ABSTRACT
Vegetative incompatibility in the chestnut blight fungus,Cryphonectria parasitica, in Europe is controlled by six unlinked vic loci, each with two alleles. Four previously identified vic loci (vic1, vic2,vic3, and vic4) were polymorphic in European vegetative compatibility (vc) types. Two new loci, vic6 andvic7, also were identified among European vc types. In one cross, vic genes segregated independently at five loci, and 194 progeny were assigned to 32 vc types; none of these loci were linked. A total of 64 vc types were identified from all crosses. All 64 genotypes possible from six vic loci, each with two alleles (26 = 64), were identified and assigned to vc types. Based on our model, vc types v-c 5 and v-c 10, which had been used in previous genetic studies, differ by only five vic genes. Future studies of vc types in C. parasitica can use knowledge of vic genotypes for analysis of population genetic structure based on vic allele frequencies and to determine the effect of each vic gene on virus transmission between vc types.
Vegetative (or heterokaryon) incompatibility is a self-nonself recognition system in filamentous fungi that regulates the formation of heterokaryons (5, 13, 19) and the transmission of cytoplasmic elements between strains (8, 12, 21). In most filamentous ascomycetes, incompatibility is controlled by allelic interactions; two strains are incompatible when they have different alleles at one or more vegetative incompatibility (vic [or het for heterokaryon incompatibility]) loci (5, 13, 19). Eight to 17het loci have been found in Neurospora crassa,Aspergillus nidulans, and Podospora anserina(reviewed in references 5, 13, and19), while segregation of large numbers of vegetative compatibility (vc) types suggests the existence of multiplevic loci in other ascomycetes as well (2, 7, 14, 16, 24). Among ascomycetes, most vic loci have only two alleles, although multiple alleles have been found for some loci inA. nidulans and N. crassa (11, 15).
Vegetative incompatibility has been a valuable phenotype for studying genetic diversity and population biology in fungi (reviewed in references 13 and 19). Although vc type diversity has been determined in numerous populations, more-detailed analyses would be possible if the vicgenotypes of vc types were known. However, with the exception of a few vc types used in laboratory genetic studies, there has been no attempt to assign vic genotypes to vc types. If vicgenotypes were known for most of the vc types in a population, population genetic analyses that require estimates of allele frequencies would be possible by using vc type data. For example, the multilocus genetic structure of populations could be analyzed to make inferences about recombination (23), or differentiation between populations could be estimated to study gene flow (22). Furthermore, in fungi in which mycoviruses may be a significant factor in population biology, as in the chestnut blight fungus, Cryphonectria parasitica (27, 32), the potential for virus transmission at the population level could be assessed (17, 21), especially if knowledge of the effect of each vic gene on virus transmission were available (16, 17). Without vic genotype data, however, it is difficult to link laboratory findings on virus transmission to field populations.
The simplest populations in which to determine vic genotypes are those in which vc type diversity is relatively low. InC. parasitica, only 31 vc types were found in samples of over 1,000 isolates from Italy and Switzerland, and most subpopulations had 10 or fewer vc types (10). This limited number of vc types could be explained by a minimum of five polymorphicvic loci (assuming two alleles for each locus). Vegetative incompatibility in C. parasitica is postulated to be controlled by allelic interactions at five to seven vic loci (1-3, 16). Anagnostakis (2) clearly identified two vic loci (vic1 and vic2), and Huber (16) recently identified three more (vic3,vic4, and vic5). Our objectives in this study were to determine the number of polymorphic vic loci inC. parasitica populations in Europe, to identify all possible vic genotypes, and to determine the genotypes of vc types used in previous genetic studies.
MATERIALS AND METHODS
The 31 vc types found to date in Europe are referred to as EU-1 to EU-31 (10). Single-conidial field isolates of vc types EU-1 to EU-20 were used as vc testers (9). These tester isolates, and additional single-conidial field isolates, were used as parents in crosses (Table1); names for field isolates are preceded by two-letter codes (PC, VO, TE, LI, FI, VA, and SA). Isolate JA17, used as a parent in cross MJ1, is a single-conidial field isolate from Japan (26). Field isolate vc testers were later replaced with ascospore isolates of the same vc type; ascospore isolates with novel recombinant vc types were given new EU numbers. Cross numbers (except MJ1) were designated with the letter P (Table 1). Ascospore isolates were designated with the letter P followed by the cross number and the ascospore isolate from that cross (e.g., P1-11 is ascospore isolate 11 from cross P1).
