B. PESTS AND DISEASES

11. Genomics of two banana pathogens: genetic diversity, diagnostics, and phylogeny of Mycosphaerella fijiensis and M. Musicola - Molina, C.M.^[13],[14], G. Kahl^[15]

Abstract

In the last few decades, the generation of informative banana and plantain DNA markers has been focused to catalyse progress in Musa breeding and germplasm characterisation. Important knowledge has been gained about the development of varieties resistant to various biotic and abiotic stresses, the identification of the A and B genomes, and the diagnosis of varieties and cultivars and their levels of heterozygosity. In addition, useful genes and promoters have been identified for the effective transformation of target cultivars. Despite all these achievements, intensive research has not been done on the most important fungal banana pathogens, Mycosphaerella fijiensis and M. musicola, which are still major causes of loss in banana-growing regions worldwide. The direct comparison of their population structure is expected to lead to a better understanding of the extent of their genetic diversity. Our initial efforts focused on the development of polymorphic locus-specific SSR markers for M. fijiensis and M. musicola. This technique, along with other PCR-based DNA fingerprinting techniques such as DNA Amplification Fingerprinting (DAF) and Selective Amplification of Microsatellite Polymorphic Loci (SAMPL), allowed us to carry out a comparative survey. Here we discuss the differences in genetic diversity and structure of both pathogen populations as detected by the different techniques, and the distribution of genetic diversity at regional and local levels in Central America and northern South America.

1. INTRODUCTION

The last decade has witnessed a series of advances in banana and plantain research that are expected to catalyse progress in a major field of banana breeding: the development of Musa varieties resistant to various biotic and abiotic stresses. The contributions of molecular biologists to this advancement included the generation of informative DNA markers for the identification of the A and B genomes, the detection of genetic diversity in germplasm, the diagnosis of varieties and cultivars and levels of heterozygosity, the origin of the B genome in hybrids, the establishment of genetic maps [1] and the construction of a bacterial artificial chromosome (BAC) library with approximately fivefold genome coverage. Useful genes and promoters have also become available for the effective transformation of target cultivars using, for example, specific Agrobacterium tumefaciens strains and optimised protocols for regeneration of embryogenic cell suspension cultures. Transgenic plants have been tested for the presence and expression of foreign genes [2,3]. Transgenic bananas expressing vaccines against human intestinal pathogens show potential [4]. Also, expression levels of a series of genes involved in the fruit-ripening process were determined [5]. Finally, the nucleome of Musa has now been characterised in more detail, using flow cytometry and fluorescence in situ hybridisation (FISH; [6]). Telomeres of the Arabidopsis type, several gene clusters and retrotransposons have been localized, and the ploidy of any banana material can be estimated unequivocally [7,8]. Notwithstanding all these (and other) achievements, a bias towards the host plant becomes obvious: a similarly intensive research activity with comparable results is lacking for the most important banana pathogens, especially the fungal pathogens. Although several laboratories are devoting their work to unravelling, for example, the infective power and potential of both pathogenic Mycosphaerellas, and though the results of this work over two decades are impressive, we lack a cooperative, dedicated, constant and engaged research effort on Mycosphaerella, especially since both ascomycetes, M. fijiensis and M. musicola, still are the major troublemakers in most banana-growing regions of the world (though Fusarium oxysporum f.sp. cubense, FOC, can also be included). Whereas the collection of isolates and their culture is routine, their genetic and pathological characterization is less common worldwide. The physiology of both fungi is more or less reduced to toxins, and the genetic make-up of Mycosphaerella is virtually unknown. For these reasons we have switched exclusively to Mycosphaerella fijiensis and M. musicola, after having isolated a series of Resistance Gene Analogues (RGAs) from banana, developed a series of informative microsatellite markers for Musa and employing them for the characterization of genetic diversity in wild and cultivated Musa [9,10].

Our initial research has focused on the development of polymorphic, locus-specific microsatellite markers for M. fijiensis [11] and M. musicola [12]. We have used these highly informative markers to study population structure and its dynamics in both pathogens. The comparison of the two pathogen populations is expected to lead to a better understanding of the extent of genetic diversity and genetic differentiation at local and regional levels, as well as the influence of environmental pressures on their spread to potentially new colonisation sites. Moreover, it could also predict the behaviour of new epidemic forms, thus helping breeders and farmers with basic information on genetic diversity, population structure, and fungal dynamics.

To our knowledge, only one study has assessed the genetic diversity of both pathogens in parallel, but attempts to transfer molecular markers from M. fijiensis to M. musicola and vice versa were not successful [13]. However, the polymorphic SSR markers developed for M. fijiensis and M. musicola in the course of this project, along with other PCR-based DNA fingerprinting techniques such as DNA Amplification Fingerprinting (DAF) and Selective Amplification of Microsatellite Polymorphic Loci (SAMPL), have allowed us to carry out a truly comparative survey of the genetic diversity of both pathogens.

