Polymorphism of Alternative Splicing of Major Histocompatibility Complex Transcripts in Wild Tiger Salamanders.

Full text

Page 1

Polymorphism of Alternative Splicing of Major
Histocompatibility Complex Transcripts in Wild Tiger
Salamanders
Zafer Bulut Æ Æ Cory R. McCormick Æ Æ
David H. Bos Æ Æ J. Andrew DeWoody
Received: 17 March 2008/Accepted: 14 May 2008/Published online: 10 June 2008
? Springer Science+Business Media, LLC 2008
Abstract
is increasingly recognized as a source of transcriptome
diversity. To date, most AS studies have focused either on
comparisons across taxa or on intragenomic comparisons
across gene families. We generated a novel data set that
represents one of the first population genetic comparisons
of AS across individuals. In ambystomatid salamanders,
AS of the major histocompatibility complex (MHC) class
IIb gene (Amti-DAB) produces two transcripts, one full-
length and one truncated. The full-length transcript is
functional, but the truncated transcript is missing the crit-
ical b1 domain that forms half of the peptide binding
region in the intact MHC class II molecule. We captured
wild salamander larvae (Ambystoma tigrinum tigrinum)
and genotyped them at Amti-DAB via DNA sequencing.
From these same larvae, we extracted RNA from gill and
spleen and evaluated the relative expression level of Amti-
DAB in each tissue. Across individuals, 21% of the tran-
scripts were truncated (alternatively spliced), and the
absolute level of alternative transcript expression was
higher in gill. The high level of nucleotide variation among
seven Amti-DAB alleles provides the ability to detect sub-
stitutions (or linked DNA polymorphisms) that might have
Alternative splicing (AS) of mRNA transcripts
influenced AS. The data reveal no correlation between AS
and haplotype, allele, or zygosity. However, indirect evi-
dence (comparative expression patterns across 3 million
years of evolution) suggests that the truncated Amti-DAB
transcript may be functional and maintained by natural
selection.
Keywords
RT-PCR ? Branch point sequence
Ambystoma tigrinum tigrinum ? mRNA ?
Introduction
Nascent mRNAs are synthesized in the eukaryotic nucleus
and must be processed before transport to the cytoplasm,
where they ultimately serve as templates for translation
into polypeptides. Upon initial synthesis, primary mRNA
transcripts faithfully mirror the structure of the gene which
encodes them, but introns are subsequently removed during
the maturation of the mRNA molecule. Some gene tran-
scripts undergo alternative splicing (AS), where specific
exons are excluded from the mature transcript. In the exon
cassette model of alternative exon splicing, a particular
exon is either included or excluded from the mature tran-
script, resulting in ‘‘short’’ and ‘‘long’’ molecules (Gravely
2001; Cartegni et al. 2002). AS has generated interest
among geneticists because of its roles in intron evolution,
exon duplication, and gene regulation (Artamonova and
Gelfand 2007). Nevertheless, the evolutionary mechanisms
underlying AS remain largely unexplored.
Most AS studies have employed either comparative
analyses of transcripts from a single genome or interspe-
cific comparisons across taxa (Alekseyenko et al. 2007).
These comparative data have revealed that AS can be
affected by synonymous substitutions (Xing and Lee
Zafer Bulut, Cory R. McCormick and J. Andrew DeWoody
contributed equally to this work.
Z. Bulut ? C. R. McCormick ? D. H. Bos ? J. A. DeWoody
The Bindley Bioscience Center and Department of Forestry &
Natural Resources, Purdue University, West Lafayette, IN
47907-1159, USA
J. A. DeWoody (&)
Department of Forestry & Natural Resources, Purdue University,
West Lafayette, IN 47907-1159, USA
e-mail: dewoody@purdue.edu
123
J Mol Evol (2008) 67:68–75
DOI 10.1007/s00239-008-9125-1

