Introduction

Tomato (Solanum lycopersicum Mill.) is a highly popular horticultural crop cultivated worldwide, although it is prone to several plant pathogens (Singh et al. 2017). Phytophthora infestans (Mont.) de Bary, the causal agent of late blight of tomato, is one of the most aggressive pathogens of tomato and causes crop loss (Nowicki et al. 2012). P. infestans is a member of the Peronosporaceae family of the phylum Oomycota, and it infects the entire plant, including the stems, leaves and fruits of tomato (Erwin and Ribeiro 1996). In 2007, the late blight epidemic episode in inner Mongolia, China, caused damage to approximately 638,900 tons of tomato production (Li 2008). In addition, this disease is also reported to cause continual annual losses of winter tomato production in Florida, which is a USD 464 million industry and produces 36% of the total fresh tomatoes of USA (Schultz et al. 2010).

The major symptom associated with late blight in tomato is the formation of small blackish/brown lesions on leaves, fruit and stems that progress to water-soaked chlorotic spots, which ultimately lead to necrosis in the entire plant (Fig. 1). P. infestans reproduces by both sexual and asexual means, but infection (Fig. 2) via sexual spores is relatively rare (Fry 2008). Asexual spores are dispersed via air and seed, while sexual spores are dispersed via soil (Seifu 2017). Aerial dispersion of the pathogen occurs via asexual spores known as sporangia, which is the primary mode of infection for P. infestans (Leesutthiphonchai et al. 2018). Sporangia are formed in sporangiophores that are dispersed by wind or water and facilitate the rapid dispersal of the pathogen (Aylor 2003). Sporangia settle on the leaf surface, and under favourable conditions (20–25 °C and available nutrients), sporangia germinate directly and cause infection, while under unfavourable conditions (lower temperature of 10–15 °C and low nutrients), sporangia cleave their multinucleate cytoplasm and release 3–8 biflagellate mononucleate motile zoospores (Fry 2008; Grenville-Briggs et al. 2005). Motile zoospores rapidly encyst and produce a germ tube that invades the plant leaves and stems through stomata or develops an appressorium that invades the cuticle and epidermal cell wall (Kots et al. 2017). Following the invasion, intracellular hyphae of the pathogen ramify within the host cells (Fig. 2). P. infestans has a hemibiotrophic lifestyle, with an initial biotrophic infection phase that mainly relies on living host cells followed by a necrotrophic phase in which host cell death is induced (Botero et al. 2018).

Fig. 1
figure 1

Phytophthora infestans infected tomato foliage and fruits exhibiting necrosis

Full size image
Fig. 2
figure 2

Disease cycle of Phytophthora infestans in tomato. The diagram illustrates infection through leaves, while fruit or oospores in the soil also act as infection repositors between seasons

Full size image

In addition to aerial dispersion, dispersion via seeds is another concern. Often, transport of asymptomatic infected fruits can disperse infected seeds to new locations. Asymptomatic infected fruit dispersion was the factor underlying the late blight epidemic in tomato in the eastern United States (Leesutthiphonchai et al. 2018). Dispersion via soil mainly occurs by oospores, in which sexual spores are produced by the two known mating types, A1 and A2 (Judelson 2007). Oospores are more tolerant than sporangia and can easily survive adverse conditions in soil between growing seasons (Drenth et al. 1995).

Control of P. infestans mainly relies on chemical fungicide applications and resistance breeding (Poudel et al. 2020). However, P. infestans is difficult to manage because of its high adaptability to overcome the resistance of host plants (Fry 2008; Hass et al. 2009). Although few resistance genes have been identified in wild relatives of tomato (Van der Vossen et al. 2003; Pel et al. 2009; Zhang et al. 2013), several of the resistance genes have been overcome by different strains of P. infestans (Vleeshouwers et al. 2011). The ability of the pathogen to overcome the resistance was explained as a consequence of unique genome organization, which emphasizes the extraordinary adaptation ability of the pathogen against host plant resistance (Haas et al. 2009; Raffaele et al. 2010). In addition, host plants have evolved several strategies to shield themselves from pathogen attack. Hence, knowledge of the molecular basis of plant–pathogen interactions is essential for designing an effective strategy for controlling fast-evolving pathogens, such as P. infestans. In this review, we focus on recent advancements in our understanding of the molecular basis of pathogenicity between P. infestans and its host plant tomato and discuss the progress made regarding the identification of resistance in the host genome, effector proteins, metabolomic models, signalling process and quantitative trait loci (QTLs).

Molecular pathogenesis

Plants present several layers of immune defence. The interactions between pathogens and plant immune defence responses determines the fate of the pathogen in the host plants (Gilroy et al. 2011). The first line of immune defence against invading pathogens relies on a large family of pattern recognition receptors (PRRs), which recognize the unique evolutionarily conserved structures of pathogens known as pathogen-associated molecular patterns (PAMPs) (Altenbach and Robatzek 2007). The recognition of PAMPs by host plants induces PAMP-triggered immunity, which results in the generation of reactive oxygen species (ROS) and hypersensitive cell death in the host plant (Furuichi et al. 2014). During the infection process, host signalling processes, such as mitogen‐activated protein kinases (MAPKs), are activated and modulate the expression of pathogenesis‐related (PR) genes. However, P. infestans suppresses the host immune system while minimizing damage to plants because host cell integrity is crucial for its initial biotrophic lifestyle phase (Leesutthiphonchai et al. 2018). A schematic overview of the molecular pathogenesis of P. infestans is presented in Fig. 3. P. infestans has evolved several strategies to overcome PAMP-triggered immunity (PTI) by delivering immunity-suppressing molecules known as effectors to the plant. However, several of these effectors are recognized by the host plant and activate effector-triggered immunity (ETI) and subsequently programmed cell death in the host plant (Gilroy et al. 2011). The host plant ETI is mediated by a group of highly specific and conserved plant disease resistance (R) genes (Saunders et al. 2012a). We have discussed the molecular interaction between pathogens and host plants from two perspectives: from the pathogen and from the host plant.