Here we report the results of such a study with populations from different Latin American countries. We discuss (a) differences in genetic diversity and structure of the two populations as detected with different techniques, (b) the distribution of genetic diversity at regional and local levels in Central America and northern South America, and (c) evidence for isolation by geographical barriers and distances.

2. MATERIALS AND METHODS

2.1. Construction of microsatellite-enriched libraries

Microsatellite-enriched libraries were constructed using a slightly modified method following the affinity capture protocol of Fischer & Bachmann [14] (Figure 1).

2.1.1. Isolation of high molecular weight DNA

A single-ascospore culture of the isolate MmCol-LM9.5.1, collected in a mid-altitude region of Colombia severely affected by yellow Sigatoka, was used as source material for the construction of a genomic library. DNA was isolated according to Weising et al. [15] and purified by caesium chloride gradient centrifugation.

2.1.2. Digestion and adapter ligation

Initial restrictions were separately performed in a volume of 25 µl containing 6 mg of DNA template, 10 units of Rsa I and Alu I (MBI Fermentas), 33 mM Tris-acetate (pH 7.9), 10 mM magnesium acetate, 66 mM potassium acetate, and 0.1 ng/ml BSA. After 3 h incubation at 37ºC, 10 ml of a ligation mix containing 5.5 ng of a 21-mer enrichment adapter, 10 pmol ATP, 33 mM Tris-acetate (pH 7.9), 10 mM magnesium acetate, 66 mM potassium acetate, 0.1 ng/ml BSA and 2 U T4 DNA ligase (MBI Fermentas) was added to the restriction reaction. Ligation-restriction reactions were incubated overnight at 37ºC.

The sequence of the 21-mer adapter was:

5'-CTCTTGCTTACGCGTGGACAT-3'
3'-GAGAACGAATGCGCACCTGTA-5'

Figure 1 Steps involved in the construction of microsatellite-enriched DNA libraries

2.1.3. Capture of the restricted DNA template with biotinylated probes

Pools of either biotinylated trinucleotide [(CAA)₈ and (GAA)₈] or dinucleotide [(GA)₁₀, (CA)₁₀ and (TA)₁₀] repeats were used. Capture reactions were performed by mixing 10 pmol of each biotinylated probe and 6 mg of DNA template, and incubating the mixture at probe-specific annealing temperatures (T_A): 74ºC for (CA)₁₀, 64ºC for (GA)₁₀, and 75ºC for (CAA)₈ and (GAA)₈. To capture the annealed probes, 350 mg of M-280 Streptavidin-Dynabeads (Dynal) were added to the reaction in a final volume of 100 ml. After incubation at the specific T_A for a further 20 min, a magnetic particle capturer (Nunc) was used to separate the probe/target hybrids from the solution. For final elution of the hybrids the magnetic beads were captured, and 20µl elution buffer (0.2M HCl, 0.1M NaOH, 0.1M NaCl. and 0.1M Tris-HCl, pH 7.4) were added to break the complex between magnetic beads and fragments (in solution) were stored for subsequent steps.

2.1.4. Amplification of captured fragments

Amplification of captured fragments was performed in 50 ml reactions containing 2 ml of eluted DNA template, 1 mM of 21-mer adapter primer, 75 mM Tris-HCl (pH 9), 20 mM (NH₄)₂SO₄, 0.01% Tween-20, 1.5 mM MgCl₂, 0.2 mM each of dATP, dCTP, dGTP and dTTP, and 0.4 U Taq-DNA-Polymerase (Biotherm, Gene Craft). PCR was performed in a Perkin Elmer 9700 thermocycler with an initial denaturation of 95°C (180 s), 28 cycles at 95°C (48 s), 56°C (60 s), 72°C (60 s), and a final elongation at 72°C (3 min). Amplified products were run on a 1.5% agarose gel and stained with ethidium bromide.

2.1.5. Cloning of microsatellite enriched fragments

DNA fragments obtained after the enrichment procedure were ligated into pGEM-T (Promega). Ligation reactions (final volume 10 ml) were prepared in 1 × Rapid Ligation Buffer (Promega) using three units of T4 DNA-ligase per microlitre of purified PCR product, and were incubated at 4ºC overnight. Fragments were transformed into competent E. coli DH-5a (Promega) by a modified heat-shock protocol [16]. Bacterial clones were plated, and amplified by colony PCR using vector-specific primers. Reactions were performed in 50 ml final volume, containing 5 ml of bacterial lysate, 0.5 mM of each forward and reverse primer, 75 mM Tris-HCl (pH 9), 20 mM (NH₄)₂SO₄, 0.01% Tween-20, 1.5 mM MgCl₂, 0.2 mM each of dATP, dCTP, dGTP and dTTP, and 0.4 U Taq-DNA polymerase (Biotherm, Gene Craft). PCR was performed in a Perkin Elmer 9700 thermocycler with an initial denaturation of 95°C (60 s), 30 cycles of 95°C (45 s), 56°C (45 s), 72°C (60 s), and a final elongation at 72°C (5 min). PCR products were run on 1.5% agarose gels and stained with ethidium bromide.

