North American Journal uf Fisheries MunaKfinent
American Fisheries Society 1997
17:929-938. 1997
Species Identification of Pacific Salmon by Means of a
Major Histocompatibility Complex Gene
RUTH E. WITHLER, TERRY D. BEACHAM, TOBI J. MING,
AND KRISHNA M. MILLER
Department of Fisheries and Oceans, Science Branch
Pacific Biological Station, Nanaimo, British Columbia V9R 5K6. Canada
Abstract.—A rapid genetic test to identify Pacific salmonid tissue samples to the species level
is described. An exon (coding DNA) and its adjacent intron (noncoding DNA) of a major histocompatibility complex (MHC) class II gene were amplified by the polymerase chain reaction from
eight Oncorhynchus species and the two Salmo species that have been transplanted to British
Columbia. Among Pacific salmonids, the length of the amplified sequence was between 809 and
826 base pairs (bp) for cutthroat trout Oncorhynchus clarki. rainbow trout and steelhead O. mykiss.
chinook salmon O. tshawytscha. coho salmon O. kisutch, masu O. mason, and some sockeyc salmon
O. nerka: it was between 993 and 1,034 bp for pink salmon O. gorbuscha, chum salmon O. keta.
and other sockeyc salmon. Sequence length ranged from 1,000 to 3,000 bp for brown trout Salmo
trutta and from 1.500 to 3,000 bp for Atlantic salmon S. salar. Amplified sequences from all
Pacific salmonids except rainbow trout-steelhead and cutthroat trout displayed species-specific
restriction fragment length polymorphisms (RFLPs) after independent digestion with three restriction enzymes (Avrll. Bell. B s o f l ) . Restriction patterns of Pacific salmon sequences between 993
and 1,034 bp distinguished them from the 1,000-bp brown trout sequences. Intraspecific RFLP
variability revealed regional differentiation in phenotypic frequencies in three species: coho salmon
populations in southern British Columbia differed from those in northern British Columbia and
the Frascr River; sockeye salmon from Kamchatka and Bristol Bay differed from those of British
Columbia; and Japanese and North American chum salmon were well differentiated, enabling an
accurate classification to continent of origin.
Correct species identification is of fundamental
importance in the management and conservation
of Pacific salmon Oncorhynchus spp. Identification
of species is routine when the fish are approaching
sexual maturity, given the distinctive physical
changes that occur. However, species identification
can be more problematical when juveniles are to
be identified, even for experienced observers. In
some cases, species identification is required when
only a portion of a fish (e.g., a fillet) is available
after having been processed and possibly cooked,
or when a sample has been preserved in formalin
or ethanol.
Species identification of fish independent of
morphological characteristics has centered on the
use of genetic and immunological variation. When
adequate tissue samples of suitable quality arc
available, electrophoretic analysis of isozymes can
be an effective method of species discrimination
(Mork el al. 1983; Keenan and Shaklee 1985;
Graves et al. 1989). However, when there is an
inadequate amount of tissue for analysis or the
tissue is of poor quality, amplification of selected
DNA fragments via polymerase chain reaction
(PCR) may be required (Pendas et al. 1995; Unseld
et al. 1995). Species identification of tunas Thunnus spp. was accomplished initially by direct se-
quencing of an amplified mitochondrial DNA fragment (Bartlett and Davidson 1991). Subsequently,
analysis of restriction fragment length polymorphisms (RFLPs) of a longer mitochondrial fragment was developed for tuna species identification
(Chow and Inoue 1993), enabling relatively simple
laboratory analysis. The PCR amplification of a
DNA sequence and subsequent RFLP analysis of
the fragment has also been applied to a number of
other species identification problems (Silberman
and Walsh 1992; Chow et al. 1993).
Substantial species differences have been reported in the nucleotide sequence of a major histocompatibility complex (MHC) class II pi gene
in Atlantic salmon Salmo salar and Pacific salmon
(Hordvik et al. 1993; Grimholl et al. 1994; Miller
and Withler 1996). The MHC genes are involved
in the function of the immune system; T cells recognize a foreign antigen only in association with
class I or II molecules encoded by the MHC. The
MHC class II gene sequenced in Atlantic and Pacific salmon displays both intra- and interspecific
polymorphism. Species-specific DNA sequences
occur in both the DNA sequences that encode proteins (exons) and the interspersed noncoding DNA
sequences (introns). In this study, we developed
techniques to amplify a contiguous MHC class II
929
930
WITHLER ET AL.
pl exon and intron by PCR and then to identify
known sequence variants (Miller and Withler
1996) from RFLPs. This proved to be a relatively
simple method of differentiating Salmo from Oncorhynchus species and of identifying tissue samples from six of the eight Pacific salmonids to
species. We then conducted a broad geographic
survey of the five Oncorhynchus species that are
harvested commercially to confirm the accuracy
of species identification over the species' range.
Finally we applied the method to several species
identification problems, including a collection of
juvenile salmon sampled at sea, a group of 24 fish
confiscated by a fishery officer and the liquid residue from a cooler used to transport fish, and a
putative Atlantic salmon carcass recovered from a
British Columbia stream.
