Infections caused by necrotrophic fungi in host plants during susceptible interactions are related to unrestricted cell death visible as spreading lesions, the appearance of which is strictly correlated to a fungal infection cycle and the production of a wide set of virulence factors [28, 29]. The cabbage cultivar 'Glory of Ehinkhuzen' used in the presented study has been considered to be susceptible to A. brassicicola infection under laboratory conditions and moderately resistant in the field . Macroscopically visible necrotic lesions on leaves during A. brassicicola infection spread gradually from small brown spots at 24 hpi to larger necroses surrounded by a chlorotic area in a leaf position-dependent manner similarly as described for other Brassicas [3, 9, 30].
Timing of A. brassicicola infection cycle depends on host leaf surface
In our study, the A. brassicicola infection cycle was delayed on the second leaf of 'Glory of Ehinkhuzen' compared to the susceptible white, red and Savoy cabbage cultivars infected with the same strain. The fungal germination began about 2 h later than on the white cabbage cultivar 'Stone Head', however the appearance of the first appressoria at 8 hpi was in accordance with the previous study . The delayed germination of A. brassicicola conidia may be related to the thicker wax layer on the 'Glory of Ehinkhuzen' leaf surface that might have caused difficulties in conidial adherence to the leaf surface and/or a delayed recognition of the host surface signals by the conidia [31-33]. Following germination, the fungus penetrated host epidermal cells through appressoria, stomata or directly without any preferential mode of penetration, but on other susceptible cabbage cultivars, A. brassicicola penetrated the leaf surface mainly through appressoria and rarely through stomata . Nowakowska et al.  have claimed that the cabbage cultivars were penetrated mainly directly and through appressoria, and only rarely through stomata or without any preferential mode of penetration independently of a cultivar.
Host cells respond differentially to penetration
Using a SEM, we found that the fungus attempts to penetrate leaf surface, regardless of the mode of penetration, were accompanied by a bright 'halo' formation as early as 12 hpi (Fig. 3a, Additional file 1: Figure S3), indicating that at first, the host cell reaction to penetration was defensive. Such clear 'halos' around penetration sites have been observed during a resistant interaction of a biotrophic fungus Blumeria graminis f. sp. tritici on a wheat cultivar carrying effective resistance genes and also on a susceptible one, using cryoscanning electron microscopy . Possibly these 'halos' were described as papillae formed around penetration sites of A. brassicicola on the leaves of both susceptible and moderately resistant cabbage cultivars after double staining with trypan blue and aniline blue using confocal laser scanning microscopy . The biochemical analysis revealed that the 'halo' contains callose, phenolic compounds and an elevated level of calcium ions . The 'halo' phenomenon indicates that the host cells are trying to combat fungal invasion at this point of the susceptible interaction and stay alive. It also shows that the A. brassicicola infection cycle probably contains a very short biotrophic phase, although there is a general agreement that a necrotrophic fungus first kills a host cell by secreting toxins, and then invades it. Interestingly, it has been postulated that another necrotrophic fungus, B. cinerea, should be considered as hemibiotroph having a short biotrophic phase. Botrytis cinerea suppresses early host defense reactions by the secretion of small RNAs (sRNAs), thus leading to the silencing of host genes, but to achieve this, the host cell must be alive . It has been also shown that during invasion and the establishment of a necrotrophic interaction, B. cinerea 'sacrifices' many of its invading hyphae by subjecting them to cell death induced by plant-secreted cell death inducing factors, to get a chance to release fungal cell death inducing factors by the surviving hyphae into plant cells . It could also be a mechanism used by A. brassicicola and an explanation for the presence of hyphae with deteriorating protoplasts (Additional file 1: Figure S5e).