Page 2

2005), that highly networked proteins are subject to
selection against AS (Hughes and Friedman 2008), and that
selection may not be evident at the DNA level yet still act
on RNA transcripts (Xing and Lee 2005). These insights
notwithstanding, there is yet another level of comparative
analyses that could prove fruitful in understanding both the
proximate and the ultimate reasons for AS: intrapopulation
assessments of wild, outbred individuals. This population
genetic approach could reveal aspects of AS that have
heretofore remained cryptic in genome-wide comparisons
or in analyses of inbred laboratory stocks. For example,
how plastic is the phenomenon of alternative splicing?
Does the genotype of the expressed gene influence the
relative proportions of a given gene’s transcripts, or are
RNA transcripts often produced in fixed ratios? In other
words, do polymorphisms in the DNA sequence of a gene
influence RNA splicing of that gene? To properly utilize a
population approach, an alternatively spliced gene(s) must
be characterized in an outbred (i.e., genetically variable)
population, ideally using multiple tissue types.
We have identified a tractable system for population
genetic analyses of AS in a key immune system gene using
larval salamanders that can be produced in the laboratory
or collected from the field. Major histocompatibility com-
plex (MHC) genes produce proteins that help differentiate
between self and nonself. In particular, highly polymorphic
class I and class II genes display short peptides to sur-
veillance cells, which can in turn stimulate the immune
response if foreign peptides are detected. MHC class II
molecules (Fig. 1) are dimers composed of alpha and beta
chains produced by two different genes, class IIa and class
IIb. The class IIb gene seems particularly subject to strong
balancing (and sometimes sexual) selection (Hughes and
Nei 1989; Potts et al. 1991; Satta et al. 1994).
Tournefier et al. (1998) reported that the primary full-
length transcript of the MHC class IIb gene in mole sala-
manders (genus Ambystoma) encodes a leader peptide (26
amino acids [aa]), a b1 domain (93 aa), a b2 domain (93
aa), a connecting peptide and transmembrane region (34
aa), and a cytoplasmic tail (20 aa). The cDNA sequence of
this gene has been fully characterized (Tournefier et al.
1998; Bos and DeWoody 2005), but the genomic sequence
is unknown and thus intron/exon boundaries are not
defined. In other vertebrates the b1 domain, which contains
the critical peptide-binding region (Fig. 2A), corresponds
to the second exon of the class II gene (Edwards et al.
1990). Nearly all of this region, all but the first 3 of 93 aa is
missing in the short (alternative) RNA transcript of the
axolotl (A. mexicanum [Tournefier et al. 1998]); exactly
the same region is missing in the short transcript of the
eastern tiger salamander (A. tigrinum tigrinum [Bos et al.
2005]). These two species diverged from one another
roughly 3 million years ago (Shaffer and McKnight 1996).
Both the full-length transcript and the alternative transcript
are transcribed in each species, as gauged by gene
expression assays and by cloning (Laurens et al. 2001; Bos
and DeWoody 2005). However, because these assays are
typically conducted on single individuals, there are no
population data on conspecific variability in AS. For
example, it is unknown if other aspects of genetic diversity
(such as heterozygosity or genetic divergence between
alleles in a heterozygote) affect gene splicing. Our work
tests whether cis-acting molecular signatures (Cartegni
et al. 2002) are involved in the evolution of splicing. We
do so by comparing MHC class IIb population genetic data
(DNA genotypes) with cDNA expression data (derived
from mRNA) on the Amti-DAB gene, using wild tiger
salamanders as a model.
Methods
Sample Collection, MHC Genotyping, and cDNA
Synthesis
In June 2006, we collected 34 wild larval tiger salamanders
(Ambystoma tigrinum tigrinum) of similar size (*7–9 cm)
from the Purdue Wildlife Area in Tippecanoe County,
Indiana. Individuals were collected using seines and min-
now traps, then transferred (live) to the laboratory, where
they were euthanized with MS-222. Spleen, gill, and
muscle tissue were immediately collected from each
specimen. Muscle tissue was flash-frozen in liquid N2,
whereas spleen and gill tissue were added to RNAse-free
1.5-mltubescontaining 500 llof TRIzolreagent
Fig. 1 Model of an MHC class II molecule displaying a short peptide
in the peptide-binding region
J Mol Evol (2008) 67:68–75 69
123

