Genomic risk for severe canine compulsive disorder, a dog model of human OCD

Abstract

Dogs naturally suffer the same complex diseases as humans, including mental illness. The dog is uniquely suited as a model organism to explore the genetics of neuropsychiatric disorders. Historical breed demographics have enriched purebred populations for founder effect mutations with tractable architectures, making genotypic analyses advantageous. Over a pet’s lifetime, owners observe the animal’s stress tolerance, arousal, and anxiety, and can inform on rich behavioral profiles for phenotypic analyses. Here we leverage these strengths in a search for inherited fac-tors that exacerbate canine compulsive disorder (CCD), the dog counterpart to human obsesssive compulsive disorder (OCD). Our rationale is that identifying pathways that predispose to disease severity will expand therapeutic options, and ultimately bring relief to those patients suffering the most. We have performed aGWAS of purebred Doberman pinschers that compares severely affected cases to moderately affected cases (24:70). This GWAS identified two statistically sig-nificant risk loci, on CFA34 and CFA11, and a third with suggestive evidence on CFA16. The locus on CFA34 includes a cluster of 5-HT3 receptor genes (HTR3C, HTR3D, and HTR3E) that implicate a serotonergic pathway that is routinely targeted by anti-OCD medications. The locus on CFA11 is syntenic with human CTXN3-SLC12A2 (5q35.1), an inherited risk factor for schiz-ophrenia. The third locus harbors teneurin-3 (TENM3), a modulator of the hypothalamic-pituitary-adrenal (HPA) axis, with effects on stress tolerance and stress-related behavior. We dis-cuss candidate genes and putative functional variants in light of pharmacological responsiveness, psychiatric comorbidity, and the potential for gene-by-environment interactions in the genetic etiology of OCD and CCD.

Full text

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 1
KEY WORDS: anxiety, obsessive-
compulsive disorder, CDH2, CTXN3,
HTR3C, HTR3D, HTR3E, SLC12A2, stress
response, Teneurin-3
ABBREVIATIONS: CCD, canine compulsive
disorder; CRF, corticotropin-releasing
factor; HPA, hypothalamic-pituitary-adrenal
axis; OCD, obsessive-compulsive disorder.
ABSTRACT
Dogs naturally suffer the same complex
diseases as humans, including mental ill-
ness. The dog is uniquely suited as a model
organism to explore the genetics of neuro-
psychiatric disorders. Historical breed demo-
graphics have enriched purebred populations
for founder effect mutations with tractable
architectures, making genotypic analyses
advantageous. Over a pet’s lifetime, own-
ers observe the animal’s stress tolerance,
arousal, and anxiety, and can inform on rich
behavioral proles for phenotypic analy-
ses. Here we leverage these strengths in a
search for inherited fac-tors that exacerbate
canine compulsive disorder (CCD), the dog
counterpart to human obsesssive compulsive
disorder (OCD). Our rationale is that iden-
tifying pathways that predispose to disease
severity will expand therapeutic options,
and ultimately bring relief to those patients
suffering the most. We have performed a
Genomic Risk for Severe Canine
Compulsive Disorder, a Dog Model of
Human OCD
Nicholas H. Dodman1*
Edward I. Ginns2
Louis Shuster3
Alice A. Moon-Fanelli1
Marzena Galdzicka2
Jiashun Zheng4
Alison L. Ruhe5
Mark W. Neff6
1Tufts Cummings School of Veterinary Medicine, 200 Westboro Road, N. Grafton, MA 01536
2University of Massachusetts Medical School, 55 N Lake Avenue, Worcester, MA 01655.
3Tufts University School of Medicine, 136 Harrison Avenue, Boston, MA 02111
4University of California, San Francisco, 1700 4th St., San Francisco, CA 94143-2542
5ProjectDog, 636 San Pablo Avenue, Albany, CA 94706
6Van Andel Institute, 333 Bostwick Ave N.E., Grand Rapids, MI 49503
Corresponding author:
Dr. Nicholas Dodman
Tufts Cummings School of Veterinary Medicine
200 Westboro Road
North Grafton, MA 01536
508.887.4640
Nicholas.Dodman@tufts.edu

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
2
GWAS of purebred Doberman pinschers that
compares severely affected cases to mod-
erately affected cases (24:70). This GWAS
identied two statistically sig-nicant risk
loci, on CFA34 and CFA11, and a third with
suggestive evidence on CFA16. The locus on
CFA34 includes a cluster of 5-HT3 recep-
tor genes (HTR3C, HTR3D, and HTR3E)
that implicate a serotonergic pathway that is
routinely targeted by anti-OCD medications.
The locus on CFA11 is syntenic with human
CTXN3-SLC12A2 (5q35.1), an inherited risk
factor for schiz-ophrenia. The third locus
harbors teneurin-3 (TENM3), a modulator
of the hypothalamic-pituitary-adrenal (HPA)
axis, with effects on stress tolerance and
stress-related behavior. We dis-cuss candi-
date genes and putative functional variants
in light of pharmacological responsiveness,
psychiatric comorbidity, and the potential
for gene-by-environment interactions in the
genetic etiology of OCD and CCD.
AUTHOR SUMMARY
We previously identied a locus on canine
chromosome 7 that confers susceptibility
to obsessive-compulsive disorder in ank
and blanket sucking Doberman pinscher
dogs. The chromosome 7 locus contains a
gene involved in the normal development of
glutamate receptors, dysfunction of which
is involved in the expression of obsessive-
compulsive disorder. This current study is
directed at identifying additional genetic
factors determining the severity of the con-
dition in our animal model. To this end, we
conducted testing in severely affected versus
mild-moderately affected dogs to explore
genetic differences. We found 2 distinct
regions on canine chromosomes 11 and 34
that appear to be genetic modiers affect-
ing the severity of the condition. The rst,
a locus on chromosome 11, contains a gene
increasing the risk of another psychi-atric
condition (schizophrenia) in humans. The
second, a locus on chromosome 34 harbors
sero-tonin receptor genes. That serotonin
genes are involved in determining the
severity of the condition seems particularly
relevant because drugs targeting the sero-
tonin pathway are routinely used in the treat-
ment of obsessive-compulsive disorder. We
hypothesize that the gene on chromosome 7
is essential for susceptibility to compulsive
disorder, and that other genes, notably ones
affecting the serotonin pathway, affect its
severity. These ndings have relevance in
furthering understanding the pathophysi-
ology of obsessive-compulsive disorder
in mammalian species and point the way
toward more effective treatments that target
both glutamate and serotonin pathways.
INTRODUCTION
Human obsessive-compulsive disorder
(OCD) is a mental illness characterized by
intrusive, distressing thoughts (obsessions)
and time-consuming, repetitive behaviors
(compulsions). OCD is one of the most
prevalent neuropsychiatric disorders, af-
fecting 1-3% of the worldwide population 1.
The World Health Organization (WHO) lists
OCD among the 20 most disabling diseases2.
Current therapies are not optimally effective
and extend medicinal benet to roughly half
of all patients3. OCD is a multifactorial dis-
order with a phenotypic spectrum. Patients
suffering from severe OCD report a greater
loss of time to persistent compulsions, and
experience sub-stantially greater emotional
distress and psychological impairment.
Severely affected patients also respond
much less frequently to available therapies,
and with greatly reduced benets in quality-
of-life outcomes3, 4. Understanding the
general etiology of OCD may lead to broad
improvements in diagnosis, treatment, and
possibly prevention. A high clinical prior-
ity is to alle-viate disease severity, and to
bring relief to those patients who currently
have the greatest unmet medical needs.
Understanding the genetic basis of severe
OCD holds promise for identifying novel
pathways, thereby expanding options for
improved diagnosis and therapeutic inter-
vention.
The apparent genetic heterogeneity of
human OCD has been a major obstacle to
genetic studies with human subjects. Ad-
ditionally, imprecise diagnosis and pheno-

