A single-amino acid substitution in the sixth MLA LRR alters Rar1 signaling specificity
Although the structures of plant R-protein LRR domains remain unknown, the crystal structures of several other LRRs have been established (Kobe and Kajava, 2001). Comparison of these structures illustrates the tendency of LRRs to form a horseshoe-shaped molecule with β-sheets on the concave side. A central xxLxLxx motif, where ‘x’ represents any amino acid, forms the β-sheet with the leucines buried in the center of the protein and the adjacent residues exposed to the solvent (Jones and Jones, 1997; Kobe and Deisenhofer, 1994). These solvent-exposed residues are hypervariable and targets for diversifying selection in plant R proteins – a major factor in the determination of disease resistance specificity (Ellis et al., 2000; Meyers et al., 1998; Michelmore and Meyers, 1998; Mondragon-Palomino et al., 2002; Noel et al., 1999; Parniske et al., 1997; Wang et al., 1998). In contrast, amino acids within the LRR that define interactions with downstream components are likely under conservative selection as many of these proteins are required by diverse R-protein pathways (Aarts et al., 1998; Austin et al., 2002; Liu et al., 2002; Muskett et al., 2002; Parker et al., 1996; Tör et al., 2002; Tornero et al., 2002). Therefore, it is probable that these residues are located outside of the xxLxLxx motif where interactions between LRR repeats determine the overall structure of the domain (Jones and Jones, 1997; Kobe and Deisenhofer, 1995).
Only two amino acids are conserved among identified RAR1-dependent MLA proteins: an asparagine at position 712 and a glycine at position 721. Constructs harboring the MLA1-R712N mutation retained moderate resistance without altering RAR1 independence. Yet, a significant reduction in the number of hyphal colonies was observed in mlo-31/rar1-2 cells bombarded with Ubi:Mla1-R712N as compared with mlo-5 cells (Figure 4). This observation suggests that either the MLA1-R712N mutation compromises resistance specificity, possibly through an alteration in tertiary structure, or the genetic background of mlo-31/rar1-2 cells confers a slight increase in resistance. Reciprocal mutation of residue 712 in MLA13 also did not alter RAR1 dependence, suggesting that this amino acid does not have a role in determining RAR1 specificity in this case. However, the G721D substitution in MLA6 and MLA13 alleviated their requirement for RAR1, while retaining resistance specificity. Moreover, these data indicate that RAR1 independence requires the presence of an aspartate at position 721, as mutation of this residue to a structurally similar, but uncharged, asparagine did not alter RAR1 dependence in MLA6. Therefore, it appears possible that the aspartate in MLA1 and MLA7 has a greater influence in defining Rar1 independence than the glycine in MLA6 and MLA13 in determining Rar1 dependence.
Amino acid 721 in MLA6 and MLA13 lies halfway between the xxLxLxx motifs of LRRs 6 and 7. The role of this residue in determining Rar1 independence is corroborated by independent tests of MLA1/MLA6 chimeras, which delimited the region required for RAR1 independence of MLA1 to between the second and eighth LRRs (Shen et al., 2003). As hydrophilic and potentially charged residues are unlikely to be buried within the hydrophobic core of the protein, we hypothesize that this aspartate residue in MLA1 and MLA7 is solvent exposed and could be involved in intra- or intermolecular charge-based interactions. A prediction of secondary structure using psipred (McGuffin et al., 2000) suggests that residue 721 lies within a helical domain (Figure 3). In order to elucidate the role this mutation might have in defining molecular interactions, a predicted tertiary structure of this LRR region was determined by alignment to the known structure of porcine ribonuclease inhibitor (PRI), an LRR-containing protein (Kobe and Deisenhofer, 1993). Although it has been suggested that plant R protein LRRs do not share the same tertiary structure as PRI (Jones and Jones, 1997), there was sufficient homology to form a three-dimensional model of the region containing the sixth LRR. As illustrated in Figure 6, amino acid 721 in MLA6 potentially lies on the convex surface of the domain where it may have a role in intermolecular interactions between repeat motifs. Modeling of this domain against another LRR protein, RNA1P (Gtpase-activating protein for Spi1, the S. pombe ortholog of Ran) (Hillig et al., 1999), also places this residue on the outer surface of the helix where it could be involved in intramolecular interactions (data not shown). While it is difficult to determine whether a glycine to aspartate change is significant enough to alter the overall structure of the LRR, the presence of a hydrophilic and potentially charged amino acid could affect intramolecular folding as well as intermolecular interactions with other members of the MLA signal transduction pathway. As residue 721 lies within the region thought to define interactions between subunits of the LRR (Jones and Jones, 1997; Kobe and Deisenhofer, 1995), this mutation could transform the overall conformation of the MLA LRR domain and, in turn, modify its dependence on RAR1. This is consistent with the observation that RAR1 specificity does not appear to be conferred by CC or TIR domains as, in Arabidopsis, RAR1-dependent and -independent R proteins were identified from classes containing either of these motifs (Muskett et al., 2002).
