Magical mystery tour: MLO proteins in plant immunity and beyond

Authors


Summary

Stable heritable restriction of the ubiquitous powdery mildew disease is a desirable trait for agri- and horticulture. In barley (Hordeum vulgare), loss-of-function mutant alleles of the Mildew resistance locus o (Mlo) gene confer broad-spectrum resistance to almost all known isolates of the fungal barley powdery mildew pathogen, Blumeria graminis f.sp. hordei. Despite extensive cultivation of barley mlo genotypes, mlo resistance has been durable in the field. Mlo genes are present as small families in the genomes of all higher plant species. The presumed negative regulatory role of particular members in plant immunity is evolutionarily conserved, as powdery mildew resistant mlo mutants have also been described in Arabidopsis thaliana, tomato (Solanum lycopersicum) and pea (Pisum sativum). Barley Mlo encodes a plasma membrane-localized seven-transmembrane domain protein of unknown biochemical activity. Here, we review the known requirements for mlo-mediated disease resistance in barley and Arabidopsis and reflect current views regarding Mlo function. We discuss additional mlo mutant phenotypes recently discovered in Arabidopsis and present a meta-analysis of the phylogenetic relationships within the Mlo family. Finally, we consider the novel versatile tools for functional analysis and targeted genome modification that can be used to induce mlo-based powdery mildew resistance in virtually any plant species.

Yesterday: Mlo genes and powdery mildew resistance

Powdery mildew is a widespread plant disease of temperate climates that is caused by ascomycete fungi of the order Erysiphales (Glawe, 2008). The disease is macroscopically characterized by ‘powdery’ fungal reproduction structures on the surface of plant organs (Fig. 1a). It is an important threat for both agri- and horticulture and can cause significant harvest losses in crop plants such as wheat, barley and tomato (Dean et al., 2012), and severely impact ornamental plants such as roses (Linde et al., 2006). Accordingly, the generation of plant breeds that exhibit robust immunity to this disease is of great economic interest. One major step in this direction was the discovery of barley mutant plants that display near complete resistance to the powdery mildew pathogen, Blumeria graminis f.sp. hordei (Bgh). These plants, which carry recessively inherited loss-of-function mutations in the gene Mildew resistance locus o (Mlo), show durable broad-spectrum resistance against virtually all Bgh isolates (Jørgensen, 1992; Büschges et al., 1997). On mlo mutant plants, powdery mildew pathogenesis is terminated at the stage of cell wall penetration and host cell entry; consequently, fungal sporelings do not form haustoria inside host cells and fungal colonies cannot develop (Aist et al., 1987). Subsequent studies revealed that (1) Mlo genes are restricted to plants and green algae, and occur as small to medium-sized families in the genomes of higher plant species (Devoto et al., 1999, 2003; Jiwan et al., 2013) and (2) that mlo-based powdery mildew resistance is not restricted to the monocot barley, but is also found in the distantly related eudicot plant species Arabidopsis thaliana (Consonni et al., 2006). Mutant alleles of Arabidopsis thaliana AtMLO2 (note that the gene nomenclature varies between species – Mlo in barley, MLO in Arabidopsis; we hereafter use MLO as the generic spelling), one out of the 15 MLO genes present in the Arabidopsis genome, confers partial resistance to the adapted powdery mildew species Golovinomyces orontii and G. cichoracearum. Moreover, the triple mutant Atmlo2 Atmlo6 Atmlo12 is completely resistant to these pathogens, restricting fungal development at the host cell entry level (Consonni et al., 2006). These discoveries, together with the knowledge that the MLO family is ubiquitously present in higher plant species (Devoto et al., 2003), led to the identification of powdery mildew resistance on the basis of natural mlo loss-of-function alleles in tomato (Bai et al., 2008; Zheng et al., 2013) and pea (Humphry et al., 2011; Pavan et al., 2011). Conservation of mlo resistance across monocot and dicot plant species implicates a common mechanistic basis for this type of plant immunity. It further implies potential application of this trait for plant breeding of many other agriculturally and economically important plant species.

Figure 1.