Crosses of C. parasitica for studying the genetics of vegetative incompatibility
Vegetative incompatibility was assayed as the appearance of dark discoloration and/or barrage formation between colonies on agar medium as described previously (9, 28). Crosses were made on autoclaved chestnut (Castanea sativa or C. dentata) stems embedded in water agar as described previously (1, 20). For most crosses, we picked approximately 20 to 50 random ascospore progeny, although 194 ascospores (ca. 50 from each of four perithecia) were analyzed in cross MJ1.
Our strategy for determining vic genotypes was to start with vc types that had been assigned genotypes at four vic loci. D. Huber gave us isolates representing 13 of 16 possible genotypes determined by vic1, vic2, vic3, andvic4 (16, 17). Six of these isolates were of vc types v-c 5, v-c 8, v-c 16, v-c 39, v-c 56, and v-c 71 which had been analyzed previously (1, 2, 29, 30). Our initial set of crosses was done to determine genotypes at vic1 tovic4 for the three remaining vc types that had not been assigned genotypes. Subsequent crosses were made to analyze thevic genotypes of additional vc types. Based on previous studies of C. parasitica (2, 16), we expected to find 2n vc types in the progeny of each cross, with alleles segregating at n vic loci. Therefore, we used the number of progeny vc types to infer the number of vic genes segregating in each cross. Interpretation of crosses was done sequentially and depended on results from each previous cross.
We also compared isolates of vc types v-c 10, v-c 17, and v-c 40 which had been used in crosses to study vc genetics (1, 2) with vc types from our crosses to determine their vic genotypes. Incompatibility caused by vic5 is weak and cannot be detected on potato dextrose agar (16) or on potato dextrose agar containing bromocresol green (9, 28; unpublished observations); therefore, this locus was not considered in our genetic analyses.
RESULTS
In our initial set of crosses among vc types with genotypes previously defined by vic1 to vic4 (Table2) we determined the genotypes of the three remaining vc types. In each cross (P33, P16, and P52), three of the four vc types in the progeny had known vic genotypes, making it possible to infer the fourth genotype. Progeny vc types from five additional crosses conformed to those expected from Huber’s model (16). For example, in cross P25, the parents (EU-10 and EU-13) differed at all four vic loci and 15 of the 16 predicted vc types were found in the 30 progeny. Eight of the 31 vc types from Europe were compatible with testers in this set of 16 genotypes.
Segregation of vc types defined by vic1 tovic4 in crosses of C. parasitica
In our second set of crosses, we found 48 additional vc types, for a total of 64, suggesting segregation at six vic loci (Table3). Crosses are discussed in the order in which they appear in Table 3 to explain how eachvic genotype was determined. Crosses with singlevic genes segregating were found between EU-1 and EU-2 (cross P2), EU-1 and EU-5 (cross P4), and EU-13 and EU-19 (cross P27).
Segregation of vc types in crosses ofC. parasitica
Progeny from cross MJ1 (EU-22 × EU-24) were of 32 vc types, indicating segregation at five vic loci in this cross. Only eight progeny vc types were among the 16 genotypes defined byvic1 to vic4 (i.e., those shown in Table 2); these eight types all shared allele vic3-1. From this result, we concluded that all of the progeny in this cross had allelevic3-1. We also concluded that alleles were segregating atvic1, vic2, and vic4 and at two newvic loci, which we designated vic6 andvic7. Alleles vic6-1 and vic7-1 were arbitrarily assigned to the 16 genotypes defined by vic1 tovic4.
Three crosses (P24, P17, and P21) were analyzed in which there were 16 vc types in the progeny; we concluded that four vic genes were segregating in each cross. In each of these crosses, 4 of the 16 progeny vc types were in the subset of 16 vc types whose genotypes are defined by vic1 to vic4 (Table 2). We interpreted this to mean that there was segregation at only two of locivic1 to vic4 and that there was segregation at both vic6 and vic7. Based on the known genotypes of progeny vc types, we inferred that all of the progeny of P24 had alleles vic2-2 and vic3-1 because no segregation was evident at these loci; similarly, all of the progeny of P17 hadvic3-1 and vic4-2, and progeny of P21 hadvic1-2 and vic4-2. Since EU-1 was a progeny vc type in these three crosses and in cross MJ1, we determined its genotype for vic1 to vic4 (as described above). Because there was segregation at vic6 andvic7 in cross P21, and parent EU-10 has allelesvic6-1 and vic7-1, the other parent, EU-1, had alleles vic6-2 and vic7-2. Therefore, EU-1 has the genotype vic1-2 vic2-2 vic3-1 vic4-2 vic6-2 vic7-2. To simplify the notation, genotypes hereafter are designated simply by their allele numbers for these six loci, e.g., 2212-22 for EU-1; the hyphen indicates that the allele for vic5 is not known. Lightly shaded cells in Table 3 indicate the cross in which a genotype was assigned to a vc type.