Methods
DNA samples.—Blood or liver samples were
collected from individual Pacific salmon representing, for most species examined, a broad geographic range of populations (Table 1). The Oncorhynchus species examined were chinook salmon O. tshawytscha, coho salmon O. kisutch. sockeye salmon O. nerka, chum salmon O. keta. pink
salmon O. gorbuscha, the masu and amago races
of O. masou (cherry salmon), cutthroat trout O.
clarki. and rainbow trout and steclhead O. mykiss.
Liver samples were also obtained from 20 Atlantic
salmon being reared on commercial farms in British Columbia and eight hatchery-reared brown
trout Salmo trutta from a naturalized population in
the Cowichan River. The methods of DNA extraction have been outlined by Withler et al. (1994)
and Beacham et al. (1995).
Once the methods for species identification had
been determined, we applied them to three problems of species identification. Another investigator
provided us with fins from 15 salmon caught on
the high seas, among which were two fish of
known (initially only by the investigator) identity
and 13 fish of uncertain species identity. A fishery
officer provided us with one scale from each of 24
salmon carcasses that required identification to
species and a water sample containing blood and
scales collected from a cooler in which fish had
been carried. Finally, we extracted DNA from the
scales of a fish that was recovered dead from a
British Columbia stream and tentatively identified
as an Atlantic salmon.
PCR analysis.—An MHC class II pi exon and
the succeeding intron sequence were amplified
with a sense primer located at the 10th codon of
pi (5'-CCGATACTCCTCAAAGGACCTGCA-3').
The antisense primer incorporated the last nucleotide of the intron and the first seven codons of
the P2 exon (5'-TCAGTCTGACATGGGGCTCAACT-3'). Each PCR reaction contained 1 mg of
template DNA, 10 pmol of each of the forward
and reverse primers, 200 mmol of each deoxynucleotide, and 2.5 units of Taq polymerase. Amplification of the gene was achieved by running 40
cycles of 1 min at 94°C, 2 min at a 55°C annealing
temperature, and 4 min at 72°C. The amplified
fragment was then digested with 5 units/mg DNA
of each of the restriction endonucleases Avrll,
Bell, or Bsofl. Five to 10 mL of each reaction
were analyzed by gel electrophoresis on a 2% agarose gel, with a 100-base-pair (100-bp) size marker
ladder run in three lanes of a 20-lane gel. The gel
was stained with cthidium bromide, and sizes of
the resultant DNA fragments were recorded.
Data analysis.—After digestion of the amplified
gene with the appropriate endonuclease, frequencies of occurrence of the observed restriction fragment phenotypes were recorded. Because the allelic products of two (or more) loci were amplified
in at least some species (Miller and Withler 1996),
the restriction fragment patterns could not be used
to provide allele frequencies (i.e., there was no
indication of copy number for each allele producing a restriction pattern). Thus, regional differentiation in the frequencies of restriction fragment
phenotypes. rather than genotypes, for coho salmon, sockeye salmon, and chum salmon was evaluated by the chi-square test with Monte Carlo simulations of the distribution of the chi-square statistic (Roff and Bentzen 1989). We made 1,000
simulations of the original data using the MONTE
program of the REAP software package (McElroy
et al. 1992).
Results
PCR of the MHC Gene
The lengths of the MHC class II sequences amplified in the 443 Oncorhynchus samples examined
in this study were between 809 and 826 bp
("short" alleles) in cutthroat trout, rainbow troutsteelhead, chinook salmon, coho salmon, masu,
and some sockeye salmon; they were between 993
and 1,034 bp ("long" alleles) in pink salmon,
chum salmon, and other sockeye salmon (Figure
1). The primers used in this study likely amplified
alleles from at least two closely related loci in most
Oncorhynchus species, as described by Miller and
Withler (1996).
SPECIHS-SPECIFIC H1STOCOMPATIBILJTY GENE
931
TABU- I.—List ofsa/monid populations sampled in survey of variation in me 0i exon and intron of a major histocompatibility complex class II gene. Numbers of fish sampled in each population are in parentheses.
Geographic area
Region
Kamchatka
Alaska
British Columbia
Kuril Lake
Iliamna Lake
Skecna River
Fraser River
Vancouver Island
Alaska
Yukon Territory
British Columbia
Bristol Bay
Yukon River
North Coast
South Coast
Fraser River
Vancouver Island
Populations (N)
Sockeye salmon
Gavrushka Bay (10), Vichenkiya River (8)
Woody Island (8). Gilbrallar Creek (10)
Fulton River (W)
Shuswap River (10), Adams River (8)
Sproat Lake (9). Great Central Lake (9)
Chinook salmon
Unknown (10)
Whitehorse Hatchery (10)
Kilimat River (10)
Bute Inlet (10)
Nechako River (10)
Robertson Creek (10). Conuma River (10). Cowichan River (10).
Nanaimo River (10), Big Qualicum River (10). fish farms (20)
Coho salmon
Kamchatka
Yukon Territory
British Columbia
Yukon River
Queen Charlotte Islands
North Coast
South Coast
Fraser River
Vancouver Island
Unknown (8)
Unknown (10)
Pal lam Creek (10)
Kitimat River (10)
Tenderfoot River (10)
Chilliwack River (10), Chchalis River (10)
Nilinat River (10). Shawnigan Lake (10). Big Qualicum River (10).