At later stages of an A. brassicicola infection (from 12-16 hpi), the successful penetration sites appeared on SEM images as electron-dense collapsed epidermal cells (Fig. 3b) or brownish possibly dead cells under bright field light microscopy (Additional file 1: Figure S2), which have been previously described during A. brassicicola infection . Brownish cells are also frequently found at penetration sites of other necrotrophic as well as biotrophic fungi . Browning of successfully infected epidermal cells may have been caused by the activation of peroxidases and phenolics within the cell wall and/or a production of ROS by dying host cells, indicating the suppression of host cell defense reactions and a highly susceptible response at this stage of infection [11, 22, 38, 39].
brassicicola colony formation is triggered by successful penetration
Successful penetration of the susceptible host epidermal cells by A. brassicicola was a signal triggering colony formation at inoculation sites. Connections between different regions of an expanding mycelial network were established via CATs and anastomoses (Fig. 4). Their formation is a typical feature of many filamentous fungi [40, 41], but these phenomena are mostly described when fungi grow in media rather than in planta during infection. CATs are usually formed during the initiation of a colony, and facilitate the transport of water and nutrients as well as horizontal gene transfer [42, 43]. Similarly to CATs, the formation of fusions between hyphae (anastomoses) originating from the same or different conidia, allow them a proper distribution of nutrients, transduction of chemical signals and even exchange of genetic material . Moreover, anastomosis formation is prerequisite for A. brassicicola virulence, as indicated by the development of an A. brassicicola aso1 mutant that is unable to form anastomoses, and when tested in planta appears to be unable to spread beyond the inoculation site .
Changes in host cell ultrastructure and transcriptome reprogramming
The appearance of necrotic spots indicated successful invasion and colonization of the host tissues by A. brassicicola, although host cells were differentially affected within the inoculation site depending on the distance from the invading hyphae, as evidenced by the gradual degradation of organelles. Similar individual changes in the ultrastructure of infected plant host cells, such as cell lysis, disintegration of the nuclei and chloroplasts or the presence of osmiophilic granules have been observed in other pathosystems during infection with viruses, bacteria and fungi [46-49]. In the case of infected B. oleracea mesophyll cells, these changes could be a result of the action of toxins and secondary metabolites secreted by A. brassicicola  and general reprogramming of the host transcriptome and metabolome in response to the infection [50, 51]. Alternaria species secrete various non-host- (nHSTs) and host-selective toxins (HSTs) which are responsible for the degradation of different organelles within an infected cell in over 200 plant species. For example, Alternaria alternata pathotypes produce different HSTs depending on the plant species, such as the AM-toxin that degrades the plasma membrane and chloroplasts in apple; the ACR-toxin that causes damage of the mitochondria in lemon; the AK-toxin, AF-toxin and ACT-toxin that target the plasma membrane in susceptible cultivars of Japanese pear, strawberry and tangerine, respectively [52-54]. However, a possible effect of identified A. brassicicola secreted compounds during infection on plant cell structures or their influence on the host transcriptome has not been described yet. It has to be emphasized that visualized differential B. oleracea cell responses to A. brassicicola infection, from defensive to highly susceptible, were confirmed by our microarray results (Additional file 2: Table S4), and were also concordant with the previous studies on plant gene expression during infection of susceptible B. oleracea and B. napus with A. brassicicola [55, 56]. The most up-regulated categories revealed in the analysis of Gene Ontology of Biological Processes were aging, the detection of ethylene stimulus, thiamine biosynthetic processes, cell wall macromolecule catabolic processes and defense response to fungus. Additionally, analysis of stress-related genes revealed up-regulation of the individual genes associated with hormone metabolism (brassinosteroids, ethylene and salicylic acid biosynthesis), peroxidases, glutathione S-transferases and secondary metabolism connected to flavonoid biosynthesis (Additional file 2: Table S4). Many of these genes have been previously described as important factors in general plant cell response to biotic stress and identified in transcriptome profile analyses of other susceptible hosts infected with Alternaria species [57, 58]. Some of these genes undoubtedly are characteristic for the B. oleracea response to infection by A. brassicicola [50, 55], albeit confirmation of their roles requires further investigations based on comparative transcriptomics and proteomic and metabolomic approaches.