Phenotypes of Arabidopsis mutant plants with defects in different AtMLO genes. (a) Powdery mildew resistance. Infection phenotypes of Col-0 wild-type (wt; left) and Atmlo2 Atmlo6 Atmlo12 triple mutant plants (right) at 10 d post inoculation with Golovinomyces orontii. (b) Aberrant root thigmomorphogenesis. Undulating growth of Col-0 roots (left) compared with the tight curling of Atmlo4 roots (right). Seedlings (1 wk old) were grown on horizontal plates and viewed from underneath. (c) Pollen tube (PT) overgrowth. Normal PT reception in the ovule of accession Ws-2 (wild-type, left) and PT overgrowth in the ovule of Atmlo7 (nta-1, in Ws-2 genetic background; right). Pollen tubes were stained with aniline blue and visualized by epifluorescence microscopy. Bars, 25 μm.

The broad-spectrum, nonrace-specific mlo-mediated powdery mildew resistance in barley is an excellent example of a (mutant)gene that has been widely used in plant breeding and agriculture. For more than 30 years natural (Piffanelli et al., 2004) and induced (Reinstädler et al., 2010) mlo alleles have granted stable powdery mildew resistance in the field (Lyngkjær et al., 2000). Nevertheless, the use of plants carrying mlo alleles is associated with certain limitations. First, in the absence of pathogens, barley mlo plants spontaneously form callose-containing cell wall appositions (papillae), predominantly in the short cell type of the leaf epidermis (Wolter et al., 1993). Additionally, leaf mesophyll cells in mlo mutants undergo spontaneous cell death (Peterhänsel et al., 1997), which has been recognized as an indication of accelerated leaf senescence (Piffanelli et al., 2002). Secondly, barley mlo alleles show enhanced susceptibility to the hemibiotrophic fungus Magnaporthe oryzae (Lyngkjær et al., 2000) and higher sensitivity to necrosis-inducing toxin from Bipolaris sorokiniana (Kumar et al., 2001). Furthermore, mlo mutations seem to have an environment-dependent effect on the leaf spot disease caused by the necrotrophic ascomycete Ramularia collo-cygni (McGrann et al., 2014). The enhanced susceptibility to certain hemibiotrophic or necrotrophic plant pathogens might be a direct consequence of deregulated cell death in mlo plants. These mlo-associated pleiotropic effects result in a reduction in grain yield (Schwarzbach, 1976), a penalty that seems to be overcompensated by the benefit of reduced powdery mildew infection. Notably, not only the powdery mildew resistance trait, but also some of the undesired side effects are phenocopied by Atmlo2 mutant plants (Consonni et al., 2006, 2010).

Here we review the different phenotypes that have been described for mlo mutants and the identified players required for the most prominent mlo mutant phenotype, broad-spectrum powdery mildew resistance. We discuss known aspects of MLO function and describe the MLO family and its phylogenetic relationships. Finally, we reflect on established and novel plant biotechnological methods for the targeted manipulation of MLO genes.

Do you want to know a secret: MLO function is not restricted to plant–powdery mildew interactions

Traditionally, MLO function has been associated with susceptibility/resistance to the powdery mildew disease (Panstruga, 2005b) and its known adverse effects, for example, premature leaf senescence (Piffanelli et al., 2002; Consonni et al., 2010). More recently, however, MLO genes have been implicated in additional developmental processes based on the phenotypes of Arabidopsis mutant plants with defects in different MLOs. Besides the powdery mildew resistance of the Atmlo2 Atmlo6 Atmlo12 triple mutant (Consonni et al., 2006; Fig. 1a), two additional mlo-associated phenotypes have been discovered: aberrant root thigmomorphogenesis (Fig. 1b) and pollen tube overgrowth in the embryo sac (Fig. 1c). When grown in vitro, Atmlo4 and Atmlo11 null mutants display unusual root curvature, evident as severe root curling, upon a tactile stimulus (Chen et al., 2009; Bidzinski et al., 2014; Fig. 1b). This phenotype is modulated by nutrients and requires intact auxin transport machinery, but is independent of the heterotrimeric G-protein complex (which has been previously proposed to play a role for MLO function – see below). Atmlo4 and Atmlo11 single mutants show a similar root curling pattern, which is not enhanced in the Atmlo4 Atmlo11 double mutant. Interestingly, the closely related AtMLO14 gene does not seem to be involved in this physiological process because no altered or further exaggerated phenotype was observed in either the Atmlo14 single mutant or the Atmlo4 Atmlo11 Atmlo14 triple mutant, respectively.

