INTRODUCTION

The major histocompatibility complex (MHC) is a multigene family encoding cell surface glycoproteins, which are responsible for recognition of antigenic proteins and intimately involved in the orchestration of an immune response (reviewed by J. Klein, 1986). MHC loci include some of the most genetically diverse regions in the vertebrate genome, and several aspects of their polymorphism are unique (reviewed by Nei and Hughes, 1991). Alleles at individual MHC loci differ greatly in nucleotide sequence, exhibiting levels of variation around the order of 0.11 in mammals (Nei and Hughes, 1991). At the nucleotide level the higher polymorphism is within the sequence encoding the peptide-binding region of MHC proteins. This suggests that natural selection is maintaining a high level of diversity at this locus, allowing populations to maximize detection and defense against an array of pathogens (reviewed by D. Klein et al., 1993).

Several methods are available for genetic analysis of the MHC II locus in salmonids, including direct sequencing, restriction fragment length polymorphism (RFLP) analysis, denaturing gradient gel electrophoresis (DGGE), reference strand–mediated conformational analysis (RSCA), and single-stranded conformation polymorphism (SSCP) analysis. Although DDGE, RSCA, and SSCP analysis are all based on polymerase chain reaction (PCR), DDGE, and RSCA require specialized equipment and more extensive development than SSCP analysis. DDGE involves the electrophoresis of PCR products through denaturing gradient gel and requires specialized gel preparation. RSCA requires an automated sequencer. It utilizes fluorescently labeled DNA of known sequence (called the reference strand) that is annealed to unlabeled PCR products prior to electrophoresis. In contrast, SSCP analysis directly employs denatured PCR products, synthesized by incorporating 32P, for electrophoresis through a nondenaturing gel.

MHC class II loci have been identified in several teleost species including carp (Hashimoto et al., 1990), zebrafish (Ono et al., 1992), cichlids (D. Klein et al., 1993), and several salmonids (Juul-Masden et al., 1992; Hordvik et al., 1993; Grimholt et al., 1994; Miller and Withler, 1996; Dorschner et al., 2000). Teleost nucleotide sequences are useful for investigation of the evolution of the MHC loci and have also been used to compare the structure of fish MHC genes with those in higher vertebrates (Okamura et al., 1993; Hordvik et al., 1993; Grimholt et al., 1994). Salmonids are one of the oldest families of teleost fishes and have been extensively studied because many have high commercial value.

Genotypic analyses of the MHC class II locus in salmonids have previously been performed by nucleotide sequencing (Miller and Withler, 1996; Miller et al., 1997; Dorschner et al., 2000; Langefors et al., 2000), RFLP analysis (Miller and Withler, 1996; Langefors et al., 1998, 2000), DDGE (Miller and Withler, 1996; Langefors et al., 2000), and SSCP analysis (Garrigan and Hedrick, 2001). Miller and Withler (1996) identified 29 MHC class II alleles by sequencing the B1 and B2 chains from 35 individuals representing 7 Oncorhynchus species. RFLP analysis was subsequently used to determine whether the amplification of multiple loci was common among species. RFLP analysis has also been used to characterize MHC class II B haplotypes in Atlantic salmon (Langefors et al., 1998, 2000). Sequencing of MHC class II B1 and B2 chains from 35 individuals representing 10 species of Pacific salmon identified 29 alleles (Miller and Withler, 1996), In chinook salmon, 21 MHC class I A1 alleles were sequenced from 36 individuals, while only 3 class II B1 alleles were identified in 47 sequences (Miller et al., 1997). DGGE identified an additional 10 chinook salmon alleles. However, of these 32 alleles only 14 were consistently distinguished by DGGE (Miller et al., 1997). In comparison, Dorschner et al., (2001) identified 43 class II B1 alleles from the sequences of between 2 and 6 cloned PCR products from 74 individual lake trout.

