Fusarium oxysporum f. sp. phaseoli genetic variability assessed by new developed microsatellites

Abstract Fusarium oxysporum f. sp. phaseoli (Fop) J.B. Kendrich & W.C. Snyder is the causal agent of Fusarium wilt of common bean (Phaseolus vulgaris L.). The objective of this study was to develop microsatellite markers (SSRs) to characterize the genetic diversity of Fop. Two libraries enriched with SSRs were developed and a total of 40 pairs of SSRs were characterized. Out of these, 15 SSRs were polymorphic for 42 Fop isolates. The number of alleles varied from two to ten, with an average of four alleles per locus and an average PIC (Polymorphic Information Content) of 0.38. The genetic diversity assessed by microsatellites for Fop was low, as expected for an asexual fungus, and not associated with geographic origin, but they were able to detect enough genetic variability among isolates in order to differentiate them. Microsatellites are a robust tool widely used for genetic fingerprinting and population structure analyses. SSRs for Fop may be an efficient tool for a better understanding of the ecology, epidemiology and evolution of this pathogen.

Introduction Fusarium oxysporum f. sp. phaseoli (Fop), causal agent of fusarium wilt in common bean, is currently considered one of the most important bean diseases (Ramalho et al., 2012). The Fop fungus is the main bean soil pathogen and has been established in all bean producing areas of Brazil. Losses caused by this disease have been increasing mainly in areas under successive and irrigated plantations (Toledo-Souza et al., 2012). In addition to the lack of chemical control, this scenario is aggravated by the production of chlamydospores, fungal resistance structures that survive for many years in the soil, even in the absence of the host. For these reasons, the main alternative to control this disease is to obtain resistant cultivars (Cross et al., 2000) that are easily adopted by producers and do not present environmental risks (Gonçalves-Vidigal et al., 2013).
Fusarium species that cause vascular wilting are all classified as Fusarium oxysporum (Pereira, 2009), which forms a complex of soil fungi composed of patotypes classified into various forma specialis, based on pathogenic criteria. They are responsible for the disease in more than 100 plant species (Pantelides et al., 2013). Each forma specialis group is pathogenic to specific plant group, demonstrating the degree of host specificity (Nelson et al., 1983).
Fusarium oxysporum sp. phaseoli can penetrate an intact root tissue, but also penetrate into more developed parts of the root and hypocotyl tissues also occurs, usually through injury or natural openings (Paula Júnior et al., 2006). Fop penetrates plants through the root system and colonizes the xylem, causing wilting, vascular discoloration, chlorosis, dwarfism and premature plant death (Nelson et al., 1983).
In resistant plants, symptoms are few or less expressive; occluding material is observed in the xylem vessels of inoculated plants (Pereira et al., 2013) and it has been associated to a delay in fungal colonization in the host. physiological races are described, but all authors report that further studies are needed.
There is a preference for microsatellite markers or simple sequence repeats (SSRs) in contrast to other types of markers, since they use the agility of the PCR technique, are codominant and randomly scattered in the genome with a relatively high frequency as well (Karaoglu et al., 2005). DNA sequences flanking microsatellites are generally conserved among individuals of the same species, or even between related species. These sequences are made up of one to six nucleotide repeats that occur naturally in the genome.
Microsatellites in fungi are more difficult to isolate and exhibit a lower polymorphism than in other organisms (Dutech et al., 2007). Bogale et al. (2005) described nine SSR markers developed for the study of Fusarium oxysporum. According to Cruz et al. (2018) there is a positive correlation between Fusarium diversity and its virulence in common bean. A major drawback of this study was that they used only seven polymorphic microsatellites out of eighteen tested to build the clustering and sustain their conclusions of positive correlation between virulence and diversity. A limitation of the use of SSRs in F. oxysporum f. sp. phaseoli, in general, is that very few markers were developed for the study of this fungus. Developing SSR markers is very laborious and costly because it traditionally involves the screening of enriched genomic libraries (Zane et al., 2002). Accurate identification and knowledge of the genetic diversity of pathogenic Fusarium oxyporum f. sp. phaseoli is important in the management of the disease and a large set of microsatellites in needed for providing full genome coverage.

