Genetic diversity of Brazilian triticales evaluated with genomic wheat microsatellites

The objective of this work was to determine the genetic variability available for triticale (X Triticosecale Wittmack) crop improvement in Brazil. Forty-two wheat genomic microsatellites were used to estimate the molecular diversity of 54 genotypes, which constitute the base of one of the major triticale breeding programs in the country. Average heterozygosity was 0.06 and average and effective number of alleles per locus were 2.13 and 1.61, respectively, with average allelic frequency of 0.34. The set of genomic wheat microsatellites used clustered the genotypes into seven groups, even when the germplasm was originated primarily from only two triticale breeding programs, a fact reflected on the average polymorphic information content value estimated for the germplasm (0.36). The 71.42% transferability achieved for the tested microsatellites indicates the possibility of exploiting these transferable markers in further triticale genetic and breeding studies, even those mapped on the D genome of wheat, when analyzing hexaploid triticales.


Introduction
Triticale (X Triticosecale Wittmack) is a synthetic self-pollinated crop derived from wheat (Triticum sp., AABB or AABBDD) and rye (Secale cereale L., RR), crossed to bring together in a single species the technological quality and yield potential of wheat with the rye stress resistance and rusticity.Octoploid triticales, comprising 56 chromosomes, are derived by crossing hexaploid wheats (T.aestivum L., AABBDD) and rye, while hexaploid triticales, with 42 chromosomes, are mostly complete cariotypes, partially or totally deprived from the D genome of wheat, resulting from the cross between Triticum durum L. (AABB), for instance, and rye (Ammar et al., 2004;Oetler, 2005).
The combination of wheat and rye genomes allows triticale to show characteristics of good adaptability to poor or harsh environments, such as acid or waterlogged soils, metal toxicity, salinity, high elevation and adverse climatic conditions, besides greater tolerance (more than wheat) to common wheat diseases (Horlein & Valentine, 1995).X Triticosecale constitutes also a valuable genetic resource for transferring genes of interest from rye into wheat, particularly those related to biotic and abiotic stresses (Vaillancourt et al., 2007).
Molecular markers such as microsatellites or SSRs (Simple Sequence Repeats) constitute an important tool for studies on genetic diversity, population structure, genetic mapping and crop breeding due to their abundance, codominance, level of polymorphism, reliability and easiness to assay (Röder et al., 1995(Röder et al., , 1998)).Additionally, these markers are not influenced by the environment or by genotype x environment interactions, contrary to what is verified for morphological and phenological characteristics.Biochemical and molecular markers most commonly used include the polymorphism of storage proteins, alozymes and DNA markers such as SSRs, RAPD (Random Amplified Polymorphic DNA), AFLP (Amplified Fragment Length Polymorphism), and others.
Microsatellites of wheat or rye have been efficiently employed in analysis of the genome of triticale (Kuleung et al., 2004(Kuleung et al., , 2006;;Tams et al., 2004) and the estimates revealed on the genetic diversity of triticales from distinct continents indicate that the genetic base of this cereal must be extended (Darvey, 1986, Nascimento Junior et al., 2004).Analyzing the genome of five triticales from Russia, Mexico and the United States, using 176 microsatellites from wheat (148) and rye (28), Kuleung et al. (2004) identified polymorphism in 31% of the used markers.Tams et al. (2004) determined polymorphism information content (PIC) of 0.54, indicative of moderate variability, in the evaluation of 128 triticale cultivars from five continents using three to five microsatellites per chromosome.The same value was also found by Kuleung et al. (2006), when 80 hexaploid triticales were analysed, employing 43 wheat and 14 rye microsatellites.
Until this moment, studies to estimate the genetic diversity of Brazilian triticale germplasm have not been carried out, neither on genotypes already recommended for cultivation nor on those still in the process of development and evaluation.