Page 3

(Invitrogen). Total RNA extractions for both spleen
(n = 34) and gill (n = 34) samples were conducted in a
sterilized, RNase-free laminar flow hood using the TRIzol
protocol. Total RNA pellets were resuspended in 75 ll of
0.01% DEPC ddH2O. Genomic DNA was isolated from
muscle tissue of each individual (n = 34) using standard
tissue digestion and phenol-chloroform protocols.
We previously characterized nine different Amti-DAB
alleles from this population using primers internal to the b1
domain (Bos and DeWoody 2005; Bos et al. 2008). The
same primers (MHC-B1F and MHC-B1R) were used to
amplify, from genomic DNA, the hypervariable peptide
binding region (PBR) of the Amti-DAB gene for the indi-
viduals in this study. These primers produce a 264-bp
amplicon that can be sequenced and used to genotype
individuals at this locus.
PCRs (n = 34) were performed in 25-ll reaction vol-
umes as follows: *50 ng genomic DNA template, 1.5 mM
MgCl2, 10 mM Tris-HCl, pH 8.9, 50 mM KCl, 0.5 mg/ml
BSA, a 0.45mM concentration of each dNTP, 0.24 lM
B1F, 0.24 lM B1R, 2.5 9 10-3U native Pfu DNA poly-
merase (Stratagene), and 1 U Taq DNA polymerase (New
England Biolabs). The thermal profile consisted of an ini-
tial denaturation step of 92?C for 2 min, followed by 34
cycles of 92?C for 30 s, 51?C for 30 s, 72?C for 30 s, with a
final extension step of 72?C for 10 min. All PCR reactions
were cleaned using QIAquick (Qiagen) columns and the
products were eluted in 30 ll Qiagen Buffer EB. PCR
products were bidirectionally sequenced using BigDye
v3.1 and an ABI3730xl. All sequences were analyzed using
Sequencher 4.7 software (GeneCodes). We used PHASE
v2.0 to resolve sequences into phase-ordered alleles
(Stephens et al. 2001). This Bayesian approach is both
precise and accurate (Bos et al. 2007), in part because
allelic variation in this population has been characterized
previously (Bos and DeWoody 2005).
We measured relative expression of transcripts between
tissues within an individual and within tissues between
individuals. Total RNA was isolated from spleen and gill
samples using TRIzol (Invitrogen). Prior to cDNA syn-
thesis with a dT20 oligo and the SuperScript III First-
Strand Synthesis System (Invitrogen), residual DNA was
removed from the RNA templates using amplification
grade DNase I (Invitrogen). cDNA was synthesized using
the SuperScript III First-Strand Synthesis System (Invit-
rogen) with oligo dT20and quantified using a NanoDrop
spectrophotometer.
RT-PCR
There are two different transcripts from the Amti-DAB
gene: a full-length transcript and a splice variant with a
270-bp deletion (Fig. 2A) (Bos and DeWoody 2005). The
deletion removes 97% of the region coding for the b1
domain, which forms half of the PBR of the mature protein,
a region that binds and presents antigens to T cells. We
designed PCR primers (DAB 40, 50-TAG GAG GGT TCG
TGT GGA TG-30; and DAB 405, 50- TCG ACC GAG ACA
CTT TCA CTT-30) that flank the b1 domain and produce a
366-bp amplicon from the full-length transcript and a 96-
bp amplicon from the alternative transcripts (Fig. 2B). The
melting temperatures (Tm) of the two amplicons differ by
approximately 4?C, making this assay amenable to disso-
ciation curve analysis.
Fig. 2 (A) Alternative splice
variants (isoforms) of the Amti-
DAB gene in A. t. tigrinum. The
amino acids encoding the b1
domain are thought to be
requisite for the formation of a
functional peptide binding
region. LP, leader peptide;
CP/TM, connecting peptide and
transmembrane region; CY,
cytoplasmic tail (see Tournefier
et al. 1998). (B) Schematic of
the expression assay, including
the location of RT-PCR primers
(DAB40 and DAB405),
expected amplicon sizes, and
expected melting temperatures
70J Mol Evol (2008) 67:68–75
123

