expression of fused reporter genes (Harigai et al., 1996; Steel et al., 1996). The significance of two 17 nt repeated sequences directly upstream of the uORF is unclear although each of these repeats contain a complete TATA box (data not shown), thus suggesting a possible transcription start site for the uORF. Interestingly, these repeats are not present in the 5′ UTR of Mla1.

Model for Mla6-mediated resistance

Based on the lack of a signal peptide, we would predict that Mla6 and Mla1 encode proteins that are likely to be localized within the cytoplasm. Since both genes act predominantly prior to haustorium development, the corresponding Avr products must be released before or concomitant with the switch from leaf surface growth to invasive growth. Therefore, recognition of avirulence gene products is likely to occur within the cytoplasm early in the infection process. Experimental evidence for a direct interaction between a leucine-rich R protein/Avr pair has been observed so far only for the rice NBS-LRD Pi-ta and the cognate Magnaporthe grisea Avr-Pita (Jia et al., 2000). Interestingly, transiently expressed Arabidopsis RPS2 co-immunoprecipitates with the cognate Pseudomonas AvrRpt2 protein, a 70 kDa plant protein, and an Avr protein (AvrB), which does not elicit a RPS2-dependent resistance response (Leister and Katagiri, 2000). One interpretation of the latter finding is that RPS2 recognizes indirectly, via the 70 kDa host protein, the presence of its cognate Avr product.

The signaling gene, Rar1, is required for resistance mediated by several Mla alleles and also for other unlinked powdery mildew resistance loci (Jørgensen, 1996). Despite the amino-acid similarity between the MLA6 and MLA1 proteins (91.2%), MLA6 signals in a RAR1-dependent manner, whereas MLA1 does not (Jørgensen, 1996; Peterhänsel et al., 1997; Zhou et al., 2001). This provides evidence for at least two separate Mla-signaling pathways. This is comparable to RPP (resistance to Peronospora parasitica) signaling pathways in Arabidopsis. Resistance mediated by several RPP genes (RPP2, RPP4, RPP5 and RPP21) requires a signaling component encoded by EDS1 (Aarts et al., 1998; Parker et al., 1996). However, RPP7 and RPP8 appear to function independently of EDS1 to confer resistance (Aarts et al., 1998; McDowell et al., 2000). It has been suggested that EDS1-dependent signaling is primarily dependent on R protein structure because RPP5 and other EDS1-dependent R genes encode proteins belonging to the TIR-NBS-LRR class of resistance proteins while EDS1-independent R genes like RPP8 encode CC-NBS-LRR proteins (Aarts et al., 1998; McDowell et al., 2000). In contrast, the Mla6 and Mla1 alleles provide an example of two highly sequence-related CC-NBS-LRR proteins, recognizing Avr genes of the same fungal pathogen that depend on different downstream signaling components for their function.

Exploiting similarities among functional Mla alleles

The introduction of novel resistance specificities into cultivated species has traditionally relied on conventional plant breeding. In contrast to the highly recombinogenic maize Rp1 rust-resistance locus (Hu et al., 1996; Richter et al., 1995), combining multiple Mla specificities has been problematic, especially in light of the suppressed recombination at the locus that has been observed between cultivars (Wei et al., 1999). The ability to isolate R genes now makes it possible to introduce them directly into the desired plant via transformation. However, positional cloning of resistance genes has proven to be a lengthy process, especially in large-genome cereal crops. The Mla6 and Mla1 alleles are 94.4% identical at the nucleotide level. This strong similarity is also shared by the Mla13-candidate allele (F. Wei and R. Wise, unpublished results). It should be possible therefore to exploit similarities shared between the functional Mla6 and Mla1 alleles, which are not conserved among the non-functional Mla1–2 or Morex Mla-RGH1 sequences, to design PCR primers to amplify homologous genes from other cultivated barley species or wild relatives. These genes can then be simultaneously introduced into a single barley variety, a feat not previously possible using conventional breeding.

