Global nutrient profiling by Phenotype MicroArrays: a tool complementing genomic and proteomic studies in conidial fungi

Abstract

Conidial fungi or molds and mildews are widely used in modern biotechnology as producers of antibiotics and other secondary metabolites, industrially important enzymes, chemicals and food. They are also important pathogens of animals including humans and agricultural crops. These various applications and extremely versatile natural phenotypes have led to the constantly growing list of complete genomes which are now available. Functional genomics and proteomics widely exploit the genomic information to study the cell-wide impact of altered genes on the phenotype of an organism and its function. This allows for global analysis of the information flow from DNA to RNA to protein, but it is usually not sufficient for the description of the global phenotype of an organism. More recently, Phenotype MicroArray (PM) technology has been introduced as a tool to characterize the metabolism of a (wild) fungal strain or a mutant. In this article, we review the background of PM applications for fungi and the methodic requirements to obtain reliable results. We also report examples of the versatility of this tool.

1. Introduction

Conidial fungi (mitosporic Dikarya, also informally called imperfect fungi or Fungi Imperfecti), commonly known as molds and mildews, represent a diverse and economically important biological group of organisms that include animal and plant pathogens, as well as organisms used in biotechnology. Some conidial fungi are responsible for the production of industrially important enzymes, antibiotics, secondary metabolites and foods (e.g., cheeses, sake, and soya sauce), whereas others produce toxins contaminating food products and/or resulting in food spoilage. Conidial fungi are capable of metabolizing a wide variety of nutrients. This characteristic has been exploited extensively to study metabolic pathways and their regulation in model fungi such as Aspergillus nidulans and Neurospora crassa. Their importance for humankind has consequently led to an increasing number of fungi for which genomic sequences are now available (>130 at http://www.ncbi.nlm.nih.gov/genomeprj as noticed on Dec. 15, 2009). Genome sequencing, however, is only the first step towards the identification and validation of the function of the genes of an organism. To understand the information stored in the genome, one must understand the production of knockout mutants and the consequence for the organism biology, learn the conditions for the expression of the gene, and have knowledge of the stability, compartmentation, and modification of the gene’s translation product. To this end, “high-throughput methods”, capable of studying all these traits, have and are being developed. For many conidial fungi, facilitated methods for production of knockout mutants have already been established (Sweigard et al., 1999; Chaveroche et al., 2000; Hamer et al., 2001), but new approaches for wide-domain functional characterization of the gene of interest need further elaboration.

In addition to investigation systems, which focus on specific genes, pathways, or processes, functional genomic methods attempt to study the impact of altered genes on the phenotype of the organism, using technologies that provide a cell-wide perspective. The most well known “global” approaches are DNA-microarrays for transcriptomics (Fodor et al., 1993; Kahmann and Basse, 2001) and two-dimensional gel electrophoresis (O′Farrell, 1975) for proteomics. These technologies allow for global analysis of the information flow from DNA to RNA to protein. To comprehend, however, how the gene initially is encoded in the genome is ultimately displayed at the cellular level; the phenotype must be considered. Phenotype is the manifested attribute of an organism, the joint product of its genes and their environment during ontogeny. As such, phenotypes are the ultimate goal in strain improvement in biotechnology during the screening of fungi for new processes or products. Therefore, a (semi)high-throughput and comprehensive phenotypic assay would greatly accelerate a functional genomic approach.

Since macronutrients are the major determinants of the fungal phenotype, a comprehensive profiling system should ideally include the maximum number of various nutrient sources in a single assay. This would contain some hundred carbon sources, and fifty to hundred nitrogen sources, sulphur and phosphorous sources, and eventually also include different pH and aeration conditions. Such a maximized system could be prepared in 5–7 microtiter plates with each well containing a standard medium specific for one particular phenotype (e.g., utilization of a sucrose), and allows the quantitative analysis of growth and/or metabolism. Unfortunately, custom synthesis of this type of comprehensive phenotype analysis system is laborious, and probably only feasible with the aid of laboratory robotics. The Phenotype MicroArray™ (PM) system (Bochner et al., 2001; Bochner, 1988; 2003) (Biolog Inc., Hayward, CA), however, offers a reasonable compromise and has frequently been used for the analysis of bacteria and yeast mutant strains (http://www.biolog.com/mID_section_13.html for bibliography). Kubicek et al. (2003) applied carbon source PMs to soil fungi, thereby detecting species-diagnostic characters in a collection of Trichoderma (teleomorph Hypocrea) isolates from Southeast Asia. Tanzer et al. (2003) demonstrated how global nutrient utilization analysis can be used to elucidate the effects of either genetic alterations or chemical treatments on Aspergillus nidulans, Aspergillus fumigatus, Magnaporthe grisea, and Mycosphaerella graminicola. In addition, Druzhinina et al. (2006) have applied PMs to investigate phenotypes of various wild-type and mutant strains of Hypocrea jecorina (anamorph Trichoderma reesei). The results presented in these three studies demonstrate that PMs indeed offer an ideal complement for the phenotypic characterization of gene knockout strains towards understanding genomic and proteomic data.

Here we review the applications of PMs to conidial fungi, and discuss the utility of this system for the high-throughput analysis of global phenotypes and utilization of particular nutrients.

2. Principle and reliability of Phenotype MicroArrays

During the initiative on functional genomics of Saccharomyces cerevisiae as a model system, two groups pioneered the testing of a large number of strains against 96 (Ross-Macdonald et al., 1999) or 288 (Rieger et al., 1997; 1999) phenotypes. In these studies, growth of yeast strains was tested on agar surface. It was problematic because scoring of growth on agar is prone to subjectivity and difficult to scale for high-throughput analysis. Several other researchers have tested their strains by cultivating them on a hundred or more different nutrient media compositions. Bochner (1989) first demonstrated a technique for global analysis of cellular phenotypes using microplates. He advocated the testing of phenotypes using cell respiration as a reporter system. The principle of this assay is the use of a tetrazolium dye (tetrazolium violet), which is reduced by the action of succinate dehydrogenase, enabling the quantification of respiration. Reduction of this dye results in formation of a purple color with a maximum absorbance at 490 nm. The reaction is essentially irreversible; therefore, the resulting dye accumulates in the well over a period of incubation, amplifying the signal and integrating the amount of respiration over time. Although this assay works well in prokaryotes and yeasts (Bochner et al., 2001; Singh, 2009), color formation in imperfect fungi does not always coincide with growth (Fig. 1). Consequently, both Tanzer et al. (2003) and Druzhinina et al. (2006) quantified growth by mycelial production reading the optical density at 750 nm (OD₇₅₀).

