Polymorphic Microsatellite Markers for the Tetrapolar Anther-Smut Fungus Microbotryum saponariae Based on Genome Sequencing

Background Anther-smut fungi belonging to the genus Microbotryum sterilize their host plants by aborting ovaries and replacing pollen by fungal spores. Sibling Microbotryum species are highly specialized on their host plants and they have been widely used as models for studies of ecology and evolution of plant pathogenic fungi. However, most studies have focused, so far, on M. lychnidis-dioicae that parasitizes the white campion Silene latifolia. Microbotryum saponariae, parasitizing mainly Saponaria officinalis, is an interesting anther-smut fungus, since it belongs to a tetrapolar lineage (i.e., with two independently segregating mating-type loci), while most of the anther-smut Microbotryum fungi are bipolar (i.e., with a single mating-type locus). Saponaria officinalis is a widespread long-lived perennial plant species with multiple flowering stems, which makes its anther-smut pathogen a good model for studying phylogeography and within-host multiple infections. Principal Findings Here, based on a generated genome sequence of M. saponariae we developed 6 multiplexes with a total of 22 polymorphic microsatellite markers using an inexpensive and efficient method. We scored these markers in fungal individuals collected from 97 populations across Europe, and found that the number of their alleles ranged from 2 to 11, and their expected heterozygosity from 0.01 to 0.58. Cross-species amplification was examined using nine other Microbotryum species parasitizing hosts belonging to Silene, Dianthus and Knautia genera. All loci were successfully amplified in at least two other Microbotryum species. Significance These newly developed markers will provide insights into the population genetic structure and the occurrence of within-host multiple infections of M. saponariae. In addition, the draft genome of M. saponariae, as well as one of the described markers will be useful resources for studying the evolution of the breeding systems in the genus Microbotryum and the evolution of specialization onto different plant species.


Significance
These newly developed markers will provide insights into the population genetic structure and the occurrence of within-host multiple infections of M. saponariae.In addition, the draft

Introduction
Anther-smut fungi belonging to the genus Microbotryum sterilize their host plant by aborting ovaries and replacing pollen by fungal spores.Sibling Microbotryum species are highly specialized on their respective host plants and show strong post-mating reproductive isolation [1][2][3].Microbotryum fungi have become model species in ecology and evolution of plant pathogenic fungi [4,5].Although anther-smut disease occurs in hundreds of plant species in the Caryophyllaceae [6] and at least nine other plant families [7,8], most studies to date have focused on M. lychnidis-dioicae, which parasitizes the white campion Silene latifolia.This is the first fungal species in which distinct mating types were described [9], it has been used in classical genetic analyses for decades (e.g., [10][11][12]), and more recently as a model in disease transmission and metapopulation dynamics (e.g., [13,14]).
More recently, population genetics studies using microsatellite markers allowed elucidation of the genetic population structure of M. lychnidis-dioicae in Europe [15][16][17], patterns of postglaciation expansion [18], and the invasion history of M. lychnidis-dioicae in North America [5,19].The use of microsatellite markers also revealed a high prevalence of multiple infections by different fungal genotypes, which segregated in different stems of the same Si.latifolia host [20].The co-infecting genotypes also revealed a high level of relatedness, suggesting competitive exclusion between unrelated fungal genotypes [21][22][23].In addition, a reference genome of M. lychnidis-dioicae is now available [24,25], which provides opportunities for examining genomic changes within the genus.In particular, in this species, the two independent mating-type loci typical of basidiomycetes are trapped within a very large region of suppressed recombination on what have thus become the sex chromosomes, spanning ca.90% of the chromosome length.These genomic data are consistent with previous findings based on marker segregation analyses and optical maps [26][27][28].This large non-recombining region explains why M. lychnidis-dioicae is bipolar, with a single mating-type locus.Bipolarity is the rule for most Microbotryum species studied so far, suggesting similar linkage between mating-type loci [29].
Many resources are therefore available for the species M. lychnidis-dioicae, while the other anther-smut fungi remain poorly studied.Other Microbotryum species, however, deserve further investigations [7,30], and in particular M. saponariae, parasitizing Saponaria ocymoides and Sa.officinalis [2,8].The common soapwort Sa. officinalis, with a broad distribution in Europe, is a large long-lived perennial plant with multiple stems, which makes it a particularly interesting subject for the study of multiple infections (i.e., multiple pathogen genotypes coexisting within a host) and phylogeography.In addition, M. saponariae has been recently identified as one of the few tetrapolar species (i.e., maintaining the independence of the two matingtype loci), recently identified, within the anther-smut Microbotryum fungi [29].Of the available microsatellite markers for various Microbotryum fungi, very few cross-amplify on M. saponariae [15,17,31], and only four were polymorphic.Therefore, our aim here was to develop polymorphic microsatellite markers by generating a draft genome of M. saponariae using mate-pair Illumina sequencing.A similar approach to develop SNP markers has been used in other plant pathogenic fungi for studying multiple infections [32].
Microsatellite loci have been widely used as genetic markers due to their ubiquity, reproducibility, neutrality and high level of polymorphism [33].These features make them valuable molecular markers for inference of evolutionary and demographic parameters [34].However, for each new species, microsatellites often need to be isolated de novo and the characterization process of each locus can be time consuming and expensive given the need for numerous, locus-specific fluorescent primers.Recent methods using more simple and inexpensive approaches allow the isolation of polymorphic markers for large-scale multiplex assays.Here we isolated microsatellite markers using a high resolution and rapid approach with generic vector-specific fluorescent primers and unlabelled locus-specific primers [35], which makes the costs considerably lower.
The isolated microsatellite markers will allow studies for unravelling patterns of infection (i.e.single or multiple genotypes) within host, population structure and phylogeography of M. saponariae.In addition, the molecular and genomic resources obtained here will contribute to increasing our knowledge on the evolution of reproductive systems and host specialization of anther-smut fungi.