Page 4

RT-PCR was performed using 25-ll reaction volumes in
a Bio-Rad iQ5. Three identical replicates of each cDNA
sample (n = 34) were run in both gill and spleen tissue
using *500 ng cDNA template, 0.2 9 SYBR Green I, 10
nM fluorescein, 1.5 mM MgCl2, 10 mM Tris-HCl, pH 8.0,
50 mM KCl, pH 8.0, 0.5 mg/ml BSA, 0.05% DMSO,
0.025% glycerol, 0.72 lM DAB 40, 0.72 lM DAB 405, a
0.45 lM concentration of each dNTP, and 4.5 U Taq. The
thermal profile was initial denaturation at 92?C for 1 min,
followed by 39 cycles of 92?C for 30 s, 54?C for 30 s, nd
72?C for 30 s, with a final extension at 72?C for 5 min
followed by a melting curve profile of 175 cycles of 60?C
for 90 s (+0.2?C each cycle). The melt curve data were
analyzed using the Bio-Rad iQ5 v2.0 software and a
spreadsheet. The iQ5 software automatically generates
melting curves (Fig. 3) by plotting the negative derivative
of the relative fluorescent units over time (–d/dtRFUs). The
relative abundance of the two transcripts was determined
by calculating the peak height ratios of the two amplifi-
cation peaks associated with the regular and alternative
transcripts (i.e., P1/[P1 + P2] and P2/[P1 + P2]) accord-
ing to Busi and Cresteil (2005). Hereafter, we refer to
relative transcript ratios (RTR), the quantity of the alter-
native transcript relative to the full-length transcript.
Statistical Analyses
We used the Wilcoxon signed-rank test for paired samples
to test for differences between the RTRs for spleen tissue
samples (n = 34) and gill tissue samples (n = 34) within
individuals. However, our primary interest was to deter-
mine whether or not the expression of the alternative
(truncated) transcript varied according to a given Amti-
DAB allele and/or zygosity at this locus. For the effect of
zygosity on RTR, we tested this possibility by conducting
the two-sample Mann-Whitney test with samples divided
into two groups: homozygotes and heterozygotes. The
entire sample was ranked according to RTR, regardless of
grouping, and the test statistic, T, was calculated from
rank-ordered data. The conversion of data to ordered ranks
avoids distributional anomalies, and because of sufficient
sample numbers in both groups, significance can be
approximated using the normal distribution. The test was
performed separately for gill and spleen tissue.
To test for an association between a given allele and the
RTRs, we considered each of the seven Amti-DAB alleles
separately in linear regression analyses. Individuals were
coded as to whether they contained zero, one (heterozyg-
otes), or two (homozygotes) copies of an allele. For
example, an AB heterozygote would receive a score of 1
for the A allele, 1 for the B allele, and 0 for all other alleles,
whereas an AA homozygote would score 2 for the A allele
and 0 for all others. Single-nucleotide polymorphisms
(SNPs) that defined alleles were partitioned into 12 hap-
lotype blocks and similarly analyzed using regression to
test for local effects.
Results
MHC Genotyping (Genomic DNA)
We genotyped all 34 salamanders at the Amti-DAB locus
and discovered seven alleles corresponding to DAB-01,
Fig. 3 Example melt curve
from which RTRs were
calculated as described in the
text
J Mol Evol (2008) 67:68–7571
123