Fig. 3
figure 3

Schematic model showing the molecular pathogenesis of Phytophthora infestans. Zoospores released from sporangium encysted and then germinated. P. infestans releases pathogen-associated molecular patterns (PAMPs), which are recognised by host pattern recognition receptors (PRRs), and induces PAMP-triggered immunity. To suppress PTI, P. infestans secretes cell wall degrading enzymes (CWDE), toxins and RXLR. Some RXLR effectors are recognised by the host resistance (R) gene and induce effector-triggered immunity (ETI). For simplicity, all possible steps are not shown, and only some of the characterised protein targets and activities are presented in this diagram. AVR avirulence, MAPK mitogen-activated protein kinase, CMPG1 ubiquitin E3 ligase CMPG1, HR hypersensitive response, ICD infestin 1-triggered cell death, ROS reactive oxygen species

Full size image

Viewpoint from the pathogen

In the initial biotrophic phase, during which the host’s immune defence and cell death are actively suppressed, pathogens produce invasive hyphae or haustoria within living plant cells (Koeck et al. 2011). Through haustoria, the pathogen derives nutrients from the host and releases a large array of effector proteins (Panstruga and Dodds 2009). Effectors can be extracellular (apoplastic) or intracellular (cytoplasmic) based on their location in plant cells (Fig. 3). Extracellular effectors are secreted in the apoplast and interact with extracellular defence-associated factors, whereas cytoplasmic effectors are transported inside the host cytoplasm and localised in various plant subcellular regions (Sharpee and Dean 2016; Wang et al. 2017).

Extracellular effectors

Extracellular effectors released by P. infestans are mainly of two types. The first type mediates the protection of the pathogen from host defence, and the second type mediates host invasion. Effectors that mediate protection of the pathogen from the host defence include protease inhibitors and glucanase inhibitors. Protease inhibitor effectors inhibit host-resistant proteases (Wang et al. 2019). For example, EPI1 and EPI10 effectors released by P. infestans inhibit the tomato subtilisin-like protease P69B (Tian et al. 2004). Similarly, EPIC1 and EPIC2B inhibit the tomato defence cysteine protease Rcr3pim (Song et al. 2009). Glucanase inhibitor proteins (GIPs) secreted by Phytophthora bind and inhibit the release of host extracellular endo-β-1,3-glucanases, which target the β-1,3 glucan polymers of the oomycete cell wall (Damasceno et al. 2008).

The second type of effector mediates host invasion using several approaches, including secretion of cell wall degrading enzymes (CWDEs) (Ospina-Giraldo et al. 2010; Wawra et al. 2012) and toxins (Liu et al. 2005). P. infestans secretes CWDEs, such as carbohydrate esterase, glycosyl hydrolases and polysaccharide lyases, which degrade the plant cell wall, thereby allowing entry of the pathogen into host tissue (Ospina-Giraldo et al. 2010). P. infestans also releases a group of toxins to establish infection inside the host cell. For example, P. infestans releases the small S locus cysteine-rich protein (SCR) family (Liu et al. 2005). The SCR gene family is highly similar to the PcF gene family, which is known for its phytotoxic necrosis-inducing role in Phytophthora cactorum (Chen et al. 2016). Another toxin secreted by P. infestans belongs to necrosis and ethylene-inducing protein 1 (Nep1)-like family (NLPs), which are involved in the damage of plasma membranes and successive cytolysis (Ottmann et al. 2009). NLPs were proposed to possess dual functions in plant-pathogen interactions, acting both as elicitors of immune responses and as toxin-like virulence factors (Qutob et al. 2006).

Elicitor is a general term that refers to molecules that act as PAMPs for the plant and subsequently induce defence responses against pathogens (Huet et al. 1994; Maffei et al. 2012). Functional characterization of the NLP protein PiNPP1.1 in P. infestans showed that rather than being elicitor, it acts as a toxin in Nicotiana benthamiana and tomato (Kanneganti et al. 2006). P. infestans secretes another group of toxins called elicitins (Huet et al. 1994; Derevnina et al. 2016). Elicitin are elicitors with highly conserved sterol-binding proteins with characteristics of PAMPs, secreted by oomycete pathogens (Derevnina et al. 2016). Elicitins were first demonstrated by Ricci et al. (1989) in oomycete pathogens Phytophthora cryptogea and Phytophthora capsici that elicit HR response and systemic acquired resistance in Nicotiana species (Ricci et al. 1989) and some radish cultivars (Keizer et al. 1998). Following that several studies have demonstrated elicitins from a diverse family of oomycete, such as cryptogein from P. cryptogea, capsicein from P. capsici, parasiticein from, and INF1 from P. infestans (Derevnina et al. 2016). In tomato, elicitins INF1 and INF1S3 induce basal defence, such as activation of jasmonic acid‐and ethylene‐mediated signalling pathways, but do not induce HR cell death or SA-mediated pathway activation (Kawamura et al. 2009). Therefore, INF1 might act as a PAMP in tomato; however, such defence responses were not enough to suppress host colonization by P. infestans because tomato is susceptible to the pathogen. In addition to the abovementioned toxin, during the process of plant–pathogen interactions, P. infestans also secretes proteins with RGD (Arg-Gly-Asp)-containing effector protein IPI-O, which weakens the plant defence responses by damaging the interconnections between the plant cell wall and plasma membrane (Senchou et al. 2004).

Intracellular effector secreted in the cytosol of host plants

Compared with extracellular effectors, intracellular effectors are translocated after secretion into host cells (Kamoun et al. 2015). Plants have evolved resistance (R) genes that encode R proteins that are capable of recognizing several of these effector genes, leading to the induction of effector-triggered immunity (ETI) (Jones and Dangl 2006). This immune reaction involves a hypersensitive response, followed by programmed cell death. Among the intracellular effectors in P. infestans, the RXLR effector has been extensively studied, followed by the CRN effector.