Genome Sequencing
The Microbotryum saponariae genome sequenced in this study was originated from a fungal individual (strain 1053) collected on a host plant Sa.officinalis growing on the University Paris-Sud campus (Orsay, France) in 2012 and is available upon request.Diploid teliospores produced in the anther of a diseased plant were collected and cultivated on a nutrient-rich medium, as Potato Dextrose Agar (PDA) at 23˚C for a few days.On nutritive media, fungal teliospores germinate and undergoes meiosis, each producing four haploid products, called sporidia, that replicate clonally.The sporidia harvested from PDA plates represent therefore the haploid products of many independent meiosis events of a single diploid fungal genotype.After adequate growth on PDA, sporidia were collected and stored in silica gel at -20˚C until use.Fungal DNA was extracted using a hydraulic press (Carver, Inc., Wabash, USA) to lyse the cell walls and the Genomic-Tip 100/G kit (Qiagen 1 , CA, USA) following the manufacturer's protocol.DNA purity was assessed by measuring 260/280 and 260/230 absorbance ratios with a NanoDrop 1 spectrophotometer (2000; Thermo Scientific, Wilmington, USA), and DNA concentration was measured using a Qubit 1 fluorometer (2.0; Thermo Scientific).Preparation of DNA libraries was performed by the GeT-Place INRA platform, Toulouse, France.Mate-pair libraries with insert sizes of 3kb were prepared with Illumina Nextera TM Mate Pair Sample Prep kits, and sequencing was performed on a HiSeq 2000 Illumina TM sequencer at a depth of coverage of 100X on average.The genome sequence of M. saponariae is available in GenBank under the accession number PRJEB11435.

