Differences in Selection Drive Olfactory Receptor Genes in

Different Directions in Dogs and Wolf

Rui Chen,1,2 David M. Irwin,1,3 and Ya-Ping Zhang1,4,*
1

State Key Laboratory of Genetic Resources and Evolution, and Yunnan Laboratory of Molecular Biology of Domestic Animals,
Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming, China
2
School of Life Science, University of Science and Technology of China, Hefei, China
3
Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Canada
4
Laboratory for Conservation and Utilization of Bio-resource, Yunnan University, Kunming, China
*Corresponding author: E-mail: zhangyp@mail.kiz.ac.cn.
Associate Editor: John H. McDonald

The olfactory receptor (OR) gene family is the largest gene family found in mammalian genomes. It is known to evolve through a
birth-and-death process. Here, we characterized the sequences of 16 segregating OR pseudogenes in the samples of the wolf and
the Chinese village dog (CVD) and compared them with the sequences from dogs of different breeds. Our results show that the
segregating OR pseudogenes in breed dogs are under strong purifying selection, while evolving neutrally in the CVD, and show a
more complicated pattern in the wolf. In the wolf, we found a trend to remove deleterious polymorphisms and accumulate
nondeleterious polymorphisms. On the basis of protein structure of the ORs, we found that the distribution of different types of
polymorphisms (synonymous, nonsynonymous, tolerated, and untolerated) varied greatly between the wolf and the breed dogs.
In summary, our results suggest that different forms of selection have acted on the segregating OR pseudogenes in the CVD since
domestication, breed dogs after breed formation, and ancestral wolf population, which has driven the evolution of these genes in
different directions.
Key words: dog, wolf, polymorphism, purifying selection, tolerated, untolerated.

Introduction
Olfaction plays a large role in mammals, helping these animals
discriminate noxious foods, mates, prey, marking territories,
etc. (Firestein 2001; Mombaerts 2004). Olfaction is initiated by
the binding of ligands to an olfactory receptor (OR) in the
nasal cavity, with the OR gene family being the largest gene
family in mammals (Glusman et al. 2001). Although mammals typically have a large number of OR genes, the number
of intact genes varies considerably between species (Hayden
et al. 2010). The previous studies have identified OR pseudogenes, and the generation of these genes have been correlated
with evolutionary changes in specific groups. For example, in
primates, the rapid and specific loss of OR genes (Gilad et al.
2003) is associated with the acquisition of trichromatic vision
(Gilad et al. 2004). The fraction of the OR gene complements
that are pseudogenes is determined by selective constraints
(Rouquier et al. 2000), and these constraints have been found
to vary between human populations (Gilad and Lancet 2003).
Unlike Drosophila, where specific OR genes correspond to
specific olfactory neurons (Ray et al. 2007), in mammals,
the expression of OR genes is stochastically determined in a
specific olfactory neuron, and an alternate OR gene will be
activated if the first gene selected is a pseudogene (Shykind
2005). It is also known that specific amino acid residues in the
OR sequences are required for guiding axons to the Glomeruli
in the olfactory bulb (Feinstein and Mombaerts 2004).

The dog diverged from the wolf more than 10,000 years
ago (Vila et al. 1997; Savolainen et al. 2002; Pang et al. 2009).
Mitochondrial DNA studies indicate that the dog was domesticated from several hundred wolves in southern China (Pang
et al. 2009). During dog evolution, two bottlenecks have been
described; the first accompanied the domestication process,
while the second occurred during the recent formation of the
multiple modern breeds (Lindblad-Toh et al. 2005). The village dogs, including the Chinese village dog (CVD), are
thought to represent an ancestral state of dog before breed
formation and thus would not have experienced the second
bottleneck associated with breed formation (Boyko et al.
2009; Boyko et al. 2010). Thus, analysis of village dogs
should yield insight into the ancestral populations of dogs
and, in comparison with modern breed dogs, the process of
breed formation. After domestication, the dog has experienced both artificial and natural selections, where the selection on some genes has been relaxed (Bjornerfeldt et al.
2006), whereas for others, it has been strengthened (Sutter
et al. 2007). During breed formation, selection on some genes
has been greatly strengthened by artificial selection (Vila et al.
1997; Vila et al. 2005; Mosher et al. 2007), and extensive linkage disequilibrium has been found in dog breeds (Sutter et al.
2004). As there have been two major bottlenecks in dog
evolutionary history, most of the phenotypes of breed dogs
can be explained by a small number of quantitative trait loci

The Author 2012. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: journals.permissions@oup.com

Mol. Biol. Evol. 29(11):34753484 doi:10.1093/molbev/mss153 Advance Access publication June 7, 2012

3475

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012
Research
article

Abstract

MBE

Chen et al. . doi:10.1093/molbev/mss153

Materials and Methods

DNA Samples and PCR Amplification of Segregating
OR Pseudogenes
Blood from six wolves (three from Heilongjiang province,
China, two from Tver province, Russia, and one from
Krasnoyarskiy Krai, Russia) and six CVDs were collected,
and genomic DNA was extracted using a phenolchloroform
method. The primers (available on request) and conditions
for the polymerase chain reaction (PCR) amplification are
3476

from a previous study (Robin et al. 2009). PCR products

were purified using the Watson PCR Purification Kit
(Watson BioTechnologies, Shanghai). PCR products were analyzed on an ABI PRISM 3730 DNA Analyzer (Applied
Biosystems), and DNA sequences were edited using
DNASTAR 5.0 (DNASTAR).

Data Acquisition
We focused on 16 segregating OR pseudogenes that were
previously identified from a sample of 109 different OR
genes in dogs of diverse breeds (the segregating OR pseudogenes are listed in table 8 of the study by Robin et al. [2009])
(supplementary table S1, Supplementary Material online). OR
gene nomenclatures are from the study by Quignon et al.
(2005). OR gene sequences from the wolves and CVD were
obtained in our study, with haplotypes acquired through the
use of PHASE (Scheet and Stephens 2006). The reference sequences were from the database of a previous study (http://
dogs.genouest.org/ORrepertoire.html) (Quignon et al. 2005);
haplotypes and SNP frequencies for 16 segregating OR pseudogenes in dogs of different breeds were obtained from a
previous study (Robin et al. 2009), and the sequences for
breed dogs were recovered with a program written in C
language.