RXLR effector

RXLR effectors are secreted from haustoria and translocated to the host cell (Ackerveken 2017). The RXLR effector consists of an N-terminal signal peptide and RxLR motif (where R is arginine, X is any amino acid, and L is leucine) (Wawra et al. 2017) and is often followed by another conserved motif, i.e., a dEER motif (Asp-Glu-Glu-Arg) (Win et al. 2012). The C-terminal domain mostly consists of a variable number of motifs that appear in a repeated manner (Jiang et al. 2008). The N-terminal domain of the RXLR motif is essential for secretion and targeting, while the C-terminal domain is essential for effector functions (Bos et al. 2006). Phylogenetic analyses have indicated that RXLR motifs are highly conserved and signatures of positive selection have been identified in the C-terminal region of a number of RXLR-class effectors (Win et al. 2007). Genome-wide analysis of the RXLR effector gene family in P. infestans has shown the presence of approximately 563 RXLR effector genes (Hass et al. 2009). Studies have shown that thirty-one RXLX effector genes are expressed during interactions between tomato and P. infestans (Zuluaga et al. 2016), and fourteen of these genes were novel compared to RXLX effector genes expressed during the interaction between potato and P. infestans. Some of these effectors have proteins that can be recognized directly or indirectly by the host plant’s resistance (R) gene and are subsequently termed avirulence (AVR) genes (Petit-Houdenot and Fudal 2017). The identification of RXLR effectors and recognition that some of them act as AVR factors have driven the identification of several effector genes and their molecular interaction with AVR (Amaro et al. 2017). Knowledge of R-AVR interactions is very important for designing strategies to control oomycete diseases in plants.

Avirulence 3 (AVR3a) is the first functionally characterized AVR effector gene of P. infestans (Bos et al. 2010). The AVR3a gene has a virulent allele and an avirulent allele that differ by only two amino acids: AVR3a (KI) and AVR3a (EM) (Huang et al. 2019). AVR3a (KI) is recognized by the host resistance gene R3a, which strongly suppresses infestin 1 (INF1)-triggered cell death (ICD), and AVR3a (EM) eludes recognition by R3a and thus weakly suppresses host ICD (Bos et al. 2003; Torto et al. 2003). AVR3a targets and stabilizes the host ubiquitin E3 ligase CMPG1, which is necessary for inducing ICD. A mutation in the C-terminal tyrosine residue of AVR3a failed to suppress ICD (Bos et al. 2010). Other than AVR3a, another AVR effector protein of P. infestans, AVRblb2, is recognized by host plants. AVRblb2 interferes with host immunity by targeting the host plant defence protease papain-like cysteine protease C14, which elevates the susceptibility of host plants to P. infestans by preventing its secretion into the apoplast (Bozkurt et al. 2011). The RXLR effectors SFI1, SFI2, SFI3, SFI4, SFI5, SFI6, SFI7, and SFI8 also manipulate host immunity by targeting PTI (Zheng et al. 2014). However, the exact molecular mechanism and plant proteins targeted by these effectors have yet to be explored.

Some effector proteins require an association with host plant proteins to initiate interactions with resistance genes. For example, AVR2 effectors associate with the plant phosphatase BSU-LIKE PROTEIN1 (BSL1) and mediate the interaction of BSL1 with plant R2 genes that activate ETI in host plants (Saunders et al. 2012b). However, virulent strains of P. infestans express an unrecognized form of AVR2-like protein that does not facilitate the association of BSL1 with R2 despite interacting with BSL1. On the other hand, some effector proteins directly manipulate defence-associated transcription regulators. For example, the RXLX effector PITG_03192 interacts with plant NAC transcription factors (NTP1 and NTP2) at the endoplasmic reticulum and prevents their release from the ER to the host nucleus to enhance host susceptibility (McLellan et al. 2013). Some RXLR effectors target host susceptibility factors to promote virulence. For example, Pi04314 interacts with phosphatase 1 catalytic isoforms (Boevink et al. 2016), and Pi04089 interacts with the plant RNA-binding protein KRBP1 (Wang et al. 2015) to promote colonization. AVRblb2, another RXLX factor of P. infestans, interacts with host papain-like cysteine protease C14 and blocks its release to the apoplast to inhibit the degradation of virulence proteins of Phytophthora (Bozkurt et al. 2011). In addition, RXLR effectors such as AVR1, which is perceived by R1, interact and stabilize the exocyst component Sec5 (a subunit of the exocyst protein complex that is associated with vesicle trafficking) to enhance resistance against P. infestans (Du et al. 2015). Although much progress has been made in identifying several RXLR effectors, the function and molecular mechanism of many of the effectors are still unknown.

Crinkler (CRN) effector

P. infestans secretes a group of effector gene families known as Crinkler (CRN for CRinkling and Necrosis). The name is based on a typical leaf crinkling phenotype detected upon ectopic expression of two cDNAs of this group in host plants (Torto et al. 2003). Bioinformatic analysis of Phytophthora genomes has revealed the presence of 196 full-length genes and 255 pseudogenes in the CRN gene family and 10 CRN genes were actively expressed during P. infestans infection of potato (Hass et al. 2009). On the other hand, an analysis of CRN expression during P. infestans infection of tomato identified 51 novel CRN effector upregulations that were not found during P. infestans–potato infection (Zuluaga et al. 2016). This finding highlights that CRN effector expression is strongly influenced by the nature of the host. Similar to RXLR effectors, CRN proteins consist of conserved N-terminal signal peptides followed by diverse C-terminal domains (Hass et al. 2009). The N-terminus of the CRN protein includes a characteristic ∼ 50-amino-acid LXLFLAK motif and is responsible for CRN secretion and translocation into the host. The N-terminus of the CRN protein also contains an adjacent diversified DWL domain and HVLVXXP motif (Amaro et al. 2017). The C-terminus of the CRN is composed of diverse domain structures with effector and virulence functions (Schornack et al. 2010; Amaro et al. 2017). Compared with the N-termini, the C-terminal domains are highly diverse and often resemble enzymes, such as restriction endonuclease (REase), protein kinase, NTPease, HNH endonuclease, LK-nuclease and peptidase, which are predicted to be involved in toxicity determinants (Zhang et al. 2016). NTPase coupled with REase domains is common in prokaryotic organisms and mainly associated with transposable elements (Amaro et al. 2017). In P. infestans, the CRN coding gene PITG_23144 showed the presence of a gypsy retrotransposon, which indicated the prokaryotic origin of the C-terminus of the CRN protein (Haas et al. 2009). Compared with the C domains, no evidence has been found for the presence of the N-terminal domain of the CRN protein among prokaryotes (Amaro et al. 2017). Although recent studies have identified and characterized several CRN virulence functions, the molecular mechanisms required for CRN secretion and translocation remain largely unknown.