RSCA has been described for obtaining genotypic information on human MHC genes (Arguello and Madrigal, 1999). RSCA utilizes fluorescent primer technology and has been adapted for use with automated sequencers. A locus-specific fluorescently labeled reference DNA fragment is generated by PCR and then hybridized with the PCR products of the samples to be tested. A duplex is formed containing loops and bulges corresponding to the number and position of base pair mismatches between the reference strand and each test strand. When electrophoresed through a nondenaturing polyacrylamide gel, each different duplex formed will migrate slightly differently from the others. An automated sequencer shows the relative positions of the fluorescent duplexes as different bands, 2 bands per heterozygous individual and a single band for each homozygote (Arguello and Madrigal, 1999).

Similarly, SSCP requires electrophoresis through a nondenaturing gel; however, this technique does not require the design of a reference strand, and no duplexes are formed. Each sample to be tested is simply amplified by PCR, then the products are denatured and electrophoresed. The single-stranded DNA migrates through the gel as determined by its 3-dimensional shape. The tertiary structure of the single-stranded DNA is determined by nucleotide sequence, and a single base change is sufficient to cause changes in the tertiary structure. Hence SSCP analysis is capable of detecting changes as slight as a single nucleotide polymorphism. Each heterozygote is characterized by 4 bands representing the unique, complementary sequences from the single strands, each of 2 different alleles, and homozygotes are identified by 2 bands, duplicates of 2 single strands from a single allele.

This article describes our application of RSCA and SSCP analysis for the genotypic characterization of an MHC class II locus in lake trout (Salvelinus namaycush). We compared these methods with those previously applied to the analysis of this locus in other salmonids.

MATERIALS AND METHODS

Materials

Genomic DNA was extracted from the fin clips of 80 lake trout from the Apostle Islands in Lake Superior. Of these, 39 were of the siscowet morphotype (samples 1 through 47) and 41 were of the lean morphotype (samples 50 through 99). The siscowet morphotype is found in deep water and has a much deeper belly than the lean morphotype, which has a more pronounced torpedo body shape and inhabits shallow waters at the edge of the lake. Previous work had indicated that Lake Superior lean and siscowet lake trout populations are reproductively isolated (R.B. Phillips, unpublished data).

Methods

RSCA

The RSCA procedure described by Arguello and Madrigal (1999) was adapted for the genotypic analysis of MHC class II in lake trout. Two primers (Miller et al., 1997),

MHC6 (5′-ccgatactcctcaaaggacctgca-3′) and

MHC9 (5′-ggtcttgacttgatcagtca-3′),

were used to amplify exon 2 of the MHC class II β1 Sana gene in lake trout, an area that represents the peptide-binding region.

Three reference strands were amplified from clones representing different alleles (GenBank accession numbers AF455255, AF130030, and AF129979) using primers labeled with fluorescent FAM, TET, and HEX nucleotides (PerkinElmer), respectively, in separate reactions. The following PCR regimen was employed for amplification of labeled reference strands and unlabeled samples: template DNA (50 ng), PCR reaction buffer (Promega; 50 mM KCl, 1% [wt/vol] Triton X-100, 10, mM Tris-HCl, pH 9.0), MHC6 (300 nM), MHC9 (300 nM), MgCl2 (1.5 mM), dNTPs (200 nM each), Taq DNA polymerase (Promega, 1.25 U). Amplification was carried out in a PTC-200 thermocycler (MJ Research) after denaturation at 94°C for 3 minutes and followed by 35 cycles of 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds. A final extension was performed at 72°C for 5 minutes. For labeled reference strands, multiple reactions were performed for each label. The PCR products were pooled respectively then purified using QIAquick PCR Purification columns (QIAGEN).