Defense-related genes are activated also in susceptible interaction
The up-regulation of several stress-related genes involved in the defense response to fungus in the examined susceptible cultivar of B. oleracea even at 48 hpi, when necrotic lesions were fully developed, indicates that some of the host cells still attempted to combat the invasion of the fungus. The fungal hyphae spreading beyond inoculation site, even at a later stage of infection, apparently induced the host defense response through PTI (pattern-triggered immunity) . Probably, the perception of DAMP (damage-associated molecular pattern, e.g. products of the A. brassicicola cell wall degrading enzymes or plant secreted peptides) or PAMP (pathogen-associated molecular pattern, e.g. chitin) through a PRR (pattern recognition receptor) triggered the PTI-associated signaling cascade (albeit transcription of MEK1 was negatively regulated) and activated WRKY33 in infected B. oleracea cells, similarly to the situation observed in Arabidopsis during fungal infection . Interestingly, the only up-regulated gene encoding PRR in our microarray analysis was RLK7 (Additional file 2: Table S4). RLK7 belongs to category XI of RLKs and acts as a receptor for PIP1 secreted peptide. Elevated expression of PIP1 has been described in Arabidopsis guard cells during Pseudomonas-induced PTI, and thus the peptide plays a role in stomatal immunity . Moreover, RLK7 also contributes to Arabidopsis resistance to B. cinerea  and Phytium irregulare . However, LYK5, the other gene encoding RLK, which is an important receptor engaged in chitin perception in Arabidopsis , was down-regulated. The up-regulated genes involved in defense signaling such as WRKY33  and PDF1.2  have been identified in A. thaliana signaling pathways and are required for signaling resistance to necrotrophic fungi A. brassicicola and B. cinerea. In Arabidopsis, WRKY33 transcription factor is responsible for the activation of camalexin biosynthesis , which efficiently inhibits development of necrotrophic fungi. Possibly, WRKY33 activates de novo biosynthesis of B. oleracea fungus-induced phytoalexins, such as brassinin, which can be metabolized by A. brassicicola, and thus the fungus suppresses the first line of host defense . Subsequently, JA-dependent signaling, which is characteristic for host cells infected by fungal necrotrophs, activates transcription of JA-responsive defense genes encoding PR (pathogenesis-related) proteins such as PDF1.2 involved in resistance against B. cinerea  and A. brassicicola . However, the down-regulation of LOX2 chloroplast lipooxygenase required for JA accumulation may be also responsible for the impaired defense of the host and in turn, salicylic acid-dependent genes become activated, which promote host cell death . Moreover, local fortification of the host cell walls microscopically observed within inoculation site in our study was reflected in the up-regulation of several stress-responsive cell wall and secondary metabolism-related genes, such as FAH1 involved in lignin biosynthesis and an important component of A. thaliana resistance mechanism to B. cinerea  and CCoAMT, which is involved in the biosynthesis of cell wall-bound phenolics and lignin . The strengthening of host cell walls through lignin biosynthesis and its deposition at a pathogen entry site also constitutes the first line of the host defense reaction in order to slow down or restrict pathogen development, especially as lignin is not metabolized by most pathogens [72, 73]. Thus, even successful invasion of host tissues within inoculation sites does not signify that every single host cell immediately surrenders to a fungal invader. However, down-regulation of numerous stress-related genes found in our study indicates that A. brassicicola effectively overcomes the host arsenal of defenses (Additional file 2: Table S4). Although it has been reported that A. brassicicola secretes phytotoxin brassicicolin A, histone deacetylase inhibitor depudicin, siderophore N,N-dimethylcoproge [15, 17], a proteinaceous host-specific toxin - AB-toxin [74, 75] and many low molecular weight secondary metabolites e.g. brassicenes A to F , there is not yet any known putative A. brassicicola effector(s) targeting also still unknown Brassica receptor(s) and triggering ETS (effector-triggered susceptibility) or ETI (effector-triggered immunity), albeit many research groups all over the world work on A. brassicicola pathogenicity factors .
Down-regulation of photosynthesis is probably not only a part of susceptible interaction
Further, we have focused on changes in the ultrastructure of chloroplasts in infected mesophyll cells, mostly due to the observed clear stages of their degradation (Fig. 6) and the fact that analysis of our microarray data pointed out photosynthesis as the most negatively regulated process during infection of B. oleracea leaves with A. brassicicola (Additional file 2: Table S3). Chloroplasts are energy and carbon source organelles, and play an important role in plant immunity as a compartment for ROS generation and the production of phytohormones, secondary metabolites and their precursors . As sensors of environmental changes, chloroplasts can shape nuclear gene expression and activate defense responses through redox flux . Moreover, many pathogen-derived effectors target chloroplast-localized proteins, including components of the photosynthetic electron transport chain . Therefore, changes in chloroplast ultrastructure are often used as good indicators of biotic/abiotic stress .