Data Analysis
On the basis of the phylogenetic tree of OR genes (Quignon
et al. 2005), we identified and obtained the sequences of the
29 most closely related genes of each of the segregating OR
genes amplified in this study (supplementary table S2,
Supplementary Material online). The 30 OR sequences, segregating pseudogene and 29 close relatives, were then aligned
with MEGA5 (Tamura et al. 2011). Alignments were then
analyzed with SIFT (http://sift.jcvi.org) (Ng and Henikoff
2001) to predict whether the observed amino acid changes
in the OR protein sequences are likely to be tolerated.
Insertions and deletions were not considered. Due to the
poor alignment of the N-terminal and C-terminal regions,
we only considered sequences between amino acids 32 and
300, in accord with the previous study (Menashe et al. 2006).
The JC69 model was used to count the numbers of synonymous and nonsynonymous sites and tolerated and untolerated sites. (Jukes and Cantor 1969). If all the nonsynonymous
polymorphisms between alleles at a specific position were
tolerated, then we defined this polymorphism as a tolerated
polymorphism. If at least one of the nonsynonymous polymorphisms at a position was found to be untolerated (damaging), then this is an untolerated polymorphism. The
nucleotide diversity and numbers of synonymous and nonsynonymous sites were calculated with DnaSP 5.1 (Librado
and Rozas 2009), and the raw data produced from SIFT were
used to count the numbers of tolerated and untolerated sites.
The Fishers exact test was applied to determine whether the
polymorphism ratios were significantly different from the site
ratios. We divided the dataset into a wolf population, a CVD
population, and the dog breed (dogs of different breeds) population. If a polymorphism existed in only one population,

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

(Lindblad-Toh et al. 2005; Gray et al. 2009; Boyko et al. 2010).

The dog is an olfactory sensitive animal and has an OR gene
repertoire of more than 1,000 genes; however, about 20% of
these OR genes are pseudogenes (Quignon et al. 2005). The
fraction of OR genes that are pseudogenes in the dog is much
lower than the 60% found in humans (Glusman et al. 2001)
and is similar to the 20% found in the rats (Quignon et al.
2005). The small fraction of pseudogenes implies that the OR
genes of dog have been under fairly intense selection pressure
through most of their evolutionary history (Gilad et al. 2003,
2004). Whether degeneration of olfaction has occurred since
domestication is a fascinating question.
Although pseudogenes are generally considered to be
under neutral selection (Balakirev and Ayala 2003; Podlaha
and Zhang 2010), it has been found that some pseudogenes
play a regulatory role for their paralogs (Balakirev and Ayala
2003). The benefit or harmfulness of a gene can change depending on environment (Zhang 2008), and gene function
can be lost or regained at different times (Bekpen et al. 2009).
Segregating pseudogenes, genes that are intact in some individuals but pseudogenes in others within a population, were
used by Gilad and Lancet (2003) to study the selection acting
in different human populations. Segregating pseudogenes
have a very recent origin (Walsh 1995), thus are good candidates to study the loss and gain of gene function. Investigation
into OR genes has found that different domains of the sequence are under different selection pressures (Emes et al.
2004). Studies in the Pea Aphid indicated that most of the
positively selected sites in OR genes are in the transmembrane
(TM) domain (Smadja et al. 2009). OR ligand-binding sites are
located in the extracellular regions (ECRs) between TM domains 3 and 6 (Vaidehi et al. 2002), and these residues are
hypervariable and have variable Ka:Ks ratios in rodents (Emes
et al. 2004). Differences in the selection pressure acting on
different domains of a protein make it possible to determine
whether a gene is under a neutral selection. Previous studies
have shown that some amino acid substitution in OR gene
sequences are deleterious, and the encoded protein does not
function despite appearing to be an intact gene (Menashe
et al. 2006), thus it is necessary to discriminate nondeleterious
polymorphisms from deleterious polymorphisms. In this
study, we used SIFT (Ng and Henikoff 2001) to predict
whether mutations in an OR gene are untolerated, thus
impair function, or tolerated and compatible with function.
We then analyzed the dynamics of the polymorphisms of the
segregating OR genes among wolves, CVDs, and dogs of different breeds.

MBE

Selection in OR Genes in Dogs and Wolf . doi:10.1093/molbev/mss153

Results
Differences in the Levels of Gene Diversity between
the Wolf and Dogs
To examine gene diversity in dogs and wolves, we investigated
16 segregating OR pseudogenes that had been identified in a
previous study (Robin et al. 2009). Gene sequences were obtained from all CVDs and wolves for each of the segregating
OR pseudogenes except OR gene CfOR04c05 in two of the
CVDs. Combining our new sequences with those from dog of
different breeds, which were from a previous study (Robin
et al. 2009), a total of 118 nonsynonymous and 85 synonymous polymorphisms and 14 indels (insertions and deletions)
were found. Of the polymorphisms and indels identified, 65
polymorphisms and 1 indels were newly identified in the wolf
and the CVD sequences (supplementary tables S3 and S4,
Supplementary Material online). Of the 9 nonsense

Table 1. Pseudogene Frequency in Wolf, CVD, and Dog Breeds for

Segregating OR Pseudogenes.
OR Name
CfOR0004
CfOR0043
CfOR0135
CfOR0180
CfOR0401
CfOR0438
CfOR04C05
CfOR0519
CfOR0565
CfOR0821
CfOR08G02
CfOR12F06
CfOR14A11
CfOR16C11
CfOR3109
CfOR5912

Wolf
0.58
1.00
0.00
0.08
0.00
0.08
0.17
1.00
0.08
0.00
0.17
1.00
0.08
0.00
0.75
0.08

