Two types of plant defenses were defined; the pathogen-associated molecular patterns PAMP-triggered or innate immunity (PTI), and the pathogen effector-triggered immunity (ETI) [18]. PTI defenses are associated with the recognition of the pathogen followed by a series of downstream events. These include the alteration of ion fluxes across the host’s plasma membrane, the release of reactive oxygen species and nitric oxide as signaling molecules, the activation of MAP kinases, SA and JA/ET, and cross-talk between these pathways or with others such as ABA. ETI reactions, on the other hand, are elicited upon perception of effectors [19,20] released by the pathogen. Resistance gene-mediated defense follows, but often depends on the type of genes involved and the direct/indirect recognition of effectors [18]. Defense responses usually take place at sub-cellular levels, with cell structure rearrangements and transcriptional regulations that lead to the accumulation of defense-related molecules. The latter fortify the host cell wall [21–23], affect the integrity of the invading hyphae [24–26], and restrict progress of the pathogen [27,28]. Cell communication establishes among the tissues surrounding the initial infection sites and remote sites, hence leading to systemic acquired resistance (SAR) or induced systemic resistance (ISR) [29].

In host-pathogen interactions, proteomics have been truly complementary to functional genomics, and will continue to be so even more as entire genomes of plants and microorganisms are increasingly accessible. Vast Genbank databases with billions of sequences are daily being browsed, annotated or corrected/replaced by researchers from around the globe. It is interesting to note that proteomic investigation of plant defenses confirmed many of the results gathered using classical forward and reverse genetics, and biochemistry or molecular biology. It also provided a new generation of dataset, descriptive yet to be further analyzed and mastered for an overall view of the processes triggered in plants while fighting pathogens. It is possible through these approaches to now focus on an entire pathway involved in resistance and determine its interaction with other signaling and defense networks.

A number of pathosystems where proteomic approaches have been successfully used revealed either an increase or a decrease in protein abundance in response to infection by pathogens or to their pathogenicity factors. These proteins can be associated with photosynthesis and energy metabolism, anti-fungal activity, phytohormone homeostasis, ROS generation, detoxification and signaling, phenylpropanoid and lignin biosynthesis, and more (see below).

Proteomic studies of plant-pathogen interactions in planta remain challenging. Data retrieval is important in determining proteins that accumulate differentially and are important for defense, pathogenicity, or counter-defense. Such studies often rely on (i) empirical comparison with databases and among pathosystems, (ii) complementary studies using the plant or pathogen individually as controls, or (iii) differential analysis using cultivars with various levels of resistance against highly aggressive isolates. Most studies, if not at a preliminary stage [4], clustered proteomes by nature, function, or sub-cellular localization of peptides [15,30], and reported on the technical challenges encountered in: (i) characterizing proteins in the absence of fully sequenced genomes of the host plants or their pathogens [31], and (ii) analyzing functions and regulation of the proteins encoded by the sequenced genomes.

To understand resistance of Brassica carinata to Leptosphaeria maculans, the causal agent of phoma leaf spot and stem canker, a proteomic investigation compared this species to a susceptible one, Brassica napus [16]. Forty-eight hours post-inoculation, 28 proteins accumulated differentially in the resistant phenotype, including superoxide dismutase, nitrate reductase, and carbonic anhydrase, as well as many photosynthesis-related enzymes. In cotton (Gossypium hirsutum L.), Coumans et al. [32] reported on proteins involved in plant defense against the black root rot fungus, Thielaviopsis basicola, especially PR-proteins and those implicated in ROS production and scavenging or in the isoprenoid biosynthesis pathway. In response to Verticillium dahliae, Wang et al. [33] also identified 51 upregulated and 17 downregulated proteins mainly involved in defense and stress responses, primary and secondary metabolism, lipid transport, and cytoskeleton organization. As such, analyzing proteomes during host-pathogen interactions will always generate differential expression of genes/proteins involved in disease or resistance. However, comparing proteomic research advances in pathosystems is complex, given the diversity of approaches used in proteome analysis and the genetic and biochemical backgrounds of the hosts or pathogens to be compared. Nonetheless, two categories can be distinguished, and involve either proteome data acquired only on the pathogen, or those gathered during host-pathogen interaction in planta [34]. One of the challenges relates to the distinction between proteins originating from the host and the pathogen. However, judicious experimental setups and recent technical and bioinformatics advances have been providing ways to better interpret such complex data. Data specifically related to the lifestyle of the pathogen, as biotroph, necrotroph or hemibiotroph are also valuable for proper result analysis and interpretation, but may bring their own complexities.