2.1.6. Detection of positive clones

Positive clones were verified by Southern blot analysis with radiolabelled, microsatellite-specific oligonucleotide probes as described by Weising & Kahl [17]. Labelling of microsatellite probes was performed with 10 pmol of 16-mer repeat oligos (e.g. [CA]₈), 5 ml of [a³²P]-dATP (10 mCi/ml), 1 × polynucleotide kinase buffer B (MBI Fermentas) and 5 U of polynucleotide kinase). Labelling reactions were incubated at 37ºC for 30 min, and then heated at 70ºC for 10 min. Labelled probes were purified using Spin Chromatography Columns (Biorad) according to the manufacturer's instructions. Nylon membranes were treated with 6 × SSC and pre-hybridised for 5 h in pre-hybridisation buffer (5 × SSPE, 5 × Denhardts, 10 mg/ml E. coli DNA, 0.1% SDS). Hybridisations were performed overnight after adding 500 ml of purified labelled probe to the pre-hybridisation reaction. Clones giving positive signals were sequenced using the chain termination BigDye kit (PE Biosystems), and products were analysed on an ABI PRISM 373 automated sequencer. Microsatellite-flanking primers were designed with the computer program Primer3 [18].

Figure 2 (a) Latin American sampling sites of Mycosphaerella fijiensis (a) and M. musicola (b) populations used in the present study.

Figure 2 (b) African sampling sites of Mycosphaerella fijiensis populations used in the present study

Table 1a Populations of M. musicola genotyped with three different molecular marker techniques: DAF, SAMPL, and STMS

Table 1b Populations of M. fijiensis genotyped with three different molecular marker techniques: DAF, SAMPL, and STMS

Designed primers were tested for their Polymorphism Information Content (PIC) in Mycosphaerella fijiensis and M. musicola, using two representative sets containing single ascospore isolates collected from plant material of affected areas in Latin America, Africa and Asia. PCR was performed in a Perkin Elmer 9700 thermocycler with an initial denaturation at 95ºC (60 s), 30 cycles at 95ºC (30 s), 45-55ºC (45 s), and 72ºC (45 s), and a final elongation at 72ºC (7 min). Amplifications were performed in 10 µl reactions containing 5 ng of genomic DNA template, 0.5 µM each of forward and reverse primer, 75 mM Tris-HCl (pH 9), 20 mM (NH₄)₂SO₄, 0.01% Tween-20, 1.5 mM MgCl₂, 0.2 mM each of dCTP, dGTP and dTTP, 0.02 mM dATP, 0.065 µl of [a³²P]-dATP, and 0.4 U Taq-DNA polymerase (Biotherm, Gene Craft). Products were separated on 6% sequencing gels and autoradiographed [16].

2.2. Analysis of population structure at continental, regional and local levels

Developed STMS primers were used to genotype the M. fijiensis and M. musicola populations according to the conditions cited above. The different populations were hierarchically collected in different Latin American countries and West Africa as shown in Figure 2a and b. The individual isolates and their origins, together with some characteristics, are listed in Table 1a and b. Two fingerprinting techniques (Selective Amplification of Microsatellite Polymorphic Loci [SAMPL], and DNA Amplification Fingerprinting [DAF]) were optimised to increase the coverage of the analysis and to compare the efficiency of the different techniques for analysis of fungal populations. The original DAF protocol was optimised for filamentous fungi, based on the conditions reported by Caetano-Anolles et al. [19]. The SAMPL protocol was optimised for M. fijiensis and M. musicola, modifying a procedure reported by Paglia et al. [20].

2.3. Data analysis

Matrices indicating absence or presence of characters were constructed for all scored DAF and SAMPL loci. For STMS allelic data, alleles were scored using the M13 sequencing reaction to size fragments. The NTSYS software package [21] was employed to calculate DICE similarity coefficient matrixes, and the corresponding dendrograms grouped by the UPGMA agglomerative method. Genetic diversity indices for each locus in all populations [22] were calculated with POPGENE 1.21 [23]. Mantel tests to find correlations between the data matrices generated by the three different marker techniques were performed with the program Mantel 2.0 [24].