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 3
typing, comorbidity and misclassication
with other disorders, and societal stigma and
privacy concerns further confound human
studies. Research to understand genetic risk
factors of OCD have met with limited suc-
cess5, 6. A recent large scale GWAS involv-
ing thousands of human cases and controls
failed to detect any loci signicantly
associated with OCD6. To our knowledge,
a replicated locus associated with human
OCD remains elusive.
Animal models represent an important
complementary strategy for gaining access
experimentally to causative mechanisms.7,
8 It is widely accepted that compulsion is
biologically conserved across mammals,
and that experimental results with naturally
occurring animal models are indeed relevant
to human OCD9, 10, 11. The literature supports
canine compulsive disorder (CCD) as a nat-
urally occurring counterpart of OCD. CCD
shares phenomenological aspects with OCD,
including the repetitious nature of basic be-
havioral patterns and the increased anxious
state of patients10, 12, 13. The early adult onset
of OCD in human patients14 is also observed
in peri-pubertal canine patients13. The neuro-
anatomical sites of OCD and CCD also ap-
pear to share overlap. Magnetic resonance
imaging in dogs showed both anterior cin-
gulate cortex and anterior insula gray matter
density reductions, implying altered activ-
ity15. An fMRI study in humans with OCD
and hoarding disorder indicated abnormal
activity in these same brain regions16. Lastly,
human and canine patients respond similarly
to therapy. As is clinical practice in human
medicine, veterinary medicine combines
behavioral modication with anti-OCD
drugs for the treatment of CCD17. These
include drugs developed for human patients,
such as uoxetine, a serotonergic agonist
via serotonin reuptake inhibition, and
memantine, an NMDA-based glutamatergic
antagonist.4, 18 The effectiveness of these
treatments in dogs suggest that clinical trials
in veterinary medicine will be predictive of
medicinal benets for human patients. In
this way, the dog also has enormous poten-
tial as a medical model for improving the
diagnosis and treatment of psychiatric disor-
ders in human and canine patients alike.
The genetic basis of CCD is expected to
be tractable in breed isolates. Breed predi-
lection implies an inherited predisposition,
and breed differences in the specic com-
pulsive behaviors co-opted by CCD further
suggests a genetic basis. Examples of breed
predisposition to specic compulsions in-
clude excessive grooming in certain breeds
(i.e., acral lick; 12,19), repetitive tail chasing
in terrier breeds10, 11, 20, and light-chasing
behavior in herding breeds.21 Thus genetic
inuences on multiple aspects of CCD can
be mapped serially in breeds to exploit
independent mutations in diverse popula-
tions. Results from multiple breed studies
may recapitulate in aggregate the biological
complexity observed in human OCD.
Each breed is naturally suited to genetic
analysis. Phenotypic variation within a breed
is attributable to founder effect variants
acting in concert with a relatively constant
(purebred) genetic background. This limits
phenotypic noise from modifying genes,
and increases the relative phenotypic effect
of a small number of segregating loci. Each
causal variant is located on an ancestral
haplotype that is 50- to 100-fold larger than
haplotype blocks comprising the human
genome. Ancestral haplotypes are readily
detectable by GWAS with SNP densities and
cohort sizes that are modest relative to hu-
man experimental standards.
Three previous studies have addressed
the genetics of CCD.22, 23 The rst study
identied a locus on chromosome 7 (CFA7)
that was associated with ank and blanket
sucking behavior in Doberman pinschers.22
The mapped interval spanned several
megabases but contained only a single gene,
neural cadherin (N-cadherin; CDH2). In the
second study, CDH2 was implicated in a dif-
ferent CCD, compulsive tail chasing, in an-
other unrelated breed, Belgian malinois (Cao
x, et al, PLOS One, 2014). CDH2 encodes
a cell-adhesion molecule expressed in the

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
4
hippocampus and cerebellum. N-cadherin
inuences many neuronal processes but its
role in the formation of NMDA receptors at
glutamatergic synapses is of particular inter-
est24. The risk conferred by this locus ap-
peared dose-dependent, with the risk allele
frequency greater among dogs exhibiting
multiple compulsive behaviors22. Finally, a
follow-up study on the Dobermans re-
analyzed the primary data using the MAGIC
algorithm23, increasing the SNP density four-
fold relative to the earlier study. A hybrid
mapping approach that incorporated systems
biology was used to detect three intervals
with genes (CTNNA2, ATXN1, and PGCP)
that could be assembled into the CDH2
network. This network emphasized synaptic
function, NMDA receptors, and glutama-
tergic neurotransmission. Polymorphisms
in CDH2 are also associated with a severe
human OCD and Tourette’s Syndrome 9,
two disorders that show comorbidity. The
mechanism by which CDH2 variation
increases risk for CCD, OCD, and Tourette’s
has not been established. Taken together,
the results are compelling as glutamatergic
and serotonergic neurotransmission are two
pathways routinely targeted using anti-OCD
medications25.
Here we extend experimental observa-
tions on CCD by focusing on the genetics of
disease severi-ty in the Doberman pinscher
breed. By mapping modiers that exacerbate
CCD, we aim to un-cover new factors that
implicate novel pathways for therapeutic
targeting. We present evidence that 2-3 loci
govern CCD severity in Doberman pinscher
dogs. Strong candidate genes in the mapped
intervals implicate pathways that may
explain important and fundamental features
of human OCD. These include response to
medications that target both the serotonergic
and gluta-matergic neural pathways, the
comorbidity of neuropsychiatric disorders,
and the role that stress tolerance and envi-
ronmental triggers play in the etiology and
severity of CCD and OCD.
RESULTS
GWAS of severe CCD in the Doberman
pinscher breed
We performed a GWAS comparing severely
affected cases with moderately affected
cases (here-after referred to as controls
for disease severity). These cohorts were
established through a com-bination of clini-
cal evaluation, owner-based surveys, and
telephone interviews conducted by a profes-
sional veterinary behaviorist. DNA from
cases (n = 24) and controls (n = 70) were
geno-typed at 174,376 SNPs genome-wide.
These genotype data were subjected to con-
ventional QC: 8,628 SNPs were excluded
due to poor replication in duplicate genotyp-
ings; 2,305 SNPs were excluded for low
call rates (< 10%); 75 SNPs were excluded
for failing a test of Hardy-Weinberg Equi-
librium; and 76,179 SNPs were excluded
due to a low minor allele frequency (MAF <
5%). The informative markers that remained
(95,817 SNPs) were tested for allelic asso-
ciation. Results from this GWAS are shown
in the Manhattan plot in Fig. 1A. Two loci,
on CFA11 and CFA34, were detected with
statistical signicance after correcting
for genome-wide testing by permuta-tion
p-value Odds Ratio Candidate Gene(s)
1.9 x 10-10 33.4 HTR3C, HTR3D, HTR3E
2.7 x 10-8 11.9 CTXN3, SLC12A1
1.2 x 10-5 5.4 TENM3
1 Loci taken from Illumina Canine BeadChipHD array.
2 Physical coordinates from the Boxer reference genome (canFam3).
3 A second SNP (BICF2G630815717) exhibited the same p-value.
This was an adjacent marker at chr16:48162506
Table 1. Summary of GWAS results

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 5
analysis. A third locus, on CFA16, showed
suggestive evidence of association. Fig. 1B
shows localized mapping results for each
chromosomal region of interest. A summary
of mapped intervals is provided in Table 1.
Serotonergic genes at the CFA34 risk
locus
The locus most strongly associated with se-
vere CCD was found on CFA34. This inter-
val spanned ~1.5 Mb (chr34:16.0-17.5 Mb)
and contained 39 annotated genes. The SNP
showing the strongest allelic association
(BICF2P816458) was 65 kb upstream of a
cluster of three paralo-gous genes (HTR3C,
HTR3D, and HTR3E), each encoding an
isoform subunit of the 5-HT3 re-ceptor. The
5-HT3 receptor is the lone ligand-gated ion
channel receptor in the serotonergic sys-tem
(26). This pathway is routinely targeted in
treatment of both human OCD (12, 27) and
ca-nine CCD (28). It is not known if SSRIs
are particularly effective in treating severe
cases, alt-hough it has been observed that
not all patients (human and canine) respond
positively to sero-tonergic-based treatment.
A syntenic locus with schizophrenia risk
A second locus, on CFA11, also showed
signicant allelic association. This mapped
interval spanned ~2.5 Mb and included
13 annotated genes. The canine locus is
syntenic with human 5q35.1, a chromo-
somal region recently found associated with
human schizophrenia (29). Two adjacent
genes (CTXN3 and SLC12A2) in this inter-
val are candidates for schizophrenia risk.
CTXN3, a member of the cortexin gene
family, is involved in intercellular signaling
for forebrain development (30). SLC12A2
encodes a Na-K-Cl symporter involved in
maturation of the gam-ma-amino-butyric
acid (GABA) signaling pathway (31) GABA
is an anxiolytic neurotransmitter (32), and
the GABAergic system has been implicated
in multiple human mood disorders (33).
Though the peak signal for association (SNP
BICF2P816458) was nearly a megabase
Figure 1A: Manhattan plot of GWAS for CCD Severity
Chromosome markers are plotted on the x-axis in order and alternately shaded. The -log10
(p-value) is plotted on the y-axis (and inset). Two loci, on CFA11 and CFA34, showed statisti-
cal signicance after correcting for genome-wide testing by permutation analysis. A third
locus, on CFA16, showed suggestive evidence of association.
Figure 1B: Localized mapping results for each chromosomal region of interest.