Plant pathogens specialize in attacking their respective hosts and feeding on their tissues, thus causing disease and, potentially, epidemics. They possess a large number of features that allow them to infect and establish in/on plant tissues [1,65,66]. Upon infection, necrotrophs release an array of enzymes that allow them to dispose of the organic matter of dead tissues [67,68]. Biotrophs, on the other hand, require their host cells to remain alive and have evolved other pathogenicity determinants, such as haustoria, following host cell penetration [1]. Similarly, pathogens producing toxins may secrete these molecules at the forefront of the infection site to lay the ground for mycelial invasion [65,66,69–71]. Recently, advances in understanding the mechanisms of biotrophic, necrotrophic and hemibiotrophic lifestyles were reviewed [72], with special emphasis on the biotrophy-necrotrophy switch of hemibiotrophic pathogens. During the last few years, a new field of research focusing on pathogen effectors has emerged [19,20]. Pathogen effectors enable plant defense suppression and can reprogram the metabolic machinery of infected tissue to benefit the pathogen in its growth and development [73].

Effectors were deemed to be widespread among invading pathogens and can be either invasive or defensive [74,75]. Invasive effectors allow a given pathogen to attack, i.e., polygalacturonases [4,68], whereas, the defensive ones protect it from degrading enzymes (i.e., chitinases; [20]) and toxic metabolites synthesized by the host (i.e., phytoalexins and defense-related metabolites; [2,21,22]). Suppressors of plant defenses are also among the pathogenicity determinants/effectors in fungi and oomycetes as they were documented in bacteria and viruses [1]. Biotrophs needing to establish a trophic relationship with their live host cells have evolved these type of mechanisms to temporarily turn off the plant guards and defenses and avoid detection [1]. For hemibiotrophs and nectrotrophs, more sophisticated strategies to suppress defense mechanisms have recently been reported. These include the manipulation of signaling pathways [2,63], the hijacking of essential defense molecules [76], the detoxification of major defense-related metabolites [21,22] and others yet to be determined. In a recent study involving V. dahliae, a fungus with multiple lifestyles, varying from necrotroph, to hemibiotroph, to endophyte, on a wide host range [77], El-Bebany et al. [4] established a proteomic map (pH 4–7) using in vitro-cultures of weakly- versus highly-aggressive isolates. The comparison revealed 25 differentially-regulated proteins, most of which were important determinants for host penetration and colonization, development of microsclerotia, and resistance to stress and antibiotics in the highly-aggressive isolate tested. The authors also reported on an increase in protein abundance of the isochorismatase hydrolase, thought to be implicated as a suppressor of plant defense responses and signaling [4]. Proteome maps are also being developed for other important fungi, such as Cladosporium fulvum and Mycosphaerella fijiensis, the causal agents of leaf mould in tomato and black Sigatoka in bananas and plantains, respectively. Four isolates of Verticillium albo-atrum from two geographical zones were used to compare the proteomes of two pathotypes on hop and establish the first mycelial proteome map of this fungus [10]. Among the 53 proteins identified, 17 were ascribed functions, and about 30 differentially accumulated in the mild versus the lethal pathotype. Among the proteins differentially abundant in the lethal pathotype were peroxiredoxin and ascorbate peroxidase, cytoskeleton components and regulators, and proteins involved in protein synthesis and energy metabolism.