Detection of Microsatellite Loci in the Genome and Primer Design
Simple repeats were detected with RepeatMasker (v4.0.5; [36]) on the assembled genome.Only perfect di-and tri-nucleotide repeats of 8 to 12 repetitions were considered.Flanking regions of 200 bp were selected on each side of the nucleotide repeat.Thus, repeats located less than 200 bp from the contig extremity were discarded.Primers were designed using Primer3 [37] on 100 to 210 bp long fragments containing the target microsatellite sequences (S1 Table ).For each microsatellite locus, up to five primer pairs were mapped to the target genome with GMAP (version 2014-10-16; [38]) using default parameters, and only pairs mapping to a single locus were further considered.), and DNA was extracted with the NucleoSpin1 Soil kit (Macherey-Nagel, Germany).To detect polymorphic microsatellite markers, we used a simple and inexpensive approach that avoids generating fluorescent labels for all tested markers [35].The eleven DNA samples, each at 5 ng.μL -1 , were pooled as a template DNA for downstream applications.PCR amplifications were performed for each microsatellite marker with unlabelled specific primers using a high fidelity Taq DNA Polymerase Taq Pfu1 (Promega, USA).We used a PCR touchdown program with 94˚C for 5 min, 35 cycles of 94˚C for 30 s, 62˚C for 30˚C with a decrease of 1˚C at each cycle during 12 cycles and 50˚C during 23 cycles, 72˚C for 45 s, followed by 72˚C for 5 min.Successful PCR amplifications were confirmed by 2% (w/v) agarose gel electrophoresis.PCR products for multiple individual microsatellite loci, amplified from the same DNA pool, were combined and purified using NucleoSpin1 96 PCR Clean-up (Macherey-Nagel, Germany), before ligation into pDrive vector (Qiagen 1 , USA).Ligation products were diluted 1/10 with deionised water and amplified again for each marker using the specific 'reverse' microsatellite primer and a universal fluorescently labelled primer, M13 forward (-40; 5' GTTTTCCCAGTCACGAC 3'), targeting the plasmid insert flanking region.We used the same PCR touchdown program as described above.Labelled amplicons were diluted 1/225 and fractionated by capillary

PCR Multiplex Assembling
For the assignment of PCR multiplexes we defined four amplification size ranges: 1) 100 to 115 bp; 2) 125 to 150 bp; 3) 155 to 170 bp; and 4) 180 to 210 bp.These groups were defined to separate the microsatellite markers by size and thus avoid interference between markers within each multiplex.For each multiplex, compatibility of the primer pairs for each locus was tested with the software Multiplex Manager (1.2; [39]).As only four different colours of dye are available we set the maximum number of loci per reaction to four and a complementary threshold equal to seven (default parameter).Primer pairs were then successfully combined into six multiplexes, ranging from three to four markers per multiplex (Table 1).

Polymorphism of the Markers in Natural Populations
We further tested the polymorphism of the 22 microsatellite markers retained after the first screening described above, using 97 fungal individuals collected from 92 populations of Sa. officinalis and 5 populations of Sa. ocymoides across Europe (Fig 1, S2 Table ).In addition, 24 individuals of a population of M. saponariae (population 1188, in Bures-sur-Yvette near the University Paris-Sud Campus, S2 Table ) were also genotyped in order to assess within-population deviation from Hardy-Weinberg equilibrium.Diploid teliospores were harvested from one anther of a diseased sporulating flower.Fungal DNA was extracted in 100 μL using Chelex (Bio-Rad 1 , USA) protocol previously described [40], and ten-fold diluted for PCR amplification.Fungal samples were genotyped using the 6 multiplexes of microsatellite markers developed above.PCR reactions were performed separately for each multiplex in 12,5 μL volume containing 3 μL of DNA, ¼ of the recommended volume of multiplex PCR Kit (Qiagen 1 , USA), and 1.25 μL of the primer mix.The primer mix included 2 μM of unlabelled forward and reverse primer and 0.5 μM or 0.75 μM of labelled forward primer depending on the dye label.FAM and Yakima Yellow dyes were at 0.75 μM, while ATTO 550 and ATTO 565 dyes were at 0.5 uM.The PCR cycling program used was the same as described above.After checking for successful PCR amplifications on 2% (w/v) agarose gel electrophoresis, PCR amplicons were diluted and fractionated by capillary electrophoresis at the GENTYANE INRA platform as previously described.Alleles of each locus were scored using GENEMAPPER 5.0 (Applied Biosystems TM , USA).

Cross-Species Amplification
To test the microsatellite markers developed for M. saponariae on other Microbotryum species, we sampled fungal individuals (S2  ).The success of amplification was checked on 2% (w/v) agarose gel electrophoresis and PCR amplicons were submitted to capillary electrophoresis at GENTYANE platform.For each locus, alleles were scored as previously described.

Descriptive Statistics
The observed and expected heterozygosity (H O and H E ), departure from Hardy-Weinberg expectations (F IS ), linkage disequilibrium (LD) among locus pairs and genotypic disequilibrium were calculated using GENEPOP [41] on the web at http://genepop.curtin.edu.au using the sample of one individual per population.Furthermore, F IS was calculated within the population of M. saponariae for which all individuals were genotyped.The nominal P-value of 0.05 was adjusted for multiple comparisons using a Bonferroni correction (i.e., 2.3x10 -4 based on 219 tests).