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
6
down-stream from the gene tandem, CTXN3
and SLC12A2 remained the most compelling
candidates in the interval (Fig 1B).
A modulator of the HPA axis and stress
tolerance
A third locus, on CFA16, showed sug-
gestive evidence for association with
severe CCD. The sig-nal for association
spanned a large interval (5 Mb). Two
adjacent SNPs (BICF2G630815658 and
BICF2G630815717) provided equally strong
statistical support. The chromosomal region
con-tained 36 annotated genes. Among these
was TENM3, a strong candidate for inte-
grating stress response with stress-related
behavior. The teneurins are transmembrane
proteins that serve as cell adhesion mol-
ecules (34). In addition, the extracellular
domain of teneurins is cleaved by pro-teol-
ysis and secreted (35, 36). These C-terminal
associated proteins (CTAPs) can counter
the ef-fect of corticotropin-releasing factor
(CRF) (37, 38) in the stress response path-
way, thereby at-tenuating the hypothalamic-
pituitary-adrenal (HPA) axis, most likely as
an adaptive response to chronic stress (39).
Chronic stress and anxiety are believed to
be important environmental inu-ences in
both CCD and OCD. Two SNP markers
showed equally strong statistical support for
association, and both were within 100 kb of
TENM3.
CDH2 and disease severity
This GWAS did not detect a signicant
association at CDH2, the locus on CFA7
previously found associated with CCD (22).
This result is consistent with the previously
published observa-tion because the present
study addressed disease severity. We did nd
modest evidence for CDH2 involvement.
Specically, at marker BICF2G630563196,
the strongest associated SNP from Dod-
man et al (22), the risk allele frequency was
greater in our severe cases (57%) than in
our moderately affected controls (42%). It
is thus possible that CDH2 contributes to
disease se-verity, but that additional power
Source Cohort Sample
Size
Risk Allele
Frequency
This study Mild/Moderate CCD Controls 70 42%
This study Severe CCD Cases 24 57%
Dodman et al, 2010 Non-Affected CCD Controls 67 22%
Dodman et al, 2010 Affected CCD Cases, a single compulsion 68 43%
Dodman et al, 2010 Affected CCD Cases, multiple compulsions 20 60%
Table 2. Risk Allele Frequencies at CDH2 1
1 Frequency of risk allele at BICF2G630563196; showed strongest association in Dodman et al (2010).
Risk allele,T; Alternate allele, C.
Locus
(Mb)
Haplotype
Length
(Kb)
Comprising
SNPs1
Risk
Haplo-
type2
Freq. of
Case
Haplotypes
Freq. of
Control
Haplotypes
p-Value3
Chr34:16.9 96.8 6 AGGGGG 33.3% (16/48) 2.8% (4/140) 1.6 X 10-14
Chr11:18.6 121.4 10 AGAG-
CACGGG
27.1% (13/48) 2.8% (4/140) 6.4 X 10-10
Chr16:48.1 51.1 5 GGGGA 39.6% (19/48) 10.7% (15/140) 1.9 X 10-7
1 Information for SNPs comprising each risk haplotype is provided in Table 4.
2 Allelic conguration for inferred risk haplotype.
3 p-value calculated from binomial distribution, as described in Materials and Methods.
Table 3. Summary of haplotype analyses

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 7
Chromosome
(CFA)
Prelim. Position
(of 21 SNPs)
Array
SNP1
Base
Coordinate2
Allele 13Allele 2
11 12 BICF2P962745 18601737 A G
11 13 BICF2P298471 18607272 G A
11 14 BICF2P1255374 18617008 A C
11 15 BICF2P1423859 18645289 G A
11 16 BICF2S23553865 18662547 C T
11 17 BICF2P912457 18672446 A G
11 18 BICF2P1037814 18685096 C A
11 19 BICF2S23519359 18701992 G A
11 20 BICF2P1047236 18716127 G A
11 21 BICF2S23158677 18723056 G A
16 12 BICF2G630815658 48111441 G A
16 13 BICF2G630815667 48123834 G A
16 14 BICF2G630815674 48135347 G A
16 15 BICF2S23110272 48145249 G A
16 16 BICF2G630815717 48162506 A G
34 6 BICF2P1061643 16833175 A G
34 7 BICF2P69046 16858896 G A
34 8 G1457f42S203 16885802 G A
34 9 BICF2S23646017 16905048 G A
34 10 BICF2S23751509 16911851 G A
34 11 BICF2P185055 16930008 G A
Table 4. SNP loci in haplotype analyses
1 SNP loci are taken from the Illumina canineBeadchipHD array.
2 Physical coordinates from the Boxer reference genome (canFam3).
3 Specic allele associated with severe CCD risk haplotype.
Mapped
Locus (Mb)
Interval
(Mb)
No. Observed
Variants2
Variant
Freq.
(per Kb)
No.
Conserved3
No. Putatively
Functional4,5
16.50-19.00 2.5 7,648 3.1 814 116
44.00-49.00 4.0 16,933 4.2 2,006 404
16.00-17.50 1.5 1,341 0.9 193 59
Table 5. Summary of whole genome sequencing1
1 Data generated using the Illumina HiSeq platform with a paired-end library and two lanes of ow cell.
2 Variants called relative to the Boxer reference genome (canFam3).
3 Variants with phastCons scores greater than 0.20.
4 Variants with phastCons scores greater than 0.70, and/or high to moderate SIFT scores.
5 No variants having high to moderate SIFT scores were detected in candidate genes.
6 Gene list from within mapped loci that harbor potentially functional sequence variation in Table 9

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
8
(i.e., larger cohorts) is needed to detect a sig-
nicant effect. Table 2 shows the compari-
sons of CDH2 risk allele frequency among
cohorts and across studies.
Haplotypic analysis of risk loci for disease
severity
Haplotype signatures facilitate population
genetics, genetic epidemiology, and replica-
tion stud-ies. In this study, haplotyping was
used to select an individual that was geno-
typically selected for mutation discovery
by next-generation sequence analysis. We
inferred the risk haplotype at mapped loci
using available SNP genotype data. Initially,
21 markers were phased at each of lo-cus.
These markers were centered on the stron-
gest associated SNP marker at the locus.
Based on initial haplotype lists, we selected
a subset of markers to further dene a core
haplotype, which best differentiated severely
affected cases from moderately affected
controls. Table 3 summariz-es these results.
We applied these multi-marker signatures
to infer the presence of causal variants in
dogs under consideration for whole genome
sequencing.
DNA variants of interest
We performed whole genome sequencing
to search for mutations that might inuence
disease severity in CCD on a case dog that
was informative at all three loci of interest.
This dog was ho-mozygous for the CFA34
and the CFA16 risk haplotypes. This dog
was also predicted to be het-erozygous for
the CFA11 risk haplotype. Table 4 shows
the parameters for whole genome se-quence
analysis. The data comprised a ~32x cover-
age of the genome. Variants were obtained
in comparison to the reference genome that
was derived from a purebred Boxer (40).
The Boxer breed does not show predilection
for CCD. This suggested that causative risk
factors in the Do-berman pinscher could be
detected as variant alleles relative to this ref-
erence genome. Table 5 summarizes DNA
variant discovery.
A total of 25,922 DNA variants were
detected across chromosomal regions of
interest. Of these, roughly 600 variants
were likely to have functional effects based

Table 6. DNA variants in three chromosomal regions of interest (phastCons > 0.2)

