For fungi and many other colonial organisms, defining the units of natural selection is not always straight-forward [1,2]. The concept of the “individual” has been contentious in fungal biology [3] because filamentous fungi form interconnected, multinucleate networks often composed of genetically heterogeneous elements. Historically, two ideas have dominated most discussion of genetic individualism for fungi. An early view of the fungal individual known as the “unit mycelium” was that genetically distinct mycelia undergo frequent somatic fusion and anastomosis to form genetically heterogeneous networks as the individuals [4].

In this view of the mycelial individual, different genotypes cooperate to acquire nutrition, but not all genotypes contribute equally toward producing progeny. This idea was strongly contested by observations that showed successful somatic fusions occurred only between hyphae that were genetically identical or nearly so (the “individualistic mycelium”, Rayner and Todd [3], Rayner [5]. Under this view, genetic identity of the fungal colony is maintained by somatic incompatibility responses among unrelated mycelia [3,6,7]. The emergence of the “individualistic mycelium” concept resulted in the dismissal of somatic fusion as a major means of generating genetic variation, and rather, sexual recombination is generally thought to be the dominant mechanism generating genetic variation in fungi [8].

Mating systems in fungi may be loosely divided into several types, including obligate outcrossing (heterothallism) and self-fertile (homothallism). Homothallic strains can produce basidiospores that germinate into mycelia that may fuse freely and are self- fertile [9]. In contrast, outcrossing occurs when two genetically distinct, mating compatible strains undergo hyphal fusion to produce a genetically stable dikaryon. In outcrossing species, mating compatibility initially overrides somatic incompatibility among monokaryotic strains [10]. However, somatic incompatibility still acts to maintain identity of the newly formed dikaryotic “secondary mycelium” which is dominant life phase in basidiomycete fungi [6]. In heterothallic basidiomycetes, the successfully mated mycelium produces meiotically recombinant haploid basidospores, which must fuse with appropriate mating compatible strains in order to regenerate the dikaryotic phase.

Fungi in the genus Septobasidium all share an unusual symbiotic life history, which offers unique opportunities for exploring mating biology and fungal individuality. Unlike other Urediniomycetes (which include rusts and their relatives), Septobasidium spp. live symbiotically with scale insects growing parasitically on woody plants, and individual colonies vary in size from a few mm in diameter to nearly 2 m. The lifecycle of Septobasidium was first elucidated by Couch [11-13], who interpreted its life cycle as follows based on the “unit mycelium” concept (Figure 1): 1) Septobasidium basidiospores infect first instar insects that settle on the plant surface; 2) The fungi grow out of infected insects and fuse with fungi from adjacent infected insects to form a mat of hyphae that covers both infected and uninfected insects; 3) Mature colonies produce basidiospores that can infect the offspring of the uninfected insects associated with the colony. Because the growth of the Septobasidium colony depends on fusion between multiple fungal mycelia (each derived from independent basidiospore infections), colonies may be composed of multiple genetically distinct nuclei. Couch [11-13] initially interpreted his observations in line with a genetic mosaic concept with many strains making up each colony.

However, a single genetic unit concept may also be consistent with Couch’s observations. If infective spores do not actually represent different genotypes then the genetic identity of a unified mycelium would be maintained. This could result from homothallism or cryptic mitotic production of basidiospores. Alternatively, apparent fusion of genetically distinct strains may cause any of several antagonistic somatic incompatibility reactions resulting in either a) distinct genotypic patches within the larger colony or b) replacement of nuclei throughout the entire colony by a dominant pair of nuclei.

Figure 1: The lifecycle of the Septobasidium symbiosis. First instar scale insects newly emerged from eggs (A) may settle without encountering fungal spores and develop into healthy male (C) or female (D) scale insects. Insects that encounter budding basidiospores (B) become infected as the fungus invades the body cavity eventually filling it with haustoria (E). When the fungi have fully invaded the insect body hyphae begin to grow out of the insect and form a mat of mycelium that may cover adjacent insects (F). Fungi from infected insects fuse to form healthy mature fungal colonies composed of uninfected and infected insects (G).