Viewpoint from the plant

Plant pattern recognition receptors (PRRs) are either mostly surface-localised receptor kinases or receptor-like proteins containing several ligand-binding ectodomains that recognize PAMPs (Zipfel 2014). Several PAMPs have been reported in P. infestans, such as PiPE (a mycelial wall-derived PAMP derived from the surface-existing glycoprotein fructose-1,6-bisphosphate aldolase of P. infestans) (Furuichi et al. 2013), Nep1-like proteins (Kanneganti et al. 2006), arachidonic and eicosapentaenoic acids (Bostock et al. 2011) and elicitin (Noman et al. 2020). PTI includes a wide range of responses to impede disease progression. Initial PTI responses include the generation of ROS, activation of calcium ion signalling (Cheval et al. 2013), activation of mitogen-activated protein kinases (MAPKs), and the expression of defence-related genes (Boller and Felix 2009). However, P. infestans releases an array of effector proteins contributing to virulence, among which some suppress PTI, while others reprogram host cell physiology and metabolic processes to establish host colonization (Bozkurt et al. 2012; Toruño et al. 2016). On the other hand, to restrict pathogen progression, the host plant uses the second layer of the recognition system, intracellular immune receptor NB-LRR proteins, also known as R (resistance) proteins. R genes account for approximately 1–3% of the genome of tomato, potato, pepper and tobacco (Wei et al. 2016).

Resistance protein of host plant

Plant resistance proteins are encoded by a family of R genes (Malik et al. 2020). R proteins recognise pathogens directly by binding to effectors or indirectly by sensing effector-induced alterations in other proteins of the host plant (Qi and Innes 2013). The R protein is composed of a C-terminal leucine-rich repeat (LRR) domain, a highly conserved middle nucleotide-binding site (NBS) domain and a diverse N-terminal domain (Takken and Goverse 2012). The LRR domain was also reported to play a role in the autoinhibition of the receptor preceding effector interactions to keep the R proteins in the “off” state (reviewed in Qi and Innes 2013; Bentham et al. 2018). The conserved C-terminal NBS domain functions as a molecular switch for R gene activation via nucleotide-dependent conformational changes mediated by ADP/ATP exchange (Takken et al. 2006). The N-terminus is composed of either the TIR (Toll/interleukin-1 receptor) domain or the CC (coiled-coil) domain (Elmore et al. 2011). Both of these domains are believed to function as receptor modules required for downstream signal transduction following R gene activation (Takken and Goverse 2012).

The interaction of the R protein and effector is a complex process, and several models, such as the elicitor-receptor model, guard model, and decoy model, have been proposed to explain the interaction mechanism. In the elicitor-receptor model, the R protein directly recognises its corresponding AVR protein and activates defence responses in host plants (Petit-Houdenot and Fudal 2017). This model was supported by the direct binding of few R-AVR combinations. However, for many R-AVR combinations, no direct physical interactions between effectors and R proteins were observed (reviewed in van der Hoorn and Kamoun 2008). To explain the indirect interaction mechanism between the R protein and effectors, the Guard model was proposed. According to the Guard model, R proteins do not directly detect the presence of the pathogen effector but rather monitor or guard the effector’s target protein in the host (Dangl and Jones 2001). Any modifications of the target by the effector lead to R gene activation. This model was initially proposed to explain Pseudomonas syringae AVRPto recognition by the tomato proteins Pto and Prf (Van der Biezen and Jones 1998), and subsequently, it was found in another effector protein (Jones and Dangl 2006). According to the Guard model, the guarded effector target is essential for the virulence function of the effector protein in the absence of the cognate R protein. However, further exploration of additional targets of AVRPto (AVR proteins of Pseudomonas syringae pv tomato) and AVRBs3 (family bacterial Avr proteins) demonstrated that some targets of effectors in the host act as decoys to identify pathogen effectors via R proteins, and a decoy model was proposed (Van der Hoorn and Kamoun 2008). According to the decoy model, the guarded host protein has no defence function but imitates as a functional effector target and traps the pathogen effector and redirects it from its true targets (Grund et al. 2019). In this model, guardees will be subject to two opposing selection pressures dependent on the presence or absence of the guarding R protein. In the presence of the R protein, the guardee would be optimised for AVR interactions and hence enhance the recognition of effectors, while in the absence of the R protein, the guardee would be under pressure to evade interaction with pathogen effectors to reduce the virulence of the pathogen (Van der Hoorn and Kamoun 2008; Champouret 2010).

Interaction of Phytophthora infestans with the host signalling pathway

Interaction of P. infestans with the host activate mitogen-activated protein kinase (MAPK) cascades (Pitzschke et al. 2009). The activation of MAPK cascades plays a crucial role in activating multiple signal transduction pathways in the host plant (Murphy et al. 2018). MAPKs are composed of 3 protein kinases: MAP kinase (MAP3K), MAP2K, and MAPK. These kinases are phosphorylated in a cascading series, where MAPK is phosphorylated by MAP2K, which itself is phosphorylated by MAP3K (Ren et al. 2019). MAPKs phosphorylate several downstream molecules, including transcription factors and RESPIRATORY BURST OXIDASE HOMOLOG D, which generates ROS to induce defence responses (Asai et al. 2008). The most studied PTI activated signalling pathway is the flg22 (a fragment of bacterial flagellin)-induced signalling pathway, which recruits MAP kinase cascades (Chinchilla et al. 2007; Jelenska et al. 2017). To suppress PTI, P. infestans releases several effector proteins. Eight RXLR effectors, SFI1-SFI8 (suppressor of early Flg22-induced immune response, SFI), from P. infestans showed the potential to suppress PTI in protoplast-based assays in tomato (Zheng et al. 2014). Among these, SFI5, SFI6 and SFI7 have been shown to suppress flg22-dependent MAP kinase activation upstream of MAPK and/or at the time of MAP3K activation. This demonstrated that these effectors target the earliest stages of the MAPK signal transduction pathway in tomato. However, P. infestans does not possess any flagellin; therefore, the function of these RXLX effectors in attenuating flg22-mediated MAP kinase activation and induction was presumed to be related with the similarity in early targets associated with both bacterial and oomycete PAMP recognition (Zhang et al. 2014).