Material and Methods
For the present study, 42 pathogenic isolates of F.oxysporum f. sp. phaseoli were collected in different states of Brazil. The study of diversity and genetic structuring was carried out with the 42 Fop isolates, 3 isolates collected in the State of Goiás, 24 isolates collected in the State of São Paulo, 1 isolate collected in the State of Pernambuco, 10 isolates collected in the State of Minas Gerais, 2 isolates collected in the State of Santa Catarina and 2 isolates collected in the State of Paraná (Table 1).
Among the 42 Fop isolates used, 29 isolates belong to the mycoteca of the Phytopathology Laboratory of the Grains and Fiber Centre of the Agronomic Institute (IAC, Campinas, SP), five isolates belong to the micro-library of the Phytopathology Laboratory of EMBRAPA Rice and Beans (Goiânia, GO), and eight isolates belong to the Phytopathology Laboratory of the Federal University of Viçosa -UFV (Viçosa, MG, Table 1). The fungi were maintained in PDA (potato-dextrose-agar) culture medium in a growth chamber incubator at 24°C.
In this study, two monosporic isolates of the pathogen were classified into physiological races according to the classification system proposed by Alves-Santos et al. (2002b). The same classification system and races were also used in the evaluation of 26 common bean genotypes for resistance to Fusarium oxysporum f. sp. phaseoli by Azevedo et al. (2015). Regarding the origin of the isolates, the first one named IAC 11173 (ID:3) was characterized as being from the American race or race I and the second one, named IAC 11233 (ID:5), characterized as a Brazilian race or race II, obtained from common bean plants with characteristic symptoms of the disease, collected in the municipalities of Angatuba and Capão Bonito, in the State of São Paulo, respectively.
Isolates IAC 11848 (ID:10) and IAC 12802 (ID:12) were used by Silva et al. (2014) in phylogenetic studies from molecular analyses of partial sequences of the elongation factor gene (EF-1a gene) and ribosomal intergenic spacer of Fusarium oxysporum f. sp. phaseoli identified as races 2, 3 and 6, respectively, which were previously classified by  using the methodology of Alves-Santos et al. (2002b).
Two enriched microsatellite libraries (adapted from Billotte et al., 1999) were developed, one for race I (IAC 11173, American) and one for race II (IAC 11233, Brazilian) for Fop. Genomic DNA was extracted from fungi mycelium, according to the instructions of the Wizard Genomic DNA Purification kit (Promega). DNA was digested with an Afa I restriction enzyme (10 u/mL) (Invitrogen), followed by the ligation of the Afa 21 adapter fragments (5CTCTTGCTT ACGCGTGGACTA3) and Afa 25 (5TAGTCCACGCGT AAGCAAGAGCACA3). The enrichement process was performed by means of the hybridization of the probes conjugated with biotin, (CT) 8 and (GT) 8 .
The fragments enriched for microsatellites were cloned into the plasmidial vector pGEM-T and the transformation to competent cells was performed using a protocol adapted from Avi Levy, 1991 (personal communication). For transformation, 100 mL of E. coli competent cell solution (JM109, Promega) was taken from the -80°C freezer and placed on ice. Then 2 mL of the ligation and 8 ml of Transfobuffer (10X KCM and 10% PEG) were added, gently shaken, and left on ice for 30 min. After that, the solution was removed and kept at room temperature for 10 min. Then 450 mL of S.O.C. medium (Thermo Fisher Scientific) was added and incubated at 37°C for 1 hour in a shaker at 225 rpm. Transformed cells were plated in solid LB medium containing ampicillin (50 mg/mL), 60 mL IPTG (24 mg/mL), 60 mL X-Gal (20 mg/mL). One volume of each (bacteria, X-Gal, IPTG) was placed on an opposite portion of the plate, spread with a Drigalsky handle until dry, and incubated overnight at 37°C. The plates were then refrigerated for 1-2 h so that the colonies turned blue. Positive clones were selected using the bgalactosidase gene. The clones from each library (race I and race II) were sequenced in a 3730 DNA Analyzer (Applied Biosystems). The SSRs were designed using Primer3 Software.
The amplification reactions of the multiplex SSRs were performed in a final volume of 15 L, containing 6.5 mL PCR Master Mix 2x kit (Fermentas), 1 mL DNA (50ng) of each Fop isolate, and concentrations of the individual primer pairs (10 pmol each, Table 2), depending on the intensity of the amplified product. The amplifications were carried out in a thermocycler, with an initial denaturation step of the DNA at 94°C for 4 minutes, followed by 34 cycles at 94°C for 30 seconds, with an annealing temperature for 1 minute and extension of 72°C for 1 minute. At the end of the cycles, there was another extension at 72°C for 10 minutes.
For the automated analyzer genotyping, multiplex sets were formed, containing three to four microsatellite loci (1st multiplex: FOP 01-B01, FOP 10-B01, and FOP 13-B01; 2nd  multiplex: FOP 07-B1, FOP 15-B01, and FOP 20-B01; 3rd multiplex: FOP08-B02, FOP11-B02, FOP18-B02, and FOP23-B02; the other primers were used alone). The selection of the oligonucleotides used in the multiplex assembling considered that the same primers do not have complementarity between their bases and the fluorescences had different colors. After assembling the multiplex system, the 5' end labeling of the oligonucleotides was performed with the 6-FAM, NED, PET, and VIC fluorophores so that only the forward primer was labeled with a fluorophore. The samples used in the genotyping were prepared in a reaction containing 8.85 mL formamide (Applied Biosystems), 0.15 mL standard marker (LIZ 500®, Applied Biosystems) and 1 mL PCR product diluted 20´. The amplified fragments were genotyped on a 3730 DNA Analyzer automated sequencer (Applied Biosystems). Product analyses were performed using the Peak Scanner TM v.1 program (Applied Biosystems).