Puntledge River (10), Quinsam River (9)
Chum salmon
Japan
Yukon Territory
British Columbia
Kamchatka
British Columbia
Hokkaido
Yukon River
North Coast
Fraser River
Vancouver Island
Fraser River
Vancouver Island
Japan
Hokkaido
Hokkaido
British Columbia
Fraser River
Skeena River
British Columbia
Vancouver Island
Vancouver Island
Fraser River
British Columbia
Vancouver Island
British Columbia
Vancouver Island
Chitose River (10), Horonai River (6). Abashin River (8)
KJuane River (5)
Atnarko River (10)
Chilliwack River (7)
Nilinat River (2). Big Qualicum River (4)
Pink salmon
Unknown (3; odd-numbered brood year)
Chilliwack River (19; odd-numbered brood year)
PunUedge River (9; even-numbered brood year)
Masu and amago
Unknown (6; masu)
Unknown (7; amago)
Rainbow trout and steelhead
Domesticated (6; rainbow trout)
Unknown (7; steelhead)
Cutthroat trout
Fulford River (5)
Oyster River (5)
Taylor River (5)
Brown trout
Cowichan River (8)
Atlantic salmon
Fish farms (20)
The sequences amplified encode 72 amino acids
of an MHC pi exon followed by an intron between
548 and 773 bp long in Pacific salmon. All Pacific
salmon long allcles are distinguished from short
ones by insertion of a 199-bp Hpal short interspersed repetitive DNA element (SINE) near the
3' end of the intron (Miller and Withler 1996).
Sockeye salmon is the only species known to be
polymorphic for long and short alleles, and individual sockeye salmon possessing both long and
short allcles have been observed (Figure 1). No
cutthroat trout MHC class II sequences were examined by Miller and Withler (1996), but three
cutthroat trout alleles sequenced in the present
932
WITHLER ET AL.
TABLH 2.—Fragment si/.es of DNA (in base pairs) observed after amplification of a major histocompatibility
complex class II gene and restriction with Avrll. Bcl\. or
Bsofl for eight Oncorhynchus species and brown trout
(the brown trout 1 ,(XX)-basc-pair allelcs only).
Species
Avrll
Bell
Chinook salmon
821
82)
203. 246. 372
Coho salmon
260.561
821
278. 543
246. 575
203. 246. 372
Sockeye salmon
Long allele
Shon allele
1 35. 858
809
281.712
281.528
206. 258, 529
206. 603
Chum salmon
1 .034
1 .034
175. 178.285. 396
1 75. 285. 574
600
Fici/'RH I.—Amplified Salmo and Oncorhynchus genes
of major histocompatibility complex class II (($1 exon
and adjoining intron). Lanes are as follows: molecular
weight markers ( 1 , 2 . 12, 13), brown trout (3), Atlantic
salmon (4). pink salmon (5). chum salmon (6), sockeye
salmon (7). rainbow trout (8). coho salmon (9), chinook
salmon (10), and masu ( 1 1 ) . The molecular weights arc
in base pairs.
study (unpublished data) were all short allelcs (i.e.,
they lacked the Hpal SINE insert).
The MHC sequences amplified in Atlantic salmon were between 1,500 and 3,000 bp long, resulting in a clear distinction between the Pacific
salmon species and Atlantic salmon (Figure 1).
However, brown trout sequences of 1,000 bp and
1,500-3,000 bp were obtained. Individual brown
trout possessing only 1,000-bp sequences, only sequences longer than 1,500 bp, and both length categories were observed. Thus, neither Pacific nor
Atlantic salmon could be distinguished invariably
from brown trout on the basis of length of amplified products. The undigested 1,000-bp brown
trout sequences were indistinguishable from the
long Pacific salmon alleles (Figure 1), but they
lack the Hpal SINE that characterizes the Pacific
salmon long alleles (unpublished data).
RFLP Analysis of Oncorhynchus and Salmo trutta
All of the Avrll, Bell, and Bsofl restriction
patterns predicted on the basis of known sequences
of the MHC class II gene in seven Oncorhynchus
species (Miller and Withlcr 1996) were observed
in the geographic survey (Figure 2; Table 2). Digestion with Avrll produced monomorphic patterns for all species except sockeye salmon, in
which the short allele (undigested) differed from
the long allele (one restriction site in the SINE
Bsofl
Pink salmon
147.852
999
267. 732
Masu-amago
823
197.626
823
206.617
Rainbow trouisieelhead
825
825
201.625
102,247,476
247. 578
Cutthroat irout
826
Brown trout
1.000
826
201.625
1.000
KM). 247.479
247. 579
16,204,231,549
204, 247. 549
45 1 . 549
insert). Only Avrll consistently distinguished chinook from coho salmon. None of the 120 chinook
salmon that were examined produced an MHC sequence that was digested with Avrll, but the amplified products from all 117 coho salmon surveyed had a single Avrll restriction site, resulting
in two fragments of 260 and 561 bp. Similarly,
pink salmon and chum salmon were readily discriminated by the presence of an Avr II restriction
site in pink salmon (Table 2; Figure 2). However,
digestion with Avrll alone failed to distinguish
among cutthroat trout, rainbow trout-steelhead,
chinook salmon, sockeye salmon (short allele), and
masu. It also provided no differentiation between
the amplified products of sockeye salmon (long
allele) and pink salmon. Moreover, Avrll failed to
distinguish the brown trout 1,000-bp sequences
from those of chum salmon (Figure 2).