CVD
0.42
0.25
0.00
0.17
0.00
0.50
0.33
1.00
0.00
0.25
0.08
0.50
0.00
0.00
0.92
0.17

BM
0.63
0.88
0.00
0.06
0.00
1.00
0.00
1.00
0.06
0.00
0.38
0.50
0.44
0.19
1.00
0.19

ESS
0.56
0.81
0.06
0.06
0.06
0.88
0.00
1.00
0.06
0.00
0.13
0.38
0.25
0.44
0.63
0.00

GREY
0.88
1.00
0.00
0.00
0.00
0.81
0.00
0.94
0.00
0.00
0.19
0.25
0.50
0.25
0.56
0.19

GSD
0.88
0.81
0.00
0.00
0.00
0.63
0.00
1.00
0.00
0.00
0.81
0.88
0.06
0.13
1.00
0.25

LR
0.56
0.94
0.00
0.00
0.00
0.56
0.00
1.00
0.00
0.00
0.44
0.75
0.91
0.19
0.88
0.31

PEK
0.56
0.44
0.00
0.00
0.00
0.94
0.13
1.00
0.00
0.13
0.06
0.38
0.25
0.00
1.00
0.06

NOTE.Data for dog breed are from the haplotypes from a previous study (Robin
et al. 2009). BM, Belgian Malinois; ESS, English Springer Spaniel; GREY, Greyhound;
GSD, German Shepherd Dog; LR, Labrador Retriever; PEK, Pekingese.

mutations and 13 indels identified in the previous study

(Robin et al. 2009), 5 of the indels and 2 of the nonsense
mutations were not found in either the wolves or the
CVDs. The new indel that we found was a deletion in a
CVD, and two of the new polymorphisms were nonsense
mutations found in wolves (supplementary table S3,
Supplementary Material online). Base C at position 702 of
CfOR5912, from the OR gene repertoire (Quignon et al.
2005), is absent in all our new sequences. This base is present
in the 1.5x Poodle genome (Kirkness et al. 2003) but absent in
the 7.5x boxer genome (Lindblad-Toh et al. 2005) raising the
possibility that this is a sequencing error and that this gene
should be 900 bp long. Due to missing data after base 872, we
have included only 872 bases of this gene in our analysis.
We also calculated the proportion of the alleles at the
16 segregating OR genes that were pseudogenes in the
wolves, CVD, and 6 different breeds of dogs (table 1).
The proportion of the alleles that were pseudogenes
ranged from 0% to 100% for the different subpopulation,
with average pseudogenes proportions ranging from 28.7%
for the genes in CVD to 40.9% in the Labrador retriever.
However, the differences in pseudogene proportions was
not statistically significant different among the eight groups
(Friedman test P = 0.629).
The gene CfOR04c05 failed to produce an intact sequence
from two individuals of CVD. In these two individuals, a PCR
product of the expected size was generated for CfOR04c05,
but the inner sequencing primer failed to generate a clear
sequence due to the deletion of one base in one of the alleles
in each individual. To calculate nucleotide diversity, we combined all gene sequences, except CfOR04c05. Wolves showed
higher nucleotide diversity (0.00248) than CVD, who had intermediate nucleotide diversity (0.00228), whereas breed dogs
3477

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

then it was defined as a population-specific polymorphism,

and thus, it was likely acquired after the foundation of that
population. If a polymorphism exists in two populations, then
it was defined as a polymorphism that was lost in the third
population. Polymorphisms that were shared by more than
one population were defined as ancestral polymorphisms.
The topology of the ORs proteins was predicted using
HMMTOP (Tusnady and Simon 2001). The numbers of
each type of polymorphism in the TM domains 17 (TM1
7), intracellular regions (ICRs), and ECRs were counted. The
distributions of each type of polymorphism were defined as
the number of that type in TM17, ICRs, and ECRs. As most
polymorphisms in the NH2 and COOH tails are removed
from the tolerated and untolerated polymorphisms analysis,
we did not consider polymorphisms in the two tails during
the polymorphism distribution analysis. We analyzed the distributions of polymorphisms using two approaches. First,
polymorphisms were divided into four groups: ancestral polymorphisms, wolf-specific polymorphisms, CVD-specific polymorphisms, and breed dog-specific polymorphisms. In each
group, the distributions of synonymous polymorphisms were
compared with the distributions of the other types (nonsynonymous, tolerated, and untolerated polymorphisms).
Second, the numbers of synonymous, nonsynonymous, tolerated, and untolerated polymorphisms and sites were
counted in each of the three domains (TM17, ECRs, and
ICRs) of the protein, and then the distributions of the numbers of polymorphisms of each type in the four groups (ancestral, wolf-specific, CVD-specific, and breed dog-specific
polymorphisms) were compared with the distribution of
the sites of that type. A chi-square test with 2 degrees of
freedom was used to determine whether the distribution of
synonymous, nonsynonymous, tolerated, and untolerated
polymorphisms across groups is significantly different from
distribution of sites or synonymous polymorphisms.
Tajimas D (Tajima 1989), Fu and Lis D*, Fu and Lis F*, Fu
and Lis D, and Fu and Lis F tests (Fu and Li 1993; Fu 1997)
were used to test whether the genes in the different populations were evolving neutrally using DnaSP 5.1 (Librado and
Rozas 2009).

MBE

Chen et al. . doi:10.1093/molbev/mss153

Table 2. Test Statistics from the Neutrality Analysis of All Segregating OR Pseudogenes Except CfOR04c05.
All samples
Wolf
CVD
Breeds
BM
ESS
GREY
GSD
LR
PEK

Tajimas D
0.40907
0.43860
0.15074
0.72885
0.35535
0.29364
0.76230
0.51345
0.45971
0.46256

Nucleotide Diversity
0.00232
0.00248
0.00228
0.00201
0.00170
0.00198
0.00150
0.00160
0.00181
0.00170

Fu and Lis D*
0.95052
0.46486
0.20661
0.58063
0.12812
0.06332
0.20054
0.20012
0.53464
0.06292

Fu and Lis F*
0.83822
0.52246
0.21890
0.77230
0.22285
0.04436
0.41620
0.33429
0.59341
0.10008

Fu and Lis D
0.46892
0.15941
0.72911
0.29702
0.11361
0.76024
0.24500
0.49581
0.00942

Fu and Lis F
0.57499
0.19310
0.56698
0.23305
0.03280
0.97887
0.36662
0.57840
0.16172

showed the lowest diversity (0.00201) (table 2). These results

are in accord with dogs representing a bottlenecked subpopulation of wolves, with breed dogs being further bottlenecked.
These results are consistent with the CVD being more representative of an ancestral dog population before breed formation (Brown et al. 2011). When the ratio of nonsynonymous
polymorphisms to synonymous polymorphisms was examined, it was found that wolves have the highest ratio, whereas
breed dogs have the lowest ratio (supplementary table S4,
Supplementary Material online).

