Molecular Analysis of Hypervirulent Somatic Hybrids of the Entomopathogenic Fungi Beauveria bassiana and Beauveria sulfurescens

Abstract

Protoplast fusion of diauxotrophic mutants of a Beauveria bassiana entomopathogenic strain (Bb28) and a Beauveria sulfurescens toxinogenic strain (Bs2) produced hybrids which were significantly different from the parents in pathogenicity. Some of the hybrids were hypervirulent and killed insects more quickly than the Bb28 strain, probably because these hybrids had acquired the toxic activity of the Bs2 strain. By using six nuclear genes and a telomeric fingerprint probe, the molecular structures of the hybrids were studied. The results demonstrated the occurrence of parasexual events. Hybrids appeared to be diploid or aneuploid, with portions of the genome being heterozygous. A mitochondrial molecular marker indicated homoplasmy of the hybrids and inheritance of mitochondria from strain Bs2 or Bb28. The pathogenicities and the ploidies of the hybrids remained stable after passage through the host insect, showing that somatic hybridization provides an attractive method for the genetic improvement of biocontrol efficiency in the genus Beauveria.

The entomopathogenic fungus Beauveria bassiana (Bals.) Vuill. is widely used as biological control agent for crop pests (6, 7, 14). Despite the agricultural importance of some strains, the genetics of this imperfect fungus has rarely been studied. Like many fungi, B. bassiana lacks a conventional sexual cycle but exhibits parasexual recombination (31). The term parasexual cycle was first used by Pontecorvo et al. (32) to describe the genetic process in Aspergillus nidulans, which involves (i) heterokaryon formation, (ii) fusion of two unlike haploid nuclei to give a diploid heterozygous nucleus, and (iii) mitotic recombination. Mitotic recombination may include crossing over leading to intrachromosomal recombination and nondisjunction, whereby the amount of genetic information per nucleus is reduced to the haploid level (36). Since its first demonstration by Pontecorvo et al., parasexual recombination has been detected in many ascomycetes, basidiomycetes, and deuteromycetes, suggesting that parasexual processes are widespread in fungi (5). Nevertheless, heterokaryon formation, a prerequisite for the parasexual cycle, seems to be limited by the existence of vegetative incompatibility in many fungi (4, 18, 25), including B. bassiana (9). This limitation can be overcome by protoplast fusion in many fungi, and this technology is a valuable method for intra- and interspecific hybridization (16).

Previously, we have obtained, through protoplast fusion, one somatic hybrid which was a cross between a B. bassiana strain (Bb28) that is pathogenic toward the European corn borer (Ostrinia nubilalis) and a Beauveria sulfurescens strain (Bs2) producing an insecticidal toxin (8). This hybrid was hypervirulent toward O. nubilalis because of the combination of the two parental insecticidal activities. This preliminary result suggested that protoplast fusion could be a useful tool for increasing the biocontrol ability of Beauveria. To confirm this assumption, several experiments were conducted to enhance the number of recovered hybrids and analyze their virulence toward O. nubilalis. Moreover, molecular tools were developed to understand the genetic exchanges involved in protoplast fusion. The aims of this work were (i) to obtain a high number of Beauveria somatic hybrids and (ii) to determine the molecular nature of the hybrids obtained and the possible mitotic recombination involved by combining the information given by different independent molecular markers.

MATERIALS AND METHODS

Beauveria strains.

The Beauveria strains used in this investigation were selected from the culture collection of the Institut National de la Recherche Agronomique at La Minière, France. The B. bassiana strain (Bb28) was isolated from the Colorado potato beetle (Leptinotarsa decemlineata) in France. This strain has a low level of virulence toward O. nubilalis. The B. sulfurescens strain (Bs2) was previously described by Couteaudier et al. (8). This strain is nonpathogenic toward any known insect but produces an entomotoxic glycoprotein (28). Double auxotrophic mutants, i.e., Bb28 arg⁻ ino⁻, requiring arginine and inositol, and Bs2 lys⁻ leu⁻, requiring lysine and leucine, were selected after treatment of wild strains with mutagens sequentially as follows: 3% (vol/vol) ethylmethane sulfonate (2 h of incubation) and 10% (wt/vol) nitrosoguanidine (2 h of incubation).