Fig. 1.

Correlation between formazan dye production and mycelial density

Formazan dye production was estimated as difference between optical density (OD) values at 490 and 750 nm. Mycelial density directly corresponds to the OD value at 750 nm. Grey circles show the result of the experiment with Hypocrea jecorina QM 6a, and black diamonds indicate values from an experiment with Hypocrea atroviridis P1 strain. Vertical and horizontal error bars illustrate standard deviation calculated for three independent experiments per each strain

To achieve reproducible OD₇₅₀ measurements for certain fungal species, both the media for inoculum production and inoculum concentration had to be optimized. As an example, for A. nidulans and M. graminicola, the inoculum densities that resulted in the least variation ranged from 1.25×10⁵ to 5×10⁵ spores/ml, while for M. grisea the optimal inoculum concentration was strictly at 4×10⁵ spores/ml (Tanzer et al., 2003).

Furthermore, contrary to endpoint assays absorbance data need to be collected over the incubation period to generate complete growth curves for each nutrient source. This is necessary because, for example, different carbon sources result in different growth kinetics (Fig. 2), and assessing growth only at a single time point would eventually be indicative of the early growth phase in one case and the phase of already terminated growth in another.

Fig. 2.

Four types of growth patterns revealed for Hypocrea jecorina (strain QM 6a) observed on Biolog FF MicroPlate

D-glucose (black diamonds) exemplifies the group of quickly utilised carbon sources, which resulted in maximal biomass values already after 40 h of incubation. D-lactose (open triangles) is typical for carbon sources with medium utilisation rates, which gave maximal mycelium production after 3 d of growth. D-tagatose (open circles) belongs to the group of carbon sources, which provided only slow growth, and still commenced for longer than 72 h. Glucuronamide (black squares) is a compound, which does not support growth of H. jecorina QM 6a. Error bars correspond to standard deviation values calculated after three independent experiments

In addition, careful visual monitoring of cultures in test wells over the incubation period is essential: several carbon sources were found to initiate the early sporulation (e.g., 48 h of growth on γ-aminobutyrate for H. jecorina), which was reflected by increase in absorbance readings (Druzhinina et al., 2006). Careful examination of the time-dependent growth patterns on every nutrient composition as recommended above helps to eliminate this problem.

Finally, a last caveat would be the hyphal morphology. There are a number of carbon sources, which lead to aberrant morphology, and one would expect that thicker or irregularly branched hyphae would yield a different ratio of OD₇₅₀ vs. true biomass. Fortunately, this problem appears to be less important than theoretically predicted: we have tested the growth of H. jecorina wild-type strain in submerged culture, on agar surface and in PMs for several carbon sources with a potential morphological effect (e.g., L-sorbose), and generally found significant correlation between them. Therefore, OD₇₅₀ values are equivalent means to measure growth of conidial fungi on PM. Considering all the preconditions discussed above, the reproducibility of PMs is very high. We have tested 17 wild-type and mutant strains of H. jecorina in three independent experiments when turbidity (mycelial production) was measured at several time points (Druzhinina et al., 2006). The subsequent statistical analysis did not detect any significant difference between corresponding plates in both cases when values for all carbon sources were averaged or tested individually, although the variable of time was always significant and has revealed three clusters attributed to spore germination, linear hyphal growth, and phase of sporulation and/or growth saturation. In spite of the high reproducibility of PMs, three or more independent tests are required as a precautionary measure against possible air-born contamination of test well(s), which might take place during readings at early growth stages or when the slow growing fungi are investigated.

2.1. Intra- and inter-specific variability in carbon source utilization in conidial fungi

Well studied filamentous fungi genera, e.g., Aspergillus, Neurospora, Hypocrea and Acremonium. are known to have a wide range of substrate assimilation (Caddick et al., 1994; Tanzer et al., 2003; Druzhinina et al., 2006; Hoyos-Carvajal et al., 2009). For species of genus Hypocrea/Trichoderma, rapid growth as well as ability to assimilate diverse substrates is a great advantage that allows this genus to colonize many ecological niches in the world. Kubicek et al. (2003) compared the inter-specific variability of carbon source utilisation profiles in Hypocrea/Trichoderma, and confirmed that isolates of most species formed defined clusters, thus proving the applicability of this method in species identification. This approach was then integrated in a description of T. brevicompactum sp. nov. by Kraus et al. (2004).

PMs were also applied to study the differentiation of two morphologically and ecologically very similar marine species Dendryphiella arenaria and D. salina (dela Cruz et al., 2006), when the qualitative difference in utilization of four carbon sources confirmed the species divergence inferred from genetic data. Moreover, PMs could also differentiate subpopulations within Dendryphiella species in relation to their geographic origin (dela Cruz et al., 2006).

Nutrient source profiling was also applied by Komon-Zelazowska et al. (2007) to determine the evolutionary context of two genetically closely related, but phenotypically very different species (Trichoderma pleuroticola and T. pleurotum) that cause green mold disease in oyster mushroom (Pleurotus ostreatus) farms worldwide. This assay revealed generally impaired growth of T. pleurotum on numerous carbon sources according to enhanced assimilation of those by T. pleuroticola, which thereby showed very similar metabolic characteristics to its phylogenetically close members in the Harzianum clade of Hypocrea/Trichoderma (T. harzianum and T. aggressivum). It was shown that the highest assimilation rates for T. pleuroticola occurred on N-acetyl-D-glucosamine and quinic acid, which could be useful for differentiation of two causative agents of Pleurotus green mold disease. T. pleuroticola, on the other hand, can be distinguished from T. aggressivum (green mold disease on Agaricus) by the inability of the latter to utilize α-ketoglutaric, L-malic, and succinamic acids (Komon-Zelazowska et al., 2007). Thus, these kinds of assays are important sources of complementary data for precise species identification and as a starting point for uncovering the ecological niche of a fungus.