Statistical Analysis
The PIC (Polymorphism Information Content) was calculated based on the number of alleles and frequency. The PIC values were determined by the expression: In the above expression, fi is the frequency of allele i in the population (Lynch and Walsh, 1998). The PIC provides an estimate of the discriminatory power of the locus, considering the number of alleles that are expressed and the relative frequencies of these alleles.
A binary data matrix 0 and 1 was generated from the coding of the presence (1) and absence (0) of polymorphic bands present in the isolates and then intrapopulational genetic diversity analyses calculated through the POPGENE (Yeh et al., 1997) were performed. Genetic diversity was in-4 Nome do autor  (1978) were performed using the ARLEQUIN 3.1 software (Excoffier et al., 2005).

Data analysis
The genetic structure of the sample was investigated using the Bayesian clustering algorithm implemented by Structure v.2.3.4 (Pritchard et al., 2000). The No Admixture model was used on the whole dataset with no previous population information and the "no-correlated allele frequencies between populations" option. Ten runs were applied with a burn-in of 200,000 interactions and a run length of 500,000 iterations performed for several clusters varying from K=1 to K=6. To determine the most probable number of clusters, the ad hoc statistic DK defined by Evanno et al. (2005) was used. The mean of the absolute values of L'(K) was divided by the standard deviation, where L (K) stands for the mean likelihood plotted over ten runs for each K. A hierarchical analysis of variance was carried out to test the significance of the differentiation among the populations and clusters as defined by Structure software.
Dendrogram trees were produced using genotyping data with 14 SSRs markers using the unweighted neighbor-joining method, as implemented in the DARwin software (version 6.0.9).