3. RESULTS

3.1. Polymorphic microsatellite markers for the banana pathogens Mycosphaerella fijiensis and M. musicola

After screening and sequencing the microsatellite-enriched libraries constructed for M. fijiensis and M. musicola, 23 STMS primer pairs were designed for M. fijiensis, and 48 corresponding markers were developed for M. musicola. After a PCR-amplification test and evaluation of the polymorphism information content (PIC) for each microsatellite locus, 11 and 26 microsatellite markers for M. fijiensis and M. musicola, respectively, yielded more than one allele in the testing population and produced a single identifiable band per assay (Figure 3a and b). Primer sequences and expected product sizes are shown in Tables 2a and b.

Figure 3 High levels of allelic resolution in two collections of Mycosphaerella fijiensis and M. musicola, respectively, from single fields in regions severely affected by Black and Yellow Sigatoka disease in Colombia. (a) M. fijiensis STMS alleles separated in a 6% polyacrylamide sequencing gel (isolates collected in a Colombian seaside commercial plantation). (b) M. musicola STMS alleles separated in a 6% polyacrylamide sequencing gel (isolates collected from a small holder banana plantation in the eastern Colombian Andes. M13: M13 phage sequencing reaction used as DNA size marker (A-C-G-T) - M. fijiensis, Fundación, Colombia

Figure 3 High levels of allelic resolution in two collections of Mycosphaerella fijiensis and M. musicola, respectively, from single fields in regions severely affected by Black and Yellow Sigatoka disease in Colombia. (a) M. fijiensis STMS alleles separated in a 6% polyacrylamide sequencing gel (isolates collected in a Colombian seaside commercial plantation). (b) M. musicola STMS alleles separated in a 6% polyacrylamide sequencing gel (isolates collected from a small holder banana plantation in the eastern Colombian Andes. M13: M13 phage sequencing reaction used as DNA size marker (A-C-G-T) - M. musicola, Cundinamarca, Colombia

Table 2a Some characteristics of polymorphic microsatellite loci of Mycosphaerella musicola

Table 2b Some characteristics of polymorphic microsatellite loci of Mycosphaerella fijiensis

Expected heterozygosities (H_E) varied from 0.09 to 0.71 in M. musicola, and from 0.25 to 0.60 in M. fijiensis. The maximum number of alleles for M. musicola was obtained with the marker Mm-SSR-07 (six alleles, H_E: 0.77), which contains a long dinucleotide repeat [(CA)₅₀], and Mm-SSR-46 (six alleles, H_E: 0.71), which amplifies the compound microsatellite [(CT)_n,(GT)_n]. As far as the motif composition is concerned, compound microsatellites were more variable than single motif repeats. If the number of nucleotides per repeat is considered, then trinucleotide repeats proved to be more variable than dinucleotide microsatellites, although the former were in a minority. Markers containing long repeats were hypervariable, whereas markers with short repeats had a lower number of alleles.

Concerning the distribution of alleles within the test population sets, markers can be catalogued in different categories: (a) slightly informative, where markers have two or three alleles unequally distributed in the population, (b) moderately informative, containing markers that amplify two or three alleles with equal distribution within the population, and (c) highly informative, which contains more than three alleles, yielding higher H_E values. Tests for differentiation of individuals were performed with the DICE Similarity Coefficient in each test population, and a minimum number of nine markers was needed to discriminate all the individuals after construction of UPGMA-clustered dendrograms.

Based on Costa Rican, Colombian and Venezuelan populations, nine M. fijiensis and ten M. musicola primer pairs, respectively, yielded an average of 3.40 and 4.0 alleles per locus, corresponding to a percentage of polymorphic loci of 80.0% and 90.9%, respectively. Expected heterozygosity values ranged between 0.26 and 0.76 in M. fijiensis, and between 0.06 and 0.83 in M. musicola. Whereas the markers proved to be highly informative for the species for which they were developed, they were of limited value if transferred to another species. For example, M. fijiensis STMS marker Mf-SSR-061 resulted in monomorphic (i.e. non-informative) patterns when used to type M. musicola populations. The same was true for M. musicola STMS marker Mm-SSR-024 on M. fijiensis isolates.

3.2. Typing of populations using DAF and SAMPL techniques

Applying the DAF technique to Colombian and Costa Rican populations, a total of 129 and 118 bands were detected for Mycosphaerella fijiensis and M. musicola, respectively. Thirty-seven percent of the total bands were polymorphic for M. fijiensis (4.4 polymorphic bands per primer), and 40% were polymorphic for M. musicola populations (4.7 polymorphic bands per primer). At the interspecific level, the DAF technique showed sufficient resolution to generate distinctive banding patterns for M. musicola and M. fijiensis, which are shown in Figure 4a.

Figure 4 Resolving power of two PCR-based DNA fingerprinting techniques, exemplified by different species of Mycosphaerella. (a) DNA Amplification Fingerprinting (DAF) patterns (1.5% agarose, ethidium bromide fluorescence).