The objective of this work was to estimate the molecular diversity of triticale genotypes that constitute the base of one of the main triticale breeding programs in Brazil using wheat genomic microsatellites.Genomic DNA was isolated from seeds (Rogers & Bendich, 1988) or seedlings (Kleinhofs et al., 1993) according to the CTAB (cetyltrimethylammonium bromide) method.A total of 42 genomic wheat microsatellites markers (Röder et al., 1998), one for each wheat chromosome arm, was used.This same set of markers (Table 2) was used by Stachel et al. (2000) to determine the genetic differentiation caused by selection for adaptation in wheat.Polymerase chain reaction (PCR), adjusted for 25 µL final volume, consisted of 1x enzyme buffer, 2 mM MgCl 2 , 0.2 mM each dNTP, 5 µM to 10 µM primers, 1 unit Taq polymerase and 75 ng of genomic DNA.The PCR amplification was performed in a MJ Research PTC-100 (Programmable Thermal Controller, MJ Research INC), as follows: initial denaturation at 94°C for 3 min, followed by 35 to 45 cycles of 1 min at 94°C,  1 min at 45-60°C, 1 min at 72°C and final extension for 3 to 10 min at 72°C.The annealing temperature for each primer (Table 2) was determined according to Röder et al. (1998) andStachel et al. (2000).Amplified DNA fragments were fractionated on 3% agarose gel stained in ethidium bromide and photographed.

Fifty
DNA segments at polymorphic loci in all 54 genotypes were scored as presence (1) or absence (0) of an allele and the results analyzed with NTSYSpc, version 2.02 (Rohlf, 1989).The scores for each allele were used to construct a binary matrix, which was then transformed to genetic similarity matrix using Jaccard similarity coefficient.The genetic similarity matrix of all genotypes was analyzed using unweighted pair group method with arithmetic average (UPGMA) algorithm and the results were used to construct a dendrogram.Null alleles were omitted in calculation, and were treated as missing data.
The average gene diversity expected for each considered marker locus in this group of genotypes was estimated according to Anderson (1993): where p is the frequency of the allele j for marker i.
Additionally, the genetic diversity was described based on the number of variants determined for the germplasm, considering: 1) the proportion of polymorphic loci (P) given by P = n pj /n total , in which n pj is the number of polymorphic loci and n total the total number of loci; 2) average expected heterozigosity (H e ), i.e., the probability of any pair of alleles in a single locus, randomly chosen in the population, be distinct from each other, given by H e = S j L H j /L, where H j is the heterozigosity per locus and L the total number of loci.Considering a locus j with i alleles, the genetic diversity in this locus is (1) Wheat chromosome: 1 to 7; genome: A, B or D; and chromosome arm: long (L) or short (S). (2)These primers did not amplify triticale DNA.
given by H j = 1 -Sp i 2 , also called PIC; 3) average observed heterozigosity (Ho), considering the proportion of heterozygote loci in all analyzed individuals; 4) the abundance of allelic variants (A); 5) the effective number of alleles (Ae) given by A e = 1/(1 -H j ) = 1/Sp i 2 , where p i is the frequency of the i-th allele in a marker locus; and 6) the average number of alleles per locus, i.e., the sum of all observed alleles in all used markers, divided by the total number of markers, a piece of information complementary to the polymorphism information: , where, K is the number of loci and n i is the number of alleles determined locus.

Results and Discussion
Out of the 42 wheat genomic microsatellites tested in the parental genotypes from the Embrapa's triticale breeding program, 30 (71.42%) amplified in the genome of triticale, suggesting good transferability of these wheat markers to triticale (Table 2).Among these, 21 (70%) were polymorphic, i.e., the frequency of the most common allele was equal or superior to 0.95.About 18% of A -B wheat genome-specific microsatellite primers did not amplify in triticale (Table 2), and some of the obtained fragments may not contain SSR sequences, as suggested by Leonova et al. (2005).Kuleung et al. ( 2004) found 57% and 39% transferability for 148 wheat microsatellites and 28 rye microsatellites, respectively, revealing low marker transferability.
The rate of transferred markers from wheat D genome was 50%, surpassing the expected values, since the analyzed triticales are hexaploids, lacking, in general, this genome.Similarly, Tams et al. (2004) and Leonova et al. (2005) obtained amplification of triticale DNA fragments using D genome-specific microsatellite primers.