The female gametophytic mutant nortia (nta, alias Atmlo7) exhibits reduced fertility and shows pollen tube overgrowth in the synergids, two small cells in the mature embryo sac that control pollen tube guidance for the double fertilization process in angiosperms (Kessler et al., 2010; Fig. 1c). Notably, mutants defective in the Feronia (Fer) gene encoding a CrRLK1L-type receptor-like kinase exhibit a similar pollen tube overgrowth phenotype (Escobar-Restrepo et al., 2007). The FER pathway seems to play a role in controlling the redistribution of NTA/AtMLO7 protein upon pollen tube arrival. Furthermore, reminiscent of Atmlo2 plants, fer mutants show enhanced resistance to G. orontii, suggesting that powdery mildew invasion and pollen tube reception share molecular components such as FER and MLO proteins (Kessler et al., 2010).

The expression of MLO family members has been shown to occur in different plant organs, tissues and cell types, and to be affected by diverse biotic or abiotic stresses (Piffanelli et al., 2002; Chen et al., 2006). One example comprises the recently found Capsicum annuum CaMLO2 gene, whose expression was shown to be strongly induced in pepper leaves upon treatment with the phytohormone abscisic acid (ABA) and under drought stress conditions. Results of virus-induced silencing in pepper plants and heterologous overexpression in Arabidopsis suggest that CaMLO2 acts as a negative regulator of ABA signalling, implicating CaMLO2 in the regulation of drought stress responses (Lim & Lee, 2013). Interestingly, CaMLO2 not only affects abiotic stresses, but also modulates biotic stress responses to bacterial and oomycete pathogens, and cell death associated with susceptibility to bacterial infection (Kim & Hwang, 2012).

Taken together, MLO genes seem to be involved in many more biological processes than thought just a few years ago. Given that so far only phenotypes for six out of the 15 AtMLO genes have been described, there are potentially still some surprises and novel phenotypes ahead of us. It remains to be seen whether existing knowledge of MLO expression profiles can assist in the identification of new phenotypes by providing leads for meaningful bioassays.

We can work it out: molecular components required for mlo-resistance

The unusual effectiveness, resistance spectrum and durability of mlo-based resistance raise the question: which molecular components are required for this remarkable trait? To address this issue, two main strategies have been pursued in the past: forward genetic screens and candidate gene approaches.

Two genes that are Required for mlo-specified resistance in barley (Ror1 and Ror2) were identified in the context of a forward genetic suppressor screen in the genetic background of a fully resistant mlo null mutant allele (Freialdenhoven et al., 1996). Recent efforts in cloning Ror1 encompassing fine mapping and chromosome walking have positioned the gene in a mapping interval of c. 0.18 cM, roughly at a third of the way down the long arm of chromosome 1H from the centromere (Collins et al., 2001; Acevedo-Garcia et al., 2013). However, the suppressed recombination, resulting in an unfavourable ratio of genetic/physical distance in the pericentromeric Ror1 region, and the low conservation of collinearity with other grass species so far have prevented the isolation and characterization of this gene (Acevedo-Garcia et al., 2013). By contrast, Ror2 was swiftly cloned through conserved collinearity with the rice genome. The gene encodes a member of the t-SNARE (Soluble N-ethylmalemide-sensitive factor Attachment protein REceptor) superfamily (Collins et al., 2003). Functionally it has been suggested to participate in the formation of ternary SNARE protein complexes, possibly for secretion of anti-microbial compounds at attempted fungal penetration sites (Kwon et al., 2008; Meyer et al., 2009; Kwaaitaal et al., 2010).