Page 5

DAB-02, DAB-03, DAB-04, DAB-06, DAB-07, and DAB-
09 of Bos et al. (2008). Eleven of the 34 larvae were
homozygous, giving an observed heterozygosity of 68%
(Table 1). Among heterozygotes, alleles differed from one
another by an average of 18.4 nucleotide substitutions. The
historical signature of natural selection is strong at this
locus, as measured by the ratio of nonsynonymous-to-
synonymous substitutions and by selection intensities (Bos
and DeWoody 2005; Bos et al. 2008).
Relative Transcript Ratios (cDNA)
Both Amti-DAB transcripts, full-length and truncated, are
expressed in larval gill and spleen tissues; we detected no
other Amti-DAB transcripts. Recently, Cocqueta et al.
(2006) cautioned that alternative transcripts can be gener-
ated in vitro as opposed to in vivo. Artificial transcripts can
be generated during cDNA preparation by template
switching of the reverse transcriptase enzyme, a phenom-
enon most often observed in splice variants with
noncanonical splice sites. We are confident this is not the
case in our assay because repeated cDNA preparations
from the same individuals resulted in very similar RTRs
(data not shown). Beyond the expression data, we also have
DNA sequences that show no signs of template switching
(Bos and DeWoody 2005; unpublished data).
The Amti-DAB RTRs are reported in Table 1. The RTRs
observed in the gill tissue samples were consistently higher
than in spleen (Wilcoxon signed-rank test, Z = -4.345,
p\0.0001), meaning that on average there was a greater
proportion of alternative transcripts in gill tissue so we
analyzed gill and spleen data sets independently. When we
tested for possible Amti-DAB allelic effects on RTRs in gill
tissue, regression analysis revealed only one significant
relationship (DAB-01, r2= 0.16, p = 0.01). There were no
significant relationships between spleen RTRs and Amti-
DAB alleles (r2= 0.08, p = 0.12). Thus, these data provide
little evidence that a given Amti-DAB allele influences the
relative ratio of truncated to full-length transcripts.
Pearson correlation coefficients revealed no correlation
between RTRs and genetic distances between alleles
(r = 0.09 for gill, r = 0.17 for spleen). Likewise, there
were no significant relationships between spleen or gill
RTRs and Amti-DAB haplotype blocks. Thus, these data
provide little evidence that a given Amti-DAB allele or SNP
influences AS. The Mann-Whitney tests revealed no asso-
ciation between zygosity and mean RTR for either gill
(Z = -1.03, p = 0.30) or spleen (Z = -1.394, p = 0.16).
All analyses reported herein rely on the entire RTR data
set; reanalyses of a culled data set without anomalous
replicates revealed exactly the same patterns.
Discussion
Several important points emerge from our data. First and
foremost, we determined that alternative (short) Amti-DAB
transcripts are expressed in all A. tigrinum sampled. We
now know that the short transcript is (constitutively?)
expressed in adult spleen, larval spleen, and larval gill
tissue. The expression of Amti-DAB is consistent with the
results of Laurens et al. (2001), but unlike our tiger sala-
manders, their axolotl salamanders (A. mexicanum) were
Table 1 Relative transcript ratios at the Amti-DAB gene (showing
three RT-PCR replicates from both gill and spleen) and genotype
based on DNA sequencing
SampleNo.GillSpleen
Amti-DAB
genotype
R1 R2 R3R1 R2 R3
6574 0.210.220.290.160.19 0.2501/01
65750.230.230.240.150.150.1701/02
65760.16 0.19 0.190.150.16–02/05
65770.090.120.150.060.100.1305/05
65780.210.210.220.140.150.1501/02
65790.360.470.570.190.210.2704/08
6580 0.120.130.16 0.100.11 0.1201/04
65810.11 0.32 0.350.130.140.1601/05
65820.140.160.17 0.170.17 0.2101/05
65830.290.320.330.190.210.2206/08
65840.680.70 0.730.10 0.110.1105/08
6585 0.42 0.440.49 0.060.11 0.1405/08
65860.280.29 0.300.180.180.1802/06
65870.150.160.16 0.570.610.7405/05
6588 0.200.200.210.150.150.1701/08
65890.180.200.200.140.150.1501/01
65900.200.200.210.160.16–01/08
65910.100.170.170.100.100.1108/08
65920.260.290.290.130.130.1401/02
65930.360.370.390.120.120.1305/08
65940.250.250.260.200.230.2603/04
65950.480.480.740.100.110.1308/08
65960.210.210.240.190.230.2502/04
65970.150.150.180.110.110.1202/08
65980.310.340.350.110.120.1208/08
65990.170.170.170.100.110.1505/08
66000.22 0.220.260.100.11 0.1301/05
66010.41 0.41 0.440.080.110.1208/08
66020.270.280.280.120.140.1601/05
66030.230.310.320.100.100.1105/08
66040.140.170.18 0.120.130.1601/05
66050.120.150.160.110.110.1308/08
66060.230.240.250.130.140.1401/01
66070.190.190.230.090.110.1102/02
Grand mean0.260.15 n/a
72 J Mol Evol (2008) 67:68–75
123