Figure 4 Resolving power of two PCR-based DNA fingerprinting techniques, exemplified by different species of Mycosphaerella. (b) Selective Amplification of Microsatellite Polymorphic Loci (SAMPL) patterns (4% polyacrylamide sequencing gel, autoradiograph). Three different banding patterns (1-3) could be distinguished by both techniques, each pattern being characteristic for a particular species. M, 100 base-pair ladder; 1, M. musicola; 2, M. fijiensis; 3, Mycosphaerella sp. (unknown)

Using SAMPL, 47% and 45% of polymorphic bands were obtained for M. fijiensis and M. musicola, respectively. The percentage of polymorphism is close to or lower than the proportion seen with DAF, but the chance of obtaining more polymorphic bands per assay is higher due to the high number of total bands per assay (40 to 60). As was already observed for the DAF technique, the high resolution of the SAMPL technique generates distinctive banding patterns for both species (Figure 4b). Although the two species are considered to be closely related, the observed proportion of shared bands was rather low.

3.3. Grouping of individuals according to DICE Similarity Coefficient

At the intercontinental level (Nigerian versus Mexican populations), Dice Similarity Coefficients completely separated the African from the American isolates (Peraza Echeverria et al., in preparation). At the regional level (Colombian and Costa Rican M. fijiensis populations), the clusters containing individuals from Sta. Marta, Fundación, Caldas and Central Valley were consistent with their geographical origin (Figure 5a). For the Colombian, Venezuelan and Costa Rican M. musicola populations, major clusters corresponding to sampling sites were also observed (Figure 5b).

3.4. Population differentiation and genetic diversity

Increasingly positive F_ST values indicate increasing population differentiation. At the regional level (comparing Costa Rican with Colombian populations), the highest pairwise F_ST values were found between Caldas and North Colombia, and Caldas and Costa Rica (average F_ST: 0.29) Differences between Santa Marta, Fundación and Costa Rican populations averaged an F_ST of 0.23. For M. musicola, comparisons between Venezuelan and Colombian (La Mesa and Cachipay), and Venezuelan and Costa Rican populations averaged an F_ST value of 0.43, whereas the lowest value was obtained within the Colombian populations of La Mesa and Cachipay (average F_ST: 0.23). The values resulting from comparisons between the Colombian and Costa Rican populations remained intermediate.

STMS markers of M. fijiensis showed a gene diversity (H) value of 0.42 for all populations. Considering the populations individually, the Costa Rican population showed the highest diversity with an H value of 0.45, whereas the population of Sta. Marta had the lowest index (H: 0.13). Comparable results were observed with the SAMPL and DAF techniques, which also detected the highest diversity index in the Central Valley population (Costa Rica). In the M. musicola populations, H values based upon data from STMS, SAMPL and DAF were: 0.14, 0.15, 0.10, and 0.18, for Cachipay, La Mesa, Merida and Costa Rica, respectively.

Nei's genetic distances based on allele frequencies were constructed at the population level, using the combined data from STMS, SAMPL and DAF. For M. fijiensis, Sta. Marta, Fundación, and Costa Rican populations had the lowest values whereas the Caldas population, being geographically more distant, showed a higher value than the comparison of the previous three populations. For M. musicola, populations from Colombia and Venezuela had the lowest value, whereas the Costa Rican population had the highest value as compared to the Colombian and Venezuelan populations.

4. DISCUSSION

The present experiments allow several conclusions:

(a) A comparison of all three techniques used in our population study revealed marked differences in time invested, costs, and percentage of detected polymorphism per assay. Highly polymorphic markers such as STMS showed a high degree of informativeness, evidenced as the percentage of polymorphic loci over all markers used (80% and 90% for M. fijiensis and M. musicola, respectively). The development of polymorphic STMS clearly is, however, time consuming, and requires substantial technical know-how. Yet they are easy to use for genotyping large populations, there is no need for an extensive PCR optimisation period, and the scoring of products (just one band per individual in the case of haploid individuals) is unequivocal.

Figure 5 (a) Similarities between isolates of four Mycosphaerella fijiensis populations. Six discrete clusters (A to F) are differentiated, in which the isolates are grouped according to their geographical origin. A, Sta. Marta, Colombia; B, Caldas, Colombia; C, Costa Rica; D, Caldas, Colombia; E, F, Fundación, Colombia.

Figure 5 (b) Similarities between isolates of four Mycosphaerella musicola populations. Five discrete clusters (A to E) are differentiated, in which the isolates are grouped according to their geographical origin. A, Cachipay, Colombia; B, Costa Rica; C, Mérida, Venezuela; D, La Mesa, Colombia; E, Costa Rica

The development of DAF markers requires no previous sequence knowledge, and is technically easy, but may need considerable optimisation [25]. The percentage of detected polymorphisms per assay is distinctly lower than for STMS, but the number of markers is considerably higher. However, a high number of assays is needed to accumulate a statistically significant number of polymorphic characters.