PCR amplification with microsatellites mapped in all wheat D genome chromosomes, but 1D and 2D, may be associated with the presence of wheat-rye (Lukaszewski & Gustafson, 1983) or wheat-wheat translocation (Hohmann et al., 1999;Leonova et al. 2005).However, studies have shown that alterations on the expression of certain genes in triticale compared to the parental wheat lines would result from the effect of rye chromosomes due to the change of triticale genomic composition, rather than being associated with the presence of wheat-rye translocation (Leonova et al., 2005).Both possibilities must be investigated in this group of triticales to detect the presence of translocations or recombination, employing molecular cytogenetic tools such as in situ hybridization and C-banding (Zhang et al., 2007b).
Besides, the presence of storage proteins subunits encoded by genes on the D genome must be determined in these lines, to confirm the presence of the D genome on the genomic constitution of the germplasm.Most triticale cultivars are derived from crossing triticale parents or from crossing wheat with triticale (Kuleung et al., 2004).Under the circumstances, the introgression of genes of interest encoded on the chromosome 1D can be widely exploited in breeding programs in order to improve important traits of hexaploid triticales (AABBRR), such as end-use quality.
Most genomic SSRs have neither a gene function nor a close linkage to coding regions and limited transferability among related species.An alternative for the development of triticale specific markers is the analysis and characterization of microsatellites in expressed sequence tags (EST) collections.Today, more than 1,050,000 ESTs for common wheat and 9,200 for rye are available in the public domain (National Center for Biotechonology Information, 2007) and are used to develop molecular markers such as EST-derived microsatellites.The use of this marker system would allow to detect variation in the expressed portion of the genome, which may increase the efficiency of markerassisted selection in crop breeding (Gupta et al., 2003;Zhang et al., 2005Zhang et al., , 2007a;;Tang et al., 2006).
Considering the set of wheat microsatellites that amplified PCR products in the analyzed triticale genotypes, the proportion of polymorphic loci (P) revealed was 0.70 and the average PIC value estimated was 0.36, similar to values determined for inbred rye (Bolibok et al., 2005) and also wheat (Zhang et al., 2006(Zhang et al., , 2007a)).The expected genetic diversity for this set of markers ranged between 0.24 for WMS408 and 0.88 for WMS389 (Table 3).However, it must be taken into account that these accessions come from only two breeding sources, Embrapa Trigo (43%) in Brazil and Cimmyt (57%) in Mexico (Table 1).The PIC values (0.54) revealed for triticale by Kuleung et al. (2006) and Tams et al. (2004) were obtained for 80 and 128 genotypes, respectively, representing a broad spectrum of historic and modern triticale germplasm from 17 countries, five continents and at least 27 breeding programs.At the same time, the PIC of 0.71 determined by Prasad et al. (2000) for wheat was calculated for 55 accessions from 29 countries, the PIC of 0.68 determined by Hai et al. (2007) for wheat included accessions from four distinct continents and the PIC of 0.51 registered by Landjeva et al. (2006) represented the wheat germplasm created in Bulgary over a period of almost 80 years.
In this work, 64 alleles were amplified with the 30 selected SSR markers, with the average of 2.13 alleles per locus and the allelic variants abundance per locus (A) ranging between one to four (Table 3).The effective of alleles (Ae) for the polymorphic markers ranged between 1.14, for WMS389, and 4.08, for WMS408, both mapped on the wheat B genome, and with average value of 1.61.The number of alleles expected for each of the markers from this set of microsatellites, under the circumstances, should not be lower than 1.61 when applied to a different group of progenitors if it is to be maintained, at least, the same level of diversity verified in this work.Otherwise, the strategy of choice of progenitors should be revised.
Among the 54 accessions analyzed, 104 heterozygote loci were detected (Table 4).Three accessions presented only homozygote loci for all used SSRs and five heterozygote loci were observed in one accession, resulting in an average observed heterozygosity (Ho) of 0.06, as expected for a group of genotypes formed exclusively by fixed lines.
