To develop EST-SSRs for yellow perch, European perch EST sequences were obtained from GenBank dbEST (http://www.ncbi.nlm.nih.gov/dbEST/index.html). All data were scanned using the software SSR Hunter version 1.3 (http://www.biosoft.net/dna/SSRHunter.htm) using search parameters set to more than seven repetitions for di-nucleotide repeats, five for tri-, four for tetra-, and three for penta-and hexanucleotide repeats.

Microsatellite-enriched libraries were conducted using the method described by Li et al. [3]. Briefly, the genomic DNA isolated from fin tissue was digested with a restriction enzyme Sau3A at 37°C for 3 hours. The fragments with the size range of 0.5 – 2 kb were recovered from an agarose gel. A synthesized adaptor SAUL (A: 5′-GCGGTACCCGGGAAGCTTGG-3′ and B: 5′-GATCCCAAGCTTCCCGGGTACCGC-3′) was ligated to the fragments using T4 DNA ligase. Microsatellite-containing fragments were selectively coupled to biotinylated repeat motifs [(CA)_n, (GT)_n, (AAC)₁₁, (GAAT)₁₀, (ACAT)₁₁, (AAAG)₁₁, (GTA)₁₅, and (AAT)₁₅], captured, and washed. Fragments containing microsatellites were ligated to a TOPO vector (Invitrogen) and transformed into competent Escherichia coli cells. Positive clones were selected for PCR amplification using M13 universal primers and the PCR products were sequenced.

To exclude duplicates, all sequences, including genomic and EST sequences, were subjected to BioEdit Sequence Alignment Editor Software for grouping clusters using multiple sequence alignment. Microsatellites with the same flanking regions were considered as the same loci. The independent sequences were submitted to the DNA Data Bank of Japan (DDBJ) for homology searches using BLASTN (http://blast.ddbj.nig.ac.jp/top-e.html) against the vertebrate DNA databases to exclude loci previously reported. Sequences with the longest perfect repeats and flanking regions were selected for PCR primer design (Primer Premier version 5.0 software; http://www.PremierBiosoft.com/faq.html). One primer of each primer pair was modified at the 5′-end with an M13 universal tail (5′-CAGTCGGGCGTCATCA-3′) as described by Boutin et al. [9].

A total of 30 adult yellow perch were collected live from a wild population in Lake Wallenpaupack in Pennsylvania, U.S. Individual fin-clips were stored immediately into 95% ethanol. For each specimen, DNA was extracted from 50 mg of tissue according to the methods described by Waters et al. [10]. Amplification of microsatellite loci was performed with three primers, the tailed primer, the nontailed primer, and the M13 universal 5′-labelled (FAM, TET, or NED) primer that contained the same sequence as the M13 universal tail. The PCR reaction mix contained approximately 50 ng of genomic DNA, 3 μL of JumpStart RedMix (Sigma), 1.5 pmol of both nontailed and labelled primers, 0.1 pmol of the tailed primer, and 100 μM of spermidine in a total volume of 6 μL. The PCR conditions were programmed as one cycle of denaturation at 95°C for 3 min, followed by 35 cycles of 30s at 95°C, 30s at locus-specific annealing temperature (Table 1), and 45s at 72°C, ending with a final step at 72°C for 5 min. Amplification products were separated using an ABI 3130 Prism DNA genetic analyzer and the genotyping results were analyzed using Genemap® 4.0 software.

For a certain locus, the allele size range (S) was directly obtained from the Genemap® 4.0 software. The number of alleles (A) and their frequency (F), the observed heterozygosity (H_o) and the expected heterozygosity (H_e) were calculated using the computer program POPGENE 32. The Markov chain method [11] was used to estimate the probability of significant deviation from Hardy–Weinberg equilibrium (HWE) and pairwise tests for linkage disequilibrium (LD) were performed using the program GENEPOP online version (http://genepop.curtin.edu.au/) using the default parameters. Significance criteria were adjusted for the number of simultaneous tests using Bonferroni correction [12].

To determine the potential for cross utility, amplification of the identified markers was assessed in one related species, the walleye (Sander vitreus). The same PCR conditions and genotyping methods were used as described above except that annealing temperature was re-optimized at each locus.