The SAMPL technique detects a percentage of polymorphism that, in the case of M. fijiensis and M. musicola populations, was lower than for DAF and STMS. However, it offsets this disadvantage by generating the highest number of bands per assay (between 40 and 60 scorable bands per selective amplification, yielding 20-40 polymorphic bands per assay). The technical requirements are comparable to those of STMS, and considerable time is necessary for adaptation to each species.

(b) We have developed a highly reliable diagnostic method for both species, based on specific SAMPL and DAF banding patterns, along with non-transferable (i.e. species-specific) STMS markers. For example, misclassification of isolates was easily discovered by inspection of a few discriminatory DAF and SAMPL banding patterns.

(c) Dendrograms based on data from M. fijiensis and M. musicola similarity analyses reveal good correlations between clusters and separated populations, with the exception of only a few isolates that escaped the clusters grouped according to their sampling sites.

(d) At an intercontinental level, the presence of multiple STMS alleles specific for Mexican and Nigerian populations, respectively (Peraza-Echeverria et al., in preparation) suggests complete separation of both; in other words, there is only a very limited gene flow between the continents. At the regional level, genetic exchange is also limited between populations, but still occurs to some extent. At a local level, Santa Marta and Fundacion populations both share haplotypes, but a high proportion of individuals from the same field clusters together. Within fields, haplotypes are randomly distributed, and individuals from the same lesion were clustered with individuals from other hierarchical levels.

(e) In M. fijiensis populations, the limited genetic diversity within fields suggests that events such as high environmental pressure (application of high doses of fungicides) might cause the fixation of certain resistant genotypes in the populations. On the other hand, in M. musicola populations, geographical distance plays the principal role in the genetic differentiation of populations, since there is no chemical control.

4.1. Where do we go from here: Genome analysis of Mycosphaerella in the future

The unequivocal identification of individual isolates of any fungus and its mating types, the characterization of whole fungal populations, their change over time ('dynamics') and their adaptation to new environments (e.g. fungicides) can now be accomplished easily and rapidly with the whole repertoire of molecular marker technologies, based on DNA amplification using PCR. Out of some 25 different techniques, of which many have also been tested with Mycosphaerella, microsatellite-based markers have proved to have the highest potential. Microsatellites are short, tandemly repeated simple sequences, one to ten base pairs in length [e.g. (CA/GT)_n or (TAC/ATG)_n] that are widely dispersed in eukaryotic genomes. These sequences are highly polymorphic, since their number of tandem repeats may change during DNA replication by, for example, a process called slipped strand mispairing [26]. Slippage, followed by replication or repair, leads to the insertion or deletion of one or more repeat units. The resulting high level of variability can also be exploited for Mycosphaerella, given that either microsatellite-complementary probes or PCR primers for the amplification of specific and polymorphic microsatellite loci are available.

The former approach, called microsatellite fingerprinting or DNA fingerprinting, has proven potential for various fungi (e.g. Ascochyta rabiei [27]) and has allowed unequivocal discrimination of a series of Nigerian M. fijiensis isolates [28]. However, this technique is quite demanding, requires highly purified template DNA for restriction, and radioactive probes for Southern hybridisation, and is therefore little used nowadays. About a decade ago it was superseded by the amplification of specific microsatellite loci with primers directed towards microsatellite flanking sequences. Notwithstanding the difficulties of generating these primers (i.e. establishment of genomic libraries, identification of microsatellite-containing clones, sequencing of positive clones, design of primers towards 5' and 3' flanks, functionality tests), such primers have been developed for M. fijiensis (Neu et al., 1999) and M. musicola (Molina et al., 2001). These elite markers (sequence-tagged microsatellite sites, STMS, also simple sequence repeats, SSRs, or simple sequence length polymorphisms, SSLPs) are currently used, and will continue to be used for the diagnosis of isolates, the estimation of genetic diversity in collections, the analysis of the structure of whole populations and their interactions, and the changes imposed on them by a changing environment (e.g. new host varieties, new fungicides, new climatic conditions). The use of these (and other) DNA markers will soon be commonplace and routine in laboratories around the world working with Mycosphaerella, but it is equally certain that these markers are underexploited in the above applications. One important application would be the mapping of the Mycosphaerella genome. We have already begun to establish a first genetic linkage map of M. fijiensis, based on mating between a strain highly resistant to propiconazole, and an isolate sensitive to the fungicide. Apart from locus-specific STMSs, selective amplification of microsatellite polymorphic loci (SAMPLs) and DNA amplification fingerprinting (DAF), also with mini-hairpin primers, which both detect multiple loci in the target genome, are employed. The expected high-density linkage map of the M. fijiensis genome, though informative in itself (e.g. for rough estimation of chromosome numbers), should allow the tagging of fungicide-resistance gene(s) and their isolation via map-based cloning [29]. Such an approach towards agronomically relevant genes will only be possible by physical mapping [i.e. the construction of Bacterial Artificial Chromosome (BAC) libraries] and identification of clones hybridising to the linked markers. A BAC library is also a prerequisite for isolating other interesting genes previously tagged with DNA markers: genes encoding pathogenicity factors, toxin-encoding genes, genes coding for secretory pathway proteins, genes for virulence factors, aggressivity genes, and others. The isolation and functional characterization of these genes will catalyse our understanding of the fungal strategy to undermine the banana host's defence system, and allow for more effective and environmentally friendly resistance strategies. Isolation may proceed through two different routes, (a) an in silico approach to find corresponding sequences in the data banks and to define conserved domains of these genes, which then can be used as probes to localize the orthologous genes on BAC clones, or (b) through the traditional map-based cloning procedure, which requires suitable matings and progeny segregating for a trait determined by the target genes (e.g. toxin synthesis, toxin secretion, virulence). The search for isolates appropriate for such matings, the establishment of a large segregating progeny, and the unequivocal typing of the segregants will be (and has to be) a continuous task for plant pathologists, necessary to overcome the obvious deficits in the Mycosphaerella research programme.