Media.

The culture media were complete agar medium (CM) and minimal medium (MM), as previously described (35). Incubation was at 25°C throughout.

Protoplast fusion.

Protoplast isolation and fusion were performed as in our previous work (8) with a polyethylene glycol concentration of 30%. Prototrophic fusion products were selected on MM supplemented with 20% sucrose (MMS). In addition, fusion products with recombinant phenotypes were identified on MMS supplemented with arginine, lysine, leucine, and inositol (each at 0.2 mg · ml⁻¹) either singly or in pairs (20). Parental protoplasts were subjected to the same polyethylene glycol treatment and regeneration process. Each putative prototrophic colony was purified from single conidia and transferred to CM. Mitotic stability was tested by five colony passages onto CM; thereafter, 10 individual conidia of each of these fusion products were tested for phenotypic stability on MM. These stable fusion products were treated with the haploidizing agents benomyl (0.5 μg · ml⁻¹), para-fluoro-dl-phenylalanine (5 μg · ml⁻¹), and chloral hydrate (1 mg · ml⁻¹) for 2 weeks on agar plates. Thereafter, 100 individual conidia of each of these fusion products were tested on MM for prototrophy.

Pathogenicity toward O. nubilalis.

The pathogenicities of the hybrids toward O. nubilalis were compared to those of the original wild-type strains Bb28 and Bs2 and diauxotrophic parental strains. Sixty newly emerged fifth-instar larvae were dipped in 20-ml conidial suspensions (10⁵, 10⁶, 10⁷, and 10⁸ conidia · ml⁻¹), and insect mortality was assessed daily (34). The 60 larvae were separated into four batches of 15 larvae in order to determine the standard error, and the experiments were performed twice. Parallel controls (treated with sterile distilled water) were included. Controls showed no mortality over the course of the experiments. To assess virulence of the wild strains, mutants, and hybrids, full logarithmic plots of insect mortality against time were analyzed by first-order linear regression equations as described by Gupta et al. (19) for B. bassiana bioassays. These equations allowed us to determine the time required to kill 50% of the insect population (LT₅₀) and the dose required to kill 50% of the insect population (LD₅₀) in 15 days.

DNA extraction.

Fungal strains were cultivated in Roux flasks containing 130 ml of liquid CM. The mycelium was collected by filtration in a sterile filter funnel and ground to a fine powder in liquid nitrogen. The DNA extraction method used was that described by Daboussi et al. (10).

DNA amplifications.

DNA amplification was performed with an Appligene (Illkirch, France) kit and model 60 Braun DNA thermal cycler. Amplifications were performed in a 100-μl reaction volume with 0.2 μM primers E23 and E24 (5′CCGAAGCAGAATTCGGTAAGCG3′ and 5′GCTGAATTACCATTGCGGAGAGG3′, respectively) (30) and approximately 50 ng of template DNA. Control amplifications, using primers only, were performed to ensure that the reagents used were not contaminated with extraneous template DNA. The PCR cycling protocol consisted of 94°C for 1 min, 60°C for 1 min, and 72°C for 1 min for 30 cycles. The PCR products were electrophoresed in 1% (wt/vol) agarose gels. The gels were stained with ethidium bromide (1 μg · ml⁻¹) and photographed under UV transillumination with Polaroid 667 type film.

Southern blot analysis.