MAPKs either act as susceptibility factors or positive regulators during interactions with pathogen effectors. Potato vascular Highway1-Interacting Kinase (StVIK), which encodes MEK kinase (MAP3K), interacts with the RXLX effector Pi17316 to suppress plant immunity (Murphy et al. 2018). Silencing of this MEK kinase in N. benthamiana attenuates P. infestans colonization, while transient expression of Pi17316 in N. benthamiana attenuates cell death triggered by Infestin1. These findings show that MAP3K (StVIK) acts as a susceptibility factor to promote disease establishment. On the other hand, the MAP3K kinase NbMAP3Kβ2 interacts with the RXLR effector Pi22926 to incite plant immunity (Ren et al. 2019). Silencing of NbMAP3Kβ2 in N. benthamiana enhanced P. infestans colonization and attenuated AVR4 (Cladosporium fulvum avirulence protein AVR4)/Cf4 (tomato resistance protein)-induced cell death, while transient and stable transgenic expression of the RXLR effector Pi22926 in N. benthamiana promoted leaf colonization by P. infestans (Ren et al. 2019). This finding showed that NbMAP3Kβ2 acts as a positive regulator. Similarly, another MAP3K, MAPKKKε, interacts with the RXLR effector PexRD2 and acts as a positive regulator of plant immunity. Silencing of MAPKKKε in N. benthamiana or expression of PexRD2 enhances susceptibility to P. infestans (King et al. 2014).

Metabolic alterations mediated in host plants by Phytophthora infestans

Plant–pathogen interactions involve the release of pathogen-effector proteins to manipulate plant cell processes and scavenge nutrients from the host cell by the pathogen to support its growth and establishment (Rodenburg et al. 2019). Although additional data have been obtained on nutrition requisition by the pathogen, much more remains to be determined. Similar to other oomycete fungi, P. infestans is considered an osmotroph that extracellularly catabolises complex host nutrients, such as proteins, sugars, and fatty acids, by secreting several depolymerizing enzymes and importing the resulting simple sugars, micronutrients and amino acids into the pathogen cell (Richards et al. 2013). In addition, P. infestans is a hemibiotroph that requires viable host cells during the initial biotrophic stages of infection (Zuluaga et al. 2016). Hence, it produces minimal symptoms in the biotrophic phase, which is followed by a necrotrophic phase in which the lesion becomes larger and sporangia start emerging from necrotic regions (Rodenburg et al. 2019). Therefore, the physiology of the host plant changes as the infection process progresses and nutrients become available for the pathogen (Ah-Fong et al. 2017). Based on the availability of the nutrient content in the host plant, P. infestans adjusts its metabolism by regulating the expression of enzyme-encoding genes related to glycolytic, gluconeogenic and amino acid pathways (Judelson et al. 2009). Transcriptome mining also showed dynamics in the expression of genes encoding nutrient transporters in P. infestans when grown in leaves or tubers or artificial media, mainly in the biotrophic stage, which indicates a rich influx of nutrients during the early stage of infection (Abrahamian et al. 2016). In addition to these transcriptomic studies, Rodenburg et al. (2018) proposed genome-scale metabolic models (GEMs) for P. infestans. By extracting information from annotated genomes (KEGG Orthology database) and introducing simulations, these authors constructed GEMs that were capable of predicting biochemical data for metabolic and genetic context and can also be used as a reference for designing future experiments to characterise the metabolism of P. infestans. To further understand the nutrient flux from host to pathogen during infection and the metabolic interaction of the host and pathogen, Rodenburg et al. (2019) reconstructed an integrated metabolic model of P. infestans and tomato (Yuan et al. 2016) by integrating two previously published models of both species. This integrated metabolic model elucidated the dynamics of pathogen–host metabolism throughout the infection process and presented information for controlling late blight infection in tomato by targeting important metabolic processes, such as thiamine biosynthesis, lipid metabolism and fatty acid biosynthesis.

Natural quantitative and qualitative resistance against Phytophthora infestans in tomato

The resistance of the host plant to P. infestans is either qualitative or quantitative. Qualitative resistance is controlled by the resistance (R) gene, while quantitative resistance is controlled by quantitative trait loci (QTLs). Approximately 68 resistance genes were identified in Solanum spp. against P. infestans (Rodewald and Trognitz 2013). Most of the resistance genes were from wild relatives of potato, such as Solanum bulbocastanum, S. venturii, S. demissum, S. verrucosum, S. microdontum and S. paucissectum (Vleeshouwers et al. 2011; Rodewald and Trognitz 2013). Additionally, several quantitative trait loci (QTLs) have been identified from both wild relative and cultivated potato species (Ghislain et al. 2001; Tan et al. 2008). Compared to potato, fewer investigations have focused on tomato because this pathogen was reported to cause more damage in potato than tomato. However, with time, P. infestans has undergone several genomic evolutions and thus has become one of the highly infectious pathogens of tomato (Foolad et al. 2008; Zhang et al. 2014). Natural genomic resistance to P. infestans has been reported in wild relatives of tomato, such as Solanum pimpinellifolium, S. habrochaites and S. pennellii. The resistance reported in S. pimpinellifolium is mainly qualitative, while the resistance reported in S. habrochaites and S. pennellii is quantitative (Merk et al. 2012). Approximately 6 resistance genes (Ph-1, Ph-2, Ph-3, Ph-4, Ph-5–1 and Ph-5–2) have been identified in tomato and its wild relatives (Panthee et al. 2017) (Table 1). The Ph-1 gene identified in S. pimpinellifolium, a wild relative of tomato, was reported to display resistance against P. infestans isolate T-0, and introduction of Ph-1 into the susceptible cultivar of tomato led to improved resistance (Bonde and Murphy 1952; Gallegly and Marvel 1955). However, the resistance was gradually overcome by predominant race T1 of the pathogen (Peirce 1971; Panthee et al. 2017). The second resistance gene, Ph-2, was identified in another accession of S. pimpinellifolium (Gallegly and Marvel 1955). Ph-2 was only able to provide partial resistance by restricting the spread of infection rather than completely eradicating the infection (Goodwin et al. 1995; Black et al. 1996). Similar to Ph-1, the resistance provided by Ph-2 was also gradually overcome by novel isolates of P. infestans (Zhang et al. 2014), which prompted further screening for new resistance genes in tomato germplasms. Subsequently, another single partially dominant gene, Ph-3, was isolated from S. pimpinellifolium L3708 (Chunwongse et al. 2002) and a hybrid of S. lycopersicum CLN2037B (containing Ph-3) × S. lycopersicum LA4084 (Zhang et al. 2013), which was found to be strongly resistant to several isolates of P. infestans and able provide resistance when Ph-1 and Ph-2-associated resistance failed. Although Ph-3 was initially considered the most effective genetic resistance resource against P. infestans (Chunwongse et al. 2002), a few novel and more aggressive P. infestans isolates have also overcome the resistance provided by Ph-3 (Foolad et al. 2008). The combination of Ph-3 with Ph-2 was found to successfully control aggressive strains (Merck et al. 2012). The combination of these two genes has been tested in recent tomato breeding lines (e.g., NC1 CELBR and NC2 CELBR) (Gardner and Panthee 2010), and hybrid cultivars of S. lycopersicum, Mountain Magic and Mountain Merit (Panthee and Gardner 2010), showed strong resistance against several isolates of P. infestans. The resistance gene Ph-4 (Kole et al. 2006) was identified in S. habrochaites LA1033, and Ph-5–1 and Ph-5–2 were identified in S. pimpinellifolium PSLP153 (Merk et al. 2012). However, a later investigation showed that Ph-4 is a QTL and a functional characterisation of Ph-5 for its biological role towards resistance against P. infestans has yet to be determined (Panthee et al. 2017).