Furthermore, as shown in Fig. 3 (biochemical and physiological groups of carbon sources are given in Fig. 1A), PM analyses of different species displayed significant intraspecies diversities (H. jecorina) or were conserved (H. atroviridis) (Seidl et al., 2006; Friedl et al., 2008b). It was interesting to see that for H. jecorina the level of this variability was in the same range as that obtained by DNA-mediated transformation (Druzhinina et al., 2006). It may suggest that the difference between H. jecorina isolates is due to the sexual recombination, which is possibly less frequent in H. atroviridis.

Fig. 3.

Metabolic profiles of H. jecorina and H. atroviridis as inferred by Biolog Phenotype MicroArray analysis after 48 h of incubation

Light grey shadow corresponds to the summed metabolic profile of five wild-type strains of H. jecorina. Dark grey shadow indicates the combined profile of three H. atroviridis strains. The solid line corresponds to the reference strain H. jecorina QM 6a, which has a unique physiological profile. Error bars indicate standard deviation values estimated based on tree independent experiments

Yet not every study could successfully confirm species differentiation: we tested the hypothesis that opportunistic strains of T. longibrachiatum isolated from the lungs of immunocompromised patients may represent specialized potentially clonal subpopulations within this species, which would allow us to identify specific genetic markers for their diagnosis. In contrast, the results from this work provide clear evidence that not only one, but two, genetically different species T. longibrachiatum and H. orientalis cause infections, and that clinical and environmental isolates reveal surprisingly consistent PMs (Druzhinina et al., 2008).

In the manner of species identification, Hoyos-Carvajal et al. (2009) recently examined the diversity of Hypocrea/Trichoderma in neotropic regions using morphological, genetic, as well as metabolic approaches, proving that the latter should not to be omitted in the recognition and characterization of new species or in the analyses of phylogenetic relationships within a fungal genus. Hoyos-Carvajal et al. (2009) reported that in T. asperellum, one of the most abundant species found in neotropical regions, two distinct genotypes were supported by carbon source utilization profiles: strains of the T. asperellum clade A originating from forests grew on very poor substrates, whereas strains from clade B in situ associated with crops preferred rather readily available substrates such as sugars. The predominance of T. harzianum can be attributed to a relatively wide variety of utilized carbon substrates, as reported by Hoyos-Carvajal et al. (2009), yet three new species morphologically similar to T. harzianum showed atypical metabolic profiles in the cluster analysis as well as a separate lineage in the sequence analysis of rRNA markers.

Moreover, a screening test for rapid differentiation of strains of the superior chitinase producing fungi attributed to T. harzianum sensu stricto clade with Biolog Phenotype MicroArrays was developed by Nagy et al. (2007), revealing that these fungi are unable to grow on N-acetyl-D-mannosamine. This fact was used to establish a simple test for directed screening for chitinase-producing strains (Nagy et al., 2007).

PM analysis aids, however, in the detection of intra- and inter-specific variability of conidial fungi, thereby offering a means for characterizing species, individual strains, and ecological groups.

2.2. Identification of phenetic differences in non-transformed mutant strains

Many of the classical fungal mutant strains used in biotechnology were selected after chemical or physical mutagenesis solely based on improved product formation, while the biochemical or genetic nature of the mutation has rarely been revealed. An example is T. reesei QM 6a (teleomorph Hypocrea jecorina), which was subjected to a classical mutagenesis by a series of exposures to radiation with a linear accelerator. The resulting mutants displayed a two- to four-fold increase in cellulolytic activity. We have tested whether PMs would be capable of identifying the possible changes, which had accompanied this mutagenesis (Druzhinina et al., 2006). The data showed that, in general, mutagenesis left the carbon source utilization profile of the respective strains remarkably unchanged, thus indicating that an improvement of cellulase formation was due to mutation of very specific intracellular targets, such as those involved in inducer formation or reception. Some subtle differences, however, could be observed: increased cellulase formation was shown to correlate with a decreased growth on adonitol (=D-ribitol), 2-ketogluconate and γ-aminobutyric acid and increased growth rates on D-sorbitol and saccharic acid. The metabolism of two of these compounds (2-ketogluconate, saccharic acid) in H. jecorina is unknown, thereby preventing an interpretation of the observed effect. The two others (γ-aminobutyric acid, D-sorbitol), however, may indeed offer interesting insights: γ-aminobutyric acid formation is up-regulated during the early phase of conidiation in Trichoderma, and exogenous addition of γ-amino-butyrate stimulates sporulation in T. reesei (unpublished data). Since cellulase formation is triggered by sporulation (Kubicek, 1987), γ-aminobutyrate may constitute a link between cellulase induction and conidiation. On the other hand, the increased utilization of D-sorbitol by the improved producer strain QM 9414 may be linked to enhanced formation of L-sorbose, an inducer of cellulase formation in H. jecorina (Nogawa et al., 2001), because it is formed from D-sorbitol by the respective nicotinamide adenine dinucleotide phosphate (NADP)-dependent ketose reductase (Seiboth et al., 2007). While the importance of these two findings for cellulase production must be verified by reverse genetics, they have nevertheless pointed to two biochemical reactions, which may have a major impact on cellulase formation, and would have remained undetected without PM analysis.

2.3. Biomining for new biocatalysts and improved producer strains

Most of the fungal strains used today in the production of enzymes, secondary metabolites, and other organic chemicals have been screened by testing for either formation of the product itself or for genetic alterations assumed to cause improved production such as carbon catabolite derepression, e.g., by using resistance to 2-desoxiglucose for cellulase overproduction in H. jecorina (Montenecourt and Eveleigh, 1979). Improved production, however, may also be due to changes of unknown or unexpected physiological properties, detection of which may therefore offer new methods for the direct screening of improved producer strains. To this end, we have recently used carbon source PMs to correlate the ability to secrete high chitinase activities with the carbon source assimilation profile in Trichoderma. In line with the comments made above, we in fact observed a statistically relevant correlation between the assimilation rate of lactic acid methyl ester and chitinase activity (unpublished data). While the physiological meaning of this correlation is not yet understood, lactic acid methyl ester is a valuable indicator of the chitinolytic potential of Trichoderma species.