Screening for Arabidopsis mutants allowing increased penetration success of the nonadapted barley powdery mildew Bgh identified three PENETRATION genes (Collins et al., 2003; Lipka et al., 2005; Stein et al., 2006). PEN1 also encodes a t-SNARE (the orthologue of barley Ror2), PEN2 an atypical myrosinase (Lipka et al., 2005) and PEN3 an ATP-binding cassette multidrug transporter (Stein et al., 2006). Based on genetic data it has been suggested that PEN2 and PEN3 cofunction in the same pathway. One conceivable scenario is that PEN3 transports indole glucosinolate-derived antimicrobial metabolites synthesized by the PEN2-associated metabolic pathway to the apoplast, thereby poisoning the fungus as it attempts to breach the plant cell wall (Bednarek et al., 2009). Intriguingly, all three genes have also been found to be required for Atmlo2-based powdery mildew resistance (Consonni et al., 2006), which spurred the hypothesis that nonhost immunity and mlo resistance rely on functionally overlapping defence pathways (Humphry et al., 2006).

Complementary to forward genetic screens, a number of candidate genes have been tested for their contribution to mlo resistance in barley or Arabidopsis. For example, results of double mutant analysis on the basis of biosynthesis or signalling mutants revealed that signal transduction via the major defence phytohormones salicylic acid, ethylene and jasmonic acid is dispensable for Atmlo2 resistance (Consonni et al., 2006). However, consistent with the demonstrated role of indolic glucosinolate metabolites in defence against nonadapted powdery mildew pathogens (Bednarek et al., 2009) and the previously noted overlap of nonhost and mlo immunity (Humphry et al., 2006), both Atmlo2 resistance and nonhost resistance were found to require AtCYP79B2 and AtCYP79B3 (Consonni et al., 2010). These genetically redundant genes encode cytochrome P450 monooxygenases that catalyse a biochemical reaction representing the entry point in the biosynthetic pathway leading to diverse indolic metabolites, including the phytoalexin camalexin and indolic glucosinolates. PAD3, another cytochrome P450 monooxygenase, catalyses the final step in camalexin biosynthesis. In contrast to mutations in AtCYP79B2 and AtCYP79B3, which have a major impact on Atmlo2 and nonhost resistance, mutations in PAD3 only have a small effect on Atmlo2 resistance and seemingly no effect on nonhost resistance, indicating that camalexin plays a minor role in defence against powdery mildews (Consonni et al., 2010). Although MLO proteins are unlikely to function as G-protein coupled receptors (see below), mutations in genes encoding the heterotrimeric G-protein β (AGB1) and γ (AGG1/AGG2) subunits led to partially compromised Atmlo2 resistance. AGB1 and AGG1 are coexpressed with a number of genes implicated in plant immunity and possibly play a role in basal plant defence (Lorek et al., 2013).

In addition to the Arabidopsis components outlined above, a number of barley genes/proteins have been associated with mlo resistance. However, in contrast to findings in Arabidopsis, which typically result from experiments with intact plants carrying loss-of-function mutations, most data in barley have been obtained upon ectopic transient gene expression in single leaf epidermal cells (Panstruga, 2004). This experimental setup revealed, for example, that the susceptibility-mediating activity of the MLO protein is enhanced by Ca+-dependent calmodulin (CaM) binding to a calmodulin-binding domain that resides in the cytoplasmic carboxy-terminal region of the protein (Kim et al., 2002; Bhat et al., 2005). The procedure also provided evidence that Calcium-Dependent Protein Kinases (CDPKs) modulate mlo resistance in barley. Constitutive activation of HvCDPK3, as well as dominant negative inhibition via the expression of the junction domain of HvCDPK4, resulted in partially restored powdery mildew host cell entry (Freymark et al., 2007). The fact that both activation and inhibition of CDPKs can affect mlo resistance suggests the antagonistic roles of individual CDPK paralogues in this type of immunity. Pharmacological interference with an actin polymerization inhibitor (cytochalasin E) or transient ectopic expression of the barley Actin Depolymerizing Factor 3 (HvADF3) gene enhanced fungal entry in the Mlo genotype (super-susceptibility) and resulted in a partial breakdown of mlo resistance (Miklis et al., 2007). By contrast, treatment with inhibitors of the microtubular network did not affect mlo resistance. These findings demonstrate that barley requires an intact actin cytoskeleton for basal defence to Bgh and for mlo-mediated resistance. Finally, differential expression of the barley BAX Inhibitor 1 (HvBI-1) gene was found in response to Bgh in susceptible (Mlo genotype) and resistant (mlo genotype) plants. Transient overexpression of HvBI-1 rendered mlo plants partially susceptible to the barley powdery mildew pathogen, suggesting a link between powdery mildew susceptibility and the control of programmed cell death (Hückelhoven et al., 2003).