Determining the mating system and genetic structure of colonies in Septobasidium is a necessary first step toward developing a model of the co-evolutionary dynamics in this potentially altruistic and mutualistic interaction. This study addresses genetic structure of Septobasidium colonies using DNA sequence data from multiple life history phases in consort with microscopic observation of nuclear status in hyphae. Two specific questions we addressed are 1) whether scale insects are infected by different fungal genotypes and 2) how genotypes are distributed within an adult fungal colony. Using microscopy and sequence data from multiple single spore isolates, we determined the minimum ploidy of spores and whether the spores represent recombinant genotypes. Using sequence data from multiple locations within a colony and from isolates of fungi within infected insects already fused to a colony we can detect intra-colony genetic variation. From infected insects that have not yet fused to the main fungal colony we can determine what fungal genotype or genotypes are the infective units. By comparing the results from each part of the fungal infection we can detect possible genetic mosaicism within the colony, and determine whether the mating system is likely to be homothallic or heterothallic. We demonstrate this approach to a single focal species, S. ramorum and also for nine other species.

Collection, culture, and DNA extraction from fungal material Septobasidium colonies were collected from Quercus, Nyssa, Acer, Cornus, Fraxinus and Liquidambar trees across the southeastern U.S. (Table 1). Infected branches were removed and stored at room temperature in paper bags for up to two weeks. Three small spatially separated sections (approximately 5mm in diameter) were sliced from the upper surface of each colony and floated on sterile water. Each rehydrated colony was then suspended, upper surface facing downwards, from the lid of a Petri dish containing Malt Yeast Agar (MYA - 15% g/L malt extract, 5% g/L yeast extract, and 15% g/L Agar). Spores that landed on the agar surface were allowed to germinate and produce colonies of fewer than 100 cells.

This usually occurred after 3-4 days. At least six yeast colonies, each derived from a single spore, were transferred to new plates and maintained as single spore isolates (SSI). From each colony that was collected, five tissue samples (TI) approximately 2 mm in diameter were removed from various locations within each colony and placed into elution buffer for DNA extraction using the Forensic DNA extraction kit (OmegaBioTek). Insects that had settled on the fungal mat of S. ramorum colonies but were not connected to it were individually removed and placed into elution buffer for DNA extraction using the Forensic DNA extraction kit (OmegaBioTek).

Insects that were embedded and fused within S. ramorum colonies were removed, slightly crushed with a needle and pulled across the surface of MYA plates. The smeared contents of the scale insects were then observed microscopically to detect growth from the liberated haustoria. Mycelia from the haustoria were then transferred to fresh MYA plates. Aerial hyphae were placed into CTAB buffer for DNA extraction (1% CTAB, 50 mM Tris pH 8, 10 mM Na2EDTA, 0.7 M NaCl). Voucher collections of each collection sample, as well as representative’s monokaryotic and dikaryotic cultures for each species are deposited at DUKE.

DNA was extracted following the Forensic DNA extraction kit (Omega Biotek) using instructions for whole blood preparation modified as follows. Tissue was disrupted by grinding with a pestle in 250 μl elution buffer in a microcentrifuge tube. Following disruption, 250 μl buffer BL and 25 μl of 100 μM Proteinase K solution were added before incubation at 70°C as described in the kit manual. DNA was collected using the spin coulombs and eluted in 50 μl of elution buffer.

DNA from SSI and aerial hyphae was extracted following the procedure of [14] with few modifications. Using a sterile loop, yeast cells were scraped from the surface of the MEA plates and placed into tubes with 400 μl CTAB and approximately 0.3 g sterile sand. Tubes were subjected to two minutes of high-speed vortexing to disrupt the yeast cells before addition of 400 μl of 24:1 isoamyl alcohol to chloroform. After brief mixing and high-speed centrifugation (10000 G) for 15 min, the aqueous layer was removed to a clean microcentrifuge tube. Then 0.6 volume of isopropanol was added and stored overnight at -4°C to precipitate DNA. After centrifugation and an ethanol wash, the DNA was in eluted in 100 μl sterile water. DNA from aerial hyphae was similarly extracted but rather than vortexing with sterile sand, the tissue was ground with a plastic pestle in a tube without sand.

PCR based identification of genotypes

We used polymerase chain reaction (PCR) to amplify at least two loci from each collection (Table 1). Primer sets were designed to amplify Septobasidium species (Table 2). Additional markers were developed in the focal species S. ramorum, by screening a clone library of genomic DNA fragments essentially as described in Morehouse et al. [15]. Genomic DNA from an SSI of S. ramorum (collection DAH045A) was partially digested using Sau3A. Fragments between 1000 bp and 500 bp were separated on an agarose gel and ligated into the p Zero plasmid vector (Invitrogen). The plasmid was used to transform TOP10 E. coli (Invitrogen) cells. After propagation, the plasmid was prepared with the QIAprep Spin Miniprep Kit (Qiagen) for sequencing with forward and reverse M13 primers. Clones from the resultant library were sequenced and those that had strong similarity to genes from other organisms were selected for primer design and use as genetic markers. Additionally, the marker STE20 was generated by an initial round of PCR with degenerate primers designed from GenBank comparisons [16] followed by cloning and primer redesign.