Southern blotting procedures were performed as described by Maniatis et al. (27). Total genomic DNA (10 μg) was digested with EcoRI, HindIII, BamHI, and AluI (Eurogentec, Seraing, Belgium) and separated electrophoretically on a 0.6% (wt/vol) agarose gel. Total DNA, cut by restriction enzymes, was transferred to a Hybond-N membrane (Amersham, Buckinghamshire, England) with a vacuum apparatus. Conserved genes previously cloned in B. bassiana were used as probes. They encode mitochondrial rRNA, β-tubulin, histone 4, and protease 1 (39). A chitin synthase gene was cloned, as described previously for the other genes, by using the degenerate primers CHS1 and CHS2 defined by Chua et al. (7). The nitrate reductase gene cloned from B. bassiana (EMBL, GenBank, and DDBJ nucleotide sequence database accession number X84950) was also used. The karyotypes of the Bb28 and Bs2 strains determined by contour-clamped homogenous electric field (CHEF) analysis and the chromosomal localizations of these genes are presented in Table 1. Double auxotrophic mutants have the same genome organization as the wild strains (unpublished results). The telomeric probe, pTel 13, cloned from the fungus Botrytis cinerea, contains a telomeric repeat, (TTAGGG)₁₁, and 153 bp of a telomere-associated DNA region (26). Plasmid DNA was radiolabelled by using a random primer kit (T7 Quick prime; Pharmacia) and [α-³²P]dCTP. Hybridizations were conducted under stringent conditions (65°C), and washes were with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% sodium dodecyl sulfate as described by Daboussi et al. (10). Blots were exposed to MP-type films (Amersham) with intensifier screens for 2 to 7 days at −80°C.

TABLE 1.

Estimation of sizes of chromosomal bands from parental Beauveria strains as determined by CHEF analysis and chromosomal locations of the probes as determined by Southern hybridization of CHEF gels (39)

Strain and chromosomal band	Size (Mbp)	Probe gene(s)
B. bassiana Bb28
I	7.7	β-Tubulin
II	7.2	rRNA, protease 1, nitrate reductase, chitin synthase
III	7.0	Histone 4
IV	4.8
V	4.1
VI	2.4
VII	1.2
B. sulfurescens Bs2
I-IIa	7.7	Histone 4, rRNA, protease 1, chitin synthase
III	6.6
IV	6.0
V	4.7	Nitrate reductase
VI	4.4	β-Tubulin
VII	2.4

^aThis band probably represents two comigrating chromosomes. 

Stability after parasitic growth on insects.

Five prototrophic hybrids and the Bb28 strain were passed through two disease cycles on successive generations of O. nubilalis. After the first disease cycle, single spores recovered from the cadavers of insects were cultured on CM before inoculation of a second insect generation. After each disease cycle, 100 individual conidia were plated on MM to determine the percentage of prototrophic conidia, and two single conidial cultures were used for DNA isolation and telomeric fingerprinting.

RESULTS

Obtaining B. bassiana-B. sulfurescens prototrophic hybrids.

Fusion of protoplasts from the diauxotrophic mutants Bb28 arg⁻ ino⁻ and Bs2 lys⁻ leu⁻ resulted in the recovery of prototrophic fusion products at a frequency of 5 × 10⁻⁴; no colonies from protoplasts of only one of the fusion partners appeared on MMS, suggesting that no back mutations occurred among 10⁷ protoplasts. Single-spore isolates originating from 48 colonies on MMS were tested for prototrophy. After five subcultures on CM without selection pressure, the fusion products remained prototrophic. The spores produced by the fusion products were larger than the parental spores. They were cylindrical and approximately 3 μm wide and 4 μm long, whereas the parental spores were spherical with a diameter of 3 μm. Since Beauveria spores are uninucleate, their size is related to their ploidy (16, 23). Consequently, the fusion products seem to be diploid or aneuploid (between n and 2n). The stable prototrophic growth and the large size of the spores suggest the hybrid status of the fusion products. Haploidization assays were performed for five hybrids (A22, C2, C17, D2, and D9) with three known haploidizing agents (benomyl, para-fluoro-DL-phenylalanine, and chloral hydrate), without any success. No auxotrophic segregants were observed, and the spores remained large.

In other fusion experiments, MMS was supplemented with growth requirements of the parental strains to allow the recovery of recombinant segregants, as suggested by Hamlyn et al. (20). Unfortunately, all the selected fusion products were prototrophic and had the same spore size as the fusion products described above.