Genomic DNA was used as the template for genotypic analyses. Each sample was amplified in triplicate (one for each labeled reference strand) using the above PCR regimen but with unlabeled primers. To the replicate PCR products for each sample was added 8 ng of FAM-labeled reference strand, 10 ng of TET-labeled reference strand, or 18 ng of HEX-labeled reference strand. The reference strands were hybridized to the unlabeled PCR products in the thermocycler with denaturation at 94°C for 5 minutes and annealing at 55°C for 5 minutes, followed by 15°C for 4 minutes. The hybridized samples were electrophoresed along with Applied Biosystems (ABI) PRISM GeneScan Size Standards through a 6% nondenaturing polyacrylamide sequencing sized gel (6% [vol/vol] Long Ranger gel solution [PerkinElmer], 1× TBE [Promega], 0.05% [wt/vol] APS, 0.06% [vol/vol] TEMED) using an ABI 373 automated sequencer for 3.5 hours.

The gels were analyzed using GeneScan software (ABI), and the data were imported into Genotyper 2.0 (ABI) for genotype scoring. Individuals were scored as homozygotes or heterozygotes, and the size of each allele was recorded relative to the size standards.

ANALYSIS

SSCP

This PCR-based technique allows amplification and separation of single-strand DNA fragments, representing both alleles at a single locus in an individual. The unlabeled forward and reverse primers (MHC6 and MHC9) were used to amplify the MHC II region described above in the same 80 lake trout from the Apostle Islands. For SSCP analysis, PCR was performed in the presence of 1.7 µCi α-32P-dCTP per 10-µl PCR reaction also containing the following: template DNA (100 ng), primers (400 nM each), dNTPs (250 µM each of dATP, and dTTP, dCTP, and dGTP), PCR reaction buffer (Promega; 50 mM KCl, 1% [wt/vol] Triton X-100, 10 mM Tris-HCl, pH 9.0), MgCl2 (1.2 mM), and Taq DNA polymerase (Promega, 1.25 U).

Amplification was carried out in a PTC-200 thermocycler (MJ Research) using the parameters given for RSCA amplification. The PCR products were denatured in 1.5 vol of formamide loading dye at 95°C for 5 minutes and snap chilled on ice before being loaded onto a nondenaturing polyacrylamide gel. The volume of gel mixture (8% acrylamide, 0.13% bisacrylamide, 5% [vol/vol] glycerol, 1× TBE, 0.05% [wt/vol] APS) was adjusted to 100 ml with Milli-Q water, and polymerization was initiated with 0.06% (vol/vol) TEMED. Sequencing size gels were cast, and 6 µl of each denatured PCR product was loaded and electrophoresed at 1500 V, constant voltage, for 35 minutes to allow the DNA to migrate into the gel, then at 6 mW, constant power, for 20 hours. The gel was vacuum dried onto blotting paper, and x-ray film was exposed overnight and developed. Animals were assessed for the presence of 2 bands, indicating a homozygote, or 4 bands, indicating a heterozygote. Alleles were labeled and recorded according to their relative position on the gel.

Allele frequencies and heterozygosity were calculated for both RSCA and SSCP analysis for comparison of the techniques.

Sequencing

Samples that were identified as heterozygotes by RSCA but as homozygotes by SSCP analysis or vice versa were sequenced in both directions in duplicate to confirm their zygosity. Automated sequencing of PCR products (approx. 200 ng) was performed on an ABI STRETCH DNA sequencer.

RESULTS

RSCA

Homozygotes were identified by the presence of 2 bands, one representing the homoduplex formed by the reference strand, and the other representing a heteroduplex between the reference strand and a single allele (Figure 1, a). The detection of 3 bands indicated a heterozygous individual. One band representing the homoduplex and 2 bands representing heteroduplexes formed between the reference strand and each of the 2 different alleles (Figure 1, b). Each allele was assigned an arbitrary number according to its apparent size, relative to known molecular size standards, calculated using Genotyper. A total of 40 alleles were identified by RSCA, representing 64 genotypes. Table 1 presents the genotypic scoring for each individual. Alleles were arbitrarily numbered, and the allelic identity (homozygote or heterozygote) of each individual is indicated.