In cross P5, EU-1 and EU-3 were found to differ at one viclocus. By the same reasoning as that used for EU-1, EU-3 has genotype2212--- because it was found in crosses MJ1, P24, P17, and P21; therefore, segregation must have occurred at eithervic6 or vic7 in cross P5. Since vic6and vic7 were not yet defined separately, we arbitrarily assigned alleles vic6-1 and vic7-2 to EU-3. When EU-3 (2212-12) was crossed (in cross P48) with EU-33 (2222-11), two vic genes segregated, atvic3 and vic7, allowing us to assign genotype2222-12 to recombinant vc type EU-35 because the other recombinant genotype, 2212-11, was already assigned to EU-28. For most of the remaining crosses (not discussed explicitly), we assigned genotypes to vc types in this way; i.e., from crosses in which all but one progeny vc type had genotypes already determined from previous crosses (Table 3). Exceptions to this strategy are described below.
Cross P12.When EU-3 (2212-12) was crossed with EU-26 (-212---), segregation at vic6 was evident because EU-1 (2212-22), which has vic6-2, was a recombinant vc type. Therefore, vic6-2 was assigned to parent EU-26 and vic6-1 was assigned to the other recombinant type, EU-30. EU-30 could not also have had allelevic7-1 because it is not among the 16 vc types defined byvic1 to vic4 that have both vic6-1 andvic7-1 (Table 2). Therefore, vic7-2 was assigned to both EU-26 and EU-30. With four different progeny vc types in cross P12, there must have been segregation at vic1 in addition tovic6 because vic2-2, vic3-1,vic4-2, and vic7-2 were common to both parents. Therefore, genotypes 1212-22 and 1212-12 were assigned to EU-26 and EU-30, respectively.
Cross P8.We deduced the genotypes for EU-2 and EU-14 from cross P8 by using a rationale similar to that used for cross P12. From crosses MJ1, P17, and P21, we knew that EU-2 and EU-14 both had allelesvic1-2, vic3-1, and vic4-2. As in P12, segregation at vic6 was evident because EU-1, which hasvic6-2, was a recombinant vc type (2212-22), while parent EU-3 (2212-12) had vic6-1. Therefore, vic6-2 was assigned to parent EU-2, whilevic6-1 was assigned to the other recombinant type, EU-14. EU-14 could not have had allele vic7-1 because it was not among the 16 vc types defined by vic1 to vic4that have both vic6-1 and vic7-1 (Table 2). Therefore, vic7-2 was assigned to both EU-2 and EU-14. Since there were four progeny types, alleles must have segregated atvic2 in addition to vic6, and vic2-1was assigned to both EU-2 and EU-14. Thus, genotypes 2112-22and 2112-12 were assigned to EU-2 and EU-14, respectively.
Cross P10.Parent EU-3 (2212-12) and recombinant type EU-1 (2212-22) differ at vic6, demonstrating segregation at this locus. Therefore, parent EU-5 must have hadvic6-2, while recombinant EU-21 must have vic6-1. From crosses MJ1 and P24, we knew that parent EU-5 had allelesvic2-2 and vic3-1. After determining the genotype of EU-4 in cross P3, we deduced from cross P1 that EU-5 hadvic1-2 because EU-4, the other parent in cross P1, hadvic1-1, but there was segregation at vic1 (Table3). EU-21 could not have had allele vic7-1 because it was not among the 16 vc types defined by vic1 to vic4that have both vic6-1 and vic7-1 (Table 2). Therefore, vic7-2 was assigned to both EU-5 and EU-21. Since there were four progeny types in cross P10, alleles must have segregated at vic4 in addition to vic6, andvic4-1 was assigned to both EU-5 and EU-21. Thus, genotypes2211-22 and 2211-12 were assigned to EU-5 and EU-21, respectively.
