Plant pathogens secrete effector proteins into host tissues to promote infection through the manipulation of host processes (Win et al., 2012). During host colonization, rust fungi form haustoria that invaginate the host plasma membrane within the host cell cavity. These structures mediate the molecular traffic between the parasite and its host, and notably the delivery of effector proteins into host cells (Rafiqi et al., 2012), although other structures such as infection hyphae are also likely to be involved in this molecular traffic (Rafiqi et al., 2010). Until now, six effector proteins have been identified in three different rust species: AvrM, AvrL567, AvrP123, and AvrP4 in the flax rust fungus Melampsora lini, the Rust Transferred Protein RTP1 in the bean rust fungus Uromyces fabae, and PGTAUSPE-10-1 in the wheat stem rust fungus Puccinia graminis f. sp. tritici (Table 1; Kemen et al., 2005; Ellis et al., 2007; Upadhyaya et al., 2014). They are all secreted proteins expressed in haustoria, with no clearly identified biochemical function. How they promote fungal growth inside host tissues remains unknown (Table 1). In contrast, their avirulence (Avr) properties (i.e., the ability to trigger specific immune responses) and/or their trafficking mechanisms (i.e., how they enter plant cells) are better understood.

The four M. lini effector proteins were first identified as effectors due to their Avr properties (Ellis et al., 2007). More recently, a screen with a bacterial protein delivery system in wheat revealed the P. graminis f. sp. tritici protein PGTAUSPE-10-1 which causes cell death in the host line carrying the resistance gene Sr22; PGTAUSPE-10-1 was thus considered as a candidate AvrRs22 effector (Upadhyaya et al., 2014). M. lini AvrL567 and AvrM are model Avrs for the study of effector recognition by immune receptors. Both proteins are recognized inside plant cells by specific immune receptors following a direct physical interaction (Table 1; Dodds et al., 2004, 2006; Catanzariti et al., 2006, 2010). For both effectors, 3D structure-driven amino acid substitutions revealed multiple contact points mediating the interaction with their cognate receptor (Wang et al., 2007; Ravensdale et al., 2011; Ve et al., 2013). Amino acid residues within these contact points are highly variable, suggesting that an arms race is taking place between these effectors and their corresponding receptors. Such knowledge of Avr-receptor interactions is valuable for engineering improved immune receptors with expanded effector recognition (Harris et al., 2013; Segretin et al., 2014), which may ultimately help to develop broad-spectrum resistance in plants (Dangl et al., 2013).

All six rust effector proteins are thought to be translocated from haustoria into host cells (Table 1). RTP1 and AvrM have been directly shown to traffic from haustoria to plant cells during infection (Kemen et al., 2005, 2013; Rafiqi et al., 2010), whereas the direct recognition of AvrM and AvrL567 by cytosolic plant immune receptors indirectly demonstrates their internalization in the plant cell (Ellis et al., 2007). Current mechanistic models based on pathogen-free assays suggest that AvrP4, AvrM, and AvrL567 proteins can enter plant cells autonomously (Catanzariti et al., 2006; Kale et al., 2010; Rafiqi et al., 2010). Rafiqi et al. (2010) further showed that AvrL567 and AvrM cell entry is mediated by divergent N-terminal uptake domains, carrying hydrophobic residues that are critical for cell entry in the case of AvrM (Ve et al., 2013). This model and the assays used to build it are currently debated, and the need to study effector trafficking during the infection has been stressed (Petre and Kamoun, 2014).