Gradual degradation of the chloroplast membrane system, such as widening of the thylakoid lumens and the disappearance of grana observed in infected B. oleracea leaves, indicated the suppression of photosynthesis light reactions and potential damage of light harvesting complexes and reaction centers. As a result, chlorophyll and carotenoid content and photosynthesis efficiency at a physiological level was decreased. Ultrastructural changes of the stroma, from its swelling to disintegration, suggested down-regulation of the photosynthesis dark reaction and a decrease of carotenoid content. In severely degraded cells, the thylakoid system was still to some extent preserved, whereas the chloroplast envelope and stroma were totally disintegrated (Fig. 6f). The observed changes in the chloroplast ultrastructure were associated with changes in the expression of photosynthesis-related genes and the physiological response of the host cells. Early down-regulation of photosynthesis-related genes could be a result of the host cell defense strategy related to the shift from photosynthesis to non-assimilatory metabolism, as observed during plant infection by various pathogens, or an effect of action of an unknown A. brassicicola-secreted effector protein . At later stages of infection (24 and 48 hpi), more genes involved in light reactions were down-regulated. Fluorescence decline ratio in steady-state (Rfd) and a significant decrease in steady-state PSII quantum yield (Fv/Fm_Lss) in infected B.oleracea leaves indicate inhibition of electron transport in light-dependent reactions. Confirmation of this was the decrease in the value of Fq/Fv, qL, as well as the qP parameter, which describes the level of energy transferred to the reaction centers and informs about the proportion of open PSII reaction centers (Fig. 10) [82, 83]. The reduction of photosynthesis, determined by a decrease in the effective quantum yield of PSII, has been previously observed in the interaction of B. juncea-A. brassicicola  and in other plants infected with necrotrophic fungi [84, 85]. Also, in wheat plants infected with Bipolaris sorokiniana, a decrease in Fv/Fm correlated with the loss of chlorophyll has been noted .
In severe stress caused by A. alternata in rice, a decrease in the electron transport rate was correlated with an increase in non-photochemical quenching (NPQ) . An early increase in NPQ took place following a decline in photosynthetic electron transport activity. A similar PSII response was also observed in plants infected with viruses, e.g. Pepper mild mottle virus (PMMoV) infecting pepper leaves . However, such an increase can be the result of both: protective processes of PSII and damage to the photosystem . In infected B. oleracea leaves, we generally observed a decrease in NPQ (Fig. 10), which indicates a reduction in the dissipation efficiency of excess excitation energy as heat. However, it should be mentioned that not all processes associated with non-photochemical quenching lead to an increase in NPQ . The decrease in NPQ in infected B. oleracea leaves may be due to reduced light absorption, e.g. as a result of the destruction of chloroplasts and a decrease in chlorophyll content (Fig. 9), rather than a lack of thermal dissipation. Also, in other plants (i.e. rice, tomato) exposed to necrotrophic fungi, a significant decrease in NPQ was observed both in the necrotic zone and in adjacent areas of the leaf blade, without significantly reducing the photosynthesis efficiency (measured as Fv/Fm) [85, 90].
In the case of progressive degradation of chloroplasts at the inoculation site, it is difficult to expect discrete changes in the photosynthesis light reactions. Therefore, the significant decrease in the values of chlorophyll a fluorescence quenching parameters is not surprising. However, a slight decrease in QY_max and QY_Lss (Fig. 10) suggests that the destruction of the photosynthetic apparatus did not completely block the electron transfer in PSII in the analyzed leaf area. It is probable that the chloroplasts of cells that have not been invaded by the fungus allow photochemical reactions to occur, although it is known that even Alternaria spp. metabolites alone can cause a reduction or complete inhibition of the electron transport chain from QA to QB [91, 92]. This is due to competition between the toxin and QB for a binding site in the D1 protein on the thylakoid PSII membrane. In general, the photosynthetic yield of plants infected with necrotrophic fungi presents complex spatial and temporal patterns, depending on the degree of colonization of the individual regions of a leaf blade by the pathogen .