Biocontrol fungi such as T. viride
are known to induce systemic resistance, ISR, and prime their host plants to become more resistant to future attack from pathogenic microorganisms [9
]. The transcriptional changes related to ISR are usually quite modest compared to systemic acquired resistance, SAR [41
]. We here found that treatment of tobacco cells with T. viride
cellulase resulted in posttranslational changes leading to altered membrane properties and alamethicin resistance. To the best of our knowledge, the presented data are the first to show that resistance to permeabilisation by the peptaibol alamethicin can be induced in any eukaryote. Interestingly, cell wall degrading enzymes and peptaibols from T. harzanium
synergistically prevented spore germination and hyphal growth of Botrytis cinerea
]. Thus, synergies that are harmful to one system (Trichoderma
on pathogen) can be protective in another system (Trichoderma
on plant), which favours a successful symbiotic relation between Trichoderma
and the plant.
The alamethicin resistance observed was mainly elicited by the enzymatic activity of T. viride cellulase. This is strongly indicated by the reduction in elicited resistance by heat inactivation and by the presence of the cellulase inhibitor cellobiose. Further, the effect of inhibitors excludes the possibility of alamethicin resistance being elicited by any of the small known contaminants of most cellulase extracts. Shortening the enzyme incubation to 20 min followed by a post-incubation in Control medium alone (until the same total of 4 h had passed) did not reduce the alamethicin resistance induced. This indicates that the cellulase elicits the resistance during the first part of the incubation and that no further stimulus is required, but that it takes a certain time for the response to develop in the plant cell. After these treatments, no visual changes could be observed by light microscopy, indicating that only a limited cell wall digestion had taken place. Interestingly, the observed resistance displays some specificity for T. viride cellulases since the effect was neither seen upon incubation with a cellulase mixture from T. reesei nor by hemicellulases of the same fungus (Figure Table ). The presence of a cellulose-binding module (frequently carried by cellulases) did not induce resistance, consistent with the inactivation and inhibition studies showing that an active enzyme was needed (Figure Figure ).
It could be argued that the resistance observed here is a part of a general defence response to cell wall degradation, intended to increase the robustness of the plasma membrane in anticipation of a fungal or bacterial attack reaching through the cell wall. It has earlier been reported that cellulase treatment can evoke defence responses, e.g
., increases in the stress-related phytoalexin capsidiol [43
] as well as the production of volatile compounds [45
]. Xylanase, which can degrade the xylan of the cell wall hemicelluloses represents a threat to cell integrity similar to that posed by cellulase [47
]. However, in contrast to the eliciting effect of cellulase in our experiments, xylanase does not need to be enzymatically active to elicit defence responses in tobacco [49
]. Also, the difference in mode of elicitation is consistent with the inability of xylanase to elicit alamethicin resistance.
If alamethicin resistance were part of a general response to pathogen attack it would be reasonable to assume that many common plant elicitors mediated a similar response. The acetylated chitin derivate chitosan is able to elicit a large range of plant defensive responses, including HR, SAR, oxidative burst and callose deposition [50
], yet we could not detect a significant difference in alamethicin resistance. Similarly, with the PAMPs flg 22 [51
] and elf 18 [52
], no elicitation of alamethicin resistance could be observed, despite their ability to trigger innate immunity. Finally, adding catalase to cells during CM did not prevent the elicitation of resistance (Additional file 2
). This indicates that the somewhat increased resistance observed after H2
incubation is not due to H2
being a putative intermediate in the cellulase-initiated signalling cascade. Instead, the presence of H2
can lead to rapid cross-linking of the cell wall proteins [53
]. The decrease in permeability of the cell wall after such cross-linking may be the reason for the moderate alamethicin resistance after H2
incubation (Figure ). In any case, this resistance at the cell wall level cannot explain the cellulase-induced alamethicin resistance, since also protoplasts devoid of cell wall were resistant to alamethicin.
Rather, the alamethicin resistance could be compared to classical R-gene-induced resistance in the sense that both might counteract pore formation activities of successful pathogens and beneficial microorganisms. Instead of manipulating the consequences of pores by deactivating the pathogen effectors that are transported through them, as is characteristic to R gene-mediated resistance, the alamethicin resistance decreases the possibility for pores to be formed.