PCR amplification used Red Hot Polymerase (ABgene) and reactions conditions as follows: 4 min at 95°C, followed by 35 cycles of 30 s at 95°C, 30 s at annealing temperature (50°C for ITS and ASP25, 48°C for EF1α, β-tubulin and STK26, 46°C for STE20) and 45 s at 72°C then a final extension step for 10 min at 72°C. PCR products were purified using PCR purification kits (Qiagen) and used directly for sequencing reactions with Big Dye V.2 or V.3. Sequencing was conducted on an ABI 3700 DNA analyzer (Applied Biosystems/ Hitachi). Sequence chromatograms were viewed and edited using Sequencher 4.2 (Gene Codes Corp Ann Arbor MI). We determined genotypes using the presence of strong “double peaks” to indicate the presence of polymorphic sites. Restriction digests were designed using Sequencher 4.2 to select appropriate enzymes, and digested PCR products from TI were run on 3% agarose gels and visualized using EtBr staining.

Table 1: Septobasidium collections sequenced and genotype for each allele.

Table 2: Primers, range of amplicon length and nucleotide variation for loci sequenced across the genus Septobasidium.

Microscopic determination of nuclear state

Material from cultures and fresh collections were prepared for fluorescence microscopy using DAPI and EtBr staining (Table 1). Sample sections approximately 5 mm in diameter were removed from fresh collections and then placed on slides with 30 μl of 10 μg/ ml DAPI or 10 μg/ml EtBr and stored at room temperature in the dark for 15 min before being viewed. The samples prepared with EtBr did not require the 15 min waiting period. The material was usually squashed, but some preparations were made as transverse sections to identify each part of the colony. Nuclear status was then determined as binucleate, uninucleate or multinucleate. Hyphae and yeasts from cultures were similarly prepared and scored for the number of nuclei present between septa and within cells.

Marker development

DNA sequences from six variable loci, ITS, EF1α, β-tubulin ASP25, STK26, and STE20, were used as genetic markers in this study (Table 2). Primers for amplification of ASP25 and STK26 were developed from a S. ramorum genomic library. The 470 bp long locus - 8 designated ASP 25 was most similar to part of aspartic proteinases (e ≤ 10), and the -550 bp long locus called STK 26 was most similar to serine-threonine kinases (e ≤ 107). We found that STK 26 could also be amplified from S. taxodii and S. burtii without further primer modification. The STE20 locus did not contain any introns and was only 339 bp long, but in the four specimens sampled, 20 polymorphic sites were detected. The ITS, β-tubulin and EF1α loci each contained introns. The ITS sequences from S. burtii, S. castaneum, S. fumigatum, S. mariani and S. pseudopedicellatum had a group I intron at position 1506 (relative to Saccharomyces) in the SSU portion of the amplicon. β-tubulin genes all contained a single splicesosomal intron varying in size from 85-92 bp. The EF1α genes contained two introns varying in size from 88-135 bp.

Nuclear status, recombination and spatial genetic structure detected using SSI and TI

Colonies from every species except S. grandisporum showed a similar pattern of polymorphism among SSI and TI samples (Table 3). SSI from each colony showed a maximum of two alleles at a locus (detected as double-peaks on chromatograms) and showed recombination among loci (Table 4). In all species except S. mariani SSI were found to possess a single allele, while TI samples possessed multiple alleles except in the case of S. grandisporum (Figure 2 and Table 5).

Single spore isolates of S. mariani sometimes possessed multiple alleles, and TI of the single S. grandisporum colony sampled possessed a single allele. Fluorescence microscopy showed that SSI growing as yeasts were predominantly uninucleate (Figure 3), though due to their small size and considerable autofluorescence this distinction was not always obvious. S. fumigatum, S. grandisporum and S. sinuosum SSI that had assumed a filamentous growth form showed a range of nuclear states, sometimes uninucleate but often multinucleate (Figure 3). Determination of the nuclear status of fresh collections was difficult due to autofluorescence and apparent lack of penetration by the stains. Where discernable, S. apiculatum, S. ramorum and S. sinuosum each showed only binucleate cells.