Restriction sites for Bel I were present in sockeye salmon allelcs (both long and short) and masu
alleles, and in some coho salmon, rainbow troutsteelhead, and cutthroat trout alleles (Table 2). Digestion with Bel I alone failed to distinguish the
amplified products of masu, rainbow trout-steelhead, and cutthroat trout, and it did not discriminate coho salmon from chinook salmon. Nor did
Bell digestion distinguish among pink salmon,
chum salmon, and brown trout or between coho
salmon and sockeye salmon.
SPECIES-SPECIFIC HISTOCOMPATIBILITY GENE
-1000
- 600
200
-1000
600
r 200
1000
600
200
933
Joint examination of the fragment patterns generated by Avrll and Bell digestion enabled the
differentiation of chinook salmon, coho salmon,
sockeye salmon, pink salmon, and chum salmon.
However, digestion with these two enzymes did
not differentiate consistently either chinook salmon or masu from cutthroat trout and rainbow troutsteelhead nor the two trout species from each other.
Chum salmon and brown trout were also indistinguishable (Table 2).
The MHC sequences of all Oncorhynchus species possessed at least one Bsof I restriction site,
and rainbow trout-steelhead, cutthroat trout, coho
salmon, chum salmon, and masu were polymorphic for Bsof I sites. Digestion of the MHC gene
with Bsof I alone did not allow complete differentiation of coho salmon from chinook salmon nor
of coho salmon, rainbow trout-steelhcad, and cutthroat trout; sockeye salmon and masu were not
differentiated consistently (Table 2). However, in
combination with the Avrll and Bel I restriction
patterns, Bsof I provided the required discrimination among chinook salmon, rainbow trout-steelhead, and cutthroat trout and between chum salmon and brown trout (Table 2). The Bsof I patterns
also differentiated masu from rainbow trout-stcclhead and cutthroat trout, although a wider survey
might reveal overlap among these three polymorphic species. The only two sympatric species not
separable on the basis of digestion with all three
enzymes were rainbow trout-steelhead and cutthroat trout.
Two rainbow trout-steelhead allcles not sequenced in the survey of Miller and Withler (1996)
were detected in this study. Two Skeena River
steelhead possessed both the known MHC sequence containing two Bsof I sites at positions 476
FIGUKH 2.—Amplified Oncorhynchus and brown trout
genes of major hisiocompaiibility complex class II digested with Avr II. Bel I, or Bsofl. Lanes are as follows:
molecular weight markers (1, 20), coho salmon (2-4).
chinook salmon (5), rainbow trout-steelhead (6. 7).
masu (8-10), cutthroat trout ( I I ) , sockeye salmon (1214), chum salmon (15-17). pink salmon (18). and brown
trout (19). For Bel I and Bsofl. sockeye salmon monomorphic for the short allele (lane 12). monomorphic for
the long allele (lane 13). and possessing both allcles
(lane 14) arc shown. Coho salmon (lanes 2, 4) and rainbow trout-steelhead (lane 7) possessing both Bel I alleles
are shown. Heterozygotes for Bsofl allcles are shown
for coho salmon (lane 4). rainbow trout-steelhead (lane
7), masu (lane 10), and chum salmon (lane 17). The
molecular weights arc in base pairs.
934
WITHLKR HT AL.
and 578 of the 825-bp sequence and a second sequence lacking the cut site at position 476 (Table
2; Figure 2). Thus, the new allele produced a twobanded (247 and 578 bp) restriction fragment pattern identical to one of the coho salmon alleles.
However, because the two stcelhead each possessed one copy of the known rainbow trout-steelhead allele, and because the new allele did not
restrict with Avrll or Bel I, these fish would have
been identified as rainbow trout-steelhead or cutthroat trout even if their identity had not been
known. Two of six domesticated rainbow trout
from the Fraser River drainage possessed both the
known MHC sequence that does not restrict with
Bel I and a sequence that digested into fragments
201 and 625 bp long (Figure 2). This Bel I restriction pattern was also observed in cutthroat trout
(Table 2).
In sockcye salmon, which were polymorphic for
the long and short alleles, and in rainbow troutsteelhead, cutthroat trout, coho salmon, chum
salmon and masu, which were polymorphic for
Bel 1 or Bsof I sites (or both), individuals possessing two restriction patterns occurred and could be
identified (Figure 2).
Identification of Unknown Samples
Amplification of MHC sequences from fin clips
of 15 juvenile salmonids sampled at sea produced
fragments that were either approximately 820 bp
(eight fish) or 1,000 bp long (seven fish). No individuals heterozygous for the short and long alleles were observed. Treatment of the eight 820bp products with Avrll resulted in no digestion,
and treatment with Bel I produced two restriction
fragments, approximately 280 and 530 bp long, in
every case. Thus, these fish were identified as
sockeye salmon homozygous for the short allele.