Results
From 384 clones that showed good sequencing quality for library 1 (American race), 175 microsatellites were from the American race and from these, 61 clones did not have microsatellites, representing an enrichment of 54.2%. From the total microsatellites found for the American race, dinucleotides were the most abundant (82 sequences), followed by trinucleotides (72 sequences); and among the motifs, the AG dinucleotides were in greater number, followed by the trinucleotide TCA. For the Brazilian race, the most abundant motifs were also the dinucleotides (52 sequences), followed by trinucleotides (48 sequences); and among the motifs, GAG and GAT trinucleotides were found in a greater number, followed by AC dinucleotides (Table 2, Figure S1).
The perfect microsatellites presented the highest number (97 sequences), followed by the compounds (63 sequences) and the imperfect ones (15 sequences) for the American race library. For the Brazilian race library, the perfect microsatellites also presented the highest number (70 sequences), followed by the compounds (35 sequences) and the imperfect ones (11 sequences).
A total of 40 SSRs were developed, and out of these, 15 SSRs were polymorphic ( Table 2). The number of alleles ranged from 2 to 10, with an average of four alleles per locus. The highest PIC (Polymorphic Information Content) value was 0.77 for Fop10-B01 (Table 2). The average PIC found was 0.38. In the genetic diversity analysis (Figure 1), the iso- lates were separated into two major clusters; each of these groups was divided into two smaller subgroups. Population analysis performed by Structure software also reinforced these results (Figure 2). The best results were obtained at K = 2 ( Figure S2) by Evanno et al. (2005). There was a great correspondence between the isolates grouped by clustering analysis and by population structure analysis except for Fop UFV 01, IAC Fop 05/13 and Fop UFV 02, which grouped externally in the dendrogram and fitted in the first group of Structure.
The percentage of polymorphic loci in isolates within each state ranged from 4.17 to 84.32%, showing low genetic variability of the isolates collected in the states of Goiás (15.62%), Pernambuco (13.54%), Minas Gerais (34%), Santa Catarina (14%) and Paraná (4.17%). The isolates derived from the state of São Paulo presented the highest genetic variability with 84% polymorphic loci and the isolates from the state of Paraná presented the lowest genetic variability with 4.17% polymorphic loci (Table S1).
The estimated total heterozygosity (H T ) was intermediate (Table 3). The mean value obtained from F ST (0.12) and G ST (0.33), indicated the genetic structure in the total sample ( Table 3). The distribution of genetic variability between and within Fop isolates was characterized by the genetic divergence of Nei (1978). The results of the AMOVA showed that 12% of the genetic variability was among the six States and 87% was within each state (Table 4).
Blasting of the expected sequence amplified by the SSRs against Fusarium oxysporum f. sp. Lycopersici genome identified sequences highly similar in chromosomes 1, 2, 4, 5, 7 and 8 (Table 5). Accurate and rapid identification of Fop is a needed for appropriate disease management. DNA-based techniques have increasingly become the tool of choice for understanding the genetic diversity and phylogeny relationships of Fusarium spp. (Huang et al., 2013;Cruz et al., 2018;Petkar et al., 2019). Microsatellites have been used because of the high resolution they provide (Dutech et al., 2007;Cruz et al., 2018); however, obtaining an acceptable level of polymorphism is generally more difficult in fungi than in other organisms (Dutech et al., 2007). According to Gao et al. (2008), most SSRs found in fungi are dinucleotides and trinucleotides, located in non-coding regions of the genome. This is in accordance to what we have found in both Brazilian and American race libraries. The size of the fungi genome (~59.9 Mb for Fusarium oxysporum) is smaller than those of plants and the longer the genome sequence, the longer the repeated units (Hancock, 2002). Indeed, the relative abundance of SSRs in fungi is low compared with the human genome, and long SSRs in fungi are rare (Karaoglu et al., 2005) The genetic diversity among the isolates of Fop in this study was low which is in accordance to literature (Bogale et al., 2006). The Fusarium oxysporum species presents asexual reproduction and does not present variation due to meiotic recombination (Ming et al., 1966). Cruz et al. (2018) showed that there was no relationship between the location and the genetic groups found by SSRs clustering. In our study, both the dendrogram and the Bayesian analysis showed that differences between the isolates were not related to their geographical origin. In fact, the introduction of the pathogen into different areas occurs mainly by infected 6 Nome do autor    (Fonseca et al., 2002). Common bean seeds of the previous crop are used for planting in the next crop, usually without control of the phytosanitary quality. The exchange and trade of these seeds is a common practice among farmers. Therefore, the absence of strong genetic structure at different spatial scales might be likely related to the spread of the pathogen through human activities. In addition to the nonrandom distribution of SSR clusters in the genome (Vieira et al., 2016), Fop transfer of specific genomic regions (Ma et al., 2010) might contribute to this low isolation by location. In relation to the intermediate estimated total heterozygosity (H T , Table 3), the species seems to have a reasonable reserve of genetic variability and the AMOVA showed higher genetic variability within each state (Table 4). In fact, Cruz et al. (2018) used SSRs to characterize the genetic diversity of F. oxysporum collected from common bean in different Brazilian states and reported that it was not possible to group them by collected location, or by their pathogenicity.