A total of 16 sequences derived from the microsatellite-enriched libraries were selected for primer design. The optimization results showed that eight primer pairs could successfully amplify target fragments of the expected sizes. All eight loci exhibited polymorphism in the individuals tested. The numbers of alleles varied from 3 – 14 with an average of 8.5 alleles per locus. The observed and expected heterozygosities ranged from 0.07 to 0.81 and from 0.20 to 0.95, respectively (Table 1). None of the loci showed significant linkage disequilibrium. After sequential Bonferroni correction for multiple tests, five loci were found to depart significantly from Hardy–Weinberg equilibrium (HWE). To exclude the impact of short allele dominance (large allele dropout), data were subject to analysis with Micro-Checker [13]. No evidence for large allele drop-out was found for any of the loci. Further tests indicated that heterozygote deficiency at these loci was responsible for the departure (Table 1). Another possible explanation for the departure from HWE is the dramatic contemporary decline in spawning populations, and consequent non-random mating and genetic bottlenecks [14, 15]. A final possibility is subpopulation structure which cannot be ruled out without further analysis.

In the process of EST database mining, a total of 2,226 EST sequences were deposited in GenBank. The mining results showed that 110 (4.93%) sequences contained microsatellites that conformed to our mining criteria (Table 2). As found in other species, di-nucleotide repeats were the most abundant, accounting for 73.64% of all repeats located. This ratio is much higher than has been reported for some other aquatic species such as shrimp, bivalves [5, 6, 16], and other freshwater fish [7]. Surprisingly, the most abundant di-nucleotide repeat type was AG/CT, which is not consistent with reported findings for other fish such as common carp Cyprinus carpio [7], pufferfish Fugu rubripes [17] and catfish Ictalurus punctatus [18] where, in general, the AC/GT repeat type is the most abundant di-nucleotide microsatellite. Similar to our findings, whole genome scanning has indicated that AG/CT is the most abundant type in some aquatic animals such as scallop [19, 20]. Biased sampling, due primarily to the small number of EST sequences examined in this study, may explain the current findings. To further confirm or refine the observation for yellow perch, more sequences or whole genome scanning are needed.

Twenty-three EST-derived sequences were chosen for PCR primer design. Among them, 13 primer pairs (56.5%) amplified products of the expected size. The presence of long introns between primers in genomic DNA, primer sequences spanning across introns and/or mutations, and indels (insertions or deletions) in the primer annealing sites between the two perch species may explain the non-amplification [5]. However, the success ratio we observed for EST-derived microsatellites is slightly higher than those in other studies using the same strategy, such as development of Japanese sea urchin (Strongylocentrotus intermedius) using the EST sequences of a related species of purple sea urchin (S. purpuratus) [8].

Although we selected relatively long microsatellite regions, the polymorphism assessment results revealed low levels of genetic diversity at these loci. Two or three alleles were detected at most loci and only one locus displayed 5 alleles in the individuals tested (Table 1). Similarly low genetic diversity was also observed in terms of heterozygosity (Table 1). In previous studies where levels of polymorphism have been compared between Type I and Type II microsatellite DNA markers in the same species, the level of polymorphism of Type I markers has usually been observed to be slightly lower [20] but not nearly so dramatic as the differences we observed in polymorphism between EST-SSRs and genomic-SSRs (Table 1). Evolutionary conservation and lower mutation rates within gene-coding sequences is a possible explanation for our observation but does not account for the prior published results. The practical implications are as described by Eujayl et al. [21], that a suite of more mutationally-stable EST-SSRs could complement highly variable genomic-SSRs to reconstruct past evolutionary events and to identify regions of genomes that are identical by descent. This would be of particular utility in genomic, gene mapping, and QTL studies across species within Perca.

Of eight genomic-SSRs and 12 EST-SSRs, three (37.5%) and eight (66.7%) loci were successfully cross-amplified in the walleye (Table 1), respectively. The cross utility results confirmed that Type I microsatellite markers have higher success ratio than that of Type II microsatellites in the cross-species amplifications among closely-related species. Although the walleye belongs to a different genus, the high cross-amplification ratio was also observed at both genomic-SSR and EST-SSR loci.