Digestion of these samples with Bsofl produced
the expected restriction pattern with fragments approximately 205 and 600 bp long. Among the seven amplified 1,000-bp products, one was digested
only with Bsofl, producing fragments approximately 175, 285, and 575 bp long (chum salmon);
four samples were not digested by Bel I but displayed two fragments about 150 and 850 bp long
after Avr II digestion (pink salmon); and two samples were digested by both Avrll and Bell, producing fragments of the sizes expected from the
sockeye long allele. Digestion of the latter six samples with Bsofl was consistent with the species
designation. One pink salmon and the chum salmon were the two samples identified independently
TABLH 3.—Regional counts of major histocompatibility
complex class II restriction fragment phenotypes in sockeye salmon; yV is total sample size. The observed phcnotypes were homo/ygous short (809-base-pair [bp] alleles
only), "heterozygous" (809- and 933-bp alleles), and homo/ygous long (993-bp alleles only). The long and short
sequences amplified from "heterozygous" individuals are
not necessarily products of the same loci.
Number with fragment phenotypc:
Region
JV
Short
Short
and long
Long
Kamchatka
Bristol Bay
Skccna River
Fraser River
Vancouver Island
18
19
5
1
10
10
12
11
9
0
10
1
9
0
0
5
to
20
IS
2
on the basis of morphology, and the genetic analysis confirmed the morphological classification.
Amplification and digestion of the DNA extracted from the 24 scales supplied by a fisheries officer
revealed that all fish were coho salmon. For each
scale sample, and for DNA extracted from the water sample collected from a cooler in which fish
had been transported, amplification with the MHC
primers produced DNA fragments approximately
820 bp long. In each case, digestion with Avr II
resulted in fragments approximately 260 and 560
bp long. For the scale samples, independent digestion with Bel I and Bsof I produced fragment
patterns indicating that each fish was either homozygous or heterozygous for the two possible
coho salmon patterns for each enzyme (Table 2).
Both coho salmon restriction patterns for both enzymes were present in DNA from the cooler sample. Thus, we determined that coho salmon had
been transported in the cooler, although none of
the fish was recovered for morphological confirmation.
Amplification of MHC sequences from the putative Atlantic salmon found dead in a stream produced fragments 2,000 bp long, confirming that
the fish was either a brown trout or Atlantic salmon. We then amplified the 5S ribosomal DNA sequences, which differ in length between Atlantic
salmon and brown trout (Pendas et al. 1995), and
confirmed that the fish was an Atlantic salmon.
Population Variation
Frequencies of the three sockeye salmon phenotypes (short sequences only, long sequences
only, short and long sequences) varied significantly (P < 0.001) among five regions ranging
from Kamchatka to Vancouver Island (Table 3).
SPECIES -SPECIFIC HISTOCOMPATIBILITY G E N E
Greater proportions of sockeye salmon from Kamchatka and Bristol Bay possessed long alleles, either exclusively or together with short alleles, than
did sockeye salmon from the Skeena and Fraser
river systems of British Columbia. Sockeye salmon from Fulton River, a Skeena River tributary,
were monomorphic for the short allele. Although
both the short and long phenotypes were relatively
common among Vancouver Island sockeye salmon, only one individual possessed both long and
short sequences. Further sampling is required to
determine whether this shortage of "heterozygotes" resulted from small sample sizes, heterogeneity in phenotypc frequencies among Vancouver Island populations, or some novel genetic
mechanism.
Frequencies of the two chum salmon MHC al-
leles detected by Bsof I digestion differed significantly between Japanese and North American
populations. Of the 24 Japanese chum salmon examined, 23 fish possessed only sequences with two
Bsof I sites and 1 fish possessed sequences with
both two and three sites. The 28 North American
chum salmon were monomorphic for sequences
possessing three Bsof I sites. Thus, all of the Japanese chum salmon possessed at least one copy of
the MHC allele possessing two Bsof I sites, and
North American chum salmon were distinguished
by its absence. Larger sample sizes of fish from
both continents will be required to determine the
frequency of the "North American'' MHC allele
(and fish homozygous for it) in Asian chum salmon
populations and to determine if the ''Japanese"
allele occurs in North American populations.
Coho salmon MHC alleles sequenced by Miller
and Withler (1996) and detected in the current
study were polymorphic for both a Bel I restriction
site and an independent Bsof I site (Table 2). Thus,
four coho salmon MHC alleles potentially exist:
one possessing no Bel I site but one Bsof I site
(allele A), one possessing a Bell and a Bsof I site
(£), one possessing no Bell site but two Bsof I
sites (C), and one possessing one Bel I and two
Bsof I sites (D). Of the five known sequences (Miller and Withler 1996), two were allele A. two were
allele B. and one was allele C. Allele D was not
present among the coho genes sequenced in that
study and was not observed in the current survey.
Coho salmon heterozygotes BD and CD would
have provided distinctive patterns after individual
digestion with Bel I and Bsof I, but the AD and BC
genotypes (each heterozygous in separate diges-
tions with Bell and Bsof I) could be distinguished
only by simultaneous (double) digestion with the
935
TABLK 4.—Regional counts of major histocompaiibiliiy
complex class II restriction fragment phenotypes in coho
salmon of British Columbia (B.C.). as identified by single
and double digestions with Bel I and Bsof I cndonucicascs.