While PM primarily provides information about the growth rate on different nutrients, the integration of these data into a biochemical map could also lead to the detection of pathways or enzymatic steps, which are unique for a particular strain, species or genus. One barrier towards achieving this is that not all of the biochemical pathways occurring in filamentous fungi are as yet understood to the level that the implementation of this type of network would be feasible. Some biotechnologically useful information can, however, be extracted already at this stage. For example, N-acetyl-mannosamine is a very poor carbon source for some species and strains of Hypocrea/Trichoderma such as H. jecorina, although they all assimilate N-acetyl-glucosamine very well. The majority of strains of H. lixii/T. harzianum also utilize N-acetyl-mannosamine. This is due to the formation of the enzyme N-acetylglucosamine epimerase, which converts N-acetyl-glucosamine into N-acetyl-mannosamine as shown in Fig. 4 (Kowal and Wang, 2002). Since the reaction of this enzyme is essentially in equilibrium, it can, vice versa, be added to the in vitro degradation of chitin by Trichoderma chitinase, which would then ultimately lead to the accumulation of N-acetyl-mannosamine, a high-value product needed for the biocatalytic synthesis of sialic acid-like compounds used in various therapies (Blayer et al., 1999; Schauer, 2000; Keppler et al., 2001). Yet, using the PM method for detecting the differentiation between chitinase-overproducing strains among H. lixii/T. harzianum (Nagy et al., 2007), the lack of N-acetyl-β-D-mannosamine utilization was present in the superior chitinase producer strains.

Fig. 4.

Conversion of N-acetyl-glucosamine into N-acetyl-mannosamine by enzyme N-acetyl-glucosamine epimerase [modified from Kowal and Wang (2002)]

Furthermore, a screening system consisting of a combination of Biolog PM and specific enzyme activity measurements using a chromogenic substrate (Seidl et al., 2006) was developed to identify carbon sources that trigger β-N-acetyl-D-glucosaminidase (NAGase) formation in Hypocrea atroviridis. Those data were compared with transcript patterns of nag1 and nag2, two genes encoding the NAGase Nag1 and Nag2, in the wild-type and an H. atroviridis Δnag1 strain (Seidl et al., 2006). No differences in the phenotype of these two strains were observed. As the result of the screening Δnag1 strain showed a strong reduction of NAGase activity on most carbon sources compared to the wild-type. A number of carbon sources that clearly enhanced NAGase activity were detected, mostly α-glucans, like glycogen, dextrin and maltotriose, and several oligosaccharides. The authors claimed that, since this kind of glycosidic linkage is present in the cell walls of the majority of ascomycetes, these carbon sources can be a part of a mechanism by which H. atroviridis senses the presence of a host cell wall containing chitin. Moreover, the comparison of NAGase formation between wild-type and mutant H. atroviridis strains showed that the reduction of this enzyme activity varied greatly among different carbon sources, and that the ratio was growth-phase dependant (Seidl et al., 2006).

Overall, the screening system of the combination of PMs and an enzymatic assay using a chromogene substrate established by Seidl et al. (2006) can be adapted for enzyme activity measurements of a variety of extracellular and cell wall-bound enzymes. Apparently, the differences in regulation between nag1 and nag2 would stay undetected without this tool. Seidl et al. (2006) propose this method as useful to monitor the expression of specific genes, even for encoding intracellular enzymes, by using appropriate promoter-fusion reporter systems.

2.4. Testing gene knockouts with expected phenotypes

DNA-mediated transformation is now a routine method applied to many fungi, and creation of knockout mutants by homologous integration, therefore, is a standard procedure. In order to obtain reliable knockouts, Southern analyses must be performed to confirm the homologous integration, and a regain of the phenotype of the parent strain by retransformation with the wild-type gene is also essential. Assessment of the wild-type and mutant phenotypes, however, is often laborious. PMs can provide a valuable tool here, as shown with an example of H. jecorina: we have previously investigated xylitol dehydrogenase gene (xdh1) for its possible involvement in D-xylose catabolism (Seiboth et al., 2003). Analysis of an xdh1-deletion mutant by PMs (Table 1) reveals confirmations that this mutant is partially defective in growth on L-arabinose and D-xylose, because its activity can be taken over by L-arabinitol dehydrogenase. PMs further show, however, that the mutant has impaired growth on D-sorbitol, whereas it has higher growths on erythritol and glycerol. The latter finding is not understood, but suggests an interplay of xylitol dehydrogenase with catabolism of shorter chain polyols, which needs to be taken into account when the physiology of these mutants is investigated.

Table 1.

Utilization of selected carbon sources by H. jecorina QM 9414 and Δxdh1 transformant strains

Carbon source	48 h			66 h
	QM 9414	Δxdh1	P value	QM 9414	Δxdh1	P value
L-arabinose	0.61	0.35	<0.01	1.14	0.46	<0.01
D-cellobiose	0.87	0.91	0.02	Not significant
i-erythritol	0.86	1.07	<0.01	1.08	1.16	0.04
Glycerol	0.44	0.66	0.03	0.74	1.04	<0.01
D-sorbitol	0.54	0.26	<0.01	0.88	0.31	<0.01
Xylitol	0.49	0.56	<0.01	Not significant
D-xylose	0.94	0.47	<0.01	0.94	0.75	<0.01

Two mutant cellulase-hyperproducing strains NG14 and its direct descendant RUT C30, initially originating from T. reesei QM 6a, were newly tested to identify mutations in the genomes that led to cellulase production (le Crom et al., 2009). These two strains were identified on the basis of their superior cellulase activity as well as resistance to 2-deoxyglucose in the presence of glycerol (Seidl et al., 2008). Biolog PMs were used to compare the mutants with their wild-type strain QM 6a. It was shown that utilization of carbon sources glucose, D-xylose, D-arabinitol, mannitol, and the β-linked disaccharides gentiobiose and cellobiose is correlated with increased cellulose production (le Crom et al., 2009). The latter two carbon sources might reflect enhanced β-glucosidase activity, yet the rest of utilized carbon sources are catabolite repressing compounds and at this point, le Crom et al. (2009) stressed that it is not clear whether this is a consequence of the cre1 mutation or it is due to the effect of other affected genes. Previously reported inverse carbon source utilization in correlation with enhanced cellulase production was detected by utilization of α-linked oligosaccharides and glycans (Seidl et al., 2008), and was complemented by the correlation between cellulase production and reduced growth on amino acids. le Crom et al. (2009) discussed this phenotype to be related to an ability of the superior cellulase production strains to use a higher portion of their amino acid pool for synthesis of secreted proteins versus growth.

