**5. Conclusion**

188 Fungicides for Plant and Animal Diseases

Promising microbial metabolites continue to be discovered using traditional activity-based screening procedures against various plant pathogenic fungi. In particular, many of the sitespecific antifungal metabolites have recently been discovered from microbial metabolites. These microbial metabolites include non-fungicidal compounds that interfere with the infection process of pathogenic fungi, and specific inhibitors of the fungal biosynthetic

Other kind of antifungal agents are proteins; antifungal proteins have been isolated from various organisms ranging from bacteria, plants, insects and amphibians to human beings. Both their fungal target site and their mode of action are extremely diverse. In order for it to be applied, an antifungal protein needs to fulfill several prerequisites such as antifungal activity *in vivo* and lack of effects on the host cells. Furthermore, resistance mechanisms need to be excluded as far as possible. Therefore, investigation of the target site and the mode of action of an antifungal protein should reveal whether the protein is suitable for an application (Theis *et al.,* 2005). The antifungal protein (AFP) are abundantly secreted by the filamentous fungus *Aspergillus giganteus*, this cysteine-rich protein have ability to disturb the integrity of fungal cell walls and plasma membranes but does not interfere with the viability

Severe membrane alterations in *A. niger* were observed, whereas the membrane of *P. chrysogenum* was not affected after treatment with AFP. The protein localized predominantly to a cell wall attached outer layer which is probably composed of glycoproteins, as well as to the cell wall of *A. niger*. It was found to accumulate within defined areas of the cell wall, pointing towards a specific interaction of AFP with cell wall components. In contrast, very little protein was bound to the outer layer and cell wall of *P. chrysogenum*. The protein was found to act in a dose-dependent manner: it was fungistatic when applied at concentrations below the minimal inhibitory concentration, but fungicidal at higher concentrations. Using an *in vivo* model system was demonstrated that AFP indeed prevented the infection of tomato roots (*Lycopersicon esculentum*) by the plant-pathogenic fungus *Fusarium oxysporum* f.

The fungal profilins, small actin-binding proteins that share limited homology to human profilin, can operate as a potential drug target, since these proteins are essential for the growth of most eukaryotic cells, including *S. cerevisiae* (Witke, 2004). Addition, the existence of structural information can support the design of structure-based ligands for profilins. Peptides are generally used as lead compounds in drug development, to design a novel peptide ligand, an *in vitro* evolution approach has often been used. Although this approach can be used without three-dimensional (3D) structural information about the target protein, it requires laborious experimental procedures, including library constructions and the screening of bioactive peptide ligands. In this respect, if information on the structure and the active site of the target protein is known, an *in silico* approach based on the 3D structure of the target protein is a useful approach to designing the peptide ligand. The validity of the profilin as antifungal drug target was evaluated by Ueno *et al*. 2010, amino acid alignments showed the low homology between human and fungal profilins. This implies that the fungal profilin could be a target with high selectivity. Furthermore, a mouse infection study showed that the suppression of profilin expression attenuated the fungal burden in the kidney and indicated that the profilin was required for survival in the host's body (Ueno *et* 

pathways for chitin, fatty acids and nucleic acids (Seok & Kook, 2007).

of other eukaryotic systems (Barakat *et al.,* 2010; Meyer, 2008).

sp. *Lycopersici* (Theis *et al*., 2005).

*al.,* 2010).

We would like to conclude by stating that antifungal targets-site are extremely diverse. However, substances that acts on these target-sites needs to fulfill several prerequisites such as antifungal activity *in vivo* and lack of effects on the host cells. Furthermore, resistance mechanisms need to be excluded as far as possible. Therefore, investigation of the target site and the mode of action of an antifungal compound can be explored by statistical learning algorithms. Performance and applicability of the statistical learning methods in studying "fungal-target likeness" may be further improved by incorporation of new information from advances in genomic, proteomics, pathogenesis and morphogenesis studies. Efficiency and accuracy of statistical learning methods in the prediction of fungal-target like proteins can also be enhanced from new progress in learning algorithms and sequence descriptors.