The three alleles possessed either one Bsof I and no Bel I
site (A allele). one Bsof I and one Bell site (B). or two
Bsof I and no Bell sites (C). Alleles from at least two loci
were amplified, hut no individuals possessing three different alleles were observed (N = sample size).
Number with genotype
Region
N
A
AB
B
AC
BC
C
Northern B.C.
Hraser River
Southern B.C.
19
20
57
9
5
5
9
6
14
1
7
4
0
|
K
0
1
0
0
15
11
two enzymes. No DD homozygotes or BD or CD
helcrozygotes were observed. Ten coho salmon
from southern British Columbia displayed the ambiguous double hcterozygote pattern, but all were
determined to be BC heterozygotes by double digestion. No fish carrying all three of the A, B. and
C alleles were detected in the double digestions,
but coho salmon carrying three pi alleles distinguished by Mho I digestion have been observed
(Miller and Withler 1996).
The distribution of MHC restriction phenotypes
varied significantly among coho salmon from three
regions of British Columbia (P < 0.001) (Table
4). Two northern populations possessed only the
A and B alleles, Frascr River populations possessed
the A, B, and (at a low frequency) C alleles, and
southern populations (Vancouver Island and mainland Strait of Georgia) were characterized by a
high frequency of the C allele.
Discussion
The MHC class II gene amplified in this study
is subject to balancing selection favoring heterozygotes, and as a result is highly polymorphic both
within and among the Pacific salmon species (Miller and Withler 1996). Nevertheless, the PCRRFLP analysis developed from the 29 known sequences proved to be a robust method for discriminating among Oncorhynchus species and between
Pacific and Atlantic salmon. Among the 443 Pacific salmon surveyed, the only unanticipated restriction patterns observed were from cutthroat
trout, a species for which no sequence data had
been available before this study, and steelhead.
Two steelhead were heterozygous for an allele with
a Bsof I restriction site loss that resulted in a Bsof I
pattern identical to one observed for coho salmon.
However, the new steelhead allele did not possess
936
WITHLER ET AL.
any Avrll or Bell sites, as would be expected of
a rainbow trout-steelhead allele but not a coho
salmon allele.
All three cutthroat trout MHC sequences obtained in this study possessed a single Bsof I site
that produced a restriction pattern identical to that
of the new rainbow trout-steelhead allele, as well
as a single Bel I site that has not yet been observed
in rainbow trout-steelhead. However, the subsequent digestion of MHC sequences of five cutthroat trout from each of three strains that have
been supplemented by hatchery production revealed that cutthroat trout also possess sequences
without the Bel I site and sequences with the common steelhead Bsof I restriction pattern. Thus, digestion of the MHC p 1 and intron sequences with
the three enzymes used in this study does not reliably differentiate rainbow trout-steelhead from
cutthroat trout. The two steelhead sequences (Miller and Withler 1996) can be distinguished from
the three cutthroat trout sequences obtained in this
study with Hinf I, which digests the rainbow troutsteelhead sequences twice and the cutthroat trout
sequences at the same two and an additional site.
However, a broader survey of purebred populations, as well as those in which introgression between the species has occurred (Carmichael et al.
1993), will be required to confirm the technique.
A broader survey of masu populations may also
reveal overlap with the trout sequences. However,
differentiation of the weakly anadromous Asian
masu from the North American trout species will
rarely be necessary.
The PCR-RFLP analysis is independent of morphological or scale characteristics, which sometimes are ambiguous, and can be conducted with
minute amounts of fresh or preserved tissue. This
makes the technique very useful for unanticipated
species identification of fresh or frozen tissue,
identification of preserved historical samples (including scales and otoliths), and forensic identification of confiscated fresh or processed tissue.
The coho salmon examined in this study for a fisheries officer were each identified to species after
amplification of DNA extracted from a single
scale. In addition, sufficient DNA for the PCR and
species identification was obtained from water
containing blood and scales of the transported fish.
Among the Pacific salmon species, most of the
variability in intron length results from the presence or absence of the 199-bp Hpal SINE, which
distinguishes the "short*' allelcs of cutthroat trout,
rainbow trout-steelhead, masu, chinook salmon,
and coho salmon from the "long" alleles of pink
salmon and chum salmon. Only sockeye salmon
is polymorphic for the SINE, possessing both long
and short alleles. Various classes of SINEs are
common in salmonid fish genomes (Kido et al.
1991; Spruell and Thorgaard 1996) as well as in
mammalian intones, including those of the MHC
(Del Pozza et al. 1991). The 1,000-bp brown trout
sequences, although the same length as the Pacific
salmon "long" alleles, do not possess the Hpal
SINE.
In British Columbia, where Atlantic salmon are
the predominant aquacultural species, discrimination between Pacific and Atlantic salmon is important because it is otherwise difficult to identify
Atlantic salmon farm escapees or hybrids that
might result from interbreeding of escapees with
native salmonids. However, the presence of a
small, introduced brown trout population in the
Cowichan River necessitates a test that can distinguish Atlantic salmon from brown trout. The MHC
exon amplified in this study is also polymorphic
and under balancing selection in Atlantic salmon
(Hordvik et al. 1993; Grimholt et al. 1994), but no
sequence data are available for the adjacent intron.