Effector proteins are anticipated to be key molecules for pathogenicity, although very little is known about how they function within host tissues. Among the six characterized rust effectors, none possess a clearly identified biochemical function or a detected virulence activity (Table 1). Indeed, M. lini transgenic lines silencing AvrL567 did not show any reduced growth on flax, suggesting that this effector is not required for full virulence (Lawrence et al., 2010). As discussed by the authors, this could be explained by a high functional redundancy in the M. lini effector repertoire (Lawrence et al., 2010). Such redundancy was also observed in the effector repertoires of bacterial plant pathogens (Kvitko et al., 2009), and represents an obstacle for the functional characterization of virulence effector functions through genetic approaches. However, recent progresses have been made regarding RTP1, a conserved rust effector that seems to work as a protease inhibitor (Pretsch et al., 2013). On the other hand, Kemen et al. (2013) reported that RTP1 accumulates within the host-parasite interface and forms filaments. The authors proposed a role as a structural effector, possibly stabilizing fungal structures during infection. A model that integrates the different RTP1 localizations and proposed functions remains to be drawn. Several methods for the genetic transformation of M. lini and U. fabae, as well as for host-induced gene silencing (HIGS) of Puccinia triticina have been reported (Lawrence et al., 2010; Djulic et al., 2011; Panwar et al., 2013). Such methods, although they are still at various stages of development, represent valuable tools to investigate the contribution of individual effectors to virulence during infection.

In the past few years, a typical profile has emerged for plant pathogen effectors. Fungal proteins are usually considered candidate secreted effector proteins (CSEPs) if they possess a signal peptide for secretion, a small size and no other targeting sequence or transmembrane domains (Stergiopoulos and de Wit, 2009; Rouxel and Tyler, 2012; Saunders et al., 2012). Such CSEPs attract more attention when they are expressed during infection or when they present signatures of rapid evolution. Besides, expression in specific infection structures such as haustoria, often considered as a major site of effector delivery, provides another level of information. Some authors also take advantage of conserved amino acid motifs or predicted protein structures to establish large CSEP classes (Godfrey et al., 2010; Pedersen et al., 2012). Homology to known rust effectors and organization in gene families or in physical clusters have also been considered to refine these sets of CSEPs (Hacquard et al., 2012; Saunders et al., 2012). In rust fungi, such criteria have been applied in the frame of effector mining pipelines that combined genome-wide analyses and transcriptomics to reveal amazingly rich catalogs of rust CSEPs (Cantu et al., 2011, 2013; Duplessis et al., 2011a; Fernandez et al., 2012; Hacquard et al., 2012; Saunders et al., 2012; Garnica et al., 2013; Zheng et al., 2013; Bruce et al., 2014; Link et al., 2014; Nemri et al., 2014; Table 2).

Various studies combined genome sequencing and transcriptomics to provide sets of CSEPs. Automated pipelines for effector mining should be unified and systematically applied to forthcoming rust fungi genomes to provide a solid foundation for future comparative analyses in Pucciniales. However, an important point to consider is the need for an accurate curation of CSEP-encoding genes in these genomes and the screening of additional time points in time-course studies and/or spore stages. Some early genome-wide surveys of CSEPs in plant interacting fungi arbitrarily focused on small proteins because of the commonly observed small size of effectors and in order to reduce manual gene curation efforts (Stergiopoulos and de Wit, 2009, Duplessis et al., 2011a). Considering that rust fungi effectors can exhibit greater size (e.g., M. lini AvrM), such an arbitrary cut-off should not be considered in future analyses of rust CSEPs.

To face the growing number of CSEPs made available by effector mining studies, and to better understand their functions in plant cells, we need tools to study them directly in planta. This relies on the ability to genetically transform the plant to perform high-throughput functional analyses (also referred to as “effectoromics”). Rust fungi hosts (e.g., wheat, soybean, flax, or poplar), are not easily amenable to molecular genetic approaches. However, non-host model plants can be used to characterize and screen CSEPs. For instance, the Agrobacterium-mediated transient genetic transformation of Nicotiana benthamiana has proven useful to rapidly express effector proteins into plant cell, but has been largely ignored in rust effector biology. This system allows combining many different approaches (cell-biology, protein biochemistry, hypersensitive response and infection assays) all in one. Thus, such approaches may help in (1) determining the sub-cellular localization of candidate effector proteins using effector-fluorescent protein fusions, (2) identifying interacting partners within protein complexes, (3) detecting candidate effector capacity to enhance susceptibility during infection with selected N. benthamiana pathogens (thus validating a role in virulence), and (4) testing their recognition by specific immune receptors.

Benjamin Petre and Sébastien Duplessis compiled data from the literature and drafted the manuscript. All the authors wrote and revised the article.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.