Table 1 Resistance genes identified in tomato and its wild relatives
Full size table

The qualitative resistance controlled by R genes is not stable because of the rapid evolution tendency of P. infestans (Li et al. 2011), whereas qualitative resistance controlled by quantitative trait loci is generally not specific to any race (Brun et al. 2010). Therefore, QTL identification, mapping, validation and functional characterization can aid in accelerating the map-based and/or positional cloning of resistance genes against P. infestans. Table 2 summarises the list of late blight resistance quantitative trait loci identified in tomato and its wild relatives. In tomato, wild S. habrochaites was assumed to be a potential source for high levels of quantitative non-race-specific resistance against several P. infestans isolates (Lobo and Navarro 1987). Resistance QTLs against P. infestans were identified on all twelve tomato chromosomes of the two reciprocal backcross populations derived from S. lycopersicum × wild S. habrochaites (BC-E, backcross to S. lycopersicum; BC-H, backcross to S. habrochaites) (Brouwer et al. 2004). Among them, six QTLs were found to be reliable QTLs (Table 2), as consistently detected in the replicate experiment. In another S. habrochaites accession, LA1777, 5 QTLs were identified, four of which were reported to be colocalised with QTLs identified by Brouwer et al. (2004), and one of them was a novel QTL. Smart et al. (2007) also reported a late blight-resistant QTL from Lycopersicon pennellii LA716 (now known as S. pennellii). Three of the QTLs reported by Brouwer et al (2004) on chromosomes 4, 5, and 11 were fine-mapped and validated using near-isogenic lines (NILs) and sub-NILs (Brouwer and St. Clair 2004), which showed that QTLs located on chromosome 5 exhibit foliar resistance while QTLs located on chromosomes 4 and 11 exhibit both foliar and stem resistance. The introgressed regions containing the resistance QTLs were also found to be linked with QTLs affecting agricultural traits, such as plant height, plant shape, maturity, yield, and fruit, which shows the possibility of linkage drag between the S. habrochaites resistance alleles and the phenotypical trait alleles (Brouwer and St. Clair 2004). Similar co-linkage between disease resistance QTLs were reported by Johnson et al. (2012) in two selected QTLs from S. habrochaites chromosomes 5 and 11 related to foliar and stem resistance against P. infestans. These QTLs present complex genomic organisations, including several loci that depicted pleiotropy and/or strong linkages. Subsequently, Haggard et al. (2015) re-evaluated the P. infestans resistance lb11 QTL identified by Brouwer and St. Clair (2004) as a probable source of quantitative resistance against P. infestans infection in tomato using the same sub-NILs in replicated field trials over 2 years for 17 agricultural traits, such as fruit size, shape, quality, yield, maturity, and plant architecture. The lb11 QTL was first reported via fine-mapping, where each QTL fractionated into multiple QTLs using higher resolution mapping. Approximately 34 QTLs were identified via fine-mapping among these traits, with 14% revealing a significant interaction between QTLs and the environment. Additionally, QTLs for several traits were found to be colocalised, indicating either pleiotropic effects or a strong linkage between genes regulating those traits. The association of disease resistance QTLs with phenotypical QTLs possesses both opportunities and constraints. The opportunity involves a favourable blend of P. infestans resistance and beneficial agricultural traits, such as that shown in the sub-NIL 08GH3999 tomato line, which can directly be used as a donor parent for marker-assisted breeding (Collard et al. 2005; Foolad and Panthee 2012) or as a parent in crossbreeding for pyramiding resistance QTL alleles with other QTL donor lines to escalate the quantitative resistance level against P. infestans infection (Collard and Mackill 2008; Brouwer and St. Clair 2010). Constraints involve the combination of P. infestans resistance and undesirable agricultural traits, such as that displayed in the sub-NIL 08GH8032 tomato lines, which show higher foliar resistance but shorter and wider fruit shapes and smaller ripe fruit perimeters with delayed maturity. Therefore, the use of such QTLs in tomato breeding requires separation from negative effects through recombination, which requires testing of thousands of progeny to recover favourable recombinants (Haggard et al. 2015).

Table 2 Summary of late blight resistance quantitative trait loci detected in tomato and its wild relatives
Full size table

Generation of resistance against Phytophthora infestans in tomato via a transgenic approach

The generation of resistant varieties using plant transformation is a faster approach than traditional breeding. The availability of several efficient regeneration systems amenable to plant transformation systems (Abu-El-Heba et al. 2008; Hoshikawa et al. 2019) and the recently developed CRISPR-mediated genome editing system (Danilo et al. 2019; Ghorbani et al. 2020) for tomato varieties has opened up wider opportunities for testing the function of several stress-related genes as well as the generation of resistant varieties. Overexpression of several antifungal genes from resistant wild relatives, transcription factors, defence-related genes, and R genes and silencing of negative regulators, such as microRNA and circular RNA, have shown promising results for the development of resistance against P. infestans in tomato (Table 3).