Selection on the Different Types of Polymorphisms

To analyze whether the segregating OR genes are evolving
neutrally in the different groups (wolves, CVDs, and breed
dogs), we compared the ratios of nonsynonymous polymorphisms to synonymous polymorphisms with the ratio
of nonsynonymous sites to synonymous sites. The ratio of
nonsynonymous polymorphisms to synonymous polymorphisms is significantly lower than the ratio of nonsynonymous sites to synonymous sites for ancestral
polymorphisms (P < 0.01, Fishers exact test one tailed) indicating that the OR genes in the ancestral populations of
the three groups (wolves, CVD, and breed dogs) were not
evolving neutrally and likely were functional (fig. 1).
Population-specific polymorphisms of breed dogs also
have a ratio of nonsynonymous polymorphisms to synonymous polymorphisms significantly lower than the ratio of
nonsynonymous sites to synonymous sites (P < 0.01,
Fishers exact test one tailed), thus purifying selection continued to occur in the breed dogs since their divergence
from the village dogs.
Because some nonsynonymous mutations are not deleterious (tolerated), whereas some are deleterious (untolerated),
the ratio of nonsynonymous polymorphisms to synonymous
polymorphisms may not be powerful enough to detect positive selection, because at best only a small proportion of
tolerated mutations are potentially under positive selection.
Therefore, we divided the nonsynonymous polymorphisms
into tolerated and untolerated polymorphisms. As described
in the Materials and Methods, we analyzed amino acid residues 32 to 300 of the OR sequences, where a total of 102
3478

nonsynonymous polymorphisms were identified. Of the 102

polymorphisms, 40 generate untolerated amino acid substitutions, whereas 62 generate tolerated amino acid substitutions (supplementary table S5, Supplementary Material
online). A total of 4192 nonvariable amino acid sites (sites
that do not have nonsynonymous polymorphisms) were
identified from amino acid 32 to 300 (from amino acid 32
to 290 in CfOR5912) of the intact copies of the 16 segregating
pseudogenes. Of these 4192 nonvariable amino acid sites, 14
were identified as being untolerated amino acids by SIFT (Ng
and Henikoff 2001). If the intact copies of these segregating
pseudogenes are indeed functional, then SIFT has incorrectly
assessed the functional impact of these 14 residues. However,
this result also indicates that the likelihood that functional
residues being assessed as untolerated by SIFT is less than
0.33% (14/4192). Further strengthening our conclusion that
SIFT identified untolerated mutations that impair function
was the observation that the mutational direction of all the
untolerated substitutions was from a tolerated amino acid
ancestor to an untolerated amino acid derived sequence.
Next, we compared the pairs of ratios. First, we compared
the ratio of tolerated polymorphisms to synonymous polymorphisms with the ratio of tolerated sites to synonymous
sites, and second, we compared the ratio of untolerated polymorphisms to synonymous polymorphisms with the ratio of
untolerated sites to synonymous sites. For the ancestral polymorphisms, we found that ratios of tolerated polymorphisms
to synonymous polymorphisms (P < 0.01, Fishers exact test
one tailed) and untolerated polymorphisms to synonymous
polymorphisms (P < 0.01, Fishers exact test one tailed) are all
significantly lower than the ratios of their sites (fig. 1). For
polymorphisms that were specific to breed dogs, we also
found that ratios of tolerated polymorphisms to synonymous
polymorphisms (P < 0.01, Fishers exact test one tailed) and
untolerated polymorphisms to synonymous polymorphisms
(P < 0.01, Fishers exact test one tailed) were significantly
lower than the ratios of their sites (fig. 2). These observations
suggest that selection is acting strongly against both tolerated
and untolerated polymorphisms within the different breeds
of dog. In the specific polymorphisms of CVD and wolf, statistical tests were not significant for either the ratio of

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

NOTE.Summary statistics for Tajimas D, Fu and Lis D*, Fu and Lis F*, Fu and Lis D, and Fu and Lis F were calculated for the entire sample and the different populations (wolf,
CVD, and different breeds of dog). BM, Belgian Malinois; ESS, English Springer Spaniel; GREY, Greyhound; GSD, German Shepherd Dog; LR, Labrador Retriever; PEK, Pekingese.

Selection in OR Genes in Dogs and Wolf . doi:10.1093/molbev/mss153

MBE

tolerated polymorphisms to synonymous polymorphisms or

the ratio of untolerated polymorphisms to synonymous polymorphisms, when compared with the ratios of their sites.
However, for specific polymorphisms within wolves, the
number of observed tolerated polymorphisms (19) was
greater than expected (11.1), whereas the number of untolerated polymorphisms (8) was fewer than expected (14.6)
under the expectation that the number of these polymorphisms should be proportional to the numbers of these sites.
A similar observation is seen within CVD. These observations
suggest that within CVD and wolves, there was a greater
likelihood that tolerated changes were maintained, but that
selection was still acting strongly against untolerated changes.
The differences in the ratios of the different types of polymorphisms and their sites ratios in the three subpopulations
(wolves, CVD, and breed dogs) suggest that different selection
pressures are acting on the different subpopulations since
their divergences.
The protein encoded by CfOR0180 was predicted by
HMMTOP to have only six rather than the expected seven
TM domains; therefore, it was removed from our analysis of
the distributions of polymorphisms. The distributions of nonsynonymous, tolerated, and untolerated polymorphisms in
the ancestral polymorphisms were different from the distribution of synonymous polymorphisms, as fewer nonsynonymous, tolerated, and untolerated polymorphisms found in
TM17 and more being located in the ECRs and ICRs of the
protein. When the polymorphisms that were specific to the
subpopulations were examined, differences were observed. In
the wolf, the distribution of the nonsynonymous polymorphisms is different from that of the synonymous polymorphisms (fig. 3); however, in breed dogs, we found that the
distributions of all four types of polymorphisms were almost
the same. These observations suggest that selection against
nonsynonymous, tolerated, and untolerated polymorphisms
in TM17 has changed in breed dogs. Then the distributions
of the ancestral, wolf-specific, CVD-specific, and breed dogspecific polymorphisms were compared with the sites distributions of the four polymorphic types (synonymous, nonsynonymous, tolerated, and untolerated polymorphisms)
(fig. 4). Synonymous polymorphisms were found only in
TM17 of CVD. For synonymous polymorphisms in the