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 9
on phylogenetic conser-vation (phastCons
> 0.7; n = 579) and/or protein structure/
function informatics (SIFT, moder-ate/high
probability scores; n = 64). No protein-cod-
ing changes (non-synonymous substitution,
frameshifts, etc) were found in candidate
genes (Table 6). This implied that causal
variants might be located in anking regions
continuing regulatory elements, in introns,
or that other ge-nes in these intervals are
responsible for the associated risk for severe
CCD.
DISCUSSION
The GWAS described here has identied
two loci strongly correlated with severe
CCD in Doberman pinscher dogs, as well as
an additional locus that yielded only sugges-
tive evidence for association. This locus was
carried forward, despite modest statistical
support, because of the compelling candi-
date gene in the interval (i.e., TENM3), and
the implications it may hold for un-derstand-
ing environmental inuences on neuropsy-
chiatric disorders. We have interpreted these
results to mean that one or more sequence
variants within each interval functionally
exacerbates CCD.
This GWAS focused on disease severity,
and as such, the loci that have been identi-
ed likely harboring disease modifying
variant, that interact with other loci confer-
ring general risk for CCD22, 23. These loci are
provisional as they require replication with
an independent cohort. This is also true of
previously published CCD loci, as these ear-
lier genomic ndings were from two studies
that utilized the same cohorts22, 23.
Further studies will identifygene-gene
interactions, which are thought to be impor-
tant in the eti-ology of mental illness, but
which have proven difcult to address in
human genetic studies.
These loci harbor compelling candi-
date genes that point to novel physiologic
pathways. Previous results emphasized
CDH2-dependent synaptic function in the
glutamatergic system (i.e., CDH2, CTNNA2,
ATXN1, and PGCP) as the principal patho-
physiology of CCD22, 23. Our results re-lating
to disease severity appear to reect distinct
aspects of CCD/OCD biology that are com-
monly recognized but poorly understood.
These include (i) the efcacy of serotonergic
agonists in roughly half of patients with
CCD/OCD (HTR3C, HTR3D, and HTR3E
on CFA34); (ii) comorbidity of OCD with
other neuropsychiatric disorders (CTXN3-
SLC12A2 on CFA11); and (iii) the inuence
of environmental factors and chronic stress
in exacerbating disease severity (TENM3
on CFA16). Taken together, the evidence
strongly suggests that the salient features of
human OCD may also be reected in the ge-
netic susceptibility of dogs to severe CCD.
Serotonin, the 5-HT3 receptor, and the
biology of compulsion
To our knowledge, this study is the rst to
implicate the serotonergic system in inher-
ited suscep-tibility to both OCD and CCD.
The seminal role of serotonin as a modulator
of human OCD has a long history, dat-
ing back to 197241, 42. This early research
demonstrated that treatment of OCD with
serotonin re-uptake blockers alone was
sufcient to generate an anti-obsessional re-
sponse in many patients. This suggested that
low serotonergic signaling was an etiologic
factor for OCD. It is now widely accepted
that serotonin dysregulation contributes
directly to OCD43.
Serotonergic agonists, particularly
SSRIs, are now a mainstay of human OCD
treatment44. These drugs have been applied
in veterinary behavioral medicine for several
decades12, with canine patients showing
similar response proles to human patients45.
Given the central role of the serotonergic
system in OCD and the efcacy of SSRIs in
treating CCD, the association of serotonin
receptor genes with severe CCD is compel-
ling.
HTR3C, HTR3D, and HTR3E gene prod-
ucts form multiple isoforms of the 5-HT3
receptor, one of seven receptor subtypes in
the serotonergic system. The 5-HT3 receptor
is the lone ligand-gated ion channel recep-
tor26, 46. Multiple psychiatric conditions have
been causally linked to changes in 5-HT3

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
10
receptor function47, and variation in 5-HT3
genes has been shown to have phenotypic
effects in human behavior (48, 49). HTR3C
is an inherited risk factor for autism and
HT3RD inuences human anxiety50. The
third paralogous gene in this cluster, HTR3E,
is res-tricted to myenteric neurons in the
peripheral nervous system51 and thus seems
an unlikely candidate for causing CCD.
However, Irritable Bowel Syndrome (IBS)
shows signicant comor-bidity with OCD52,
53, and gastrointestinal (GI) function is often
affected in human OCD pa-tients who also
suffer from major depressive illness54. Regu-
latory changes in the HTR3 gene cluster
could have pleiotropic effects on the central
nervous and GI system. To our knowledge,
the comorbidity of CCD and IBS in dogs has
not previously been investigated.
The 5-HT3 receptor genes are expressed
in brain regions that are functionally abnor-
mal in human OCD patients. Based on neu-
ro-imaging studies, the cingulate, the CA1
region of the hippocam-pus, and amygdaloid
complex are aberrant in human patients55, 56.
HTR3C and HTR3D are highly expressed
in these regions57, as well as in other brain
regions associated with cogni-tion, affect,
and modulation of sensory input58.
The 5-HT3 receptor has also been im-
plicated in the biology of addiction. Clinical
studies have shown that 5-HT3 receptor
antagonists decrease alcohol consumption
in patients with alcoholism59, 60. This has
led to the suggestion that activation of the
5-HT3 receptor may have rewarding or
reinforcing properties. It has been sug-
gested that compulsive behavior is a form
of addiction and that known comorbidity of
OCD and addictive behavior1, 61 may stem
from shared risk variants acting through
the 5-HT3 receptor. Interestingly, treatment
of obsessive-compulsive dis-order with
odansetron, a specic 5HT3 antagonist, was
associated with a signicant decrease in the
Yale Brown obsessive-compulsive scores in
one study of 8 patients62.
Relevance of OCD comorbidity
Comorbidity is a common and perhaps
telling feature of molecular mechanisms un-
derlying neu-ropsychiatric disorders63. OCD
has a signicant comorbidity with several
psychiatric disor-ders. This implies shared
etiology and common risk factors. Major
depressive illness, bipolar di-sorder, To-
urette’s syndrome, attention decit disorder,
panic disorder, generalized anxiety disor-der,
schizoaffective disorder and addiction have
all been reported to co-occur with OCD1,
64. The locus on CFA16 (CTXN3-SLC12A2)
may be relevant to this comorbidity. The
orthologous locus in the human genome
is an inherited risk factor for schizophre-
nia29, 65; two adjacent genes are candidates
for causality. CTXN3 is a member of the
cortexin family, which is involved in cogni-
tion, memory, and learning30 as well as early
forebrain development30, 66)]. SLC12A2 is a
Na-K-Cl symporter involved in GABA sig-
naling. Low GABA neurotransmission is a
common nding in mood disorders33, 66, and
there is considerable crosstalk between the
GABAergic, glutamatergic, and serotonergic
systems67. Our results, although tentative,
sug-gest that at-risk dogs may suffer from
other psychiatric conditions at an increased
rate. Whereas no known counterparts for
disorders such as major depressive illness or
Tourette’s syndrome ha-ve been described
in the dog, other mood disorders, such as
panic disorder and separation an-xiety, are
well documented68, 69. Comorbidity of these
disorders with CCD has not been re-ported.
Integrating nature and nurture
Neuropsychiatric disorders are believed to
stem from a combination of genetic and
environmen-tal risk factors. The multitude
of potentially interacting environmental
inuences presents an enormous challenge
to understanding the role that environment
plays in human mental illness. In the etiol-
ogy of OCD, there is considerable evidence
that stress may trigger and/or exacerbate the
disorder. Although neuroendocrine control
of the stress response is well understood,
much less is known of the biology of stress
tolerance and of the neural circuitry underly-
ing stress-related behavior. Both are likely to