S. ramorum genotypes infecting individual insects

DNA from free-living insects not yet fused to colonies was successfully amplified from 10 of 40 extracts. Due to the microscopic size of the infected scale insects, each DNA extract was sufficient for only one to three PCRs. For one collection, DAH042F, three fungal genotypes from infected insects were the same as the TI and three were monomorphic for the single sequenced locus. For collection DAH078B three insects were successfully amplified and each had the same genotype as the TI. However, an insect from an adjacent colony DAH078A possessed a recombinant genotype between TI from DAH078A and TI from DAH078B. All ten successfully isolated fungi from insects fused to colony DAH301 shared identical 3 locus genotypes with a TI from DAH301. Microscopy revealed that fungi isolated from infected insects of DAH301 were binucleate (Figure 3).

Table 3: Summary of data of ten septobasidiumspecies with different numbers of spores per basidium, including putative ploidy of single spore isolates (SSI) and number of alleles detected in tissue isolates (TI).

Table 4: Segregation of alleles recovered from SSI of S. ramorum colony 045A.

Figure 2: Polymorphism segregating in SSI from a fungal colony and contained in TI from that colony. PCR-RFLP shows two alleles detected from each TI with each allele segregating among SSI (A). Chromatograms showing the segregating sites in SSI and characteristic “double peaks” of heterogeneous TI. The highlighted text is the consensus, while the chromatogram labels reflect the orientation of the read (B).

Table 5: Alleles recovered from putatively diploid SSI of S. mariani colony 092.

Figure 3: Fluorescence microcopy of Septobasidium species. S. mariani (A) and S. ramorum (B) yeast isolates derived from basidiospores appear uninucleate while filamentous isolates from infected insects of S. ramorum (C) are binucleate. Filamentous isolates from basidiospores of S. fumigatum are uninucleate (D), binucleate (E) or multinucleate (F). Arrows indicate nuclei. Each scale bar is 10 μm.

Our data indicate that colonies of most Septobasidium species consist of single outcrossed dikaryotic individuals, and that basidiospores represent haploid recombinant progeny. In S. ramorum genotyping revealed that although colonies arecomposed of a single dikaryotic individual, fungi from young infected insects not yet fused to the rest of the colony represent recombinant or homozygous genotypes. Based on Couch’s observations of infection and our own results, basidiospores appear to be the most likely infective life stage in S. ramorum. These results are consistent with the modern view of genetic individualism as found in many other basidiomycetes.

Our results may also support Couch’s observations that a colony is generated via multiple basidiospores. First, the evidence from S. grandisporum suggests that this fungus produces basidiospores that have identical genotypes. Second, the data from the other species, especially S. ramorum, show that single dikaryotic genotypes persist despite frequent fusions with other genotypes. Our results do not support spatial structuring of genotypes within a fused colony or multiple genotypes sharing the entire colony freely. For S. grandisporum, our data suggest that this fungus is homothallic and produces basidiospores with identical genotypes. In other species, especially S. ramorum, single dikaryotic genotypes may sometimes persist despite frequent fusions with other genotypes. Our results do not support spatial structuring of genotypes within a fused colony or multiple genotypes sharing a common colony (genetic mosaicism).

Previous microscopic work reported by Couch [13] for S. grandisporum suggested that cells were mostly dikaryotic but sometimes uninucleate. This is consistent with our microscopic results, though we also detected multinucleate cells. It is notable that the spores from this species do not primarily grow as yeasts but instead form one or a few generations of bud cells followed by a filamentous growth habit typical of dikaryotic mycelium. Although, our results cannot provide direct evidence for mitotic production of basidiospores or self-fertilization in S. grandisporum, the lack of any detectable allelic variation within a colony suggests this species is homothallic.

The results from S. mariani also contrast with the other species by sometimes having diploid spores. This may be a result of either incomplete meiosis resulting in heterokaryotic multinucleate spores or the production of diploid spores. Cryptococcus neoformans is known to produce diploid strains from cross serotype hybridization [17] and Helicobasidium mompa, a fungus in the same sub-class as Septobasidium, was also demonstrated to have diploid nuclei with unusual segregation [18]. Our microscopy did not reveal obvious consistently binucleate yeast cells in cultures of S. mariani, but it is possible that these nuclei are cryptic because of their close spatial proximity in the small yeast cells. This issue could be resolved for S. mariani using transmission electron microscopy or light microscopic observations of the complete cell cycle. Interestingly, Olive [19] observed that S. apiculatum typically produced some binucleate basidiospores. Although we examined S. apiculatum, we did not detect recombinant binucleate spores in that species. Our S. mariani results serve to highlight the potential complexity of the fungal lifecycle.