In L. maculans, which causes economically important diseases of crucifers, including canola blackleg and stem canker [23,78], the mycelial proteome has recently been reported [79], showing several CWDEs previously identified in this and other necrotrophs. Proteomic analysis of mycelia provides important insights into the ability of pathogens to evolve new tactics to deal with plant defenses. However, there is still much to achieve regarding the spatio-temporal accumulation of peptides/proteins produced by pathogens throughout their disease cycle, and how such choices allow them to strategically take advantage of specific plant weaknesses, or surmount most sophisticated plant defenses. It is nonetheless obvious that integration of proteomic, transcriptomic, and metabolomic data over time is a major step towards understanding such intricate interactions. In their study on exoproteome of the notorious nectrotrophic fungus Sclerotinia sclerotiorum, Yajima and Kav [67] reported on 52 secreted proteins that included many CWDEs, known as pathogenicity factors in this pathogen, and highlighted the complementarity of proteomics to genomic approaches by identifying an α-l-arabinofuranosidase that was omitted in functional genomic studies involving ESTs.

Cao et al. [11] examined the proteome profile of virulent and avirulent isolates of the biotrophic Pyrenophora tritici-repentis, the causal agent of tan spot in wheat, known for its ability to produce several Ptr host-specific toxins. Virulent isolates usually produce the toxic chlorosis-inducing protein Ptr ToxB due to the expression of the ToxB gene. Avirulent isolates harbor a homolog of the gene but do not produce a functional protein [69,80]. Aside from ToxB gene, the virulent isolate actually exhibited a distinctive profile where several translated and accumulated proteins were up-regulated and ascribed to virulence, including an α-mannosidase and an exo-β-1,3-glucanase. In addition, saprophytic growth and development may account for differences in the proteome profile between virulent and avirulent isolates.

In other filamentous fungi, a proteome map has been created from mycelia collected during the dimorphic transition from budding to filamentous growth in Ustilago maydis [81]. In Blumeria graminis f.sp. hordei, proteic fractions collected from conidiospores were used to generate such a map. Among the identified proteins were enzymes involved in carbohydrate, lipid, and protein metabolism [82]. In Cladosporium fulvum interacting with tomato, three extracellularly secreted proteins, namely CfPhiA, Ecp6 and Ecp7, have been recently characterized [83]. Therefore, proteome characterization in this class remains sporadic and many studies are needed to fill the gaps for the different development stages of oomycetes and during their interactions with their hosts.

In oomycetes, especially Phytophthora spp., proteomics are informative in studying the proteic pool during various stages of growth and development, including hyphae, cysts, sporangia, zoospores, and appressoria [84–86].

Some work has been done on the use of exogenous supplementation of chemically-synthesized molecules. Shimizu et al. [87] used vanillin on Phanerochaete chrysosporium cultures and reported on the identification of up-regulated proteins involved in vanillin metabolism. A similar procedure was applied on the white rot fungus, Phanerochaete chrysosporium, using benzoic acid [88], allowing the characterization of a pool of mycelia proteins differentially responsive to the treatment. In general, not many fungal/oomycete pathogens have been extensively studied in this respect. A recent review reports on B. cinerea, S. sclerotiorum, and F. graminearum as the most studied plant pathogenic fungi in proteomics [89].

The studies available so far highlight the worth of proteomic approaches in determining pathogenicity factors and effectors of many pathogens, but also suggest that a great horizon is open to more studies to investigate the function and roles of individual microbial peptides and proteins in development stages and host-pathogen interactions.