Table 3 Transgenic tomato lines developed against Phytophthora infestans
Full size table

Expression of antifungal genes

The expression of antifungal genes is the most common approach for addressing oomycete diseases in plants. The expression of antifungal Kiwi pathogenesis-related group 5 (PR-5) proteins in tomato showed improved resistance against P. infestans (Korneeva et al. 2010). Pathogenesis-related (PR) proteins are induced and accumulate in host plants in the event of pathogen attack (Jain and Khurana 2018). Chitins are β-1,4-like polymers of N-acetylglucosamine polysaccharides and primary components of cell walls in fungus (Elsoud and Kady 2019). Although a putative chitin synthase (CHS) gene is present in the genome, the presence of chitin has not been detected in the cell wall of P. infestans (Guerriero et al. 2010; Hinkel et al. 2017). A study on ectopic expression of chitin-binding genes from Amaranthus caudatus (ac), A. retraflexus (RS-intron-Shir) and hevein-like antimicrobial peptides (amp1 and amp2) from Stellaria media in tomato plants showed enhanced resistance against P. infestans (Khaliluev et al. 2011). However, the mechanism by which chitin-binding genes enhances resistance in tomato were not well explained.

Expression of defence-related genes

The expression of defence-related genes is another approach to elevating disease resistance. Overexpression of the potato R1 resistance gene (Faino et al. 2010), S. bulbocastanum R gene (Rpi-bIb1) (Van Der Vossen et al. 2003) and S. pimpinellifolium R1 gene (Jiang et al. 2018a) in tomato showed significantly improved resistance. However, tomato plants that overexpressed the potato R1 resistance gene (S. lycopersicum cv Moneymaker and cv San Marzano) showed improved resistance only against IPO-0, an isolate of P. infestans, and displayed susceptibility to another isolate, 88,133 (Faino et al. 2010). This study provided evidence indicating that the use of a single R gene is not a sustainable approach to achieving resistance against fast-evolving pathogens, such as P. infestans. Overexpression of two phytoalexin genes of grapevine (Vitis vinifera) stilbene synthase (vst1 and vst2) in tomato showed significant improvements in resistance against P. infestans (Table 3). Stilbene synthase catalyses the biosynthesis of stilbene phytoalexin trans-resveratrol, which is known for its active role in plant defence mechanisms (Parage et al. 2012).

Manipulation (overexpression/silencing) of transcription factors

Overexpression of transcription factor (TF) genes, such as WRKY and MYB, also shows significant enhancement in resistance against P. infestans. Transcription factors are important components in the signalling system and have been involved in mainly positive but sometimes negative regulatory processes underlying stress responses in plants by inducing and/or repressing the expression of an array of downstream defence-related genes (Ma et al. 2013; Tian et al. 2015). The WRKY family is an example of a positive stress regulator TF that is mainly associated with diverse biotic stresses (Pandey and Somssich 2009). Overexpression of S. pimpinellifolium WRKY1 (SpWRKY1) in the susceptible cultivated tomato cultivar Zaofen No. 2 showed a significant reduction in cell death, ROS accumulation, lipid peroxidation and relative electrolyte leakage and an increase in ROS scavenging antioxidant enzyme activity (peroxidase, superoxide dismutase, phenylalanine ammonia-lyase) compared to the wild type (Li et al. 2015). The overexpression of SpWRKY1 also upregulated other downstream defence-related genes, such as ROS scavenging-related genes and salicylic acid/jasmonic acid-responsive genes, in transgenic lines (Cui et al. 2019). Similarly, overexpression of SpWRKY3 (Cui et al. 2018b) and SpWRKY6 (Hong et al. 2018) also showed promising enhancement in resistance in tomato plants against P. infestans (Table 3). SpWRKY3 and SpWRKY6 overexpression showed elevated resistance, which is evidence of lower necrosis, small lesion sizes, less ROS generation and a low disease index, while the silenced lines showed more severe disease symptoms than the wild-type lines. Furthermore, the overexpression lines after inoculation with P. infestans infection showed upregulated pathogenesis-related (PR) gene expression in transgenic tomato plants compared to wild-type plants.

In addition to WRKYs, MYB TFs are also known for their crucial role in increasing the susceptibility of plants to biotic stresses (Erpen et al. 2017). Overexpression of native MYB49 in the susceptible tomato cultivar S. lycopersicum Zaofen no. 2 significantly enhanced resistance against P. infestans, as evidenced by lower necrosis, small lesion size and lower disease index (Cui et al. 2018a). Compared to the wild-type control plants, the transgenic lines showed higher accumulation of ROS and malonaldehyde content and lowered relative electrolyte leakage and higher activity of ROS scavenging enzymes (peroxidase and superoxide dismutase) upon P. infestans inoculation. Although some of the TFs are reported to have a significant response against P. infestans, a vast majority of the TF family remains unexplored. For example, the role of biotic stress-related TFs, including ERF and bZIP, has been reported in response to P. infestans in potato. ERF was reported to act as a negative stress regulator in potato. Transgenic potato lines with silenced ERF (StERF3) showed significant enhancement in resistance against P. infestans (Tian et al. 2015). Moreover, bZIPs were reported to act as positive stress regulators in potato, and potato overexpressing bZIP (Stb ZIP 61) (Zhou et al. 2018) showed significant enhancement in resistance against P. infestans. Hence, the TFs ERF and bZIP can also be explored in tomato in response to P. infestans.

Silencing of susceptibility genes

Silencing of the negative regulator is another approach to improving resistance. Plant genes that facilitate infection and aid in the compatibility of the host are known as susceptibility (S) genes (van Schie and Takken 2014). Silencing of such an S gene Defense No Death 1 (DND1) in tomato plants (S. lycopersicum cultivar Moneymaker) showed remarkably increased resistance against P. infestans (Sun et al. 2016).

Manipulation of non-coding genes

In addition to functional genes, manipulation of non-coding genes, such as microRNAs (miRNAs), long non-coding RNAs and circular RNAs, was also reported to be useful for improving resistance in tomato plants against P. infestans. MiRNAs are a class of endogenous small non-coding RNAs of approximately 22 -24 nucleotides that post-transcriptionally regulate the expression of genes via translational repression or mRNA degradation (Cai et al. 2009). miRNAs can regulate both positive and negative resistance genes. Negatively regulated miRNAs are downregulated during biotic stress in resistant plants, while positively regulated miRNAs are upregulated (Chauhan et al. 2017). Therefore, miRNAs serve as a potential alternative resource to be used for the control of plant pathogens.