Table 1 Summary of Lake Trout Genotypes by RSCA and SSCP
Full size table
Figure 1
figure 1

Chromatogram showing the RSCA detection of a homozygous animal (a) and a heterozygous animal (b). A peak at 301.45 bp in both samplesrepresents the homoduplex formed by the reference strand. The upper trace contains 1 additional peak at 377.75 bp, representing the heteroduplex formed between the reference strand and the single allele of the homozygote. The heterozygote was identified by the detection of 2 additional bands, one at 370.53 bp and a second at 405.11 bp, representing different alleles.

Full size image

ANALYSIS

SSCP

For SSCP analysis alleles were scored according to the order in which they had migrated down the gel and represented alphabetically to avoid confusion with RSCA genotypes. Again the homozygotes and heterozygotes were identified and recorded. Since SSCP analysis yields 2 bands for homozygotes and 4 bands for heterozygotes, only the lower set of bands were used for scoring genotypes (Figure 2). The scoring results for SSCP analysis are presented in Table 1 alongside the information obtained from RSCA, for comparison of the techniques. SSCP analysis resolved 46 alleles representing 63 genotypes in the sample of 80 individuals.

Figure 2
figure 2

Autoradiogram showing the lower half of alleles detected by SSCP analysis. Homozygotes were detected by the presence of 2 bands and heterozygotes contained 4 bands. However, since only the lower half of the autoradiogram is shown here, homozygotes are represented by a single strong band and heterozygotes by 2 strong bands. The alleles were assigned an arbitrary alphabetical label according to their migration with respect to other samples (see Table 1).

Full size image

Heterozygosity and Allele Frequencies

Heterozygosity calculated from the RSCA data was 76%, and that calculated from the SSCP data was 70%. Table 1 shows that 84% of the individuals recorded as homozygotes were identified by both RSCA and SSCP. The allele frequencies (Table 2) calculated by RSCA indicated that the alleles 25, 2, and 22 (respectively) occurred most frequently, while SSCP analysis indicated that alleles H, A, and I (respectively) were the most common. RSCA indicated that 17.5% of alleles were shared between the siscowet and lean morphotypes, while SSCP analysis indicated that almost 20% of alleles detected were common to both morphotypes.

Table 2 Comparison of Allele Frequencies and Heterozygosity in Lake Trout
Full size table

Sequencing

Of the 12 samples in which a zygosity mismatch was observed (Table 1), sequencing confirmed that in 50% of cases the RSCA result was correct, and in the other 50% the SSCP result was correct. Thus (as indicated in Table 1 by boldface), sample numbers 10, 38, 50, 84, 93, and 95 were all heterozygotes correctly identified by RSCA. Sample numbers 37 and 43 were correctly identified as homozygotes by SSCP, and sample numbers 33, 70, 71, and 91 were heterozygotes, as initially determined by SSCP. Samples 7 and 15, representing 2 different homozygous genotypes, were sequenced as controls and aligned alongside a corresponding heterozygote (Figure 3).

Figure 3
figure 3

Chromatograms showing the alignment of a heterozygote for RSCA genotype 22/25 (b) and homozygotes of genotype 25/25 (a) and 22/22 (c). The homozygote samples were included as controls for zygosity identification by sequencing. Each letter “N” in the basecalling represents a SNP and identifies the positions of heterozygous alleles in the sequenced fragment.