Figure 6. Model of the predicted tertiary structure of the sixth MLA LRR.
This three-dimensional structure of the sixth MLA LRR was derived by sequence comparison to the determined crystal structure of porcine ribonuclease inhibitor (Kobe and Deisenhofer, 1993). Panel (a) shows an aspartic acid (circled in gold) unique to the RAR1-independent MLA1 and MLA7 proteins. Hydrogen bonds with atoms of an isoleucine in the adjacent α-helix are shown as dashed purple lines. Panel (b) illustrates a glycine at the corresponding position in the RAR1-dependent MLA6, MLA10, MLA12, and MLA13 proteins, which has only a hydrogen atom as a side chain (shown on right). The parallel β-sheets are represented by red arrows. Predicted α-helical regions are shown in green.
Download figure to PowerPoint
Single-amino acid changes within the LRRs of resistance gene products also can result in the loss of the resistance phenotype (Bent et al., 1994; Bryan et al., 2000; Grant et al., 1995; Mindrinos et al., 1994; Shen et al., 2003; Warren et al., 1998). The majority of these mutations occur within the solvent-exposed residues of the β-sheet/β-turn region of the LRR, with a concurrent loss in resistance specificity. However, in addition to evidence presented in this paper, two other findings suggest that the LRRs of some R proteins are additionally involved in resistance gene signaling. The rps5-1 mutation in the third LRR of RPS5 compromises resistance to Pseudomonas syringae pv. tomato DC3000 expressing avrPph3, as well as resistance mediated by several other R proteins, suggesting an interaction with a signal transduction element common to several R gene pathways (Warren et al., 1998). Introduction of an Rps5 transgene did not fully restore the resistance phenotypes of all the affected genes, indicating that the rps5-1 mutant protein may titrate out a signaling component common to multiple resistance pathways. Additionally, Banerjee et al. (2001) found that an Arabidopsis Po (Poppelsdorf)-1-derived allele of Rps2 is able to confer resistance to P. syringae expressing avrRpt2 in a Columbia (Col-0) genetic background, but not in a Po-1 background. Six amino acid polymorphisms within the LRR of Po-1 RPS2 were determined to limit the ability of this protein to utilize one or more other Po-1 resistance signaling components. The majority of these differences appear to lie outside of the putative ligand-binding, solvent-exposed residues. Notably, three of the six differences lie between LRR6 and LRR7 of the RAR1-dependent RPS2 protein – the same region we have defined as important for RAR1 dependence in MLA6 and MLA13. Taken together, this suggests that, in some cases, non-xxLxLxx residues are crucial in determining signaling component interactions.
RAR1 is required for a diverse subset of R proteins (Liu et al., 2002; Muskett et al., 2002; Tornero et al., 2002). While it is not known where RAR1 fits into each of these signaling pathways, a role in the stability of pre-formed R proteins has been implicated, as RPM1 (resistance to P. syringae) fails to accumulate in an Arabidopsis rar1 mutant (Tornero et al., 2002). In this case, RAR1 either actively facilitates the stability of RPM1 or functions to block actions designed to reduce the quantity of RPM1 protein within the cell. RAR1 could affect the ubiquitination of target proteins through its interaction with both barley and Arabidopsis SGT1, as well as subunits of the COP9 signalosome (Azevedo et al., 2002). Yeast SGT1 interacts with SKP1, a subunit of the SCF-type E3 ubiquitin ligase complex (Kitagawa et al., 1999). The COP9 signalosome is a multisubunit protein complex that regulates proteosome-mediated degradation of specific proteins (Wei et al., 1998). Based on differences in susceptibility of MLA chimeras in rar1-2 mutant barley leaves, Shen et al. (2003) have suggested a role for RAR1/SGT1 in activation of MLA-containing recognition complexes. However, in our experiments we did not observe a distinct difference between susceptibility of rar1-2 cells bombarded with Mla6 when compared to other susceptible controls, indicating that MLA6 complexes are likely not more active in the absence of RAR1. Therefore, it appears that RAR1 could function as an antagonist of SGT1-mediated ubiqitination and degradation of pre-formed R proteins. Resistance mediated independently of Rar1 in plants containing Mla1 and Mla7 could be because of alternative methods of regulating the expression of these genes, or by bypassing the requirement of RAR1 through changes in the tertiary structure of the MLA1 and MLA7 LRR domains.
In summary, our results indicate that RAR1 independence is determined by an aspartate residue at position 721 within the sixth LRR of MLA. Furthermore, residue 721 appears distinct from those hypothesized to determine resistance specificity and likely plays a role in determining inter- or intramolecular interactions that regulate resistance signal transduction.