The results of this study indicate that the Atlantic
salmon intron is consistently longer than those of
the Pacific species (resulting in amplified sequences at least 1,500 bp long), whereas brown trout
sequences overlap those of Pacific salmon long
alleles and Atlantic salmon alleles (1,500-3,000
bp).
Digestion with the three enzymes used in this
study failed to differentiate consistently between
brown trout and Atlantic salmon sequences that
were in the 1,500- to 3,000-bp size range (unpublished data). The MHC sequences from both species were undigested by Avrll and Bell, and the
Bsof I patterns were not species specific. Thus,
identification of a putative purebred or hybrid Atlantic salmon requires two steps: amplification of
the MHC sequences with no digestion to produce
sequences of 1,500 bp or longer; and amplification
of the 5S ribosomal sequences with the primers of
Pendas et al. (1995) to confirm that the long sequences are of Atlantic salmon rather than brown
trout origin. The putative Atlantic salmon that was
recovered dead from a stream and examined in this
study yielded highly degraded DNA from liver tissue that was unsuitable for amplification of the
2,000-bp MHC sequence, whereas high-quality
DNA extracted from scales was amplified easily
and used to confirm an Atlantic salmon identity
through the two-step amplification of the MHC and
5S ribosomal sequences.
SPECIES-SPKCIFIC HLSTOCOMPAT1BILITY G K N K
Although the emphasis of our study was on establishing a rapid, reliable method of species identification, the PCR-RFLP analysis of the MHC
class II gene revealed intraspecific regional variation in restriction pattern frequencies that might
be exploited for stock identification of sockeye
salmon, chum salmon, and coho salmon. If assignment of coho salmon to fairly broad geographic areas (e.g., northern British Columbia, Fraser
River, southern British Columbia) is sufficient for
management purposes, observed differences in
MHC RFLP frequencies may enable the same level
of stock composition estimates routinely obtained
for other species from isozyme analysis (Bcacham
et al. 1987; Wood et al. 1989). Similarly, if Fulton
River sockeye salmon (monomorphic for the short
MHC allele) represent all sockeye populations in
the Skecna River watershed, regional estimates of
sockeye salmon stock composition based on amplification of the MHC gene may be possible.
Japanese chum salmon are reportedly distinct
from Russian and North American populations at
both isozyme loci (Winans et al. 1994) and minisatellite DNA loci (Taylor et al. 1994; Beacham
1996). However, application of single-locus minisatellite DNA probes in the classification of individual chum salmon to continent of origin has
resulted in only about 80-95% accuracy (Taylor
et al. 1994; Beacham 1996). In the present study,
digestion of the MHC class II gene with Bsof I
enabled 100% accuracy in classifying Japanese
and North American chum salmon to continent of
origin. Although only 52 chum salmon were examined in this study and no Russian populations
were surveyed, the results indicate that PCRRFLP analysis may provide a relatively rapid and
effective method of distinguishing North American from Asian chum salmon.
As more classical MHC genes are isolated from
Pacific salmonids, and as more of the nucleotide
sequence variability among alleles can be detected
by sensitive techniques such as denaturing gradient gel electrophoresis (Fischer and Lerman 1983),
the usefulness of these highly polymorphic genes
for stock identification in Pacific salmonids may
increase.
Acknowledgments
We appreciate the assistance of many individuals in obtaining samples, particularly M. Kaeriyama for samples from Japan, N. Varanavskaya
for samples from Russia, Ray Billings (Duncan
Trout Hatchery) for the cutthroat trout and brown
trout samples, and W. Barner, C. Murray, A.
937
Thompson, and J. Candy for samples from British
Columbia and the Yukon Territory. D. Tuck assisted with the laboratory analysis. Funding was
provided by the Department of Fisheries and
Oceans.
References
Bartlett. S. E., and W. S. Davidson. 1991. Identification
of Thunnus tuna species by the polymerase chain
reaction and direct sequence analysis of their mitochondria! cylochrome b genes. Canadian Journal
of Fisheries and Aquatic Sciences 48:309-317.
Beacham, T D. 1996. The use of minisatellite DNA
variation for stock identification of chum salmon
(Oncorhynchus keta). U.S. National Marine Fisheries Service Fishery Bulletin 94:611-627.
Beacham. T. D., A. P. Could, R. E. Withler, C. B. Murray,
and L. W. Barner. 1987. Biochemical genetic survey and stock identification of chum salmon (Oncorhynchus keta) in British Columbia. Canadian
Journal of Fisheries and Aquatic Sciences 44:17021713.
Beacham. T. D., R. E. Withler, and C. C. Wood. 1995.
Stock identification of sockeye salmon by means of
minisatellite DNA variation. North American Journal of Fisheries Management 15:249-265.
Carmichacl, G. J., J. N. Hanson, M. E. Schmidt, and D.
C. Morizot. 1993. Imrogression among apache,
cutthroat, and rainbow trout in Arizona. Transactions of the American Fisheries Society 122:121130.