other three subpopulations (ancestral, wolf-specific, and

breed-specific polymorphisms), no statistically significant difference was found. In the distribution of the other three types
polymorphisms (nonsynonymous, tolerated, and untolerated
polymorphisms), statistically significant (P < 0.05) differences
were only found in the ancestral and wolf-specific polymorphisms, suggesting that selection has changed since
domestication.

Neutral Test on Segregating Pseudogenes in Different

Populations
To attempt to identify the type of selection that may have
occurred in the three subpopulations, we applied Tajimas D,
Fu and Lis D, and Fu and Lis F tests. However, these tests can
be influenced by population demographic dynamics in addition to selection. Negative values from these tests indicate an
excess of rare alleles, which may be caused by positive selection, negative selection, or population expansion. Positive
values indicate an excess of intermediate alleles, which may
be due to balancing selection or a population bottleneck.
Negative values for these tests were generated by the wolf
and CVD data, whereas the breed dogs yield positive values
(table 2). To distinguish selection from population dynamics,
we separated nonsynonymous polymorphisms from synonymous polymorphisms, and for the nonsynonymous polymorphisms, we separated them into tolerated and untolerated
polymorphisms (supplementary table S6, Supplementary
Material online). Synonymous polymorphisms should be neutral, so that they can be used to detect population history.
Negative values from these tests using synonymous polymorphisms for the wolf suggest that they are experiencing
a population expansion, whereas positive value for the breed
dogs suggests that they experienced a bottleneck. The test
statistics for the CVD is near zero, which suggests that the
CVD population size has remained nearly constant. Because
population dynamics do not influence the test statistics in the
CVD, in contrast to the wolf and breed dogs, we can then
attribute the positive values from Tajimas D, Fu and Lis D,
and Fu and Lis F tests for balancing selection acting in the
CVD population.
3479

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

FIG. 1. Polymorphisms in the segregating OR pseudogenes in the wolf, CVD, and breed dogs. Venn diagram shows the sharing of polymorphisms in the
segregating OR pseudogenes. (A) Pairs of numbers represent nonsynonymous/synonymous polymorphisms. (B) Pairs of numbers represent tolerated/untolerated polymorphisms.

Chen et al. . doi:10.1093/molbev/mss153

MBE

Discussion
Nucleotide Diversities of Segregating OR Pseudogenes
OR gene families adapt to the physical requirements of different species (Hayden et al. 2010), since olfaction is very
important for mammals (Szetei et al. 2003). The ecological
niche of dog has changed greatly since domestication from
the wolf (Morey 1994); thus, it is likely that changes in the
composition of the OR gene family have occurred between
wolf and dog. The dog only recently diverged from the wolf,
about 16,000 years ago (Pang et al. 2009), and there has been
some backcrossing since domestication (Savolainen et al.
2002; Vila et al. 2005), which might explain why no significant
difference in the OR pseudogene fraction between the wolf
and dog was found (Zhang et al. 2011). In this study, we
focused on 16 OR genes that had been previously shown to
be segregating pseudogenes in breed dogs (Robin et al. 2009).
We characterized these 16 genes in wolves, CVDs, and breed
dogs and examined distribution patterns of polymorphisms
to identify whether the selective forces acting on these sequences differ between the subpopulations.
3480

The nucleotide diversities of the segregating OR pseudogenes, except CfOR04c05, were found to range from 0.00201 in
the breed dogs to 0.00248 in wolves (table 2), values that are a
little higher than the nucleotide diversity (yp = 0.0018) calculated in the previous study of all 105 OR genes in breed dogs
(Robin et al. 2009). Because the expected y of genes based on
the entire dog genome is 0.0005 (based on y = 4Ne)
(Lindblad-Toh et al. 2005), there may be three reasons for
the elevated levels seen in the segregating OR pseudogenes.
First, the necessity to interact with an enormous diversity of
potential ligands may be the cause of the high nucleotide
diversity of OR genes (Hayden et al. 2010). Second, balancing
selection is known to be an important element in maintaining
high nucleotide diversity, and this type of selection might be
different between functional genes and segregating pseudogenes (Alonso et al. 2008). Third, the reduced effective population size of dog, especially breed dogs, may have led to
lower nucleotide diversity in those dogs (Lindblad-Toh et al.
2005).
The gray wolf has experienced population contraction and
expansion in the past 10,000 years (Lucchini et al. 2004), while

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

FIG. 2. Gain and loss of polymorphisms in the segregating OR pseudogenes during the evolutionary history of the wolf, CVD, and dog breeds. (A and C).
Numbers in the rectangular box are nonsynonymous to synonymous (nonsynonymous/synonymous) polymorphism ratios. Numbers above the line
are the numbers of polymorphisms gained, whereas the numbers below the line are the numbers of polymorphisms lost. (B and D). Numbers in the
rectangular box are ratios of tolerated to untolerated polymorphisms (tolerated/untolerated). Numbers above the line are the numbers of polymorphisms gained, whereas the numbers below the line are the numbers of polymorphisms lost. BM, Belgian Malinois; ESS, English Springer Spaniel; GREY,
Greyhound; GSD, German Shepherd Dog; LR, Labrador Retriever; PEK, Pekingese. Double or single asterisks indicate that the ratios of polymorphisms
(nonsynonymous, tolerated, or untolerated polymorphisms) to synonymous polymorphisms are significantly different (**P < 0.05) or show a trend
(*0.05 < P < 0.1) from the sites ratios (nonsynonymous, tolerated, or untolerated sites to synonymous sites).