Help! What could be the function of MLO proteins?

Based on genetic data it was originally proposed that barley Mlo has a negative regulatory role in plant immunity (Büschges et al., 1997). Consistent with this notion, the barley mlo-5 mutant shows enhanced differential gene expression compared to the wild-type in the presence vs absence of the powdery mildew pathogen (Zierold et al., 2005). Nevertheless, despite intensive efforts to elucidate the biochemical function of MLO proteins their core biochemical activity remains elusive. MLOs are predicted to reside in lipid bilayers with a consensus topology of seven transmembrane (7-TM) domains. In the case of barley Mlo, extensive biochemical analysis verified the supposed 7-TM topology and revealed plasma membrane localization with an extracellular amino terminus and an intracellular carboxy terminus (Devoto et al., 1999). The cytoplasmic tail harbours a binding site for the ubiquitous calcium sensor calmodulin. The presence and relative position of this calmodulin binding domain is conserved throughout the protein family (Kim et al., 2002). Absence of any other identifiable protein domains precludes further hints on MLO function to date. The topology and subcellular localization of barley Mlo is reminiscent of the G-Protein Coupled Receptor (GPCR) superfamily in metazoans (Temple & Jones, 2007). In barley, previous genetic (transient gene expression) and pharmacological data did not support a role of the heterotrimeric G-protein α subunit in Mlo function or mlo-mediated powdery mildew resistance (Kim et al., 2002). Results of a recent study based on loss-of-function mutants indicate that Atmlo2-mediated powdery mildew resistance is modulated by the heterotrimeric G-protein Gβ and Gγ subunits in Arabidopsis (Lorek et al., 2013). However, the overall findings of this and a previous study (Chen et al., 2009) are inconsistent with a role of MLO proteins as canonical GPCRs.

Until recently it remained unclear whether barley Mlo and AtMLO2 play a direct role in plant defence or whether the susceptibility-conferring activity of these proteins is an indirect consequence of their manipulation by the pathogen (Panstruga & Schulze-Lefert, 2003). A recent meta-analysis of genome-wide barley and Arabidopsis expression data revealed that barley Mlo and AtMLO2 are coexpressed with a core set of genes with proven functions in plant immunity, including Arabidopsis PEN1, PEN2 and PEN3 and barley Ror2. Results of pathogen assays showed that some of the coexpressed genes are required for nonhost resistance against powdery mildew (Humphry et al., 2010). In sum, these data suggest that Mlo and AtMLO2 are integral components of an evolutionarily conserved and transcriptionally coordinated machinery for basal and nonhost resistance. In agreement with a direct role in plant defence, AtMLO2 was recently identified as a target of the Pseudomonas syringae type III effector HopZ2 (Lewis et al., 2012).

The question remains of how MLO proteins might link to components of plant immunity. There is accumulating evidence supporting a connection of MLO proteins to Receptor-Like Kinases (RLKs). First, a number of RLKs are present in the set of genes coexpressed with Mlo/AtMLO2 (Humphry et al., 2010). Secondly, feronia (affected in a RLK) and Atmlo7 mutants exhibit similar phenotypes with regard to pollen tube perception, while feronia and Atmlo2 mutants show a similar level of powdery mildew resistance. These findings suggest that the corresponding gene products (the FER RLK and MLO proteins) may cofunction in the same biochemical pathway (Kessler et al., 2010). Thirdly, recent yeast-based interaction studies revealed evidence for direct physical contact between AtMLO proteins and various RLKs (Lalonde et al., 2010). Despite these indications it remains to be seen whether MLO proteins and RLKs are indeed interconnected in a functional sense.