Experimental procedures

CDNA library screening and sequencing

A cDNA library was constructed in co-operation with D.-W. Choi, at the T.J. Close laboratory (University of California, Riverside, CA, USA) using the Uni-ZAP XR Library kit (Stratagene, La Jolla, CA, USA). The library was constructed from mRNA isolated from both uninoculated barley seedlings and seedlings inoculated with B. graminis f. sp. hordei isolate 5874 (AvrMla6). Tissue was harvested at both 20 and 24 h post-inoculation and snap-frozen in liquid nitrogen. The cDNA library was screened using probes derived from the LRR region of previously described resistance gene homologues RGH1a, RGH1e, RGH2a and RGH3a (Table 4; Wei et al., 1999). RGH1a and RGH1e represent the Mla-RGH1 family where all members of this family have greater than 81% nucleic acid similarity. RGH2a and RGH3a are each 100% similar to other members of their respective families due to a large duplication in the Mla region of the barley genome. DNA sequencing and oligonucleotide synthesis was performed by the Iowa State University DNA Sequencing and Synthesis facility. cDNAs were named in accordance with the nomenclature guidelines for multigene families (Mcintosh et al., 1998; http://wheat.pw.usda.gov/ggpages/wgc/98/index.html).

Table 4. RGH-specific primer pairs utilized for C.I. 16151 cDNA and cosmid library screen
Primer
designation
Primer sequences
(5′ → 3′)
Fragment
size (bp)
Sequence
designation
(origin)
Region of
RGH ORF
Annealing
temperature
39F13GGTTACCATCCTCTTTCGTCACC582RGH1aLRR56
39B95GGAGGCTCGTTGTGTCTCTGAATAC (Morex)  
38F19TGGTTCCAACTGGTGTGTTGC426RGH1eLRR54
38B27CCCCAATGATTTCCACGTCC (Morex)  
38IF50GCTCTCTCACTGTTCGTATGGACC198RGH2aLRR54
38IB62AGCAGCTACCAGGCTGTATTGC (Morex)  
80H14BF30TGCTTTACCTCAAGTTGGCTGC212RGH3aLRR56
80H14BB35CGAAGGTGTGTGATTTCGATGC (Morex)  
9-1AAGCATGGGATAGCTCAC1433Mla6 cDNANBS-LRR58
53Rev3CCCAAGATTACATCGTGAC (CI 16151)  
3UTRFGCACGAGGTCATTCCAGAGATATG1616Mla6 cDNA5′ UTR-58
53Rev4GAAAGAGAGTATTCTCCGC (CI 16151)NBS 

Cosmid library construction, screening and sequencing

Cosmid library construction was done in co-operation with Cell & Molecular Technologies, Inc. (Phillipsburg, NJ, USA). High- molecular weight genomic DNA from C.I. 16151 was partially digested with Sau3A, size selected for fragments ranging between 50 and 75 kb, and ligated into the BamHI site of digested cosmid SuperCos-1 (Stratagene). Ligated cosmids were then electroporated into the XL-1 Blue strain of E. coli. The library was amplified in semi-solid medium and aliquoted into 347 pools containing between 7500 and 10 000 clones each. An aliquot (0.5 µl; ∼5 × 106 clones) of each bacterial pool was placed in a PCR reaction with Mla6 cDNA primers. Pools from which the appropriately sized PCR product could be amplified were diluted and plated onto solid media. Individual cosmids were identified by colony hybridization using the Mla6 cDNA as a probe. A plasmid library of partially digested 9589–5a DNA was constructed in pGEM-7Zf(+) (Promega, Madison, WI, USA) and 384 templates were sequenced.

Fungal isolates

B. graminis f. sp. hordei isolates 5874 (AvrMla1, AvrMla6, virMla13), A6 (virMla1, AvrMla6) and K1 (AvrMla1, virMla6) were propagated on H. vulgare cv. Manchuria, Golden Promise, and Ingrid, respectively, at 18°C (16 h light/8 h darkness).