Proteomic approaches have not yet provided a full coverage of the entire proteome of plant species such as A. thaliana or the model fungus C. fulvum. More than 35,000 translated proteins were initially predicted in A. thaliana and currently only around 1/10 of that number has been characterized [94], likely due to challenges in separating and identifying all the proteins expressed under specific conditions. The latter relate to insolubility, low abundance, size, or pH. However, proteomics findings often feed back to genomics by complementing the annotation of ORFs and correcting gene structure predictions. In addition, proteomics in model plant species allows for the determination of recurrent peptides to be used as internal controls for large-scale proteomic studies. It also allows for the examination of post-translational processes that proteins go through, such as cleavage of signal peptides to allow compartmentalization, N-glycosylation, or phosphorylation. Such knowledge accelerates our understanding of other pathosystems involving economically important crops and destructive pathogens, hence increasing the likelihood of setup control management strategies that reduce their impact and restore yield and quality.

Carpentier et al. [95] discussed the challenges and opportunities facing proteomic approaches in non-model species, such as bananas and plantains, where the proteome analysis has flourished in absence of sequencing data on a large portion of the genome and with enormous challenges in conducting large scale gene expression. El Hadrami et al. [66] had examined the isoenzymatic profile and activities of a set of cytosolic and chloroplastic proteins involved in the scavenging and detoxification of ROS in a susceptible versus a partially resistant banana cultivar interacting with the toxin juglone, one of the pathogenicity factors of M. fijiensis. The main difference between the two tested phenotypes was in the quickness in triggering their antioxidant systems on perceiving the toxin, and how long the system remains active and coordinated.

It is clear that in the years to come, a continuous move will occur from studying plant models to studying non-model species. In spite of the lack of extensive genomic data on the latter and/or how they differ from model species, they will still benefit from the available genomics and proteomics datasets. There is a need for appropriate extraction methods that minimize interferences with other analytes and maximize separation and quantification. As for identification and assigning functions, one would still rely on the background of model species unless other information can be gathered through complementary approaches.

Significant progress has been made in proteomic studies in many pathosystems using differential models or comparing inoculated with mock-inoculated controls [4,10,14,33]. Separation of the proteome is conducted by 1D or 2D-PAGE i.e., [4,10], which often lack sensitivity and reproducibility [10]. The advent of nanoflow-LC technology has made it possible to minimize some of these issues i.e., [4]. To enhance the separation and gain an in-depth understanding, new approaches, such as liquid-phase IEF coupled to high-resolution 2D, have also been used in fungi [79]. Immense challenges are still encountered in mass spectrometry detection, identification, and quantification. Current methods use nanoflow-LC coupled to mass spectrometry technologies with triple or quadruple quads such as MALDI, MALDI-Q-TOF, ESI-MS/MS, ICP-MS/MS, as well as programs specialized in data acquisition, spectral analysis and protein identification through matches with onsite or online databases (i.e., MASCOT, SEQUEST, X!TANDEM). Other quantification procedures using stable labeling of ¹³C, ²H, ¹⁵N, or ¹⁸O within techniques such as ICAT [96–99], SILAC [100], and iTRAQ™ [101] have also became popular and have greatly improved the quantitative determination of the detected proteins. iTRAQ MS/MS technology has been applied in profiling differentially translated and accumulated proteins from F. graminearum after an in vitro-induction of mycotoxins synthesis in comparison with a mock-treated control [102]. The presence of phosphorus and sulfur residues in proteins has also been used along with technologies such as ICP-MS to translate the molarity of these residues into absolute protein concentrations.

Resolving unambiguous identification of proteins among contrasting treatments remains an important challenge in proteomics. While one can conceivably manually compare various gels with a limited number of differentially up- or down-regulated protein spots, difficulties arise when proteome analysis is conducted using high-resolution 2D-PAGE, especially after the use of liquid-phase IEF or other pre-screening methods. Theoretically, on these gels, up to 15,000 protein spots could be detected but practically only a third of that number or less can accurately be resolved and analyzed. An extensive development of hardware and software to precisely analyze these images has been seen in recent years to maximize spot annotations, establish protein patterns, resolve coalescent spots, and provide an accurate comparison among tested treatments. To further minimize identification errors, spots can be excised from the PAGE and enzymatically digested and subjected to MS/MS and database matching. Matching against available databases is often hampered by an extensive presence of hypothetical proteins with no assigned function and/or with lack of full protein sequences. Combining transcriptomic and proteomic approaches often yield an accurate correlation between the variation in the gene expression and the changes in the levels of the detected proteins, allowing for an exact determination of identity and function [2]. These can also be complemented with an array of in silico analysis to predict compartment localization, structures, and function [2]. In addition, a better knowledge of the pathosystems considered for proteomic analysis through other molecular, biochemical, physiological, and epidemiological approaches allows and eases the unambiguous interpretation of the gathered data.