2.5. Analysis of wide-domain regulatory processes

PMs are particularly well suited for the investigation of wide-domain regulatory processes, e.g., carbon catabolite regulation, due to the possibility of measuring growth on 96 different conditions simultaneously (carbon sources, nitrogen sources, stress inducers, pH control, and others). These factors have been intensively studied in A. nidulans, N. crassa, H. jecorina, and Acremonium chrysogenum (Bailey and Arst, 1975; Kudla et al., 1990; Dowzer and Kelly, 1991; Caddick et al., 1994; Strauss et al., 1995; Ilmén et al., 1996; Platt et al., 1996; Ravagnani et al., 1997; Marzluf, 1997; Wilson and Arst, 1998; Jekosch and Kück, 2000; Felenbok et al., 2001). In this type of experiment, growth of a parent strain and a respective mutant would be compared on every medium composition, which would reveal those particular nutrients whose utilization depends on the mutated wide-domain control gene.

2.5.1. Nitrogen repression

Tanzer et al. (2003) used this approach to investigate the effect of the areA-dependent nitrogen regulation on the utilization of various nitrogen sources by A. nidulans. The areA gene encodes a GATA family transcriptional activator mediating nitrogen metabolite repression, thus ensuring the utilization of ammonium and L-glutamine in preference to a variety of alternative nitrogen sources. Mutations in areA exhibit an extraordinarily diverse range of phenotypes when monitored for their effects on utilization of nitrogenous compounds. One of these alleles, areA-102, displays altered specificity of target promoter activation, which ultimately leads to loss-of-function, gain-of-function, or a wild-type phenotype based on its interaction with the target structural gene promoter (Kudla et al., 1990; Ravagnani et al., 1997). In comparison to wild-type strains on agar plates, strains carrying areA-102 show increased growth when urea, L-histidine, L-citrulline, L-aspartate, or L-glutamate is used as the sole nitrogen source and decreased growth when xanthine or uric acid is used as the sole nitrogen source (Arst and Cove, 1973; Arst and Scazzocchio, 1975; Hynes, 1975; Kudla et al., 1990). Results obtained with PMs were largely in agreement with these observations. In contrast to them, however, growth on L-glutamate or L-aspartate in the areA-102 mutant was not higher than that in the wild-type, and growth on xanthine was not detectable. On the other hand, assimilation of quite a number of nitrogen sources was not altered, indicating that they are not a subject to the AreA-mediated nitrogen regulatory system. An interesting finding from this study was that several D-amino acids (D-alanine, D-valine, and D-lysine) were utilized more efficiently by the areA-102 mutant strain than by the wild-type strain, suggesting that this areA mutation elevates D-amino acid oxidase levels. Consistent data were obtained with another allele (areA-30), which is a revertant allele that also shows altered specificity of target promoter activation.

Tanzer et al. (2003) further used PMs to perform a large scale analysis of 21 A. nidulans strains containing various areA mutant alleles. The data from all test wells for each strain were used for cluster analysis based on overall nitrogen source utilization patterns. The data provided excellent insight into the severeness of the individual mutations, e.g., mutant areA-1903 or areA-2173, which contain the conservative G698A substitution (Wilson and Arst, 1998), clustered with the wild-type strain, because these mutants produce no apparent effect in growth tests on solid medium. Other mutations, particularly those leading to a truncated AreA protein, showed much more serious differences. This work therefore beautifully illustrates the use of PM to screen for essential and dispensable mutations in a regulatory protein.

2.5.2. Carbon catabolite repression

Carbon catabolite repression is another wide-domain regulatory circuit amenable to investigation by PM. Its most observed phenotype is due to the creA/cre1 gene, which encodes a C₂H₂-zinc finger-containing transcriptional repressor (Dowzer and Kelly, 1991; Strauss et al., 1995; Ilmén et al., 1996). It ensures the utilization of glucose and some other carbon sources in preference to a large variety of other carbon compounds, which either yield less energy during assimilation (e.g., C₂-components), or require a considerable amount of energy for becoming assimilable (e.g., extracellular macromolecules, for which hydrolases must be exported via a secretory pathway).

To determine the utilization of which carbon sources is actually affected by the carbon catabolite repression, Seidl et al. (2008) investigated the differences in the growth of a cre1-truncated mutant strain of H. jecorina (RUT C30) (Ilmén et al., 1996) and the respective parent strain NG 14. The results (Fig. 5) show that the carbon catabolite-derepressed mutant strain displays a higher growth rate on D-arabitol, D-mannitol, D-fructose, D-trehalose, D-mannose, D-xylose, D-ribose, N-acetyl-D-glucosamine, γ-aminobutyric acid, and glycerol, indicating that all of them are under partial carbon catabolite repression, and that they all repress their own utilization to some extent. As expected, growth on glucose was not different and therefore it can serve as a control for these investigations. Interestingly, the cre1 mutation also results in significant changes in the other direction, e.g., the carbon catabolite-derepressed mutant had decreased the ability to grow on erythritol, dextrin, maltotriose, glycogen, and β-methyl glucoside. The biochemical nature of these observations is unclear, but indicates a dependence of carbon catabolite repression on assimilation of these carbon sources.

Fig. 5.

A phenetic map of carbon utilization patterns of H. jecorina QM 6a and the carbon catabolite-derepressed strain RUT C30 after the two-ways joining cluster analysis was applied to (i) carbon sources and (ii) fungal strains as two groups of variables. Only “fast” and “medium” carbon sources are shown. The respective growth (OD₇₅₀ after 48 h) is given by a corresponding color as indicated at the color scale

Seidl et al. (2008) also showed that the H. jecorina RUT C30 and its ancestor NG 14 lack an 85-kb genomic fragment, missing additional 29 genes comprising transcription factors, enzymes of the primary metabolism and transport proteins. They reported that these mutations are not linked to the cre1 locus. Performing PM analysis, it was concluded that a greatly reduced growth of RUT C30 on L-arabinose, L-erythritol and also D-galactose indicates that one of the aldo-/keto-reductases identified as lacking in this strain could be involved in polyol assimilation (Seidl et al., 2008).