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 11
be important components underlying neuro-
psychiatric disorders, in-cluding OCD.
TENM3, a candidate gene at the CFA16
locus, is involved in integrating stress toler-
ance with stress-related behavior70. TENM3
encodes teneurin-3, a protein that forms
a heterodimer with TENM1 and mediates
cell adhesion at synapses. Moreover, the
c-terminus of this protein is clea-ved and
secreted in the CNS. Exogenous teneurin-3
suppresses stress-related behavior that is in-
duced in the rat by injecting with CRF38, 71,
72. Similarly, exogenous TENM1 can counter
the effect of CRF to reinstate cocaine-
seeking behavior in a rat model of addic-
tion38, 73. The inability to cope with chronic
stress has been suggested as a risk factor for
disease severity in OCD73, 74. We suggest that
variation in TENM3 affects a dog’s ability to
adaptively dampen the stress response, and
consequently, worsens disease severity.
An OCD/CCD model that assimilates
pharmacological response and genetic
risk
Prior pharmacological insights begin to
make sense in light of the current results
with CCD and OCD that now implicate both
glutamatergic and serotonergic mechanisms.
The distinct risk con-tributions could explain
why response to serotonergic and glutama-
tergic drugs is variable among patients. The
genetic ndings in dogs may also explain
why a glutamate-blocking strategy more ef-
fectively reduces compulsive behavior in an
animal model when combined with a SSRI74.
We propose a working model in which
generalized risk stems from variation in sev-
eral genes that inuence the glutamatergic
system. CDH2 inuences the assembly and
function of NMDA receptors. Drug block-
ers of NMDA receptors attenuate compul-
sion in both human and compan-ion animal
models18, 25, 75. Moreover, the observation
that serotonin actually decreases glu-tamate
in some brain regions67 might explain the
complementary action of SSRIs and NMDA
antagonists in treating OCD.
Breed ancestry, hallmark traits, and
pleiotropic effects on CCD
The history of the Doberman pinscher breed
is relevant to CCD susceptibility. The breed
was constructed in Germany (c.1890) to
serve as an energetic and watchful guard
dog. Genes from multiple pre-existing
breeds were intentionally introgressed to
assemble a specic work-related behavioral
prole. This prole included high-energy,
arousal, and vigilance, all of which are use-
ful for working watchdogs. In this respect,
the vigilance of Doberman pinscher dogs is
adap-tive and desirable but can also be asso-
ciated with an anxious or nervous tempera-
ment. The sero-tonergic and glutamatergic
systems govern anxiety and high arousal,
respectively.Anxiety con-tributes to compul-
sive behavior in dogs69.
The purpose for which Doberman pin-
schers were bred also establishes a context
for environmen-tal inuence in CCD. Al-
though Doberman pinschers were adapted to
a demanding and active guard dog lifestyle,
most dogs today do not receive this level of
environmental engagement. We propose that
under-stimulated dogs may be at increased
risk for developing severe CCD.
Compulsion Abbreviation Description
Blanket Sucking BS Excessive mouthing or suckling of soft objects
Flank Sucking FS Excessive mouthing of the ank
Object Fixation OF Excessive preoccupation with an object or toy
Shopping/Hoarding SH Excessive collecting and organizing of objects
Acral Lick1AL Excessive grooming of the lower extremities
Table 7. List of compulsive behaviors observed in Doberman pinschers
1 Self-injurious when severe and intense.

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
12
Chromosome
(CFA)
Array
SNP1
Base
Coordinate2
p-value Odds
Ratio
11 BICF2G630294216 15497414 3.62E-06 17.5
11 BICF2P539905 16690377 1.43E-05 7.882
11 BICF2P1310910 16694511 1.43E-05 7.882
11 BICF2G630295301 16757670 1.39E-07 10.81
11 BICF2G630295322 16786128 1.39E-07 10.81
11 BICF2G630295566 17041909 1.00E-06 7.763
11 BICF2G630295619 17110379 3.19E-06 8.798
11 BICF2G630295864 17355025 8.93E-07 10.64
11 BICF2P293356 17406145 4.28E-06 9.529
11 BICF2G630295925 17446652 4.28E-06 9.529
11 BICF2S23548912 17623884 4.28E-06 9.529
11 BICF2S23141428 17626394 4.28E-06 9.529
11 BICF2P1349400 17637729 4.28E-06 9.529
11 BICF2S2303223 17655946 8.93E-07 10.64
11 BICF2P124153 17757867 8.93E-07 10.64
11 BICF2P701 17764798 8.93E-07 10.64
11 BICF2P55387 17871634 4.28E-06 9.529
11 BICF2P461569 17879714 1.53E-06 7.984
11 BICF2P718382 17882504 3.48E-08 13.06
11 BICF2P321292 17926621 2.22E-07 8.533
11 BICF2P878837 17956937 2.87E-07 8.402
11 BICF2P218661 18237383 1.00E-06 7.763
11 BICF2P1088502 18288511 1.68E-06 6.875
11 BICF2G630296074 18300189 2.22E-07 8.533
11 BICF2G630296084 18305904 1.68E-06 6.875
11 BICF2S23643044 18328886 2.75E-08 11.91
11 BICF2G630296192 18422040 4.32E-06 7.043
11 BICF2S23418452 18591044 1.62E-08 NA
11 BICF2P962745 18601737 1.02E-07 10.13
11 BICF2P298471 18607272 4.89E-07 9.194
11 BICF2P912457 18672446 1.39E-07 10.81
11 BICF2P112299 18964693 3.19E-06 8.798
11 BICF2P235304 19098405 1.53E-06 7.984
11 BICF2S24320279 19419232 2.26E-06 8.312
11 BICF2P772375 19502128 1.53E-06 7.984
11 BICF2P1095814 20534589 9.95E-06 7.485
11 BICF2P914727 20546241 9.95E-06 7.485
16 BICF2G630815658 48111441 1.20E-05 5.357
Table 8. Estimated p-values for informative SNPs in three regions of interest for CCD

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 13
Chr. Position3
(Mb)
Proximal
Gene
Molecular Function4Potential Relevance
to CCD/OCD
CFA1
1
16.51 MEGF10 Multiple EGF-Like Domains 10
CFA1
1
16.64 PRRC1 Proline-Rich Coiled-Coil 1
CFA1
1
16.75 CTXN3 Cortexin 3, Kidney- and Brain-expressed Locus associated with
schizophrenia
CFA1
1
17.25 SLC12A2 Solute Carrier Family 12 (Sodium/Pota-
ssium/Chloride Transporter), Member A2
Locus associated with
schizophrenia
CFA1
1
17.32 FBN2 Fibrillin 2
CFA1
1
17.35 IFBN2 Not Available
CFA1
1
17.97 SLC27A6 Solute Carrier Family 27 (Fatty Acid
Transporter), Member A6
CFA1
1
18.01 IS0C1 Isochorismatase Domain Containing 1
CFA1
1
18.39 ADAMTS1 9 A Disintegrin and Metalloproteinase with
Thrombospondin Motif
CFA1
1
18.51 KIAA1024L Not Available
CFA1
1
18.62 CHSY3 Chondroitin Sulfate Synthase
CFA1
6
44.24 FAT1 FAT Atypical Cadherin 1 Member of cadherin
superfamily highly
expressed in brain
CFA1
6
44.48 F11 Plasma Thromboplastin Antecedent
CFA1
6
44.53 CYP4V2 Cytochrome P450, Family 4, Subfamily V,
Polypeptide 2
CFA1
6
44.58 FAM149A Family With Sequence Similarity 149,
Member A
CFA1
6
44.62 H6BA88 Toll-Like Receptor homolog
CFA1
6
44.88 SORBS2 Sorbin And SH3 Domain Containing 2
CFA1
6
45.14 PDLIM3 PDZ And LIM Domain 3
CFA1
6
45.18 CCDC110 Coiled-Coil Domain Containing 110
CFA1
6
45.23 ANKRD37 Ankyrin Repeat Domain 37
CFA1
6
45.24 LRPSBP LRP2 Binding Protein
CFA1
6
45.24 UFSP2 UFM1-Specic Peptidase 2
Table 9. Genes from within mapped loci that harbor potentially functional sequence varia-
tion1,2