Chow, S., M. E. Clarke, and P. J. Walsh. 1993. PCRRFLP analysis on thirteen west Atlantic snappers
(suhfamily Lutjanidae): a simple method for species
and stock identification. U.S. National Marine Fisheries Service Fishery Bulletin 91:619-627.
Chow, S., and S. Inoue. 1993. Intra- and interspecific
restriction fragment length polymorphism in mitochondrial genes of Thunnus species. Bulletin of the
National Research Institute of Far Seas Fisheries
30:207-224.
Del Po/./a, G.. and six coauthors. 1991. Alu repeats and
evolution of the HLA-DQA1 locus. Pages 231-242
in J. Klein and D. Klein, editors. Molecular evolution of the major histocompatibility complex.
Springer-Verlag, Berlin.
Fischer, S. G.. and L. S. Lerman. 1983. DNA fragments
differing by single base-pair substitutions are separated in denaturing gradient gels: correspondence
with melting theory. Proceedings of the National
Academy of Sciences of the USA 80:1579-1583.
Graves. J. E., M. A. Simovich. and K. M. Schaefer.
1989. Electrophoretic identification of early yellowfin tuna, Thunnus albacares. U.S. National Marine Fisheries Service Fishery Bulletin 86:835-838.
Grimholt, U.. I. Olsaker. C. de V. Lindstrom. and O. Lie.
1994. A study of variability in the MHC class II
31 and class I a2 domain exons of Atlantic salmon
(Salmo salar). Animal Genetics 25:147-153.
Hordvik. L. U. Grimholl. V. M. Fosse, O. Lie, and C.
Endresen. 1993. Cloning and sequence analysis of
938
WITHLER ET AL.
cDNAs encoding the MHC class II (3 chain in Atlantic salmon (Salmo salar). Immunogenetics 37:
437-441.
Keenan. C. P., and J. B. Shaklee. 1985. Electrophoretic
identification of raw and cooked fish fillets and other
marine products. Food Technology in Australia 37:
117-128.
Kido, Y., and six coauthors. 1991. Shaping and reshaping of salmonid genomes by amplification of
tRNA-derivcd retroposons during evolution. Proceedings of the National Academy of Sciences of
the USA 88:2326-2330.
McElroy, D.. P. Moran. E. Bermingham. and I. Kornfield. 1992. REAP: an integrated environment for
the manipulation and phylogenetic analysis of restriction data. Journal of Heredity 83:157-158.
Miller. K. M., and R. E. Withler. 1996. Sequence analysis of a polymorphic Mhc class II gene in Pacific
salmon. Immunogenclics 43:337-351.
Mork, J.. P. Solemdal. and G. Sundnes. 1983. Identification of marine fish eggs: a biochemical genetics
approach. Canadian Journal of Fisheries and Aquatic Sciences 40:361-369.
Pcndas, A. M.. P. Moran. J. L. Marline/, and E. GarciaVa/que/.. 1995. Applications of 5S rDNA in Atlantic salmon. Brown trout, and in Atlantic salmon
X brown trout hybrid identification. Molecular
Ecology 4:275-276.
Roff. D. A., and P. Bentzen. 1989. The statistical analysis of mitochondrial DNA polymorphisms: \ 2 and
the problem of small sample sizes. Molecular Biology and Evolution 6:539-545.
Silberman, J. D., and P. J. Walsh. 1992. Species identification of spiny lobster phyllosome larvae via ri-
bosomal DNA analysis. Molecular Marine Biology
and Biotechnology 1:195-205.
Spruell. P., and G. H. Thorgaard. 1996. SINE sequences
detect DNA fingerprints in salmonid fishes. Heredity 76:317-324.
Taylor. E. B.. T. D. Beacham, and M. Kaeriyama. 1994.
Population structure and identification of North Pacific Ocean chum salmon (Oncorhynchus keia) revealed by analysis of minisatellite DNA variation.
Canadian Journal of Fisheries and Aquatic Sciences
51:1430-1442.
Unseld, M., B. Beycrmann. P. Brandt. and R. Heisel.
1995. Identification of the species origin of highly
processed meat products by milochondrial DNA sequences. PCR Methods and Applications 4:241243.
Winans. G. A.. P. B. Aebersold, S. Urawa, and N. V.
Varnavskaya. 1994. Determining continent of origin of chum salmon (Oncorhynchus keta) using genetic stock identification techniques: status of allozymc baseline in Asia. Canadian Journal of Fisheries and Aquatic Sciences 51 (Supplement I):95113.
Withler. R. E.. T. D. Beacham. R. F. Watkins. and T. A.
Stevens. 1994. Identification of farm-reared and
native chinook salmon (Oncorhynchus tshawytscha)
on the west coast of Vancouver Island, British Columbia, using the nuclear DNA probe B2-2. Canadian Journal of Fisheries and Aquatic Sciences
51:267-276.
Wood, C. C., D. T. Rutherford, and S. McKinnell. 1989.
Identification of sockeye salmon (Oncorhynchus
nerka) stocks in mixed-stock fisheries in British Columbia and southeast Alaska using biological markers. Canadian Journal of Fisheries and Aquatic Sciences 46:2108-2120.