High-resolution genetic mapping

High-resolution mapping between the two near-isogenic lines C.I. 16151 and C.I. 16155 has been described previously (DeScenzo et al., 1994; Mahadevappa et al., 1994). The Franger (C.I. 16151) and Rupee (C.I. 16155) derived lines contain the Mla6 and the Mla13 resistance specificities, respectively. Starting from 3600 gametes (1800 F2 seed), a total of 286 recombinant barley lines were identified to be recombinant between and homozygous at the Hor1 and Hor2 loci at a resolution of 0.028 cM

Barley DNA was isolated from frozen tissue using a modified CTAB extraction. These DNA extractions, as well as DNA gel blot analyses, were conducted as described previously (Wise and Schnable, 1994). cDNA sequences were screened for RFLPs by Southern hybridization with parental DNA digested with a number of restriction enzymes to reveal which restriction endonuclease revealed a polymorphism. RFLPs were exhibited as differences between the two parental lines and were mapped by Southern hybridization on the 88 individual recombinant lines that contain recombination breakpoints in the Xmwg036–Xmwg068 interval.

Single-cell transient assay

Biolistic bombardment of leaves was carried out according to Shirasu et al. (1999a) using a particle inflow gun (Vain et al., 1993). Detached leaves of 7-day-old barley or wheat seedlings were placed onto 1% Phytoagar (Gibco) plates supplemented with 10% sucrose (w/v) and allowed to recover for 1 h at room temperature. Three barley leaves and four wheat leaves were used per plate. Gold particles (BioRad) were coated with plasmid and/or cosmid DNA at a plasmid: cosmid molar ratio of 2 : 3. The leaves were then incubated at room temperature for 4 h and transferred to 1% Phytoagar prior to fungal inoculation. The inoculated leaves were incubated at 15°C (16 h light/8 h darkness) for 5 days (barley) or 66 h (wheat).

Barley cells expressing GFP were visualized 5 days after fungal inoculation using a microscope with an excitation filter of 450–490 nm, bypass filter 515–565 nm (Leica, GFP plus). Wheat leaves were vacuum-infiltrated twice with a GUS-staining solution (containing 0.1 m Na2HPO4/NaH2PO4 pH 7.0, 10 mm Na-EDTA, 5 mm potassium hexacyanoferrat (II) and potassium hexacyanoferrat (III), 1 mg ml-1 5-bromo-4-chloro-3-indoxyl-b-d-glucuronic acid, cyclohexylammonium salt (X-Gluc), 0.1% (v/v) Triton X-100, 20% (v/v) methanol) and incubated at 37°C overnight. The leaves were rinsed briefly with water and then immersed in Coomassie blue stain (0.3% w/v Coomassie blue, 7.5% w/v TCA, 30% v/v methanol) for 5 min and rinsed again before visualization using a light microscope as described previously (Schweizer et al., 1999).

Acknowledgements

The authors thank Drs Thomas Baum and Adam Bogdanove for critical review of the manuscript. Special thanks goes to Dong-Woog Choi of the Close laboratory (University of California, Riverside, CA, USA) for assistance in preparation of the C.I. 16151 cDNA library. This research was supported in part by USDA-NRI/CGP grant 98–35300–6169 and facilitated by the North American Barley Genome Mapping Project. Research in the P.S.-L. laboratory was supported by a GATSBY foundation grant. Joint contribution of the Corn Insects & Crop Genetics Research Unit, USDA-Agricultural Research Service and the Iowa Agriculture and Home Economics Experiment Station. This is Journal Paper No. J-19010 of the Iowa Agricultural and Home Economics Experiment Station, Ames, IA, Project no. 3368, and supported by Hatch Act and State of Iowa funds.

EMBL accession numbers: Mla6 class A, B and C cDNAs (AJ302292, AJ302294 and AJ302295, respectively), Mla6 genomic (AJ302293) and Mla-RGH1 family members from cv. Morex (AJ302296, AJ302297, AJ302298, AJ302299).

Ancillary