Indexing and Analysis of Fungal Phenotypes Using Morphology and Spectrometry

Michael Edberg Hansen

AbstractMicrofungi have the ability to produce a high number of secondary metabolites, which they use for various reasons in their natural habitat. The ability to produce this huge variety of metabolites, as they can be made only by nature, make them extremely interesting subjects for many reasons. First of all fungi are some of the most industrially important production organisms in the food industry. And in the same industry, fungi also belong to the most common spoilers of food and stored cereals, resulting in significant economic losses.

Furthermore, fungi are important sources in the search for potentially new compounds used in the pharmaceutical industry. Again, while the metabolites can be used in the treatment against various diseases and infections, some metabolites, the so called mycotoxins, can be harmful causing e.g. infections and allergenic reactions to humans. Identification of the fungi is therefore important. By rapid and correct identification it is possible to investigate the food e.g. to evaluate the quality, as well as to eliminate already known isolates from drug-discovery screening programs, and perhaps even predict the presence of new drug candidates.

Unfortunately, detection and identification of the fungi is considered difficult and laborious. Though visual expressions have been and still is used as phenotype markers in the classification and identification of fungal species, one of the most successful characters used has been the profile of the secondary metabolites.

In order to evaluate the visual phenotypic characters, a method for visual clone identification of Penicillium commune the most widespread and most frequently occurring spoilage fungus on cheese was developed (Papers A, B and C). The method was based on images of fungal colonies acquired after growth on a standard medium and involves a high degree of objectivity. On a data set from 137 isolates we obtained a leave-one-out cross-validation identification rate of approximately 93%

A fully automated data processing approach for qualitative comparison of a large number of mass spectra from the direct infusion analysis (ESI-TOF-MS) of complex extracts is presented. In order to be able to compare this type of data a new matching algorithm was developed (Paper D). Based on data from analysis of extracts from 80 isolates representing 9 classes of closely related Penicillium species in the series Viridicata (associated with stored cereals), the method is capable of obtaining an almost perfect grouping. By utilising the high resolution mass spectroscopy, we have shown that we find a relationship between species, and ions responsible for the segregation. Furthermore, a practical library search procedure to automatically extract highly complex and similar ESI-MS mass spectra for identifying fungal extracts in a reference library are being developed and tested (Paper E).

Whereas mass spectrometry is one modality used in systematising the fungi, high pressure liquid chromatography combined with an UV diode array detector, is another and complementary method. A method for unsupervised chemosystematic identification of closely related fungi, using analysis of secondary metabolite profiles created by HPLC with UV diode array detection is presented. In this study the secondary metabolite production from closely related penicillia in the Camemberti series, is investigated. By full cross evaluation of data from cultural extracts, it is found that the species may be segregated into taxa in full accordance with published taxonomy (Paper G). In order to evaluate the performance, this method is tested on a group of Alternaria species (Paper H), comparing this chemosystematic method with the visual proposed in earlier studies (Papers A, B and C).

Finally, a new algorithm for detecting the presence of similar or identical UV spectra in highly complex samples of natural products is presented (Paper F). Based on a database containing profiles of known spectra, the method proves to have both high sensitivity and specificity. The great potential of this method is its ability to support the choice of the optimal organism for production of a given natural product, and for the tracking further lead compounds in scenario where a new and UV detectable bioactive compound has been discovered.
TypePh.D. thesis [Academic thesis]
PublisherInformatics and Mathematical Modelling, Technical University of Denmark, DTU
AddressRichard Petersens Plads, Building 321, DK-2800 Kgs. Lyngby
NoteSupervised by Dr. Jens Michael Carstensen (IMM) and Professor Jens Christian Frisvad (BioCentrum-DTU). Electronic full text version in PostScript available by contacting Michael E. Hansen,
Electronic version(s)[zip]
BibTeX data [bibtex]
IMM Group(s)Image Analysis & Computer Graphics