Pathogenicities of hybrids.

Pathogenic activities of parental strains and hybrids were assessed by bioassays under standardized conditions. The O. nubilalis larva mortalities obtained with the Bb28 strain and five hybrids (A22, C2, C17, D2, and D9) are presented in Fig. 1. No insect was infected when spore suspensions of the Bs2 wild strain or the Bb28 arg⁻ ino⁻ and Bs2 lys⁻ leu⁻ parental mutants were used (data not shown), but the parental strain Bb28 appeared to be weakly pathogenic toward O. nubilalis. Some of the hybrids, including C17, D9, and D2, appeared to be hypervirulent, and others, including C2, were less virulent than the Bb28 parental strain. The logarithmic transformation of the time to mortality and the percent mortality resulted in linear regression lines with correlation coefficients greater than 0.9. The resulting LT₅₀s and LD₅₀s confirmed the great variability in the pathogenicities of the hybrids and the occurrence of hypervirulent hybrids (Table 2). For example, the C17 hybrid has an LT₅₀ (at an inoculum concentration of 10⁸ conidia · ml⁻¹) approximately 2.4-fold lower than that of the Bb28 parental strain and an LD₅₀ approximately 290-fold lower than that of the Bb28 strain. These experiments were conducted twice with similar results.

FIG. 1.

Pathogenic activities of the parental strain Bb28 and hybrids toward O. nubilalis. Four batches of 15 newly emerged fifth-instar larvae were dipped in conidial suspensions with 10⁸ conidia · ml⁻¹ and insect mortality was assessed daily. The standard error is indicated for day 15.

TABLE 2.

Virulence of the parental strain Bb28 and hybrid strains toward O. nubilalis

Strain	Inoculum (conidia · ml−1)	LT50a (days)	LD50a on day 14 (conidia · ml−1)
Bb28	107	24.6	3.2 × 108
	108	16.6
A22	106	30.2	2.5 × 108
	107	18.6
	108	21.1
C2	107	>40	>109
	108	>40
C17	105	12.7	1.1 × 106
	106	10.4
	107	9.4
	108	6.9
D2	107	16.1	8.3 × 107
	108	13.3
D9	106	26.7	9.4 × 107
	108	8.7

^aData were calculated by logarithmic transformation of the mortality curves as described by Gupta et al. (19). 

Molecular characterization of the hybrids by using conserved genes.

The mapped genes (Table 1) were used as markers to study inheritance in the hybrids. PCR amplification of a part of the 28S ribosomal DNA (rDNA) (30) was done with parental strains and hybrids. The PCR products obtained with primers E23 and E24 were of different sizes in the two strains, approximately 100 and 500 bp (Fig. 2). The smallest fragment, from strain Bb28, is the same size as the homologous fragment of Saccharomyces cerevisiae (17). DNA amplification of the Bs2 strain revealed a larger fragment containing a nucleotide insertion of approximately 400 bp. Since some Beauveria strains were previously found to have 350- to 450-bp group I introns in this region (29, 30), the insertion in the Bs2 strain is probably a group I intron. This insertion allows molecular differentiation of the parental strains. The diauxotrophic mutants had the same patterns as the wild strains, but all the hybrids had additive banding patterns. This result indicates that the hybrids have both kinds of ribosomal units.

FIG. 2.

PCR amplification patterns of Beauveria parental strains and hybrids with rDNA conserved primers E23 and E24. Lanes: 1, no DNA; 2, Bs2; 3, Bb28; 4, hybrid A22; 5, hybrid C2; 6, hybrid C5; 7, hybrid C13; 8, hybrid C17; 9, hybrid D7; 10, hybrid D8; 11, hybrid D9; 12, hybrid D15; 13, hybrid D22. DNA sizes are given on the right.