The 54 accessions evaluated were divided into seven main groups (Figure 1), using UPGMA dendrogram based on Jaccard's coefficient of similarity (average similarity = 0.56).Since most of the analyzed germplasm is derived from Mexican triticale, a high similarity among them was expected.The PIC (0.36) and the average number of alleles (2.13) indicate the expected low polymorphism in the 54 genotypes and the obtained average similarity value (0.56), indicative of moderate variability, may change if a larger number of markers is used to screen these genotypes, revealing a more actual value.Tams et al. (2004Tams et al. ( , 2006) ) used wheat and rye microsatellites to study genetic diversity of European winter triticale and their results showed no distinct clusters of lines from the same breeding source.On the other hand, Kuleung et al. (2006) divided 80 hexaploid triticale accessions available in the world collection into five clusters (average similarity = 0.45).
Clusters obtained could not be clearly characterized and besides the pedigree and the breeding source, other data on the Embrapa's germplasm were used to discuss aspects of the resulting dendrogram.Group 1 was subdivided into four subgroups, 1-a, 1-b, 1-c and 1-d (Figure 1).Subgroup 1-a comprised four accessions created at Cimmyt, all short in height, early in maturity and susceptible to Fusarium graminearum, causal agent of scab.About 50% of the lines included in subgroup 1-b share the germplasm of Tatu, a cultivar created at Cimmyt and released for commercial use in Brazil.Within subgroup 1-c, around 27% of the genotypes show shorter height and earlier maturity and the same proportion of accessions did not present high susceptibility to scab.Subgroup 1-d is formed by a single cultivar, CEP 28 -Guará, created at Cimmyt and released for commercial use in Brazil.It is characterized by its intermediate cycle and height as well as good behavior against Drechslera tritici-repentis.The male parent of CEP 28 -Guará integrates subgroup 1-b and is present in the pedigree of 50% of the genotypes forming this subgroup.In turn, CEP 28 -Guará's female parent is present in the pedigree of only two other accessions among all 54 studied ones, both included in group 5, PFT 0416 and PFT 0417, created at Cimmyt, showing early maturity and short height.
Group 2 is represented by the only two accessions comprising late cycle in this set of triticales, both showing very good agronomic traits.All accessions included in groups 3 to 4 are very susceptible to F. graminearum.Group 3 was formed by a single genotype, PFT 0517, a breeding line created and introduced from Cimmyt, Mexico.PFT 0517 is the only genotype among all the analyzed accessions that has the parental GAUR_3 and BANT-1, and this might be the explanation for this separated cluster.Group 5 is formed by the Brazilian line PFT 0510 and two other sib lines, PFT 0416 and PFT 0417, mentioned before.
Most accessions adapted to warm environments are found in cluster 6, grouping accessions with good agronomic type (80%), except for the Mexican lines PFT 205 and PFT 0413 (intermediate).Finally, in group 7, characterized by clustering genotypes of intermediate maturity and height, two accessions can be found, among all the analyzed triticales, that share FD-693 in their pedigree, as well as IAC 2, the genotype with the best performance against Magnaporthe grisea.
The use of rye microsatellites with wheat microsatellites (genomic or derived from expressed Table 4. Number of heterozygote loci (h) observed in triticale (X Triticosecale Wittmack) parental genotypes from Embrapa's improvement program analyzed with 30 wheat (T.aestivum L.) genomic microsatellites (1) .
(1) Total heterozygote loci: 104 sequences) must be considered in further studies with this crop, distributing at least one pair of primers per chromosome arm of all genomes.Regardless the diversity revealed by rye SSR in triticale being lower in comparison to wheat microsatellites (Kuleung et al., 2004;Tams et al., 2004), this strategy allows the balanced inclusion of all species genomes in triticale analysis.The utilization of pooled data from both species, besides data from more microsatellite markers or other marker class, including expressed regions of the genome, would provide a broader coverage of the triticale genome, leading to more complete data about the species and its diversity patterns.

Conclusions
1.The 71.42% transferability of the chosen set of wheat genomic microsatellites to triticale raises the potential of exploiting these markers in Brazilian triticale genetic and breeding studies.
2. The average number of alleles (2.13) and PIC (0.36) indicate low variability in the set of genotypes used in Embrapa's triticale breeding program.

Table 3 .
Number and frequency of alleles, polymorphism information content (PIC) values and effective number of alleles (Ae) estimated for parental genotypes from Embrapa's triticale (X Triticosecale Wittmack) breeding program using 30 wheat (T.aestivum L.) genomic microsatellites (SSR).