Tanzer et al. (2003) used an alternative, indirect approach to examine the effect of carbon catabolite repression on carbon source utilization in A. nidulans, A. fumigatus, M. grisea, and M. graminicola. The approach was based on the CreA-dependent repression of alcohol dehydrogenase, which renders these strains unable to grow on derepressing carbon sources in the presence of allyl alcohol (Bailey and Arst, 1975; Felenbok et al., 2001). Obviously, this method can be used only with fungi that are able to use ethanol as a sole carbon source, which is not the case for others, e.g., H. jecorina (C.P. Kubicek, personal communication). The validity of the approach for the species investigated, however, arises from the fact that in all four fungal species, the addition of allyl alcohol to the media containing D-glucose had a little effect on the growth levels whereas it was strongly inhibited in media containing the derepressing carbon source L-lactic acid. A closer examination showed that the derepressing class of carbon compounds could be further subdivided by comparing growth in the presence of different concentrations of allyl alcohol. A. nidulans and A. fumigatus were generally more sensitive to allyl alcohol toxicity than M. grisea and M. graminicola, the latter two species requiring, respectively, 25- and 2.5-fold higher concentrations of allyl alcohol to reduce growth on derepressing carbon sources (Tanzer et al., 2003). These data also revealed contrasting results in different fungi, e.g., while dextrin was derepressing in the A. nidulans, it was repressing in A. fumigatus, M. grisea, and M. graminicola. Additionally, several other carbon sources such as maltose, D-sorbitol, and D-trehalose were more strongly repressing in A. fumigatus as compared to A. nidulans, and a much larger number of carbon sources were strongly repressing in M. grisea and M. graminicola, e.g., D-trehalose, lactose, and D-raffinose. The fact that Magnaporthe and Mycosphaerella are less sensitive to allyl alcohol and are derepressed by fewer compounds suggests that the plant pathogens are poised to utilize a number of compounds as carbon sources rapidly whereas the aspergilli are initially more selective and at the same time can use a wider range of carbon sources. Tanzer et al. (2003) concluded that the regulation of nutrient utilization may evolve to favour adaptation to the certain environmental niches and thus may provide insights into the nutrients typically encountered by these fungi.

Even though the allyl alcohol resistance system cannot be used for fungi like Hypocrea/Trichoderma, it is still amenable to this type of global analysis by other means: 2-desoxyglucose (2-DOG) has been used as a nonmetabolizable agent conferring carbon catabolite repression in yeast and filamentous fungi, because it accumulates as 2-desoxyglucose-6-phosphate (Franzusoff and Cirillo, 1982; Randez-Gil et al., 1995). Therefore, 2-DOG has been used for the isolation of carbon catabolite derepressed mutants in H. jecorina and Penicillium chrysogenum (Montenecourt and Eveleigh, 1979; Farkaš et al., 1981; Barredo et al., 1988). A drawback of the use of 2-DOG, however, is that it also interferes with energy metabolism (e.g., it depletes the cells of adenosine-triphosphate (ATP) because of storage of phosphate in 2-DOG-phosphate and slowing down glycolysis). In order to take that into consideration, the influence of 2-DOG is examined at several different sublethal concentrations.

Brunner et al. (2008) analyzed G protein-coupled receptors (GPCRs) in a manner to learn more about the G protein signaling. With the help of in silico exploration of the genome database of the close relative T. reesei, four H. atroviridis GPCR-encoding genes were isolated and affiliated to the cyclic adenosine monophosphate (cAMP) receptor-like family. Using Biolog PMs, it was shown that gpr-1 silenced mutants exhibited reduced growth on several carbon sources, e.g., glucose and glycerol growth of the mutants was reduced by 70%. Furthermore, in comparison to the parental strain, the mutant was not able to grow on lactose and its isomerization product lactulose, and also showed scarce growth on maltose and raffinose. There were also some carbon sources that showed similar growth rate at both strains, indicating that the mutant’s reduced growth is carbon source-dependent and generally ineffective. Overall, the characterization of the H. atroviridis Gpr-1 revealed that the GPCR influences conidiation, conidial germination and vegetative growth on many carbon sources (Brunner et al., 2008). The PM method aids to the exploration of the fundamental role of G protein signaling in Trichoderma.

2.6. Application of PMs in fungal photobiology

Contrary to the initial expectation, Friedl et al. (2008a) revealed that conidiation in Hypocrea atroviridis is primarily carbon source-dependent and that illumination plays a catalytic role in this process. They applied PMs to a wild-type strain of H. atroviridis as well as two mutant strains, with each the loss of function of the blue light regulators 1 and 2 (BLR-1 and BLR-2), respectively, responsible for stimulation of transcription activation. In the mutant strains Δblr-1 and Δblr-2, conidiation in the darkness was significantly decreased on most carbon sources. Except for strong conidiation of Δblr-1 on D-sorbitol and Δblr-2 on glucuronic acid, the carbon sources that stimulated low conidiation in the mutant strains were mostly from the pool of those that contribute to strong conidiation in the parent strain (Friedl et al., 2008a). While observing this strongly impaired sporulation, it is notable that the mutations do not completely block sporulation but only greatly decrease it, and they still show (light insensitive) sporulation on some carbon sources (notably glucuronic acid and 2-ketogluconic acid). Friedl et al. (2008a) therefore hypothesized that the observed differences in levels of conidiation on different carbon sources are due to different redox potentials in the cell during growth. Thus, BLR proteins are not only involved in the photosensing, but also act as redox and oxygen sensors.

Friedl et al. (2008a) also concluded that the addition of cAMP induces the sporulation in darkness, as previously reported by Casas-Flores et al. (2004), but also showed that effect of cAMP is carbon source-dependent. The addition of cAMP induced conidiation in the parent and both blr mutant strains on some, but not all, carbon sources. Thus, the carbon source dependence of mutant strains was different from that of the parent strain. Only a small set of carbohydrates, polyols, and sugar acids allowed conidiation in darkness, whereas for most of them conidiation was strongly expressed in light. Moreover, under every condition tested (darkness, light, addition of cAMP, the loss of function of BLR-1/BLR-2) spore formation occurred only on carbohydrates and their derivates (polyols and sugar acids). Yet, the presence of light always enabled the sporulation only on those carbon sources which also empowered conidiation in darkness by increasing the speed or intensity of sporulation (Friedl et al., 2008a). There were a number of carbon sources (e.g., rhamnose, glucosamine and xylitol) detected where light did not further stimulate the spore formation.