Results and Discussion
The search for microsatellite loci in the M. saponariae draft genome yielded 188 loci meeting our criteria.We chose 96 loci located on different scaffolds and with different amplification sizes for designing multiplexes, with half of di-and half of tri-nucleotide repeats.Out of the 96 tested markers, four yielded no amplification.We thus screened six multiplexes of three to four microsatellite markers (Table 1, S1 Table ) for polymorphism based on pooled DNA of a sample subset.We selected among those showing the highest level of polymorphism and the most easily readable peaks on the capillary electrophoresis chromatograms (S1 Fig).
We further tested the polymorphism of the markers retained after the first screening described above.A total of 22 microsatellite markers were found polymorphic and easy to score (Table 1).The number of alleles per locus varied from 2 to 11, for a total of 51 alleles observed.Descriptive statistics on the polymorphism of M. saponariae and on deviations from Hardy-Weinberg equilibrium are shown in Table 1.The mean observed (H O ) and mean expected heterozygosity (H E ) values ranged from 0 to 0.97 and from 0.01 to 0.58, respectively.Most markers (20 out of 22, i.e., 91%) displayed significantly lower levels of heterozygosity than expected under Hardy-Weinberg equilibrium.The fixation index (F IS ) ranged from -0.95 to 1 for a mean multilocus value of 0.61 (±0.10).Microbotryum saponariae exhibited only 11% of heterozygous genotypes while 21% were expected under Hardy-Weinberg equilibrium, which suggests high selfing rates as previously reported in other Microbotryum species [16,40,42].Only one marker, out of the 22, Msap_146, displayed higher levels of heterozygosity than expected (extreme FIS value of -0.95) with most individuals being heterozygous.Further analysis on the marker position in the M. saponariae genome revealed that it is located on the contig carrying the pheromone receptor gene.Most likely this marker is situated in the nonrecombining region of the mating-type chromosome, which leads to permanent heterozygosity and high differentiation between the two mating types [25,29].This marker will be useful for studies on mating types and mating system, but should not be used in analyses assuming Hardy-Weinberg equilibrium.Deficits in heterozygotes may also results from a Wahlund effect, which would be expected because we genotyped here only one individual per population and populations were sampled at a large geographical scale.If allele frequencies vary among populations, our pooled sample would necessarily have a deficit of heterozygotes [43].In fact, the genotyping of a whole population showed little deviation from Hardy-Weinberg expectations (mean F IS across loci, excluding SAP_146, of -0.03).On the other hand, despite a likely Wahlund effect, the heterozygote deficit observed here was lower than those previously reported within populations of M. lychnidis-dioicae (e.g., mean 0.96±0.06 in M. lychnidis-dioicae; [19]), suggesting lower selfing rates in M. saponariae than in M. lychnidis-dioicae.This is consistent with M. lychnidis-dioicae being bipolar and M. saponariae tetrapolar [44].Bipolarity is indeed thought to evolve under selfing mating systems while tetrapolarity is beneficial under outcrossing, increasing discrimination against self [44].
After Bonferroni correction, significant linkage disequilibrium was observed only between six pairs of markers out of the 219 comparisons: Msap_11 and Msap_94, Msap_11 and Msap_31, Msap_4 and Msap_96, Msap_4 and Msap_27, Msap_96 and Msap_27, Msap_90 and Msap_77.This linkage disequilibrium may be due to genetic linkage between these markers or to population structure.Indeed, we pooled individuals from multiple populations and differences in alleles frequencies between populations will lead to apparent linkage disequilibrium when analysed as a single population.
Cross-species amplification was further examined among nine other Microbotryum species and was successful at 6 to 19 loci (Table 2).Some markers were amplified in all species, as for Table 2. Results of cross-species amplification of the 22 microsatellite loci in nine additional Microbotryum species.No amplification (-); successful amplification is given by the allelic profile of homozygous (colour scale) or heterozygous individuals (white).The name M. violaceum s.l.refers to the species complex name and it is indicated when the fungal species has not yet been described.The colour scale illustrates the allele size: darker colour corresponds to larger amplification products.instance Msap_146 and Msap_73, while others, e.g.Msap_11 and Msap_4, were only amplified in the species most closely related phylogenetically to M. saponariae, such as the Microbotryum species on Dianthus hosts [3].In general, the size of the alleles of each marker was different in the other species, except for nine markers where the alleles had identical size compared to those identified in M. saponariae.In some species, such as M. scabiosae, few markers could be amplified, unsurprisingly, given the high genetic distance from M. saponariae species [8].Microsatellite markers in fungi are known to be difficult to amplify in other species, even closely related ones [45].Among Microbotryum species, the divergence between the sister species M. lychnidisdioicae and M. silenes-dioicae has been estimated to be 420,000 years old [46].All the other species pairs studied here had older divergence times, as shown by phylogenies [3,7,30,46,47].