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
14
CONCLUSIONS
This study focused on the genetics underly-
ing CCD severity in an innovative animal
model. Our aim was to identify novel path-
ways in OCD/CCD that would ultimately
point to more effective therapeutic interven-
tions. To our knowledge, no previous study,
in human or canine, has ad-dressed the
factors that drive severity in OCD and CCD.
To accomplish this, we leveraged the inher-
ent strengths of canine breed genetics, where
causal factors are derived from a small num-
ber of ancestral mutations. This strategy suc-
cessfully identied 2-3 novel loci contining
candi-date genes that govern CCD severity
in the Doberman pinscher breed. These
results represent important genetic leads to
pursue, and can lead to a better understand-
ings of the molecular mech-anisms underly-
ing CCD and OCD.
MATERIALS AND METHODS
Ethics Statement
Samples were obtained with informed owner
consent according to IACUC protocols
#G82706 (Cummings School of Veterinary
Medicine (CSVM) at Tufts University) and
#11-02-002 (Van Andel Research Institute).
Canine Subjects
Severe cases and moderately affected con-
trols were privately owned purebred dogs
from the American Kennel Club (AKC) reg-
istry. Subjects were recruited by researchers
at CSVM, relying both on a clinical program
in veterinary behavioral medicine and on a
breed community network. A large sample
of dogs was enrolled initially (n = 200).
Phenotypic assessment was made ac-cording
to owners’ responses on a phenotype survey
questionnaire, the same as used in our pre-
vious study (13). All of the dogs studied
exhibit anked sucking or blanket sucking,
both closely related, virtually breed-specic
CCDs, with or without other compulsions
such as object xation, shopping/hoarding,
or acral lick (Table 7). Twenty four dogs
exhibited severe CCD as deter-mined by
their displaying more than one compulsion
(12 dogs), hourly or daily frequency, and
owner-estimated duration of >8% of the
day. Mild to moderately affected dogs (n =
70) typically engaged in only one compul-
sion, exhibited the behavior daily of weekly,
and engaged in the be-havior <8 of the
day (6.4% +/- xxx, n = 59). Time spent per
day was only available for 59 dogs, others
simply stated “no access,” “situational” or
“when crated” conrming their mild-moder-
ate status.
Sample Recruitment and DNA
Informed owner consent was obtained at
the time of sample submission according
to IACUC protocols at TSVM and VARI
((#G82706 and #11-02-002, respectively).
Samples were collected randomly from
willing participants’ dogs across the United
States assuring a heterogeneous mix with no
geographical or other bias. TSVM samples
were collected as blood, and genomic DNA
was extracted with conventional methods.
VARI samples were collected by buccal
swab, saliva kit (Genotek, Ontario, CA), or
blood. DNA was extracted from blood or sa-
liva using protocols adapted to an automated
workow (AutogenFlexStar, Holliston, MA)
as previously described76. Crude extracts
were prepared from buccal swabs as previ-
ously described76. Samples were quantied
by nanodrop spectrophotometry. The integ-
rity of genomic DNA (i.e., fragmenta-tion
and degradation) was assessed by agarose
gel electrophoresis.
Genotyping and Related Analyses
Genome-wide genotyping was performed
with the CanineHD BeadChip (Illumina, La
Jolla, CA). Briey, the platform is based on
arrayed oligonucleotides, each correspond-
ing to a known physical polymorphism in
the dog genome, which can be assayed by
enzymatic single-base ex-tension and dual-
color uorescence to report bi-allelic calls.
The array offers 174,376 SNP loci at an
average density of 70 SNPs per megabase.
Raw uorescence intensity data gener-
ated with a BeadArray Reader (Illumina,
La Jolla, CA) were converted to curated
genotypes using Ge-nomeStudio software
with pre-set genotypic cluster algorithms.
Fixed array SNP data were ana-lyzed us-

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 15
ing PLINK (76). SNP data were ltered to
exclude (i) data from individual dogs with
total SNP call rates less than 98%; (ii) SNPs
with minor allele frequency (MAF) less than
0.05; (iii) individual SNP loci with call rates
less than 99.8%; and (iv) individual SNP
loci that did not pass the test for Hardy–
Weinberg equilibrium (HWE) in the control
cohort (p < 0.05). Case-control cohorts were
tested for population substructure using mul-
tidimensional scaling in PLINK. P-values
for GWAS were adjusted for multiple tests,
and the genome-wide signicance thresh-
old was set by permutation of case-control
labels (100,000 iterations);78.
Haplotype analysis was performed using
PHASE (v 2.1.1; 79). Phased haplotypes
were inferred in two steps. Preliminary
haplotypes were inferred with the use of
10 SNPs on each side of the marker that
showed the most signicant p-value for
association. This was performed in each
of the three regions of interest (Table 8). A
subset of SNPs was then selected to dene
a core risk haplotype that most strongly dif-
ferentiated the case group from the control
group. These SNPs were re-phased in cases
and controls separately. To maximize the
likelihood of detecting risk haplotypes in
the control group, 25 mock individuals that
were homozygous for the risk haplo-type
were added to the input genotype data for
controls. This ‘seeded’ the analysis toward
detec-tion of the risk haplotype. The esti-
mates of haplotype frequency excluded the
results from the mock individuals. The p-
values for enrichment of the risk haplotypes
in the cases were calculat-ed from a bino-
mial distribution of the haplotype frequency
in controls. The p-values in
Table 3 reect the probability of observ-
ing in the cases the same or higher risk
haplotype fre-quency that was observed in
the controls.
Whole Genome Sequencing and Related
Analyses
Whole genome sequencing was performed
at Beijing Genome International using two
lanes of a HiSeq instrument (Illumina, La
Jolla, CA). DNA was isolated from whole
blood obtained from a severely affected dog.
Raw HiSeq sequence data were aligned to
the reference genome (canFam3.0) using the
SNAP Sequence Aligner (80). Variants rela-
tive to the reference were called using SAM-
Tools81. Variants were assessed for putative
functional impact by SnpEff82 and SIFT 83,
and by assessing conservation across mam-
malian species (phatsCons scores);84. NGS
analytics were performed on a cloud cluster
(Amazon).
ACKNOWLEDGEMENTS
Three collaborating institutions provided
funding and supported this work: Tufts
University Cummings School of Veterinary
Medicine, the University of Massachusetts
Medical School, and the Van Andel Re-
search Institute. We thank the many dog
owners who enrolled their pet dogs in this
study. We are especially grateful to Kathy
Davieds of the Doberman pinscher com-
munity. We thank Alex Roemer, Andrew
Borgman and Nicole Cottam for technical
assistance and Lisa Kefene for performing
xed array genotyping.
REFERENCES
1. Ruscio AM, Stein DJ, Chiu WT, and Kessler RC.
(2010) The epidemiology of obsessive-compulsive
dis-order in the National Comorbidity Survey
Replication. Molecular Psychiatry 15:53-63.
2. World Health Organization (2014) Global Health
Observatory (GHO), World Health Statistics.
Available at: http://www.who.int/gho/publications/
world_health_statistics/2014/en/.
3. Pallanti S, Hollander E, Bienstock C, Korin L,
Leckman J, Marazziti D, Pato M, Stein D, Zohar
J (2002) Treatment non-response in OCD: meth-
odological issues and operational denitions. Int J
Neuropsycho-pharmacol 5:181-91.
4. Haghighi M, Jahangard L, Mohammad-Beigi
H, et al (2013) In a double-blind, randomized
and placebo-controlled trial, adjuvant emantine
improved symptoms in inpatients suffering from
refractory obsessive-compulsive disorders (OCD).
Psychopharmacology 228:633-640.
5. Potkin SG, Turner JA, Guffanti G, et al (2009) A
genome-wide association study of schizophrenia
using brain activation as a quantitative phenotype.
Schizophr Bull 35:96-108.
6. Stewart SE, Yu D, Scharf JM, et al. (2013)
Genome-wide association study of obsessive-com-
pulsive dis-order. Molecular Psychiatry 18:788-
798.
7. Insel TR (1992) Toward a neuroanatomy of