Cross P56.Parent EU-33 (2222-11) and recombinant type EU-28 (2212-11) differ at vic3, demonstrating segregation at this locus. Therefore, parent EU-7 hadvic3-1, while recombinant EU-41 had vic3-2. One parent (EU-7) and one recombinant type (EU-41) were not among the 16 vc types defined by vic1 to vic4 that have bothvic6-1 and vic7-1 (Table 2). Therefore, alleles segregated at either vic6 or vic7 in addition tovic3. However, genotypes 2212-12 and 2222-12 were already defined, leaving genotypes 2212-21 and 2222-21 to be assigned to EU-7 and EU-41, respectively.
Cross P37.Parent EU-11 (1212-11) and recombinant type EU-12 (1112-11) differ at vic2, demonstrating segregation at this locus. Therefore, the other parent, EU-8, had vic2-1, while recombinant EU-25 hadvic2-2. EU-8 and EU-25 were both assigned vic6-2because 32 vc types had already been assigned genotypes withvic6-1. With segregation at only two loci (vic2and vic6), there could not have been segregation at any other locus. Therefore, genotypes 1112-21 and1212-21 were assigned to EU-8 and EU-25, respectively.
Cross P67.Six of the eight progeny vc types, including both parents (EU-6 and EU-36), had known genotypes; genotypes were not yet known for EU-16 and EU-59. The only progeny genotypes in this cross that were not already assigned to vc types were 2111-21 and2121-22. Because EU-16 was a progeny vc type in cross MJ1 and EU-59 was not, EU-16 had vic3-1 and was assigned genotype 2111-21, leaving 2121-22 to be assigned to EU-59.
All 64 possible vic genotypes, vc types, and tester isolates are summarized and cross-listed with vc types and testers from previous genetic studies in Table 4. vc types v-c 10, v-c 17, and v-c 40, which were previously used in genetic studies but without known vic genotypes, were compatible with EU-1, EU-26, and EU-5, respectively.
Genotypes and tester isolates of 64 C. parasitica vc types in this and previous studies
There was no evidence for linkage among any of the six vicloci (Table 5). Data from multiple crosses with segregation at the same two vic loci were tested for homogeneity (α = 0.10) before pooling (31).
Analysis of linkages among six vic loci identified in C. parasitica
DISCUSSION
Genetic analyses of vc types in C. parasitica in Europe are consistent with a model of six unlinked vic loci, each with two alleles. Our results confirm that vegetative incompatibility is controlled by allelic interactions. Four of thevic loci we identified in this study were previously described by Huber (16); therefore, we have identified two additional vic loci. We concur with Huber (16) that—except for vic1 and vic2—the tentative genotype assignments made by Anagnostakis (1) and Rizwana and Powell (29, 30) should be disregarded and replaced by Huber’s nomenclature (16). Therefore, we called the two newvic loci found in this study vic6 andvic7. Incompatibility caused by an additional viclocus, vic5, cannot be detected on agar media (16; unpublished results) and was therefore disregarded in this study.
There was no evidence for multiple alleles at any vic locus for the vc types found in the field in Europe (EU-1 to EU-31). Thirty of the 31 European vc types were among the progeny of cross MJ1, in which alleles segregated at each of five vic loci; there was no segregation at vic3 in this cross. EU-10 was the only vc type found in the field in Europe that was not found among the progeny of cross MJ1 (Table 3). EU-10 was shown to have allelevic3-2 (cross P33, Table 2), but all of the other European vc types had allele vic3-1 (Table 2). Therefore, since all 64 vc types were either found in the field or derived from field isolates, only two alleles were found for each vic locus.