Southern blot hybridizations were performed after digestion of total DNA with different restriction enzymes. Examples of the different restriction fragment length polymorphism (RFLP) patterns are shown in Fig. 3. DNAs from the Bb28 strain and from the D8 and D22 hybrids were digested with EcoRI and produced one band at 7 kb that hybridized to the histone 4 gene probe, while the Bs2 strain and the A22, C2, C5, C17, and D7 hybrids displayed a different pattern with a band at 11 kb. The pattern of the C13 hybrid contained both parental bands. The same experiments were conducted with nuclear probes corresponding to the genes encoding β-tubulin, nitrate reductase, a chitin synthase, and protease 1. In each case, the diauxotrophic mutants showed the same patterns as the wild strains. All the hybrids had additive banding patterns for the β-tubulin, chitin synthase, and protease 1 genes. In contrast, the results with the histone 4 and nitrate reductase genes were additive banding patterns or unique patterns of either parental strain (Table 3). Four hybrids (C12, C13, D14, and D26) displayed additive banding patterns for the six nuclear genes, indicating that large parts of their genomes are diploid and heterozygous. The other hybrids showed only one parental pattern for the histone 4 gene, and in one case (hybrid A7) for the nitrate reductase gene, indicating that parts of their genomes are haploid or homozygous. Since we used genes on different chromosomes (Table 1), the molecular characterization of the hybrids indicates that different parental chromosomal regions are present in the hybrid genomes.

FIG. 3.

Southern hybridization analysis of histone 4 restriction fragments showing segregation of RFLPs following somatic hybridization between strains Bb28 and Bs2. EcoRI-digested genomic DNA was electrophoresed, blotted to nylon, and hybridized with the histone 4 probe. Lanes: 1, Bb28; 2, Bs2; 3, hybrid A22; 4, hybrid C13; 5, hybrid C2; 6, hybrid C5; 7, hybrid D8; 8, hybrid C17; 9, hybrid D7; 10, hybrid D22. DNA sizes (in kilobase pairs) are given on the right.

TABLE 3.

Molecular characterization of the hybrids with conserved genes

Strain(s)	Pattern for the following genea:
	Histone 4 (EcoRI)	Nitrate reductase (HindIII)	Mitochondrial rDNA (AluI)
Bs2	1, 0	0, 1	1, 0
Bb28	0, 1	1, 0	0, 1
A1, D8, D22	0, 1	1, 1	1, 0
A2, A4, A5, A6, A15, A16, A18, A19, A20, A21, A22, A25, A26, B1, B5, C1, C2, C4, C14, C17, C26, D1, D2, D7, D11, D12, D15, D16, D17, D18, D19, D20	1, 0	1, 1	1, 0
A7	1, 0	1, 0	1, 0
A23, C16, C18, D3, D6, D9, D21	1, 0	1, 1	0, 1
C5	1, 0	1, 1	1, 0
C12, C13	1, 1	1, 1	0, 1
D14, D26	1, 1	1, 1	1, 0

^aMolecular characterization was done by Southern blot procedures with the indicated restriction enzyme. 1, presence of a band; 0, absence of the band. 

The mitochondrial marker used to study the inheritance of the mitochondrial DNA is part of the small ribosomal unit cloned from the Bb28 strain. When used as a probe in Southern blot procedures, this marker indicated that the hybrids carry mitochondrial DNA from only one parental strain (Table 3).

Molecular characterization of the hybrids with telomeric fingerprints.