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
16
obsessive-compulsive disorder. Arch Gen Psychia-
try 49:739-44.
8. Feldstein Ewing SW, Sarah W, Chung T (2013)
Neuroimaging mechanisms of change in psycho-
therapy for addictive behaviors: Emerging transla-
tional approaches that bridge biology and behavior.
Psychology of Addictive Behaviors 27:329.
9. Moya PR, Dodman NH, Timpano KR, et al (2013)
Rare missense neuronal cadherin gene (CDH2)
variants in specic obsessive-compulsive disorder
and Tourette disorder phenotypes. Eur J Hum
Genet 21:850-4.
10. Dodman NH, Moon-Fanelli A, Mertens PA,
Pueger S, and Stein D (1997) Veterinary Models
of OCD. Obsessive--Compulsive Disorders: Diag-
nosis, Etiology, Treatment: (Mosby, St. Louis), pp
99-143.
11. Dodman NH, Jenike MA, Baer L, Minichiello WE
(1998) Veterinary Models of Obsessive Compul-
sive Disorder: Obsessive Compulsive Disorders.
Practical Management. (Mosby, St. Louis), pp
318-334
12. Rapoport JL, Ryland DH, and Kriete M (1992)
Drug treatment of canine acral lick. An animal
model of ob-sessive-compulsive disorder. Arch
Gen Psychiatry 49:517-21.
13. Moon-Fanelli AA, Dodman NH, and Cottam N
(2007) Blanket and ank sucking in Doberman
Pinschers. Journal of the American Veterinary
Medical Association 231:907-912.
14. Grisham JR, Anderson TM, and Sachdev PS
(2008) Genetic and environmental inuences on
obsessive-compulsive disorder. European Archives
of Psychiatry and Clinical Neuroscience 258:107-
116.
15. Ogata N, Gillis TE, Liu X, et al (2013) Brain
structural abnormalities in Doberman pinschers
with canine compulsive disorder. Prog Neuropsy-
chopharmacol Biol Psychiatry 45:1-6.
16. Tolin DF, Stevens MC, Villavicencio AL, et al
(2012) Neural mechanisms of decision making in
hoarding disorder. Arch Gen Psychiatry 69:832-41.
17. Ross S, and Peselow E (2009) Pharmacotherapy of
addictive disorders. Clinical Neuropharmacology
32:277-289.
18. Schneider BM, Dodman NH, and Maranda L
(2009) Use of memantine in treatment of canine
compulsive disorders. Journal of Veterinary
Behavior: Clinical Applications and Research
4:118-126.
19. Paradis M, de Jaham C, Page N, Sauve F, and
Helie P (2005) Acral mutilation and analgesia in 13
French spaniels. Vet Dermatol 16:87-93.
20. Moon-Fanelli AA, and Dodman NH (1998)
Description and development of compulsive tail
chasing in ter-riers and response to clomipramine
treatment. J Am Vet Med Assoc 212:1252-7.
21. Dodman NH, and Shuster L (2005) Animal Models
of Obsessive-Compulsive Behavior: A Neurobio-
logical and Ethological Perspective. Concepts and
Controversies in Obsessive-Compulsive Disorder:
(Mosby, St. Louis), pp 53-71.
22. Dodman NH, Karlsson EK, Moon-Fanelli A, et
al (2010) A canine chromosome 7 locus confers
compulsive disorder susceptibility. Molecular
Psychiatry 15:8-10.
23. Tang R, Noh HJ, Wang D, et al (2014) Candidate
genes and functional noncoding variants identied
in a canine model of obsessive-compulsive disor-
der. Genome Biology 15:R25.
24. Shoval I, Ludwig A, and Kalcheim C (2007)
Antagonistic roles of full-length N-cadherin and
its soluble BMP cleavage product in neural crest
delamination. Development 134:491-501.
25. Stewart SE, Jenike EA, Hezel DM, et al (2010) A
single-blinded case-control study of memantine
in severe obsessive-compulsive disorder. J Clin
Psychopharmacol 30:34-9.
26. Maricq AV, Peterson AS, Brake AJ, Myers
RM, and Julius D (1991) Primary structure and
functional ex-pression of the 5HT3 receptor, a
serotonin-gated ion channel. Science 254:432-7.
27. Preston JD, O’Neal JH, and Talaga MC (2013)
Handbook of Clinical Psychopharmacology for
Therapists (New Harbinger Publications, Oakland),
p 75
28. Irimajiri M, Luescher AU, Douglass G, Robertson-
Plouch C, Zimmermann A, Hozak, R (2009)
Random-ized, controlled clinical trial of the
efcacy of uoxetine for treatment of compulsive
disorders in dogs. J Am Vet Med Assoc 235:705-9.
29. Panichareon B, Nakayama K, Iwamoto S,
Thurakitwannakarn W, and Sukhumsirichart W
(2012) Associa-tion of CTXN3-SLC12A2 polymor-
phisms and schizophrenia in a Thai population.
Behav Brain Funct 8:27.
30. Wang HT, Chang JW, Guo Z, Li BG (2007) In
silico-initiated cloning and molecular characteriza-
tion of cortexin 3, a novel human gene specically
expressed in the kidney and brain, and well con-
served in verte-brates. Int J Mol Med 20:501-10.
31. Hyde TM, Lipska BK, Ali T, et al (2011) Ex-
pression of GABA signaling molecules KCC2,
NKCC1, and GAD1 in cortical development
and schizophrenia. The Journal of Neuroscience
31:11088-11095.
32. Foster AC, Kemp JA (2006) Glutamate- and
GABA-based CNS therapeutics. Curr Opin Phar-
macol 6:7-17.
33. Brambilla P, Perez J, Barale F, Schettini G, and
Soares JC (2003) GABAergic dysfunction in mood
disor-ders. Molecular Psychiatry 8:721-737.
34. Oohashi T, Zhou X-H, Feng K, et al (1999) Mouse
ten-m/Odz is a new family of dimeric type II trans-
membrane proteins expressed in many tissues. J
Cell Biol 145:563-77.
35. Qian X, Barsyte-Lovejoy D, Wang L, et al (2004)
Cloning and characterization of teneurin C-termi-
nus as-sociated peptide (TCAP)-3 from the hypo-
thalamus of an adult rainbow trout (Oncorhynchus
mykiss). Gen Comp Endocrinol 137:205-16.
36. Wang L, Rotzinger S, Chawaf AA, et al (2005)
Teneurin proteins possess a carboxy terminal
sequence with neuromodulatory activity. Brain Res
Mol Brain Res 133:253-65.
37. Al Chawaf A, Xu K, Tan L, Vaccarino FJ, Lovejoy