Furthermore, Friedl et al. (2008b) showed that enhancement of mycelia growth on various carbon sources is highly light-dependent. Carbon source utilization by PMs showed that Hypocrea atroviridis growth is light-stimulated on 17 of 95 carbon sources, which are mainly metabolically related to cellulose and hemicelluloses and can be found in the upper soil litter layer (Friedl et al., 2008b). The light stimulation depends on the function of two blue light receptor proteins BLR-1 and BLR-2. Utilization of carbon sources in darkness was nearly the same in both knock-out mutants (Δblr-1 and Δblr-2), and it was similar to that of the parent strain. But comparing carbon utilization profiles to those exposed to rhythmic and constant illumination, Friedl et al. (2008b) concluded that the mutant strains are not photostimulated, but in contrast are inhibited by light, especially in Δblr-2. Evocation of oxidative stress response in darkness imitates the photostimulation on 9 of the mentioned 17 carbon sources, and this effect is fully dependent on the function of BLR-1. The resistance of both blr mutants to conditions of oxidative stress is therefore carbon source-dependent and is notably weaker compared to the wild-type strain. Friedl et al. (2008b) concluded that light in combination with the availability of litter-specific carbon sources serves as a signal for the fungus to be above ground, thereby stimulating fast growth in order to produce a maximum of propagules in the shortest time. Furthermore, they deduced that this process involves oxidative stress response and the two blue light receptor proteins BLR-1 and BLR-2, the former playing the major role.

To determine the role of the light regulatory protein ENVOY, Schuster et al. (2007) used PMs as an aid to their global screening for genes, which are specifically affected by light in the fungus H. jecorina, an important producer of cellulases and hemicellulases. The expression of its cellulase genes is partially susceptible to carbon catabolite repression. They showed that ENVOY acts as a light-independent repressor for several genes and it is crucial for normal growth in light on several carbon sources, but is not able to fully execute its regulatory function when overexpressed in darkness. Higher growth rate of the H. jecorina QM 9414 than that of the mutant strain env1^PAS− measured in the light indicated that the growth rate is dependent on ENVOY since no such stimulation was observed in the mutant strain (Schuster et al., 2007). Moreover, Schuster et al. (2007) found that, with the exception of growth on γ-aminobutyric acid, mutant strain growth rate is always lower in light than in the dark. They also reported that the growth rates of two strains compared separately in light and darkness differ significantly more in the light, indicating that light inhibits growth of strain env1^PAS−. On the other hand, it was shown that growth in darkness (with the exception of glycerol) was poorly affected. They claimed an env1-dependent enhancement of energy metabolism, and thus biomass formation by light as well as a negative effect of light on H. jecorina in the absence of functional ENVOY. The inhibitory effect of light in the absence of ENVOY is carbon source-dependent since the inhibition by light in the mutant strain was not observed on all carbon sources (Schuster et al., 2007).

2.7. Development and analysis of action of chemical inhibitors

Many fungi are pathogens of plants, animals or human beings. Therefore, screening for novel chemicals capable of inhibiting these species is an important issue (Vitale et al., 2005). This is also of particular significance to the increasing number of cases in which non-human-pathogenic fungi such as Trichoderma or Fusarium have caused fatalities in immunocompromised patients (Yeo and Wong, 2002; Sampathkumar and Paya, 2001). Many of these isolates proved to be resistant to the common clinical fungicides such as fluconazole and 5-fluorocytosine (5FC), amphotericin B, tetraconazole and ketoconazole (Ragnaud et al., 1984; Campos-Herrero et al., 1996; Tanis et al., 1995; Guarro et al., 1999; Hennequin et al., 2000; Antal et al., 2002; Avery, 2004). Thus, the development of new agents is essential. This development also must accompany the identification of action of the compound, for which PM offers an attractive possibility. To show the utility of PM for chemical inhibitor analysis, Tanzer et al. (2003) compared the responses of A. nidulans and M. grisea to the addition of the growth inhibitor glufosinate. Glufosinate (phosphinothricin) is a herbicidal compound that has also been shown to inhibit animal and fungal growth (Avalos et al., 1989; Pall, 1993; Kutlesa and Caveney, 2001). It is an analogue of L-glutamate that inhibits glutamine synthetase, thereby preventing the synthesis of glutamine and consequently the synthesis of numerous other nitrogenous compounds required for growth. Based on its mode-of-action, the addition of glutamine should abrogate the inhibitory effects of glufosinate by bypassing the need for glutamine synthetase. Tanzer et al. (2003) tested the effect of glufosinate on growth of A. nidulans and M. grisea in the presence of different nitrogenous compounds. A. nidulans showed near wild-type levels of growth on glutamine and glutamate, which successfully competes for the inhibitor, but no growth when ammonium, nitrate or nitrite was used as the nitrogen source. Thus, Tanzer et al. (2003) showed that L-aspartate was not an effective competitor and glutamine-containing dipeptides were apparently unable to supply sufficient glutamine to overcome the toxicity. M. grisea was able to grow normally in the presence of L-glutamine or L-glutamate, while growth was inhibited when ammonium, nitrate or nitrite was present as the sole nitrogen source. In contrast to A. nidulans, M. grisea was able to utilize L-aspartate or the glutamine-containing dipeptides as the sole nitrogen source in the presence of glufosinate, thus demonstrating metabolic differences between fungi, differences potentially relevant to the action and screening for inhibitory compounds.

PM also offers an additional advantage for eventual identification of new inhibitory components. Looking at the nutrient source utilization profile, it is evident that for every fungus or strain, there is a cluster of nutrients on which the organism grows even slower than on water and a number of nutrients enabling essentially the same growth rate as water. The observation of growth on water may be mostly due to the utilization of reserve carbohydrates and polyols present in the conidia (d′Enfert et al., 1999). Consequently, those nutrients, which result in an even slower growth, are likely inhibitors, at least, of the spore germination process. Examples of this are D-lactic acid methyl ester and glucuronamide for H. jecorina, which exhibit a statistically significant inhibitory effect (Druzhinina et al., 2006), and hydroxylamine for A. nidulans (Tanzer et al., 2003). While there may not be too much interest in developing inhibitory agents for H. jecorina and A. nidulans, similar strategies could be used to screen for biodegradable inhibitors against pathogenic plant and animal fungi.