A telomeric probe previously cloned in B. cinerea (26) and used to fingerprint numerous Beauveria isolates (9, 39) was used to estimate the minimum ploidies of the hybrids. Southern blots of DNAs from strains Bb28 and Bs2 cut with EcoRI (Fig. 4) exhibited 10 and 13 bands, respectively, which hybridized to the telomeric probe. These bands were numbered 1 to 10 for strain Bb28 and 1 to 13 for strain Bs2 in order to monitor their inheritance in the hybrid patterns. The diauxotrophic mutants had the same telomeric patterns as the wild strains. Since the parental Bb28 and Bs2 telomeric patterns were different, it was possible to determine the minimum number and the origin of the telomeres in the hybrid genomes. Figure 4 presents the telomeric patterns of only eight hybrids, but all the hybrids were analyzed by using the same technique. The number of telomeric bands revealed by the blots was between 16 (in hybrid A7) and 20 (in hybrid D3), which indicates a minimum of 8 to 10 chromosomes in the hybrids. Nevertheless, these numbers could be underestimated, since (i) some telomeric bands are thought to comigrate in each parental strain (39), (ii) bands 8 and 10 from strain Bs2 cannot be differentiated from bands 4 and 8 from strain Bb28, and (iii) crossing over and chromosomic segregation during mitotic recombination may lead to heterozygous chromosomes with homozygous telomeric regions (5, 36). All the hybrids had bands 6 and 9 from Bs2 and band 5 from Bb28. No other telomeric bands were found in any of the hybrids, but all of the hybrids carried at least four bands from Bb28 and at least six bands from Bs2. These results confirm that the hybrid nuclei were heterozygous. Moreover, the telomeric patterns indicate that all the hybrids have a part of their genomes which is haploid or homozygous.

FIG. 4.

Southern hybridization analysis of telomeric restriction fragments showing segregation of RFLPs in a somatic hybridization between strains Bb28 and Bs2. EcoRI-digested genomic DNA was electrophoresed, blotted to nylon, and hybridized with the telomeric probe from B. cinerea. Lanes: 1, Bs2; 2, Bb28; 3, hybrid A22; 4, hybrid C2; 5, hybrid C5; 6, hybrid C13; 7, hybrid C17; 8, hybrid D7; 9, hybrid D9; 10, hybrid D15. The parental bands are numbered 1 to 10 for strain Bb28 and 1 to 13 for strain Bs2. DNAs size (in kilobase pairs) are given on the right. The arrow indicates a new telomeric band that was present in neither parent.

Cosegregation of telomeric bands was studied with the software Genepop (33), which is able to distinguish linked markers even when total haploidization is not realized. This analysis indicated the absence of conserved telomere pairs (corresponding to the two extremities of one chromosome) in the segregants, suggesting that intrachromosomal recombination had occurred. Six of the 48 hybrids, e.g., D7, had a new telomeric band that was present in neither parent (Fig. 4). Such a new restriction fragment could result from mutation, recombination, or rearrangement.

Stability after parasitic growth on insects.

The biological and molecular stabilities of the hybrids A22, C2, C17, D2, and D9 were evaluated through infection of O. nubilalis followed by reisolation of single conidial isolates from mycosed cadavers. After the first and second disease cycles, the 100 conidia isolated for each hybrid remained prototrophic, and the virulence of the hybrid was conserved in the two experiments. After each disease cycle, DNA was extracted from two single conidial cultures and subjected to telomeric fingerprinting. The original configuration of the banding patterns was retained in all cases, except for a single conidial culture isolated after the two disease cycles of the D2 hybrid (Fig. 5). In this case, a new telomeric band that was present in neither parent appeared.

FIG. 5.

Southern hybridization analysis of telomeric restriction fragments, showing telomeric fingerprints of the D2 hybrid before disease (lane 1) and after one (lane 2) and two (lanes 3 and 4) disease cycles on O. nubilalis. DNAs size (in kilobase pairs) are given on the right. For one culture after two disease cycles, a new telomeric band that was present in neither parent appeared (arrow).

DISCUSSION

Entomopathogenic fungi are being used for the control of many insect pests as an environmentally acceptable alternative to chemical insecticides. A key aim of recent work has been to increase the speed of killing and so improve the commercial efficacy of these biocontrol agents (38). One way that this might be achieved is by adding the toxic activity of B. sulfurescens (28) to B. bassiana pathogenic strains. The fusion frequency between B. bassiana Bb28 and B. sulfurescens Bs2 that we obtained was higher than previously reported: 5 × 10⁻⁴ instead of approximately 10⁻⁶ (8). The stable prototrophic growth of the fusion products even in the presence of haploidizing agents and the large size of the spores indicate the hybrid status of the 48 fusion products analyzed (16, 23). Some of the hybrids appear to be hypervirulent. Hybrid C17 and others have LT₅₀s and LD₅₀s significantly lower than those of the Bb28 strain and remain stable after multiple disease cycles. This rapid kill of insects combined with stability of virulence following passage through the insect constitutes a success in the engineering of the entomopathogenic Beauveria fungi.