Another step forward - which we definitely propose - is the sequencing of the complete Mycosphaerella genome, using, for example, the whole genome shotgun sequencing approach. Since the relatively small M. fijiensis genome, composed of 11 chromosomes (A. James, pers. comm.) probably harbours some 50-70 Mbp, the sequencing will be a minor challenge only (cf. the human genome, 3.2 Gbp [30]). The major problem will certainly be the functional annotation of the various sequences. However, the mere knowledge of the Mycosphaerella genome sequence will already allow us to isolate important genes in silico, will inform us of the genome structure and its various components, and help to find sequences of future use: promoters, enhancers, silencers, transposons and retrotransposons (if present). So the sequencing of the Mycosphaerella genome will be a major step towards deciphering its functions. The actual analysis of functions, however, can already be started now. There exists a series of high-throughput techniques for a genome-wide transcript analysis, e.g. serial analysis of gene expression (SAGE), rapid analysis of gene expression (RAGE), or the various expression chip technologies, that work with either oligonucleotides, Expressed Sequence Tags (ESTs), truncated or full-length cDNAs, or genic sequences spotted on glass, quartz, gold, silicon, or simple membranes, to which fluorophore-labelled cDNAs from Mycosphaerella, derived from different phases of its life-cycle or different stages during the infection cycle, can be hybridised. Such techniques recommend themselves for screening the transcriptome for mRNAs synthesized after, for example, contact with the host plant. Moreover, genes up- or down-regulated during, for example, an attack on a banana plant can be identified. Such information is a prerequisite for a deeper understanding of the molecular events during the encounter of the fungus with the host and the actual colonization phase. Crucial genes for pathogenicity can then be targeted by genetic or chemical strategies.

Apart from all these molecular approaches, there exists an acute need for the pathotyping of Mycosphaerella isolates on a host differential set. Such a set has already been designed, so that the identification of virulence types or even races can be envisaged. Coupled to the marker-based genotype, a passport for each isolate can be established, which will allow easy identification of the pathotype by simply running one identifier gel. Such experiments allow the creation of geographical maps showing the distribution of haplotypes (and in consequence, pathotypes), that can continuously be updated by a monitoring process. The input for such a tremendously important work would be comparably modest, and could even be simplified and catalysed by an easy-to-use kit for genotyping even in agricultural experiment stations.

Research into the host's biology, important as it is, has to be complemented by an equally intensive and heavily supported research programme into the biology of the pathogen. Only this intimate combination will allow breakthroughs in the near future.

ACKNOWLEDGEMENTS

The authors appreciate the contributions of their colleagues D. Fischer (now IPK, Gatersleben, Germany), C. Neu (now University of Zürich, Switzerland), K. Weising (now University of Kassel, Germany), L. Peraza Echeverria, A. James, D. Kaemmer (CICY, Merida, Mexico), M. Guzman and J. Sandoval (CORBANA, Guapiles, Costa Rica), and M.E. Garcia (Merida, Venezuela). The authors were supported by IAEA (contract CRP 302-02-GFR-8148) and EU (contract ERBIC-18/CT-970192). C. Molina gratefully acknowledges fellowships from UNESCO (Paris, France), IAEA (C6/COL/00001/R, Vienna, Austria), Stiftung für Internationale Wissenschaftliche Zusammenarbeit (Frankfurt, Germany) and COLCIENCIAS (Colombia).

REFERENCES

[1] FAURé, S., et al., A molecular marker-based linkage map of diploid bananas (Musa acuminata), Theor. Appl. Genet. 87 (1993) 517-526.

[2] REMY, S., et al., Co-transformation as potential tool to create multiple and durable resistance in banana (Musa spp.), Acta Hort. 461 (1998) 361-365.