Intern J Appl Res Vet Med • Vol. 14, No. 1, 2016. 17
DA, Rotzinger S (2007) Corticotropin-releasing
factor (CRF)-induced behaviors are modulated by
intravenous administration of teneurin C-terminal
associated peptide-1 (TCAP-1). Peptides 28:1406-
15.
38. Tan LA, Xu K, Vaccarino FJ, Lovejoy DA,
Rotzinger S (2008) Repeated intracerebral teneurin
C-terminal associated peptide (TCAP)-1 injections
produce enduring changes in behavioral responses
to corticotropin-releasing factor (CRF) in rat mod-
els of anxiety. Behav Brain Res 188:195-200.
39. Lovejoy DA, Rotzinger S, Barsyte-Lovejoy D
(2009) Evolution of complementary peptide sys-
tems: teneu-rin C-terminal-associated peptides and
corticotropin-releasing factor superfamilies. Ann N
Y Acad Sci 1163:215-20.
40. Lindblad-Toh K, Wade CM, Mikkelsen TS, et al.
(2005) Genome sequence, comparative analysis
and hap-lotype structure of the domestic dog.
Nature 438:803-819.
41. Nathan C, Rolland Y (1987) Pharmacological treat-
ments that affect CNS activity: serotonin. Annals
of the New York Academy of Sciences 499:277-
296.
42. Insel TR, Mueller EA, Alterman I, Linnoila M, and
Murphy DL (1985) Obsessive-compulsive disorder
and serotonin: is there a connection? Biological
Psychiatry 20:1174-1188.
43. Goddard AW, Shekhar A, Whiteman AF, Mc-
Dougle CJ (2008) Serotoninergic mechanisms in
the treat-ment of obsessive--compulsive disorder.
Drug Discovery Today 13:325-332.
44. Brandl EJ, Müller DJ, and Richter MA (2012)
Pharmacogenetics of obsessive-compulsive disor-
ders. Phar-macogenomics 13:71-81.
45. Kellner M (2010) Drug treatment of obsessive-
compulsive disorder. Dialogues Clin Neurosci
12:187-97.
46. Niesler B, Walstab J, Combrink S, et al (2007)
Characterization of the novel human serotonin
receptor sub-units 5-HT3C,5-HT3D, and 5-HT3E.
Mol Pharmacol 72:8-17.
47. White SW, Oswald D, Ollendick T, and Scahill L
(2009) Anxiety in children and adolescents with
autism spectrum disorders. Clinical Psychology
Review 29:216-229.
48. Lennertz L, Wagner M, Grabe HJ, et al (2014)
5-HT3 receptor inuences the washing phenotype
and visu-al organization in obsessive-compulsive
disorder supporting 5-HT3 receptor antagonists as
novel treatment option. Eur Neuropsychopharma-
col 24:86-94.
49. Denys D, Van Nieuwerburgh F, Deforce D, West-
enberg HG (2006) Association between seroto-
nergic can-didate genes and specic phenotypes
of obsessive compulsive disorder. J Affect Disord
91:39-44.
50. Jones BJ, Costall B, Domeney AM, et al (1988)
The potential anxiolytic activity of GR38032F,
a 5-HT3-receptor antagonist. British Journal of
Pharmacology 93:985-993.
51. Glatzle J, Sternini C, Robin C, Zittel TT, Wong H
(2002) Expression of 5-HT3 receptors in the rat
gastroin-testinal tract. Gastroenterology 123:217-
226.
52. Gros DF, Antony MM, McCabe RE, Swinson RP
(2009) Frequency and severity of the symptoms
of irrita-ble bowel syndrome across the anxiety dis-
orders and depression. J Anxiety Disord 23:290-6.
53. Masand PS, Keuthen NJ, Gupta S, Virk S, Yu-Siao
B, Kaplan D (2006) Prevalence of irritable bowel
syn-drome in obsessive-compulsive disorder. CNS
Spectr 11:21-5.
54. North CS, Napier M, Alpers DH, Spitznagel EL
(1995) Complaints of constipation in obsessive-
compulsive disorder. Ann Clin Psychiatry 7:65-70.
55. Szeszko PR, Robinson D, Alvir JM, et al (1999)
Orbital frontal and amygdala volume reductions
in obses-sive-compulsive disorder. Arch Gen Psy-
chiatry 56:913-9.
56. Atmaca M, Yildirim H, Ozdemir H, Tezcan E,
Poyraz AK (2007) Volumetric MRI study of key
brain re-gions implicated in obsessive-compulsive
disorder. Prog Neuropsychopharmacol Biol Psy-
chiatry 31:46-52.
57. Tecott LH, Maricq AV, and Julius D (1993) Ner-
vous system distribution of the serotonin 5-HT3
receptor mRNA. Proceedings of the National
Academy of Sciences 90:1430-1434.
58. Tecott L, Shtrom S, and Julius D (1995) Expres-
sion of a serotonin-gated ion channel in embryonic
neural and nonneural tissues. Mol Cell Neurosci
6:43-55.
59. Sellers EM, Toneatto T, Fomach MK, Somer GR,
Sogell LC, Sobell MB (1994) Clinical efcacy of
the 5-HT3 antagonist ondansetron in alcohol abuse
and dependence. Alcohol Clin Exp Res 18:879-85.
60. Johnson BA, Campling GM, Grifths P, Cowen PJ
(1993) Attenuation of some alcohol-induced mood
changes and the desire to drink by 5-HT3 receptor
blockade: a preliminary study in healthy male
volun-teers. Psychopharmacology (Berl) 112:142-
4.
61. Robbins TW, Gillan CM, Smith DG, de Wit S,
Ersche KD (2012) Neurocognitive endophenotypes
of im-pulsivity and compulsivity: towards dimen-
sional psychiatry. Trends Cogn Sci 16:81-91.
62. Hewlett WA, Schmid SP, Salomon RM (2003)
Pilot trial of ondansetron in the treatment of 8
patients with obsessive compulsive disorder. J Clin
Psychiatry 64 (9): 1025–30.
63. Witthauer C, T Gloster A, Meyer AH, Lieb R
(2014) Physical diseases among persons with
obsessive com-pulsive symptoms and disorder: a
general population study. Soc Psychiatry Psychiatr
Epidemiol. (2014 Jun 8. Epub ahead of print}
64. Cunill R, Castells X, Simeon D (2009) Relation-
ships between obsessive-compulsive symptomatol-
ogy and severity of psychosis in schizophrenia:
a systematic review and meta-analysis. J Clin
Psychiatry 70:70-82.
65. Potkin SG, Macciardi F, Guffanti G, et al (2010)
Identifying gene regulatory networks in schizo-
phrenia. Neuroimage 53:839-847.
66. Fatemi SH, Folsom TD, Rooney RJ, Thuras
PD (2013) Expression of GABAA α2-, β1- and

Vol. 14, No.1, 2016 • Intern J Appl Res Vet Med.
18
ε-receptors are altered signicantly in the lateral
cerebellum of subjects with schizophrenia, major
depression and bipolar disorder. Transl Psychiatry
3:e303.
67. Ciranna L (2006) Serotonin as a modulator of
glutamate- and GABA-mediated neurotransmis-
sion: implica-tions in physiological functions and
in pathology. Curr Neuropharmacol 4:101-14.
68. Sherman BL, Mills DS (2008) Canine anxieties
and phobias: an update on separation anxiety and
noise aversions. Vet Clin North Am Small Anim
Pract 38:1081-106, vii.
69. Overall KL (2000) Natural animal models of hu-
man psychiatric conditions: assessment of mecha-
nism and validity. Prog Neuropsychopharmacol
Biol Psychiatry 24:727-76.
70. Tan LA, Al Chawaf A, Vaccarino FJ, Boutros PC,
Lovejoy DA (2011) Teneurin C-terminal associated
pep-tide (TCAP)-1 modulates dendritic morpholo-
gy in hippocampal neurons and decreases anxiety-
like behav-iors in rats. Physiol Behav 104:199-204.
71. Chen Y, Xu M, De Almeida R, Lovejoy DA (2013)
Teneurin C-terminal associated peptides (TCAP):
modulators of corticotropin-releasing factor (CRF)
physiology and behavior. Front Neurosci 7:166.
72. Van’t Veer A, Yano JM, Carroll FI, Cohen BM,
Carlezon WA (2012) Corticotropin-releasing factor
(CRF)-induced disruption of attention in rats is
blocked by the κ-opioid receptor antagonist JDTic.
Neuropsycho-pharmacology 37:2809-16.
73. Kupferschmidt DA, Lovejoy DA, Rotzinger S, Erb
S (2011) Teneurin C-terminal associated peptide-1
blocks the effects of corticotropin-releasing fac-
tor on reinstatement of cocaine seeking and on
cocaine-induced behavioural sensitization. Br J
Pharmacol 162:574-83.
74. Wald R, Dodman N, Shuster L (2009) The com-
bined effects of memantine and uoxetine on an
animal model of obsessive compulsive disorder.
Exp Clin Psychopharmacol 17:191-7.
75. Rendon RA, Shuster L, Dodman NH (2001) The
effect of the NMDA receptor blocker, dextro-
methorphan, on cribbing in horses. Pharmacol
Biochem Behav 68:49-51.
76. Neff MW, Beck JS, Koeman JM, et al (2012)
Partial deletion of the sulfate transporter SLC13A1
is associ-ated with an osteochondrodysplasia in the
Miniature Poodle breed. PLoS One 7:e51917.
77. Purcell S, Neale B, Todd-Brown K, et al (2007)
PLINK: a tool set for whole-genome associa-
tion and popu-lation-based linkage analyses. The
American Journal of Human Genetics 81:559-575.
78. Churchill GA, and Doerge RW (1994) Empirical
threshold values for quantitative trait mapping.
Genetics 138:963-971.
79. Stephens M, and Donnelly P (2003) A comparison
of bayesian methods for haplotype reconstruction
from population genotype data. Am J Hum Genet
73:1162-9.
80. Zaharia M, Bolosky WJ, Curtis K, et al (2011)
Faster and more accurate sequence alignment with
SNAP. arXiv preprint arXiv:1111.5572.
81. Li H, Handsaker B, Wysoker A, et al (2009) The
Sequence Alignment/Map format and SAMtools.
Bioin-formatics 25:2078-9.
82. Cingolani P, Platts A, Wang le L, et al (2012) A
program for annotating and predicting the effects
of single nucleotide polymorphisms, SnpEff: SNPs
in the genome of Drosophila melanogaster strain
w1118; iso-2; iso-3. Fly (Austin) 6:80-92.
83. Ng PC, and Henikoff S (2003) SIFT: Predicting
amino acid changes that affect protein function.
Nucleic Acids Research 31:3812-3814.
84. Siepel A, and Haussler D (2005) Phylogenetic
hidden Markov models. Statistical Methods in
Molecular Evolution: (Springer), pp 325-351.