The mechanisms generating and maintaining genetic uniformity of Septobasidium colonies are not entirely clear. Although inbreeding and restricted segregating variation have the potential to generate multiple origins of shared somatic compatibility groups [20], these are unlikely to explain fusions among progeny and parents in Septobasidium, given the high heterozygosity detected at other markers. The het loci responsible for somatic incompatibility in other filamentous fungi [21] may have modified function in Septobasidium or may have a different genetic architecture or selective regime leading to an extreme reduction in genetic variation relative to other loci.

Although our results conform to the idea that basidiomycete individuals are defined by genotype, some questions remain. Each Septobasidium colony represents a single genetic individual, but its stability may be at risk. Frequent non-self-fusions may make an individual vulnerable to invasion by other nuclei and cytoplasmic elements, disrupting the dominant dikaryon. Also, almost all of the resources available to a genetic individual must be acquired through non-self-fusions, making the genetic individual entirely dependent on other genotypes for survival. In some sense, this is a community of individuals providing benefit to a single dominant individual for at least one season of reproduction. From this perspective, Septobasidium conforms closely to a “unit mycelium” in which a community of genotypes function as an individual [22-27].

The genetic life-history of Septobasidium may disconnect aspects of the fungus - insect interaction. Each insect is infected by a unique genotype, but it is the dominant dikaryon alone that draws most nutrition from living infected insects and provides any potential protection to the scale insect colony. The origin of the dominant genotype is unclear, and it may originate from a single infected insect or arises from nuclear exchange among different fusing fungal genotypes later in colony formation.

Interactions between scale insects and Septobasidium affect each other’s reproductive fitness depending on whether the fungus can or cannot infect its insect host. Since only uninfected insects are fertile and able to contribute insect progeny to the next generation, a fungus that successfully infects (and sterilizes) an insect has five potential fates: 1) The fungus may not reproduce because it is unable to gain sufficient resources to fruit or is not an infection caused by a strain or strains capable of sexual reproduction; 2) The fungus may establish a new colony and fruit successfully without encountering another fungal genotype; 3) The fungus may be met and overrun by another genotype; 4) The fungus may invade other fungi and fruit successfully; 5) The fungus may partially invade another fungus by donating only one of its nuclei or its cytoplasmic components to a new dominant dikaryon. From these potentialities, it is clear that in order for an allele to propagate through the fungal population it must be able to infect insects, but the most infective genotypes are not necessarily successful at propagating through the population because the fungal competition component occurs after infection. The effect of the insects on fungal fitness at this stage is complex.

Another category of fungus/insect interaction is it that of fused insects with existing dikaryons rather than the dispersing fungal propagules. A dominant dikaryon does not affect the fitness of the fused scale insects directly because they are already sterilized. However, the dominant dikaryon may have a strong impact the fitness of the fused scale insects indirectly by modifying the fitness of uninfected scale insects associated with the fungal colony. Fungi can modify that fitness component through multiple avenues, including protection from predators, desiccation and the funguses own infective spore production. A colony of fused infected scale insects can affect the existing dikaryon’s fitness by modifying the production of spores or the dikaryon’s competitive ability in non-self-fusions.

The Septobasidium symbiosis is potentially mutualistic or cooperative at multiple levels of genetic organization. That individual insects may be sacrificed for the greater good of the nearby insect colony is nearly mirrored by the infective fungi associated with them. The nuclei of these fungi, derived from meiotically produced basidiospores, may be sacrificed for the benefit of the fused fungal colony. The individuals sacrificed are likely to be very closely related to the individuals receiving benefit. Coevolution may have a prominent role in the loss of “individualistic mycelium” behavior because of the Septobasidium’s dependence on infecting individual scale insects [28- 31].

We wish to thank the Duke Mycology Discussion Group, Paul Manos, John Willis, François Lutzoni and Cathie Aime for helpful comments on this manuscript. We especially wish to thank Dr. Austen Ganley and Dr. Tim James for their helpful advice during the planning of this research. We also thank Lisa Bukovnik and the Duke DNA sequencing facility. We gratefully acknowledge the support of NSF grant DEB-0408011, and funding from the Duke Biology department.