Mla6 confers AvrMla6- and AvrMla14-dependent resistance specificity to Bgh
In some host–pathogen interactions, a single resistance protein can confer specificity to more than one pathogen effector (Axtell and Staskawicz, 2003; Grant et al., 1995; Kim et al., 2002; Mackey et al., 2003). In this regard, genetic evidence indicates that barley lines harboring distinct Mla alleles are also able to confer alternate resistance specificities to multiple Bgh isolates, each containing unique AvrMla genes (Brown, 2002). In C.I. 16151, AvrMla14-dependent resistance is defined as an intermediate IT that co-segregates with a rapid and absolute response conferred by Mla6 (Jørgensen, 1994; Wei et al., 1999). We have shown that the cloned Mla6 ORF, expressed transiently under control of a maize ubiquitin or native promoter can specify Rar1-dependent resistance equally to Bgh isolates 5784 and A27, which contain AvrMla6 and AvrMla14, respectively (Giese et al., 1981; Torp et al., 1978). Furthermore, the single-amino acid change in the MLA6-G721D protein alleviates dependence on RAR1 to effect both AvrMla6- and AvrMla14-dependent resistance responses. This presents two possible scenarios: either the MLA6 protein is capable of monitoring the presence of two diverse avirulence gene products, or the weakened infection phenotype induced by Bgh isolates that contain AvrMla14 is caused by an alternate AvrMla6 allele whose product is only partially recognized by MLA6. To our knowledge, there are no clear data that indicate AvrMla6 and AvrMla14 segregate independently (Jensen et al., 1995). Thus, it remains possible that these historically designated Avr genes are either (i) allelic variants, (ii) separate but closely linked, or (iii) that the genetic backgrounds of Bgh 5874 and A27 influence the phenotype of AvrMla6 in interactions with barley accessions that harbor Mla6.
There are, however, several other documented instances of barley lines with putative, tightly linked Ml genes that provide alternate responses to various Bgh isolates that harbor unique Avr genes (Caffier et al., 1996; Giese, 1981; Jørgensen, 1994; Jensen et al., 1995). For example, the C.I. 16155 (Mla13) accession displays an intermediate IT in response to Bgh isolate R189, historically designated as the Ml-Ru3 specificity (Caffier et al., 1996; Giese, 1981). This second specificity co-segregates with Mla13 in a mapping population of 3600 gametes, just as Mla14 co-segregates with Mla6 (Wei et al., 1999). Previous data corroborate the existence of a second Avr gene in Bgh that elicits the Ml-Ru3 phenotype (Brown, 2002; Brown and Simpson, 1994; Caffier et al., 1996; Jensen et al., 1995). Thus, it could similarly be that Mla13, being the only documented functional Mla copy in C.I. 16155, recognizes either a second Avr gene or an AvrMla13 allele to confer Ml-Ru3-mediated resistance. It remains uncertain whether the recognition of multiple Bgh genotypes by Mla is determined on the plant or pathogen side, or some combination of the two, thus, the molecular isolation of additional Ml genes and their cognate AvrMl genes will be necessary to complete our understanding of these dynamic host–pathogen interactions.
Parallel evolution of the Mla family
Previously, we developed a working model for the evolution of the Mla complex using a 261-kbp DNA region spanning the locus in the barley cultivar Morex (Wei et al., 2002). In the process of isolating additional Mla paralogs (Mla6-2 and Mla13-2) and alleles (Mla7 and Mla10), we identified distinct Insertion/deletion (InDel), retrotransposon and long interspersed nuclear element (LINE) insertions, in addition to the (AT)n simple sequence repeat (SSR) within the third intron (Tables S1 and S2; Halterman et al., 2003; Shen et al., 2003; Zhou et al., 2001). These polymorphisms were used as markers within and flanking Mla family members to construct a model for the diversification of Mla resistance haplotypes (Figure S2). This analysis was consistent with the interpretation that the present-day distribution of the Mla gene family has been the result of at least two evolutionary pathways. One proposed branch of the pathway contains Mla1, in addition to Morex RGH1bcd, whereas the other branch contains Mla6 and Mla13.
The timing of retrotransposon insertions within and flanking Mla ORFs suggests that an ancient Mla progenitor was duplicated prior to insertion of two LINEs at the 5′ end of these two haplotypes (Figure S2). Only one of these haplotypes was host for the differential amplification of an (AT)n microsatellite within the third intron. As the (AT)n microsatellite is present in the Mla1, Mla6, and Mla13 alleles, as well as the corresponding third intron of the non-functional Mla6-2 and Mla13-2 paralogs, it was most likely propagated very early, but after duplication of the ancestral Mla gene. Subsequent to the initial duplication, the evolution of functional Mla resistance specificities appears to branch into two pathways – one leading to Rar1-independent Mla1 (branch A), and another leading to Rar1-dependent Mla6, and Mla13 (branch B). Initially, this suggested that Rar1 signaling specificity may have diverged early in the evolution of Mla. However, when integrated with the analysis of phylogeny and reticulate evolution, the data are consistent with the conclusion that Rar1 dependence, or independence, is not limited to one evolutionary course.