Several tomato miRNAs, such as miR396a-5p and -3p, miR1916, miR482a, miR482b, miR482c and miR4773p, are actively involved in interactions between tomato and P. infestans and negatively affect tomato resistance by targeting mainly R gene (NBS-LRR) genes. Infection of P. infestans in tomato plants downregulates the expression of miR396a-5p and miR396a-3p (Chen et al. 2017) and MiR1916 (Chen et al. 2018). MiR396a-5p targets growth-regulating factor (GRF) genes (Chen et al. 2017), while miR1916 targets strictosidine synthases, uridine diphosphate‐glycosyltransferases, disease resistance protein RPP13‐like, late blight resistance protein homologue R1B‐16 and MYB 12 (Chen et al. 2018). Overexpression of miR396a-5p and miR396a-3p in tomato plants increases the susceptibility to P. infestans, as evidenced by lower necrosis and higher ROS compared to the wild type (Chen et al. 2017). Similarly, overexpression of miR1916 in transgenic tomato also increases susceptibility to late blight infection, while silencing of miR1916 improves tolerance (Chen et al. 2018). Overexpression of another negative regulator, miR482a (Jiang et al. 2020) and miR482b (Jiang et al. 2018b), in tomato plants showed impairment in resistance against P. infestans, while silencing resulted in improvement in tomato resistance. Similar impairment in resistance was observed for miR482c-overexpressing tomato lines, where the transgenic plants showed reduced activity of ROS scavenging enzymes (peroxidase, superoxide dismutase and phenylalanine ammonia-lyase) and higher lipid peroxidation than wild-type tomato plants. The overexpressed lines showed downregulation of target resistance genes (NBS-LRR): Solyc07 g049700 and Solyc11 g006530. Additionally, for another miRNA, miR1918, transgenic tomato lines overexpressing artificial pi-miR1918 exhibited a higher infection rate than wild-type plants upon inoculation with P. infestans, which indicates that miR1918 increases the susceptibility of tomato plants to P. infestans (Hong et al. 2019).

In contrast to negative regulator miRNAs, few miRNAs act as positive regulators in tomato stress resistance. Overexpression of miR172a and b in the susceptible tomato cultivar Zaofen No. 2 significantly improved resistance against P. infestans by suppressing the AP2/ERF transcription factor (Solyc11 g072600.1.1). The overexpressed line displayed a lower disease index, smaller lesion sizes, lower ROS, lower lipid peroxidation levels but higher activities of ROS scavenging enzymes (POD and SOD) and photosynthetic rates (Luan et al. 2018).

In addition to miRNAs, long non-coding RNAs also regulate plant resistance. Long non-coding RNAs are a group of non-coding RNA molecules that are longer than 200 nucleotides (Kung et al. 2013). MiRNAs cleave long non-coding RNAs and generate phased small interfering RNAs, which compete with endogenous RNAs and act as decoys for mature miRNAs, leading to re-expression of miRNA target genes (Ratti et al. 2020). In tomato, a long non-coding RNA, Sl‐lncRNA15492 (511 bp), was reported to repress the negative stress regulator miR482a, whose precursor was placed within the Sl‐lncRNA15492 antisense sequence (Jiang et al. 2020). Degradome analysis followed by RLM‐5′ RACE experiments showed that mature Sl‐miR482a might also cleave Sl‐lncRNA15492. This shows that Sl‐lncRNAS15492 and Sl‐miR482a reciprocally destroy Sl‐NBS‐LRR1 homeostasis during the process of tomato plant resistance against P. infestans.

Circular RNAs (circRNAs) are long non-coding RNAs whose 5’ and 3’ termini are covalently linked, and because of their capacity to bind microRNAs (miRNAs), they also serve as miRNA sponges (Yu and Kuo 2019). In silico analysis detected a total of 68 circRNAs, of which 9 were reported to be upregulated during the P. infestans infection process in tomato plants (Hong et al. 2020). Among them, circRNA45 and circRNA47, transiently overexpressed in tomato plants, displayed smaller lesion areas in both transgenic lines compared to wild-type plants upon P. infestans infection. CircRNA45 and circRNA47 act as sponges for the negative regulator miR477-3p, which might be the main reason behind the improvement in resistance, as evident by the low expression of miR477-3p.

Conclusion and future prospects

P. infestans has become a devastating pathogen to control even 180 years after its identification because of its tremendous ability to overcome host resistance. By mining the literature on P. infestans and tomato interactions, we observed the involvement of multiple cell wall-degrading enzymes, effectors of pathogens and host resistance genes, kinases, transcription factors and non-coding RNAs in the molecular interactions underlying pathogenesis. The molecular interactions between P. infestans virulence and host resistance drive the coevolution of the pathogen as well as the host genome. R genes exert selection pressure on pathogens to improve their virulence through the modification of the pathogen AVR gene inventory to overcome host plant defence. To date, approximately 6 resistance genes/genomic regions have been identified in tomato and its wild relatives (Table 1), most of which have already been overcome by aggressive isolates of P. infestans. At present, the combination of Ph-3 with Ph-2 is the most successful. Hence, detailed research is needed to screen new resistance genes in tomato germplasms. Several QTLs were identified, although few of them have been validated (Table 2). Several QTLs were found to be linked with other phenotypic traits. Validated QTLs with favourable phenotypic traits will be promising candidates for use in P. infestans resistance breeding programs in tomato. Much progress has been made in elucidating tomato metabolome manipulation by P. infestans. Based on that, an integrated host–pathogen genome-scale metabolic model was proposed that can be used for further exploration of plant–oomycete metabolic interactions and identification of novel defence or oomycete genes. Overexpression of several antifungal genes, defence-related genes, and long non-coding RNAs and silencing of negative regulators, such as susceptibility genes or miRNAs, showed improvement in resistance in transgenic tomato lines (Table 3). However, for a fast-evolving pathogen such as P. infestans, a more strategic approach is needed, such as stacking multiple genes with a suitable combination along with or silencing of negative regulators. With advancements in sequencing technology, genomes of both the pathogen and host plant tomato are now available in public domains and efficient transformation and genome editing systems are reported. These resources provide an opportunity to explore further the functions and networks of effectors, regulatory genes, defence genes and non-coding RNAs associated with pathogenesis, along with evolutionary insight to design targeted strategies to eliminate late blight in tomato.

Author contribution statement

PM has designed the outline of the article, composed the manuscript and figure. NR and PS provided feedback and comments to revise the content. DK has assisted with artwork and ID has provided field image. All the authors read and approved the manuscript.