Conclusion
The method used for screening polymorphism at microsatellite loci [35] was efficient and powerful.Using it enabled us to isolate 22 polymorphic markers grouped in 6 multiplexes without needing to develop the 96 specific fluorescent primers.The developed microsatellite markers will be very useful for future studies on the population structure and phylogeography of M. saponariae and for comparison with patterns in M. lychnidis-dioicae [18,40,48].Furthermore, knowing the genome sequence and population structure of M. saponariae will allow the comparison of polymorphism at genes evolving under particular selection regime, such as the mating-type genes [49,50].Microsatellites are also rapidly evolving markers that may reveal cryptic genetic subdivisions among populations parasitizing the two closely related host species Sa. officinalis and Sa.ocymoides, whose relationship remains unclear.It will also be interesting to assess whether M. saponariae is more outbreeding than the other anther-smut fungi studied so far [51].Tetrapolar species are indeed usually outcrossing [44], and M. saponariae may reveal the first known tetrapolar selfing species [29].
The microsatellite markers will also be useful for studying the occurrence of multiple infections in Sa. officinalis, for assessing whether the results found in Si. latifolia [21,22] and in Si. acaulis [34] are general.In particular, we are interested in knowing whether the large long-lived Sa. officinalis plants carry multiple genotypes of M. saponariae, if they segregate spatially in different stems, and if genotypes co-occurring within the same host plant are significantly related.
In addition, the genomic resources released here will be useful for studies of fungal comparative genomics of anther-smut fungi and even more broadly.Fungi are indeed very useful models for tackling questions of genomic evolution and adaptation [52].The published genome will help in particular studying mating-type chromosome evolution in the Microbotryum genus [25,26,29] and host specialization [1,30] in plant pathogenic fungi.
Chelex protocol (Biorad, USA) and was used to test the polymorphism of the 22 microsatellite markers in M. saponariae and in the other Microbotryum species used for cross-species amplification.Symbols ( Ã ) refers to DNA samples that were also extracted with the Nucleospin Soil kit (Macherey-Nagel, Germany) and were pooled as a template DNA for downstream applications in the first screening of markers polymorphism.(PDF)

For a first
screening of polymorphism, a sample subset of eleven M. saponariae individuals were selected from different geographical locations across Europe (Fig 1, S2 Table

Fig 1 . 1 )
Fig 1. Map of the geographical location of Microbotryum saponariae individuals sampled in 97 populations of the host plants: 1) Saponaria officinalis (red circles, n = 92), including those used for the first polymorphism screening (green circles, n = 11), and 2) Saponaria ocymoides (blue circles, n = 5) in Europe.Fungal DNA was extracted from a sporulated anther of a diseased plant and was used to test the polymorphism of the 22 microsatellite markers.doi:10.1371/journal.pone.0165656.g001

Table 1 .
Summary statistics for the 22 microsatellite markers in Microbotryum saponariae (N = 97).Repeat: Motifs and repeat numbers of the microsatellite markers in the individual genome from which the markers were developed.Range: allelic size range in number of base pairs (bp).Multiplex: number indicating to which multiplex group the marker belongs.Multiplex 3 a and b were kept separate during PCR amplification to avoid the amplification interference that occurred when all four markers were amplified together, but they were pooled for genotyping in the end.Dye: forward primers of the selected markers were labelled with the following fluorescent dyes: Yakima Yellow equivalent to VIC, ATTO 550 equivalent to NED, ATTO 565 equivalent to PET.Alleles: number of alleles observed in the sample.H O : observed heterozygosity.H E : expected heterozygosity; Symbols (*) and (+) show significant deficit or excess in heterozygosity compared to Hardy-Weinberg expectations (* P < 0.05; ** P < 0.01; *** P < 0.001).F IS : Fixation index.SE: standard error.