Full size image

DISCUSSION

Of the several methods available for genotyping salmonids at the MHC class II locus, we have selected to use and compare RSCA and SSCP analysis. There are several reasons for choosing these methods in preference to sequencing methods such as RFLP and DGGE. Most studies to identify and characterize MHC alleles in salmonids have employed sequencing (Miller and Withler, 1996, Miller et al., 1997; Dorschner, 2000). However, although sequencing is the most accurate and sensitive method, sample preparation and analysis are time-consuming and expensive. Also, most salmonid species display low to moderate allelic diversity at MHC loci, ranging from only a single MHC class II allele in O. tshawytscha (10 clones), 2 alleles in O. kisutch (5 clones), and 3 alleles in O. nerka (11 clones) (Miller and Withler, 1996), to 12 class I alleles in O. tshawytscha (44 animals) (Garrigan and Hedtick, 2001), 22 class I alleles in O. tshawytscha (36 sequences) (Miller et al., 1997), and 25 class II alleles in Salmo salar (60 clones from 7 individuals plus 41 DGGE fragments) (Langefors et al., 1998, 2000). It is possible that these studies did not identify all alleles owing to the small sample sizes involved. Recent investigations using larger sample sizes suggested that MHC class II diversity is much higher in lake trout. Fifty-two MHC class II B1 alleles were sequenced from 74 individual lake trout (approx. 300 clones) (Dorschner et al., 2000). RSCA and SSCP analysis of 80 lake trout identified several more class II B1 alleles that were sequenced and added to the GenBank database (AF455255–AF455266).

RFLP analysis may be more cost-effective, but this method is also time-consuming, labor-intensive—requiring high-grade DNA for restriction digestion—and is reported to give lower resolution than DGGE analysis (Langefores et al., 2000). PCR-based techniques are the most attractive option for large-scale genotyping as they require very little DNA. The DNA does not have to be of high molecular weight and does not need to be highly purified. Thus much time (and money) can be saved during sample preparation by using PCR-based techniques.

In considering the analysis of many samples at once, SSCP analysis has previously been dismissed because of the need for specialized, dedicated equipment and radioisotopes. DGGE has been the preferred method for genotyping MHC class II in salmonids (Miller et al., 1997; Langefores et al., 2000). Though DGGE is more convenient than RFLP for genotyping salmonid MHC alleles (Langefors et al., 2000), it appears to be less sensitive than SSCP. For example, in a sample containing 17 Atlantic salmon alleles, Langefors et al. (2000) observed that 2 (12%) of the alleles were not resolved by DGGE. In additional, of 32 chinook salmon alleles, only 14 (44%) were distinguished consistently by DGGE without further analysis (Miller et al., 1997). Another disadvantage of DGGE is the requirement of specialized equipment and training.

RSCA is an SSCP-based technique that utilizes fluorescently labeled oligonucleotide primers and has been used for the detection of human MHC class I and MHC class II alleles (Arguello and Madrigal, 1999). Our laboratory has used RSCA in conjunction with sequencing to characterize more than 60 different MHC class II B1 alleles and to genotype several hundred samples collected from both Lake Superior and Lake Michigan (Dorschner et al., 2000; R.B. Phillips, unpublished work). As a continuation of this work, a subset of samples from the Apostle Islands in Lake Superior was also analyzed by SSCP using 32P-labeled nucleotide. In these studies RSCA identified 40 alleles and SSCP distinguished 46 alleles. Unfortunately, nucleotide sequences have not been determined for all of the animals used in these studies. However, it is fair to suggest that, according to zygosity matches and sequencing of mismatches between SSCP and RSCA, most alleles were detected if not all. A recent study by Garrigan and Hedrick (2001) utilized SSCP analysis to genotype 44 winter-run chinook salmon at the MHC class I A locus. Sequences of all individuals confirmed that SSCP analysis distinguished 100% of alleles that were present. With the increasing improvement and diminishing cost of fluorescence technologies, it is possible that SSCP analysis can be adapted to utilize the benefits provided by fluorescent detection mechanisms.

Both the RSCA and SSCP results obtained from these samples clearly distinguished the lean morphotype from the siscowet morphotype. Despite a small difference between the numbers of shared alleles detected by the 2 methods, the genotype data gathered suggested that only about 20% of alleles were shared between the 2 morphotype populations. The difference in the number of alleles detected and some possible errors produced by each method were also evidenced by a 6% difference in the calculated heterozygosities by the 2 methods. However, both observations compared favorably with observations of other different Lake Superior lake trout populations by Reimer et al. (2001).