Analyses conducted with artificial lipid bilayers have suggested that alamethicin needs to be delivered from the compartment with the net positive electric potential in order to be inserted and form pores in membranes [21
]. Experimental data on biological systems are in line with this, i.e
., the vacuole (which has a positive transmembrane potential) in tobacco cells was left intact under conditions when other membranes were permeabilised [26
]. Upon cellulase treatment, the transmembrane potential of Medicago sativa
root hairs was depolarised to ca -50 mV [54
., to what probably would be the diffusion potential [55
]. However, for the resistance development described here, transmembrane potential changes could be ruled out as important since no effect was obtained by the protonophore FCCP (Figure ), an agent shown to depolarise the transmembrane potential in roots to the diffusion potential [38
]. Also, protein synthesis was not needed for the process (Figure ), showing that the resistance depended on modifications performed by pre-existing enzymes or structures. Cell wall modifications induced by the action of T. viride
cellulase may result in both chemical and mechanical signals reaching the plant cell. Cellodextrins (β-1,4 glucose oligomers), i.e
., the predominant breakdown products of cellulose, induced pathogen responses in Vitis vinifera
]. On the other hand, homologues of prokaryotic and eukaryotic mechanosensitive channels were recently identified in A. thaliana
], and an existence of mechanosensing signalling also in plants has recently been suggested [58
]. However, the lack of effect by xylanase in our experiments (Figure ) and the quite small effect induced by macerozyme (Table ) shows that if the signal is mechanical, it cannot operate simply through the degradation of classical matrix polysaccharides.
Peptide-induced pore formation depends on membrane lipid species and lipid/peptide ratio [31
]. We found that the sterol to membrane lipid fatty acid ratio (Figure ), the fraction of PS+PI (Figure ) and the acyl group 20:0 decreased as a consequence of enzyme treatment. Our analyses were performed with cells that still were indistinguishable from untreated cells with regard to shape (Additional file 1
), but when substantial alamethicin resistance could be detected. Therefore, the changes in lipid composition seen probably reflect the defence induced against T. viride
, whereas the degradative changes often associated with complete protoplastation [59
] are kept at a minimum. This also agrees with that strains of Staphylococcus aureus
, Enterococcus faecalis
and Bacillus cereus
with a five-fold increased resistance to alamethicin permeabilisation (IC50
of 2-5.5 µg ml-1
alamethicin in sensitive and 9.5 to 29 µg ml-1
in resistant strains, respectively), showed altered membrane lipid composition as well as lower alamethicin association to vesicles prepared from membrane extracts [62
The CM-induced changes in phospholipids and their corresponding fatty acids (Figure Figure ), suggest that the physical properties of the plasma membrane were altered, possibly sufficient to affect alamethicin insertion and pore formation. This agrees with that the conductance through pores made by the antimicrobial cationic peptide gaegurin 4 was larger in planar bilayers made of PE, PC and PS (80:10:10) compared to membranes composed of only PE and PC (80:20) [63
]. A role of sterols with respect to alamethicin channel activity was shown with artificial membranes, i.e
., the presence of cholesterol increased the duration of the alamethicin pore in its open state, indicating a more efficient use of created pores, while the critical concentration of alamethicin needed for pore formation increased [64
]. Oligomerisation and pore formation by Vibrio cholerae
cytolysin also depended on the presence of cholesterol [66
]. With gaegurin 4 [63
], inclusion of cholesterol in planar lipid membranes acted opposite to PS, i.e
., it prevented channel formation. This deviates from the association of increased alamethicin resistance to decreased sterol levels (relative to fatty acids) observed with tobacco cells (Figure ). However, the hydrophobic alamethicin forms pores that traverse the membrane through its hydrophobic part, whereas cationic peptides such as gaegurin 4 form pores in the membrane where peptide and membrane lipid headgroups are exposed to the inner of the pore [67
]. Besides, the presence of proteins in biological membranes adds another degree of complexity, making direct comparisons between peptide types difficult.
Large differences in lipid composition were used in the above investigations of alamethicin pore formation with artificial membranes. This might speak against direct comparisons with the smaller differences found for the tobacco plasma membrane here, also since the artificial membranes do not contain proteins as do biological membranes. However, effector-induced changes in membrane phospholipids and sterols of similar magnitudes as we found with tobacco lead to changes in membrane stability with isolated plasma membranes from oat roots [68
] and S. cerevisiae
] as seen by changes in transversal bilayer diffusion.
Another important property of especially the phospholipids is their charge, with PC and PE being uncharged and PA, PI, PS and PG being negatively charged. The charges of the lipid head groups and the membrane proteins will cause a local surface charge which will affect the attraction of ions to approach the membrane, and also modulate the spacing of lipids. In our experiments, we found that CM treatment resulted in lower PM-associated PS+PI and higher PE (+PG) compared to control cells (Figure ). Even though the surface charges depend also on e.g
., proteins and the phospholipid distribution between the respective plasma membrane leaflets, the results suggest that overall surface charge of the plasma membrane may be lower in CM-treated cells compared to control cells. With artificial membranes, lower surface charge result in less alamethicin inserted [70