Discussion
One of the problems in the characterization of isolates in the F. oxysporum species complex is the assumption that there is a link between pathogenicity and a specific host or a group of host species and sub-species taxa. In most cases this assumption is either incorrect or an oversimplification of the actual situation (Bogale, 2006). Non-pathogenic F. oxysporum isolates are genetically diverse and make up a significant component of the species complex (Bao et al., 2002). Moreover, many phylogenetic studies have shown that some pathogenic isolates are more closely related to non-pathogenic strains than to other pathogenic strains in the same formae speciales or races (Mohammadi et al., 2004;Pasquali et al., 2004).This suggests that pathogenicity may be governed by a few genes, and that most likely, pathogenic genotypes in F. oxysporum arise by transfer of "pathogenicity chromosomes" (Ma et al., 2010). Consequently, the use of pathogenicity as a sole characteristic for grouping F. oxysporum isolates is flawed (Skovgaard et al., 2002).
The average marker's polymorphic information content (PIC) was higher than the one reported by Cruz et al. (2018) and much alike to the one reported by Bogale et al. (2005), who found an allelic diversity for nine microsatellite loci of 0.003-0.895 and a total of 71 alleles among 64 F. oxysporum isolates.
According to the annotation performed, it was found homology to a nudix hydrolase (FOP28-B01, Table 5), which constitutes a large family of proteins that hydrolyze nucleoside diphosphate derivatives. Nudix effectors have been reported in plant pathogenic oomycetes, fungi, and bacteria suggesting that these effectors might be important virulence components in the "toolbox" of plant pathogens (Ge and Xia, 2008;Dong and Wang, 2016).
The Fop10-B01 sequence had high homology with Fusarium oxysporum f. sp. lycopersici 4287 26S proteasome regulatory subunit RPN-1 mRNA (99% identity, E-value 3e-164, Table 5). RPN1a is required for basal defense and R protein-mediated defense (Yao et al., 2012). Some subunits of the 26S proteasome were found to be involved in innate immunity in Arabidopsis and probably in Fusarium oxysporum too.
In Brazil, Fusarium oxysporum f. sp. phaseoli (Fop) occurs in almost all common bean-producing areas (Batista et al., 2016), significantly reducing the yield. The difficulty in obtaining resistant genotypes to Fop is due to the diversity of physiological races that the pathogen presents, making necessary the study of the genetic and physiological variability of the pathogen. Several control methods are available, but none are efficient. There is no completely resistant common bean cultivar available on the market and no dominant resistance gene has been properly identified yet. The development of tools allowing early detection of the pathogen and effective disease control relies on the knowledge of the pathogen diversity. This study is expected to serve as an im-Fop microsatellite diversity 7

Supplementary Material
The following online material is available for this article: Figure S1 -Electropherogram with the peak of the allele of the microsatellite. Figure S2 -Number ideal cluster's according to the methodology of EVANO et al. (2005). Table S1 -Mean, standard deviation of the genetic parameters obtained for the 14 microsatellites.

Associate Editor: Célia Maria de Almeida Soares
License information: This is an open-access article distributed under the terms of the Creative Commons Attribution License (type CC-BY), which permits unrestricted use, distribution and reproduction in any medium, provided the original article is properly cited.