The molecular analysis using both conserved genes, such as the protease 1 gene, which is involved in pathogenicity (21), and telomeric sequences indicated a partially heterozygous diploid structure of the hybrids and showed that whole genomes from both parents had not been successfully integrated into the progeny. So far as we know, molecular evidence for parasexual recombination is available for only a few species of filamentous fungi. Durand et al. (11) and Arnau and Oliver (2) showed mitotic rearrangements in integrated plasmid DNA during protoplast fusion in Penicillium roqueforti and Cladosporium fulvum, respectively. In P. roqueforti, randomly amplified polymorphic DNA markers showed that recombination between two transformants of the same strain during the parasexual cycle was not limited to the foreign DNA sequences introduced into the parental strain (12). In the present study evidence has been obtained, with both genes and telomeric markers, that karyogamy and genetic modifications occur after protoplast fusion involving the B. bassiana Bb28 and B. sulfurescens Bs2 strains. The resulting hybrids are diploid or aneuploid, with a minimum of 8 to 10 chromosomes. A portion of the hybrid genomes was heterozygous, while other portions were haploid or homozygous. Haploidization seems to be limited even in the presence of haploidizing agents or on the host insect. The absence of conserved telomere pairs in the segregants may indicate intrachromosomal recombination. Moreover, a new telomeric restriction fragment is present in some hybrids. During the disease cycles of the hybrids on the host insect, a few telomeric rearrangements also occurred. Since the telomeric patterns of the wild strains were previously found to be stable during mitosis (9), new telomeric bands in hybrids could be the result of recombination in the telomeric associated DNA. Such new telomeric bands were also observed after meiotic recombination in Neurospora crassa (37) and in Magnaporthe grisea (13).

In several filamentous fungi, the segregation behavior of hybrids produced after protoplast fusion seems to be influenced by the taxonomic relationship of the parental strains. In a cross between two members of the A. nidulans group (A. nidulans and Aspergillus rugulosus), diploid hybrids which gave rise to haploid segregants were obtained, suggesting that these two species were similar in overall genome organization (23). By contrast, in a cross between the distantly related species A. nidulans and Aspergillus fumigatus, an aneuploid structure of the fusion products was presumed (24). The same situation was observed in Penicillium: haploid recombinants have been isolated only from hybrid progeny derived from taxonomically closely related species. Some stable hybrids obtained from crosses between Penicillum chrysogenum and P. roqueforti had spores that were larger than those of either parent and were unaffected when grown in the presence of haploidizing agents, which implies that their chromosome configuration was probably stable (1). We hypothesize that the B. bassiana-B. sulfurescens hybrids are similar because of the different genome organizations in the parental strains.

The presence of only one parental mitochondrial ribosomal type reveals the hybrid homoplasmy of the B. bassiana-B. sulfurescens hybrids. In isogamous species, such as S. cerevisiae, uniparental inheritance of mitochondrial markers is thought to be due to vegetative segregation by random partitioning of mitochondria during cell division (22). The same vegetative segregation seems to occur after protoplast fusion between B. bassiana and B. sulfurescens. In our study, because only one mitochondrial marker was used, it was not possible to examine mitochondrial recombination similar to that demonstrated for Coprinus cinereus (3) and Lentinula edodes (15).

One of the most striking features of our results is the variability in pathogenicity among the hybrids, but no correlation between molecular pattern and pathogenicity was found. In conclusion, somatic hybridization via protoplast fusion provides an attractive method for the genetic improvement of biocontrol efficiency in the genus Beauveria, in which strains with different host ranges and toxicities are genetically isolated because of vegetative incompatibility (9).