Selection in OR Genes in Dogs and Wolf . doi:10.1093/molbev/mss153

MBE

at the same time, the domesticated dog has gone through

two bottlenecks due to domestication and breed formation
(Lindblad-Toh et al. 2005). Negative values for synonymous
polymorphisms are consistent with the expansion of the wolf
population, whereas the near zero values indicate that the
CVD population has remained nearly constant. Positive
values for synonymous polymorphisms demonstrate that
the breed dogs have experienced a bottleneck. Previous studies have shown that the bottleneck associated with breed
formation was much more intensive than the bottleneck
that occurred during domestication (Boyko et al. 2010);
thus, the much lower values for nucleotide diversity observed
within breeds in our data is in accord with this conclusion.
However, population demography alone cannot explain the
lower ratios of nonsynonymous polymorphisms to synonymous polymorphisms and suggests that selection continues
to act on the segregating pseudogenes.

Selection on the Segregating OR Pseudogenes

OR genes in humans are maintained by balancing selection
(Alonso et al. 2008). Higher values from the neutrality tests on
nonsynonymous polymorphisms indicate that the segregating OR pseudogenes may be under balancing selection; however, the values was not significant, hence, we cannot make a

strong conclusion. To refine our results, we divided the nonsynonymous polymorphisms into tolerated and untolerated
polymorphisms based on their predicted effects on protein
function. Untolerated substitutions are predicted to disrupt
function (Ng and Henikoff 2001), whereas tolerated substitutions are not incompatible with function. However, accuracy
of this analysis depends on the power of SIFT, and specific
characters of OR genes may influence the results of this analysis. Because ORs must interact with a huge number of potential ligands in nature, tolerated polymorphisms in the
extracellular domains may be beneficial for generating the
diversity of OR function (Alonso et al. 2008). Discrimination
of untolerated polymorphisms from tolerated polymorphisms in the nonsynonymous polymorphism analysis
should allow a more accurate detection of selection, as purifying selection on the deleterious polymorphisms should
reduce the ratio of nonsynonymous to synonymous polymorphisms. On the other hand, positive selection acting on beneficial polymorphisms would increase the nonsynonymous to
synonymous ratio. With this analysis, by using the ratios of
untolerated polymorphisms to synonymous polymorphisms
and tolerated polymorphisms to synonymous polymorphisms, we can distinguish purifying from positive selection.
In the wolf, selection is removing the untolerated polymorphisms and preserving the tolerated polymorphisms, whereas
3481

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

FIG. 3. Distribution of polymorphisms in three domains of OR proteins (TM17, ICRs, and ECRs). Distribution of the four polymorphism types
(synonymous polymorphisms [syn], nonsynonymous polymorphisms [nonsyn], tolerated polymorphisms [tole], untolerated polymorphisms [untole])
are shown for each domain. (A) Analysis of ancestral polymorphisms. (B) Analysis of specific polymorphisms of wolf. (C) Analysis of specific polymorphisms of CVD. (D) Analysis of specific polymorphisms of breed dogs. Significant differences in the patterns of distributions of sites to synonymous,
nonsynonymous, tolerated, or untolerated polymorphisms were determined by a chi-square test with 2 degrees of freedom. P value of the statistical
significance is shown above each significant type.

Chen et al. . doi:10.1093/molbev/mss153

MBE

in the breed dogs, purifying selection is acting on both the

untolerated and tolerated polymorphisms. Gene conversion
and recombination may help OR pseudogenes to become
functional genes (Sharon et al. 1999), and our findings demonstrate that selection may also remove pseudogenes to preserve the functional genes.
Selection acts differently on the different domains of chemosensory receptors and on the orthologous genes in different species (Shi and Zhang 2006; Smadja et al. 2009). We
analyzed the distribution patterns of polymorphisms across
the different domains of the OR protein for the segregating
OR pseudogenes and found there were significant differences
between the distribution patterns of the synonymous polymorphisms with polymorphisms of other types (nonsynonymous, tolerated, and untolerated) for the ancestral
polymorphisms. However, in the subpopulation-specific polymorphisms, within breed dogs, the patterns of distribution for
the three types of polymorphisms (nonsynonymous, tolerated, and untolerated) are almost the same as the pattern
of distribution for synonymous polymorphisms (fig. 3). In
contrast, in the wolf, the patterns of distribution for the
three types of polymorphisms differ from that for synonymous polymorphisms, with the wolf having a pattern more
similar to the pattern of distributions for the ancestral polymorphisms (fig. 3). When the distribution patterns of the
3482

polymorphisms are compared with the patterns for the distribution of the sites, our statistical analysis did not find any
significant differences for the CVD- or breed dog-specific
(fig. 4), nonsynonymous, tolerated, and untolerated polymorphisms. In contrast, the pattern of distributions for the
wolf-specific and ancestral nonsynonymous and untolerated
polymorphisms are statistically significant (P < 0.05) or have a
significant trend (0.05 < P < 0.1), that is, different from the
distribution patterns for the sites, suggesting that the selection acting on these sites has changed since domestication.
In my previous analysis, a low ratio of nonsynonymous
polymorphisms to synonymous polymorphisms in ancestral
polymorphisms and specific polymorphisms of breed dogs
was found. Our polymorphism distribution analysis suggests
that the reasons for the similar patterns for ancestral and wolf
polymorphisms compared with breed dogs are not identical.
The low ratio for the ancestral polymorphisms may be explained by accelerated fixation of beneficial mutations due to
positive selection and elimination of deleterious mutations by
purifying selection. It is not simply the result of purifying
selection. Accumulation of tolerated polymorphisms was
also enhanced in wolves, with the number of untolerated
polymorphisms reduced, in accord with this conclusion.
The low ratio observed in the specific polymorphisms of
breed dogs is likely the result of artificial selection. The history

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

FIG. 4. Distribution of sites and polymorphisms across three domains of OR proteins (TM17, ICRs, and ECRs). Distributions were analyzed in four
groups (ancestral polymorphisms [AP], specific polymorphisms of wolf [WP], specific polymorphisms of CVD [CP], specific polymorphisms of dog
breeds [BP]). (A) Synonymous (syn) sites and polymorphisms. (B) Nonsynonymous (nonsyn) sites and polymorphisms. (C) Tolerated (tole) sites and
polymorphisms. (D) Untolerated (untole) sites and polymorphisms. Significant differences in the patterns of distributions of sites to synonymous,
nonsynonymous, tolerated, or untolerated polymorphisms were determined by a chi-square test with 2 degrees of freedom. P value of the statistical
significance is shown above each significant type.