All together now: phylogenetic analysis of MLO proteins

The availability of complete plant genomes led to the realization that MLO proteins are part of a sizeable family. The 15 Arabidopsis MLO proteins were originally grouped into four major clades (Chen et al., 2006). Subsequently, a more detailed analysis including 17 MLO members of grapevine (Vitis vinifera) and some barley, bread wheat (Triticum aestivum), rice (Oryza sativa) and corn (Zea mays) MLO proteins distinguished six distinct clades (Feechan et al., 2008). This classification was, with minor variations, confirmed in subsequent studies analysing the MLO families of additional plant species (Liu & Zhu, 2008; Konishi et al., 2010; Jiwan et al., 2013; Zhou et al., 2013; Chen et al., 2014; Deshmukh et al., 2014) (Fig. 2). Several species have been subjected to detailed genome-wide phylogenetic analysis regarding MLO proteins, indicating that the actual number of MLO proteins per plant species varies considerably (Table 1). While gene expression data indicate that bread wheat seems to encode only eight members (Konishi et al., 2010), soybean (Glycine max) was recently found to possess 39 MLO members, the largest MLO protein set detected so far (Deshmukh et al., 2014).

Table 1. Mildew resistance locus o (MLO) family members of selected plant species and their phylogenetic classificationa
Plant speciesCommon nameNumber of MLO genesCladeReference
IIIIIIIVVVIVII
  1. a

    Only fully characterized MLO families are shown. Classification is based on literature.

  2. b

    One truncated MLO family member (SlMLO14) was excluded from phylogenetic analysis.

Arabidopsis thaliana Thale cress153350310Chen et al. (2009)
Cucumis sativus Cucumber144230311Zhou et al. (2013)
Glycine max Soybean3975821160Deshmukh et al. (2014)
Oryza sativa Rice122613000Liu & Zhu (2008)
Prunus persica Peach163620320Jiwan et al. (2013)
Solanum lycopersicum Tomato17b3430411Chen et al. (2014)
Triticum aestivum Wheat81313000Konishi et al. (2010)
Vitis vinifera Grapevine173321620Feechan et al. (2008)
Figure 2.

Phylogenetic relationships of mildew resistance locus o (MLO) proteins. Phylogenetic tree representing the relationship of the seven MLO clades. The unrooted tree was obtained by the Maximum Likelihood method based on the JTT matrix-based model, with 1000 bootstrap replications. Evolutionary analyses were conducted using MEGA v6 (Tamura et al., 2013) and are based on 122 amino acid sequences from published MLO proteins (barley Mlo plus those listed in Table 1 except the Prunus persica sequences, which are not accessible via GenBank and excluding the truncated SlMLO14 sequence, which might be a pseudogene). Subtrees were collapsed to highlight the respective MLO clades (black triangles). Bar, 0.1 amino acid exchanges per 100 amino acids. Known physiological associations of MLO are summarized in the balloons next to the respective clade. Note that in this particular tree clade IV is underrepresented owing to few monocot sequences in the protein set.

All MLO proteins from dicot species that are known to be associated with powdery mildew susceptibility sort into clade V, including AtMLO2, AtMLO6, AtMLO12 (Consonni et al., 2006), tomato SlMLO1 (Bai et al., 2008), pea Er1/PsMLO1 (Humphry et al., 2011; Pavan et al., 2011), and grapevine VvMLO3 and VvMLO4 (Feechan et al., 2013) (Fig. 2). Interestingly, none of the known monocot MLO gene products clusters to clade V (Feechan et al., 2008; Jiwan et al., 2013; Zhou et al., 2013). Barley Mlo is found in clade IV, which is primarily but not exclusively represented by monocot MLO proteins (Table 1, Fig. 2). In sum, MLO proteins involved in powdery mildew interaction appear to be restricted to these two clades. As an additional characteristic feature, these proteins share a specific carboxy-terminal tetrapeptide motif (Panstruga, 2005a).

Because most other MLO genes have not been associated to any mutant phenotype or molecular function, the significance of the other clades and their respective MLO members remains in part elusive. AtMLO7, a member of clade III, has been found to be required for pollen tube perception of the female germ cell (see previous section and Fig. 1c; Kessler et al., 2010). One study used a peach member of this clade, PpMLO1, for antisense expression in strawberry, which seemed to confer resistance to strawberry powdery mildew (Jiwan et al., 2013), even though this does not conclusively associate PpMLO1 with susceptibility to powdery mildew. Furthermore, Arabidopsis AtMLO4 and AtMLO11, found in clade I, were shown to be involved in root thigmomorphogenesis (see above; Fig. 1b; Chen et al., 2009).