Further comparison of these 2 techniques suggested that SSCP analysis was more sensitive than RSCA, detecting an additional 6 alleles. Also, correlations were seen between the RSCA alleles and the SSCP alleles in the genotypes of homozygotes (Table 1), at least for those alleles occurring with the highest frequencies within the sample set. That is, in homozygous individuals, RSCA allele 2 corresponds with SSCP allele H, RSCA allele 25 corresponds with SSCP allele P, and RSCA allele 22 corresponds with SSCP allele J. Some discrepancies were noticed between the RSCA genotypes and the SSCP genotypes. For example, several animals that contained RSCA allele 25 were scored by SSCP as having alleles A, AR, and P. Similarly, a few animals contained the RSCA allele 34, which was scored by SSCP as AA, H, and I (Table 1). In cases in which the RSCA allele did not match the SSCP allele, sequencing revealed that errors in zygosity identification were equal between RSCA and SSCP. However, the RSCA genotyping incorrectly scored heterozygotes as homozygotes and homozygotes as heterozygotes, whereas SSCP errors occurred only as heterozygotes scored as homozygotes (SSCP correctly identified all homozygotes).

The frequencies of the RSCA allele 5 and the SSCP allele I differed (Table 2). It seems that in the SSCP analysis, 2 alleles had migrated so closely that they were not resolved in some samples. Hence the two bands were scored as being the same. This outcome is not surprising when dealing with such a large number of alleles on a single gel, and the problem may be resolved in future studies by the inclusion of an internal standard or by the use of an allelic ladder as suggested by K.M. Miller (personal communication), and also by extending the electrophoresis time.

We noticed that RSCA allele sizes varied between gels, making the assignment of size and thus the distinction of the same allele between gels difficult. When assigning molecular sizes to RSCA alleles using Genotyper, it is easy for the operator to err by subjectively designating slightly different sizes to the same allele in different samples. The use of a more sophisticated electrophoresis system utilizing thermal control may alleviate inconsistencies in electrophoretic mobilities between gels. Another explanation for the discrepancies observed when matching RSCA alleles with SSCP alleles is the presence of extra alleles detected by SSCP analysis. It is also possible that the 6 extra alleles detected by SSCP might have been resolved by RSCA using a fourth reference strand, or by altering the electrophoresis conditions. Conversely, the same may be said of SSCP analysis for such discrepancies. It is also possible that 2 different alleles in a heterozygote could migrate to the same point on a gel, appearing as a single allele, which could help to explain discrepancies in the numbers of homozygotes detected by the 2 techniques. Since neither method provides an absolute size value, as the DNA migrates through the nondenaturing acrylamide gel according to its 3-dimensional shape, not its molecular weight, a direct comparison of individual alleles might be better accomplished by comparing fewer alleles and fewer genotypes at one time from smaller sample numbers.

Considering the merits and caveats of both RSCA and SSCP, SSCP analysis appears to be a more suitable method for genotyping salmonids for MHC class II alleles. Although both are PCR-based techniques, RSCA requires more knowledge of sequence in order to design and produce the fluorescently labeled reference strands required for hybridization to, and detection of, unlabeled test samples. The production of reference strands adds an extra level of complexity, cost, and time to this method compared with SSCP analysis. SSCP analysis is a simpler technique that provides at least the same, if not slightly higher, resolution of MHC class II alleles and genotypes in lake trout. Finally, it is possible that fluorescently labeled oligonucleotide primers can be used during the PCR amplification of SSCP products and that the standard SSCP technique could be easily adapted to fluorescent detection using an automated sequencer or other fluorescence detection system. Fluorescent technologies would also allow us to incorporate an internal standard for every sample by SSCP analysis, thus improving the resolution of alleles that migrate very close to each other. Our continuing research will strive to refine the detection of MHC class II alleles in salmonids by fluorescence SSCP as the preferred method of analysis for large sample numbers.