Gardiner et al. (2009) used PMs to assess the role of different carbon and nitrogen sources that could be crucial for regulation of trichothecene synthesis by a plant pathogenic fungus, Fusarium graminearum, which causes a head blight disease of wheat. Presence of these toxins has important consequences for human and animal health. Gardiner et al. (2009) investigated the application of PMs in combination with a transgenic strain of F. graminearum in which the promoter from the TRI5 trichothecene biosynthesis gene was fused to green fluorescent protein (GFP). Using this strain in PM profiling, a variety of amines were identified, which significantly induce TRI5 expression.

2.8. Global inducer screening

The increase in available fungal genome sequences has also led to an increase in the identification of genes for which orthologues in even closely related species are unknown (Dogra and Breuil, 2004; OBrian et al., 2003; Schmoll et al., 2004). As explained in the introduction, studies of the genome-wide role of these genes are still limited to the use of mutants and their use in DNA-microarray analysis and proteomics (Sims et al., 2004). Identifying the physiological conditions under which a newly identified gene is expressed, however, is still restricted to the analysis of its transcript formation in individual experiments. Having an array-type system available for this purpose would greatly facilitate and accelerate the assignment of a function to new genes.

Our interest in this area comes from studies on chitinase formation by Hypocrea atroviridis (Mach et al., 1999; Brunner et al., 2003; Seidl et al., 2005). While there are several reports of chitinase gene expression in fungi in response to the availability of chitin or chitin degradation products (Carsolio et al., 1999; Mach et al., 1999; Donzelli and Harman, 2001; Kim et al., 2002), not all chitinases follow this rule. We have recently identified 18 chitinases in H. jecorina, and several of them are in fact not induced by chitin (Seidl et al., 2005). In order to find conditions leading to their induction, we have constructed recombinant strains containing various promoter-reporter fusions, inoculated them on PM plates, and photometricaly quantified the formation of the reporter enzyme. In our study, we used the A. niger glucose oxidase encoding (goxA) gene, for which an orthologue is absent in Hypocrea, because the enzyme is secreted into the medium and can therefore be assayed without cell breakage (Mach et al., 1999). The glucose oxidase assay involves formation of a green color, however, which is difficult to measure in the presence of the violet formazan dye formed by the redox dye implemented in the Biolog FF MicroPlates. Therefore, other non-invasive strategies such as the use of various forms of the green fluorescent protein or the luciferase (March et al., 2003; Arnone et al., 2004; Piwnica-Worms et al., 2004; Gu et al., 2004) may be alternatives. Nevertheless, the use of the goxA reporter enables us to identify new chitinase inducers in Hypocrea/Trichoderma (Verena Seidl, personal communication).

3. Concluding remarks and outlook

In this article, we have reviewed recent applications of PM technology as an accessory tool for global and genome-wide investigations on fungi, and reported on early achievements in the area. Although, not all the potential uses of this technology have already been experimentally verified, we anticipate that this broad phenotypic testing will become a simple standard practice once the method becomes more familiar to scientists working with industrially important or pathogenic fungi. For example, in the past it was impractical to assay more than a few phenotypes and therefore it was impossible to measure phenotypic drift. One still untested use of PM technology is therefore the ability to provide researchers with a practical means to characterize each fungal isolate. To this end, the establishment of a public database of carbon and nitrogen utilization profiles for fungi (wild-type strains and mutants) would be an important instrument in understanding the comparative biochemistry and physiology of these organisms, and aid in the selection of appropriate strains if new products or strain properties are desired.

An advantage of the PM system is that it is much easier to perform than gene expression analysis. It involves only the preparation of a standardized fungal inoculum and its pipetting into the microplates of the standard PM plate. By contrast, to obtain specific and useful data on genes, one needs to perform genetic engineering work to create the strains to be analyzed. At this moment, the commercial availability of a sufficient number of nutrient conditions for fungi is still limited. In theory, one can devise hundreds or thousands of growth conditions in which specific biological pathways or physiological functions are linked to hyphal growth, but only about 2500 different assays are available for bacteria and yeasts, and only half are available for fungi.

4. Acknowledgement

We acknowledge Christian P. Kubicek (Vienna University of Technology, Austria) for the critical reading and helpful discussions on the manuscript.

Biographies

Introducing editorial board member: Irina S. Druzhinina, the corresponding author of this invited review, is an editorial board member of Journal of Zhejiang University-SCIENCE B (Biomedicine & Biotechnology). Her study is focused on phylogeny and molecular ecology of Hypocreales fungi and on the development of bioinformatic tools for molecular identification of microbes. In the other line of research she is interested in metabolomics of filamentous fungi applying Phenotype MicroArray techniques. In 2001, she gained the doctorate of natural sciences of University of Vienna (Austria). Shortly after, she founded a research group on fungal evolution and functional biodiversity in Vienna University of Technology (Austria) and obtained a position of Assistant Professor in Department of Applied Biochemistry and Gene Technology.

Lea Atanasova, the first author of this invited review, studied biology and ecology at the University of Ljubljana (Slovenia) and has graduated from the University Vienna (Austria). She was awarded for her master thesis by the Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management as well as by Ecological Society of Germany, Austria and Switzerland. Soon after graduation, she joined the Irina S. Druzhinina’s research group of fungal biodiversity and evolution in Vienna University of Technology (Austria) and started to work on her PhD thesis dedicated to the molecular ecology of soil mycobiota.

Appendix

Fig. 1A.

Carbon sources of Biolog FF MicroPlate

Numbers indicate physiological and biochemical groups: 0: control; 1: monosaccharides: 1.1 heptose, 1.2 hexoses, 1.3 pentoses; 2: monosaccharide-related compounds: 2.1 sugar acids, 2.2 hexosamines, 2.3 polyols; 3: other sugars: 3.1 poly-saccharides, 3.2 oligosaccharides, 3.3 glucosides; 4: nitrogen-containing compounds: 4.1 peptides, 4.2 L-amino acids, 4.3 biogene and heterocyclic amines, 4.4 TCA-cycle intermediates, 4.5 aliphatic organic acids; 5: others