It seems that MLO genes have been subject to both ancient and recent gene duplications, and diversification before and after divergence of monocot and dicot plants (Jiwan et al., 2013). This conclusion is supported by the appearance of several copies of some MLOs with extensive sequence similarity within the same clade. Prominent examples of the latter are AtMLO2, AtMLO6 and AtMLO12, which share partially redundant functions in susceptibility to Arabidopsis powdery mildews (Consonni et al., 2006) and are likely the result of recent segmental gene duplications in Arabidopsis. Interestingly, clades I–IV are populated by MLO members of both dicot and monocot species, whereas clades V and VI seem to be represented by dicot MLO proteins exclusively (Feechan et al., 2008; Jiwan et al., 2013; Zhou et al., 2013; Table 1). This scenario suggests that diversification of clades I–IV took place before the divergence of monocots and eudicots, whereas clades V and VI appear to be recent dicot innovations that emerged after the monocot–dicot split (Jiwan et al., 2013). With the exception of soybean, few members sort to clade VI, which could mean that this clade is the most recent addition to the MLO family. Interestingly, two studies proposed clade VII, which would be represented by CsMLO11 from cucumber (Cucumis sativus; Zhou et al., 2013) and SlMLO2 from tomato (Solanum lycopersicum; Chen et al., 2014). Our own phylogenetic analysis supports the existence of this novel clade (Fig. 2), which is seemingly not represented in all plant species.

The most ancient clade seems to be clade I. Known MLO gene products from mosses (e.g. Physcomitrella patens) and ferns (e.g. Selaginella moellendorfii) are solely found in this branch (Jiwan et al., 2013). There they group together with MLO proteins from both dicot and monocot species, suggesting that this clade was present early in plant evolution, even before the separation of vascular and nonvascular plants. The oldest acquisition of seed-producing plants (Spermatophyta) seems to be clade II, which consistently shows the closest association to the ancient clade I (Feechan et al., 2008; Jiwan et al., 2013; Zhou et al., 2013). Because MLO proteins from gymnosperm species have not yet been considered in the published studies, their phylogenetic relationships to angiosperm MLOs remain elusive, making any further conclusions about the evolution of clades II–IV impossible.

With a little help from my friends: complementation of mlo phenotypes by MLO orthologues and paralogues

Mutant phenotypes revealed apparent isoform specificity with regard to MLO functions, which could result either from tissue/cell type specific expression of MLO genes and/or isoform-specific features of MLO variants. Transient gene expression studies in leaves of a barley mlo null mutant genotype (mlo-5) revealed that wheat and rice orthologues of barley Mlo (TaMloB1 and OsMlo2, residing in the same phylogenetic clade as barley Mlo) partially complement the powdery mildew resistance phenotype (Elliott et al., 2002). Similarly, two out of four tested grapevine co-orthologues of AtMLO2 partially complemented powdery mildew resistance in transgenic Atmlo2 Atmlo6 Atmlo12 lines (Feechan et al., 2013), and the pepper CaMLO2 gene restored powdery mildew susceptibility in the ol-2 (Slmlo1) mutant (Zheng et al., 2013). By contrast, attempts to complement the aberrant root curling phenotype of the Atmlo4 mutant with either the paralogue AtMLO2 or a panel of different AtMLO2/AtMLO4 chimeras failed, although their expression in transgenic lines was driven by the authentic AtMLO4 promoter (Chen et al., 2009). Taken together, these results indicate that members from the same phylogenetic clade can (partially) complement each other, while members from different clades cannot. This suggests that primarily isoform-related features determine the specificity of MLO function.