[3] SáGI, L., et al., Production of transgenic banana and plantain, Acta Hort. 521 (2000) 51-57.

[4] ARNTZEN, C., High-tech herbal medicine: Plant-based vaccines. Nature Biotechnol. 15 (1997) 221-222.

[5] CLENDENNEN, S., MAY, G., Differential gene expression in ripening banana fruit, Plant Physiol. 115 (1997) 463-469.

[6] DOLEZEL, J., et al., Nuclear DNA amount in diploid bananas (Musa acuminata and M. balbisiana), Biol. Plant. 36 (1994) 351-357.

[7] DOLEZELOVá, M., et al., Physical mapping of the 18S-25S and 5S ribosomal RNA genes in diploid bananas, Biol. Plant. 41 (1998) 497-505.

[8] BALINT-KURTI, P.J., et al., Identification and chromosomal localization of the monkey retrotransposon in Musa sp., Mol. Gen. Genet. 263 (2000) 908-915.

[9] KAEMMER, D., et al., Oligonucleotide and amplification fingerprinting of wild species and cultivars of banana (Musa spp.), BioTechnology 10 (1992) 1030-1035.

[10] KAEMMER, D., et al., Molecular breeding in the genus Musa: a strong case for STMS marker technology, Euphytica 96 (1997) 49-63.

[11] NEU, C., et al., Polymorphic microsatellite markers for the banana pathogen Mycosphaerella fijiensis, Mol. Ecol. 8 (1999) 523-525.

[12] MOLINA, C., et al., Microsatellite markers for the fungal banana pathogen Mycosphaerella musicola, Mol. Ecol. Notes 1 (2001) 137-139.

[13] CARLIER, J., et al., DNA restriction fragment length polymorphisms in Mycosphaerella species that cause banana leaf spot diseases, Phytopathology 84 (1994) 751-756.

[14] FISCHER, D., BACHMANN, K., Microsatellite enrichment in organisms with large genomes (Allium cepa), BioTechniques 24 (1998) 796-802.

[15] WEISING, K., et al., Plant DNA fingerprinting with radioactive and digoxigenated oligonucleotide probes complementary to simple repetitive DNA sequences, Electrophoresis 12 (1991) 159-169.

[16] SAMBROOK, J., et al., Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, New York (1989).

[17] WEISING, K., KAHL, G., "Hybridization-based microsatellite fingerprinting of plants and fungi", DNA Markers: Protocols, Applications and Overviews (CAETANO-ANOLLéS, G., GRESSHOFF, P., Eds), Wiley & Sons, New York (1997) 27-54.

[18] ROZEN, S., SKALETSKY, H.J., PRIMER 3. Available at http://www.genome.wi.mit.edu/genome_software/other/primer3.html (1997).

[19] CAETANO ANOLLéS, G., GRESSHOFF, P., DNA amplification fingerprinting with arbitrary mini-hairpin primers, BioTechnology 12 (1994) 619-623.

[20] PAGLIA, G.P., MORGANTE, M., PCR-based multiplex DNA fingerprinting techniques for the analysis of conifer genomes, Molecular Breeding 4 (1998) 173-177.

[21] ROHLF, J., Numerical Taxonomy and Multivariate Analysis System. Version 1.8 (1994).

[22] NEI, M., Estimation of average heterozygosity and genetic distance from a small number of individuals, Genetics, 83 (1978) 583-590.

[23] YEH, F.C., BOYLE, T., Popgene, Version 1.2. Available at http://www.cgiar.org/cifor/research/projects/project6-2/popgene.html (1997).

[24] LIEDLOFF, A., Mantel non Parametric Test Calculator, Version 2.0. School of Natural Resource Sciences, Queensland University of Technology, Queensland, Australia (1999).

[25] CAETANO ANOLLéS, G., et al., DNA amplification fingerprinting and marker screening for pseudo-testcross mapping of flowering dogwood (Cornus florida L.), Euphytica 106 (1999) 209-222.

[26] LEVINSON, G., GUTMAN, G.A., Slipped-strand mispairing: a major mechanism for DNA sequence evolution, Mol. Biol. Evol. 4 (1987) 203-221.

[27] GEISTLINGER, J., et al., Locus-specific microsatellite markers for the chickpea pathogen Ascochyta rabiei, Mol. Ecol. 9 (2000) 1939-1941.

[28] MUELLER, R., et al., Oligonucleotide fingerprinting detects genetic variability at different levels in Nigerian Mycosphaerella fijiensis, J. Phytopathol. 145 (1997) 25-30.

[29] WINTER, P., KAHL, G., Molecular marker technologies for crop improvement. World J. Microbiol. Biotechnol. 11 (1995) 449-460.

[30] LANDER, E.S., et al., Initial sequencing and analysis of the human genome, Nature 409 (2001) 860-921.