Our results (Table 2) are fully in agreement with Huber’s (16) and agree with six of the nine crosses reported by Anagnostakis (1, 2). However, our results disagree with three of Anagnostakis’ crosses. In general, we predict fewer vc types segregating than were observed in some crosses. We speculate that the excess vc types found in previous studies were artifacts caused by vc assays. Recent improvements in vc testing techniques have reduced these ambiguities considerably (9, 28). First, Anagnostakis (1) reported finding only parental vc types in a cross between v-c 5 and v-c 16. In contrast, Rizwana and Powell (29) found 17 nonparental vc types among the progeny of a cross between v-c 5 and v-c 16; however, the ratios of progeny vc types in this cross were highly irregular. According to our interpretation, v-c 5 and v-c 16 are compatible with EU-42 and EU-46, respectively, which differ by three vic genes and should produce progeny in eight vc types; Huber (16) also found that these two vc types differed by vic1, vic2, andvic3. The second cross Anagnostakis (1) reported that differed from our results was between v-c 8 and v-c 17 (EU-15 and EU-26, respectively). Anagnostakis reported finding 22 vc types in the progeny, but we would predict only 16 because the parents differ at four vic loci (Table 4). Finally, the third cross, with which our results are not fully consistent, is between v-c 5 and v-c 10 (2). This cross has often been cited because it was the basis for estimating the minimum number of vic loci controlling vc types in C. parasitica. Anagnostakis (2) estimated segregation at seven vic loci after observing 106 vc types in the progeny of this cross. A lower estimate of five vic loci, however, was based on the observation that approximately 1/25 (11 of 263 and 37 of 973) of the progeny from this same cross were of one parental type. One thirty-second of the progeny would have been expected to be compatible with each parent if alleles had segregated independently at five vic loci (2, 3), whereas 1/128 would have been expected to be compatible if alleles had segregated at seven loci. Our results show that v-c 5 and v-c 10 (compatible with EU-42 and EU-1, respectively; Table 4) differ by only five, not seven, vic genes.
Six vic loci, each with two alleles, may be sufficient to explain much of the diversity of vic genotypes, or vc types, found in populations of C. parasitica in North America. Anagnostakis and Kranz (4) found 48 vc types in a small forest plot in Connecticut; other studies have shown similar levels of vc type diversity in North America (18, 20, 25). Our preliminary studies have shown that most of the 31 vc types found in one population in Maryland (20) are compatible with testers in the 64 genotypes found in this study (unpublished results). More testing is required to determine if additional vic loci are polymorphic in North America or if multiple alleles occur at known loci. In Asia, there are probably additional polymorphicvic loci since 131 vc types were identified among 231 isolates of C. parasitica (33). Furthermore, only two of the 71 vc types found in Japan are compatible with the 64 types from this study (21a). Unfortunately, the effort required to identify additional vic genotypes doubles with every locus that is identified. For example, one additionalvic locus in C. parasitica would bring the total number of possible vc types to 128 and two loci would raise it to 256. Alternatively, a third allele at any vic locus would increase the number of possible genotypes by 50%. Multiple alleles at several loci—in relatively high frequencies in populations—would be necessary to explain the diversity of vc types observed in Asia and the lack of compatibility with the 64 vc types we found. However, we found no evidence for multiple alleles in European populations.
Our study has significantly extended the findings of previous genetic studies on vc types in C. parasitica because we have successfully assigned 64 (26) vic genotypes to vc types (Table 4). To our knowledge, such extensive information onvic genotypes is not available for any other fungus. All 31 of the European vc types (EU-1 to EU-31) found in the field to date (10) have been assigned a vic genotype. Thirty of the 32 possible vc types (all except EU-42 and EU-43) defined by fivevic loci (vic1, vic2, vic4,vic6, and vic7) have been found in the field. Interestingly, allele vic3-2 was only found in two populations in southern Italy, in vc type EU-10 (9, 10); all of the other vc types found in Europe have vic3-1. Knowledge of vic genotypes will allow us to reanalyze vc type survey data more thoroughly (e.g., references 6, 9, and10). For example, we will be able to test whether vc types occur at frequencies that would be expected from random mating; this analysis requires estimates of vic allele frequencies along with vc type frequencies (see reference 23). In addition, comparison of the 64 vc types from this study with vc types found in the United States is revealing a high proportion of successful matches (unpublished results), which opens the possibility of reanalyzing vc type population data in the United States based onvic genotypes. Finally, with our knowledge of vicgenotypes, we can now examine the effect of each vic allele on virus transmission (17, 21; unpublished results) to make inferences about the potential of virus transmission in populations of C. parasitica.
ACKNOWLEDGMENTS
This research was supported by a NATO Cooperative Research Grant, USDA NRI competitive grant 93-37303-9035, and a fellowship from the National Research Council of Italy, Special Project RAISA, to M.G.M.
We thank David Huber for sharing his laboratory isolates and prepublication drafts; Sandy Anagnostakis for sending us vc testers from her collection; Paola Allegra, Luca Rancati, Sabrina Senzacqua, and Kazue Takeuchi for assisting with vc tests; and Marco Bisiach for helpful discussions.
FOOTNOTES
- Received 30 March 1998.
- Accepted 12 June 1998.
- Copyright © 1998 American Society for Microbiology