Let it be: breeding for the future

The directed creation and subsequent utilization of mlo alleles seems an exciting strategy to confer broad-spectrum powdery mildew resistance to crop species. Identification of the respective MLO paralogues that confer powdery mildew susceptibility in a given species is a prerequisite for this approach. This might be achieved either by bioinformatic analysis on the basis of characteristic MLO sequence motifs (e.g. Panstruga, 2005a) or experimentally via heterologous complementation analysis (e.g. Elliott et al., 2002). Once the correct target(s) have been recognized, different experimental approaches can be deployed to generate mlo loss-of-function genotypes. Barley stripe mosaic virus-induced gene silencing (BSMV-VIGS) and antisense approaches have been used to downregulate Mlo expression in barley (Schweizer et al., 2000; Delventhal et al., 2011), wheat (Várallyay et al., 2012) and peach (Jiwan et al., 2013), demonstrating that targeting MLO genes via RNA interference can result in enhanced powdery mildew resistance. However, this approach relies on sustained transgene expression, which is often undesirable (e.g. in agricultural settings). Alternatives comprise methods that either do not involve any transgenes (TILLING) or that make only temporary use of them (TALEN, CRIPSR), followed by their outcrossing in subsequent generations.

TILLING (Targeting Induced Local Lesions IN Genomes) combines chemical mutagenesis with high-throughput genome-wide screening for lesions in genes of interest (McCallum et al., 2000). The procedure can be applied to any plant species, irrespective of its ploidy and genome size. In addition, the mutations are stably inherited. The feasibility of TILLING has been demonstrated for a large number of agronomically important crops including rice, barley, wheat and tomato (reviewed in Kurowska et al., 2011). Typically a wide range of mutations including nonsense mutations, missense changes and alterations in splice junction sequences can be obtained by TILLING. With regard to mlo resistance this might be particularly powerful because it has been shown that the extent of its pleiotropic effects can be balanced by the choice of the mutant allele (Piffanelli et al., 2002).

In the past few years, methods for targeted genome modification in plants have become practical. The discovery of Transcription Activator-Like Effectors (TALEs) in Xanthomonas spp. which bind specific genomic sequences in their host plant and manipulate expression of pathogenesis-related genes, was groundbreaking in this respect (Kay et al., 2007; Römer et al., 2007). TALEs possess a central DNA-binding domain, consisting of a series of near-identical repeats of 33–35 amino acids, with each repeat coding for a target DNA base pair via Repeat Variable Diresidues (RVDs) situated at position 12 and 13 (Boch et al., 2009; Moscou & Bogdanove, 2009). Fusion of TALEs with the sequence-nonspecific bacterial endonuclease FokI (TALEN) have been shown to induce aberrations at genomic target sites in species of all kingdoms, including several plant species such as Arabidopsis, barley, rice, and Brachypodium (reviewed in Chen & Gao, 2013). This tool has proven to be very useful for introduction of desired genomic alterations in many plant genes and promises efficient application for targeted modification of MLO genes in many plant species.

One disadvantage in the application of TALENs is that the assembly of custom TALEN modules is complex and requires experience with the required nonclassical cloning methods. However, TALENs are not the only tool that allows targeted genome modification. Clustered, Regularly Interspaced, Short Palindromic Repeats (CRISPR) are short repetitive stretches of DNA that, with the help of bacterial Cas proteins, confer adaptive defence against foreign DNA molecules such as plasmids and DNA viruses (Jinek et al., 2012; Wiedenheft et al., 2012). The type II CRISPR system from Streptococcus pyogenes consists of a mature CRISPR RNA (crRNA) and a trans-activating crRNA (tracrRNA), which guide the endonuclease Cas9 as a duplex to bind and cleave a specific DNA target sequence (Jinek et al., 2012). This system can be applied for targeted genome modification, using a single custom guide RNA (gRNA) including a protospacer that is complementary to the target sequence (Jinek et al., 2012). Indeed, Cas9/gRNA systems have been successfully adapted for application in many species including plants (reviewed in Belhaj et al., 2013). These systems appear to be highly specific for their target sequence and as efficient as TALENs but easier to handle (Jinek et al., 2012). Accordingly, CRISPR/Cas9 is a promising tool for the generation of mlo mutant plants. Altogether, the methods presented here represent versatile instruments to generate formerly unavailable mlo mutant plants, holding great promises for both future research and agriculture.

Acknowledgements

We acknowledge The Beatles for usage of some of their song titles in our section headings. We thank Sharon Kessler (University of Oklahoma) and Chiara Consonni for providing unpublished nortia micrographs and Arabidopsis Col-0 and Atmlo2 Atmlo6 Atmlo12 plant photographs, respectively. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG PA861/11-1).

Ancillary