Selection in OR Genes in Dogs and Wolf . doi:10.1093/molbev/mss153

of most dog breeds is not more than 400 years (Parker et al.
2004). A great amount of linkage disequilibrium is found in
the breed dogs (Sutter et al. 2004). Artificial selection, a form
of purifying selection in our study, is not as complex as natural
selections (Boyko et al. 2010) and may not be powerful
enough to discriminate all tolerated and untolerated mutations. Our results suggest that selection on the OR genes has
changed a great deal since the divergence of the dog breeds,
whereas the analysis of the segregating OR pseudogenes in
wolves indicates that they are still undergoing selection that
favors functional genes.

Conclusion

Supplementary Material
Supplementary tables S1S6 and sequence information are
available at Molecular Biology and Evolution online (http://
www.mbe.oxfordjournals.org/).

Acknowledgments
The authors thank Guodong Wang for valuable advice and
Stephanie Robin and Francis Galibert for their kind help. They
also thank Dr John H. McDonald and anonymous reviewers
for their comments and suggestions. This work was supported by grants from the National Basic Research Program
of China (973 Program), Chinese Academy of Sciences, Bureau
of Science and Technology of Yunnan Province, and National
Natural Science Foundation of China.

References
Alonso S, Lopez S, Izagirre N, de la Rua C.. 2008. Overdominance in the
human genome and olfactory receptor activity. Mol Biol Evol. 25:
9971001.
Balakirev ES, Ayala FJ. 2003. Pseudogenes: are they "junk" or functional
DNA? Annu Rev Genet. 37:123151.
Bekpen C, Marques-Bonet T, Alkan C, Antonacci F, Leogrande MB,
Ventura M, Kidd JM, Siswara P, Howard JC, Eichler EE. 2009.
Death and resurrection of the human IRGM gene. PLoS Genet. 5:
e1000403.
Bjornerfeldt S, Webster MT, Vila C. 2006. Relaxation of selective constraint on dog mitochondrial DNA following domestication.
Genome Res. 16:990994.
Boyko AR, Boyko RH, Boyko CM, et al. (15 co-authors). 2009. Complex
population structure in African village dogs and its implications for

inferring dog domestication history. Proc Natl Acad Sci U S A. 106:

1390313908.
Boyko AR, Quignon P, Li L, et al. (24 co-authors). 2010. A simple genetic
architecture underlies morphological variation in dogs. PLoS Biol. 8:
e1000451.
Brown SK, Pedersen NC, Jafarishorijeh S, Bannasch DL, Ahrens KD, Wu
JT, Okon M, Sacks BN. 2011. Phylogenetic distinctiveness of Middle
Eastern and Southeast Asian village dog Y chromosomes illuminates
dog origins. PLoS One 6:e28496.
Emes RD, Beatson SA, Ponting CP, Goodstadt L. 2004. Evolution and
comparative genomics of odorant- and pheromone-associated
genes in rodents. Genome Res. 14:591602.
Feinstein P, Mombaerts P. 2004. A contextual model for axonal sorting
into glomeruli in the mouse olfactory system. Cell 117:817831.
Firestein S. 2001. How the olfactory system makes sense of scents.
Nature 413:211218.
Fu YX. 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:
915925.
Fu YX, Li WH. 1993. Statistical tests of neutrality of mutations. Genetics
133:693709.
Gilad Y, Lancet D.. 2003. Population differences in the human functional
olfactory repertoire. Mol Biol Evol. 20:307314.
Gilad Y, Man O, Paabo S, Lancet D. 2003. Human specific loss of olfactory
receptor genes. Proc Natl Acad Sci U S A. 100:33243327.
Gilad Y, Przeworski M, Lancet D. 2004. Loss of olfactory receptor genes
coincides with the acquisition of full trichromatic vision in primates.
PLoS Biol. 2:E5.
Glusman G, Yanai I, Rubin I, Lancet D. 2001. The complete human
olfactory subgenome. Genome Res. 11:685702.
Gray MM, Granka JM, Bustamante CD, Sutter NB, Boyko AR, Zhu L,
Ostrander EA, Wayne RK. 2009. Linkage disequilibrium and demographic history of wild and domestic canids. Genetics 181:
14931505.
Hayden S, Bekaert M, Crider TA, Mariani S, Murphy WJ, Teeling EC.
2010. Ecological adaptation determines functional mammalian
olfactory subgenomes. Genome Res. 20:19.
Jukes TH, Cantor CR. 1969. Evolution of protein molecules. In: Munro
HN, editor. Mammalian protein metabolism. New York: Academy
Press. p. 21123.
Kirkness EF, Bafna V, Halpern AL, et al. (11 co-authors). 2003. The dog
genome: survey sequencing and comparative analysis. Science 301:
18981903.
Librado P, Rozas J. 2009. DnaSP v5: a software for comprehensive analysis
of DNA polymorphism data. Bioinformatics 25:14511452.
Lindblad-Toh K, Wade CM, Mikkelsen TS, et al. (234 co-authors). 2005.
Genome sequence, comparative analysis and haplotype structure of
the domestic dog. Nature 438:803819.
Lucchini V, Galov A, Randi E. 2004. Evidence of genetic distinction and
long-term population decline in wolves (Canis lupus) in the Italian
Apennines. Mol Ecol. 13:523536.
Menashe I, Aloni R, Lancet D. 2006. A probabilistic classifier for olfactory
receptor pseudogenes. BMC Bioinformatics 7:393.
Mombaerts P. 2004. Genes and ligands for odorant, vomeronasal and
taste receptors. Nat Rev Neurosci. 5:263278.
Morey DF. 1994. The early evolution of the domestic dog. Am Sci. 82:
336347.
Mosher DS, Quignon P, Bustamante CD, Sutter NB, Mellersh CS, Parker
HG, Ostrander EA. 2007. A mutation in the myostatin gene

3483

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

Segregating pseudogenes exist during a subtle period of time,

where they could become a functional or pseudogene within
a population. In our investigation, we found that in natural
populations, segregating OR pseudogene and pseudoalleles
showed a trend to be eliminated in competition with functional alleles, whereas within domesticated dog, this trend has
changed. In the CVD, these genes are evolving neutrally,
whereas in breed dogs, they are evolving under artificial purifying selection. Genes can be beneficial in one period of time
and then become deleterious in another period (Olson 1999;
Zhang 2008). Our analysis demonstrates the selection acting
on segregating OR pseudogenes has changed during the domestication of dogs, and thus, these genes are evolving in
different ways among wolves, CVD, and breed dogs.