Pathosystems involving a host, a pathogen, and a biocontrol agent are more challenging to study in terms of discerning the role of each one of them in the interaction. Proteomics-based approaches have been suggested to study these complex systems [119] and unveil the three-way signaling cross-talks. Marra et al. [119] used an experimental set-up that allowed them to separate the individual proteomes from beans, B. cinerea or Trichoderma solani and the antagonistic biocontrol agent Trichoderma atrviridae. The approach was instrumental in showing an up-regulation of bean’s defense-related proteins such as PRs and that T. atroviride qualitatively and quantitatively alters the protein profile of the plant regardless of infection by a pathogen. The study also revealed proteins in T. atroviride with homology to fungal hydrophobins and ABC transporters, and the up-regulation in the pathogen proteome of virulence factors such as cyclophilins, when interacting with the host alone or in the concomitant presence of the biocontrol antagonist as well. Many other studies need to be investigated in such an integrated mode in order to gather more valuable information that is closer to the events that happen in nature.

Between pathogens and their hosts, the balance was for many decades leaning towards studying plant defenses. In recent years, with the advent of genomics-based technologies that made more sequenced pathogen genomes available, more studies burgeoned on characterization of pathogenicity factors and pathogen-based strategies that avoid host resistance. Such strategies included ways to avoid detection by the host, suppress or cope with host defenses, and strategically manipulate host pathways and signaling cascades, thereby leading to unnecessary and energy-costly defenses or ones more beneficial to the pathogen [1,2]. These seem to be evolutionary strategies for co-existence that allow the pathogens to make the best of their trophic interactions with the hosts. While many biotrophs seem to have adapted strategies to either remain undetectable by plant cognate guardees and PRRs or to cope with plant defenses by detoxifying phytoalexins and other defense-related metabolites [1,21,22], hemibiotrophs and necrotrophs tend to raise the bar in sophisticating their counter-defense strategies [1,2,63]. For instance, necrotrophic fungi such as S. sclerotiorum suppress the host’s ROS signaling cascade in planta via catalases, superoxide dismutases, or other scavenging molecules, such as mannitol or oxalic acid [120,121], hence leading to an enhanced hypersensitive response locally and/or preventing signal transduction to remote healthy sites. Botrytis cinerea forces its host to produce ROS, which kill the tissues, thereby enabling their colonization [122,123]. Fusarium oxysporum, on the other hand, has been reported to use the COI1-mediated JA-signaling to promote disease development in A. thaliana [76]. This manipulation of signaling pathways has also been shown in potato interacting with V. dahliae and tomato with B. cinerea [2,63]. In potato, plant defenses include an increase in the synthesis and accumulation of rutin. This defense-related flavonol-glycoside represents a carbon source for the pathogen, but cleaving the sugar moiety, using enzymes such as glucosidases and rhamnosidases, exposes V. dahliae to quercetin, the rutin’s aglycone. During its coevolution with its hosts, V. dahliae has evolved a mechanism that allows it to detoxify this flavonol using its quercetinases [2]. The byproduct of detoxification, the 2-PCPGCA, is composed of phloroglucinol, a potent antibiotic that theoretically expands the pathogen’s competitiveness in the rhizosphere, and protocatechuates that get converted into benzoates and salicylates, potential triggers of the SA-pathway, known to interfere with the JA-pathway. In the years to come, these reports on counter-defense strategies evolved by hemibiotrophs and necrotrophs will benefit to unveil new mechanisms and interplays using proteomics-based technologies.