MBE

Chen et al. . doi:10.1093/molbev/mss153

3484

Shykind BM. 2005. Regulation of odorant receptors: one allele at a time.

Hum Mol Genet. 14:R33R39.
Smadja C, Shi P, Butlin RK, Robertson HM. 2009. Large gene family
expansions and adaptive evolution for odorant and gustatory receptors in the pea aphid. Acyrthosiphon pisum. Mol Biol Evol. 26:
20732086.
Sutter NB, Bustamante CD, Chase K, et al. (21 co-authors). 2007. A single
IGF1 allele is a major determinant of small size in dogs. Science 316:
112115.
Sutter NB, Eberle MA, Parker HG, Pullar BJ, Kirkness EF, Kruglyak L,
Ostrander EA. 2004. Extensive and breed-specific linkage disequilibrium in Canis familiaris. Genome Res. 14:23882396.
Szetei V, Miklosi A, Topal J, Csanyi V. 2003. When dogs seem to lose their
nose: an investigation on the use of visual and olfactory cues in
communicative context between dog and owner. Appl Anim Behav
Sci. 83:141152.
Tajima F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585595.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S.. 2011.
MEGA5: molecular evolutionary genetics analysis using maximum
likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol. 28:27312739.
Tusnady GE, Simon I. 2001. The HMMTOP transmembrane topology
prediction server. Bioinformatics 17:849850.
Vaidehi N, Floriano WB, Trabanino R, Hall SE, Freddolino P, Choi EJ,
Zamanakos G, Goddard WA III. 2002. Prediction of structure and
function of G protein-coupled receptors. Proc Natl Acad Sci U S A.
99:1262212627.
Vila C, Savolainen P, Maldonado JE, Amorim IR, Rice JE, Honeycutt RL,
Crandall KA, Lundeberg J, Wayne RK. 1997. Multiple and ancient
origins of the domestic dog. Science 276:16871689.
Vila C, Seddon J, Ellegren H. 2005. Genes of domestic mammals augmented by backcrossing with wild ancestors. Trends Genet. 21:
214218.
Walsh JB. 1995. How often do duplicated genes evolve new functions?
Genetics 139:421428.
Zhang H-h, Wei Q-g, Zhang H-x, Chen L. 2011. Comparison of the
fraction of olfactory receptor pseudogenes in wolf (Canis lupus)
with domestic dog (Canis familiaris). J Forest Res. 22:275280.
Zhang J. 2008. Positive selection, not negative selection, in the pseudogenization of rcsA in Yersinia pestis. Proc Natl Acad Sci U S A. 105:
E69.

Downloaded from http://mbe.oxfordjournals.org/ at Indian Institute Of Chemical Biology (Iicb) on November 2, 2012

increases muscle mass and enhances racing performance in heterozygote dogs. PLoS Genet. 3:e79.
Ng PC, Henikoff S. 2001. Predicting deleterious amino acid substitutions.
Genome Res. 11:863874.
Olson MV. 1999. When less is more: gene loss as an engine of evolutionary change. Am J Hum Genet. 64:1823.
Pang JF, Kluetsch C, Zou XJ, et al. (14 co-authors). 2009. mtDNA data
indicate a single origin for dogs south of Yangtze River, less than
16,300 years ago, from numerous wolves. Mol Biol Evol. 26:
28492864.
Parker HG, Kim LV, Sutter NB, Carlson S, Lorentzen TD, Malek TB,
Johnson GS, DeFrance HB, Ostrander EA, Kruglyak L. 2004.
Genetic structure of the purebred domestic dog. Science 304:
11601164.
Podlaha O, Zhang J. 2010. Pseudogenes and their evolution. In:
Encyclopedia of life sciences (ELS). Chichester (United Kingdom):
John Wiley & Sons, Ltd.
Quignon P, Giraud M, Rimbault M, et al. (11 co-authors). 2005. The dog
and rat olfactory receptor repertoires. Genome Biol. 6:R83.
Ray A, van Naters WG, Shiraiwa T, Carlson JR. 2007. Mechanisms of odor
receptor gene choice in Drosophila. Neuron 53:353369.
Robin S, Tacher S, Rimbault M, Vaysse A, Dreano S, Andre C, Hitte C,
Galibert F. 2009. Genetic diversity of canine olfactory receptors. BMC
Genomics 10:21.
Rouquier S, Blancher A, Giorgi D. 2000. The olfactory receptor gene
repertoire in primates and mouse: evidence for reduction of the
functional fraction in primates. Proc Natl Acad Sci U S A. 97:
28702874.
Savolainen P, Zhang YP, Luo J, Lundeberg J, Leitner T. 2002. Genetic
evidence for an East Asian origin of domestic dogs. Science 298:
16101613.
Scheet P, Stephens M. 2006. A fast and flexible statistical model for
large-scale population genotype data: applications to inferring
missing genotypes and haplotypic phase. Am J Hum Genet. 78:
629644.
Sharon D, Glusman G, Pilpel Y, Khen M, Gruetzner F, Haaf T, Lancet D.
1999. Primate evolution of an olfactory receptor cluster: diversification by gene conversion and recent emergence of pseudogenes.
Genomics 61:2436.
Shi P, Zhang J.. 2006. Contrasting modes of evolution between vertebrate sweet/umami receptor genes and bitter receptor genes. Mol
Biol Evol. 23:292300.

MBE