LETTERS

Highly effective SNP-based association mapping and

© 2008 Nature Publishing Group [Link]

management of recessive defects in livestock

Carole Charlier1, Wouter Coppieters1, Frédéric Rollin2, Daniel Desmecht3, Jorgen S Agerholm4,
Nadine Cambisano1, Eloisa Carta5, Sabrina Dardano6, Marc Dive, Corinne Fasquelle1, Jean-Claude Frennet,
Roger Hanset1,11, Xavier Hubin7, Claus Jorgensen8, Latifa Karim1, Matthew Kent9, Kirsten Harvey5,
Brian R Pearce5, Patricia Simon1, Nico Tama1, Haisheng Nie1,10, Sébastien Vandeputte2, Sigbjorn Lien9,
Maria Longeri6, Merete Fredholm8, Robert J Harvey5 & Michel Georges1

The widespread use of elite sires by means of artificial deficiency affected B0.2% of newborn calves, at an estimated annual
insemination in livestock breeding leads to the frequent cost in the United States of $5 million3. More recently, B25% of
emergence of recessive genetic defects, which cause significant Holstein-Friesian bulls were estimated to be carriers of a missense
economic and animal welfare concerns. Here we show that mutation in the gene encoding the Golgi UDP-N-acetylglucosamine
the availability of genome-wide, high-density SNP panels, transporter SLC35A3, causing complex vertebral malformation
combined with the typical structure of livestock populations, (CVM)4. Genealogical studies determined that both genetic defects
markedly accelerates the positional identification of genes and were likely disseminated in the population by a founder sire of the
mutations that cause inherited defects. We report the fine-scale modern Holstein-Friesian breed: Carlin-M Ivanhoe Bell3,4. It is crucial
mapping of five recessive disorders in cattle and the molecular that the breeding industry implement effective strategies to control
basis for three of these: congenital muscular dystony (CMD) such outbreaks. A path toward that goal is to establish surveillance
types 1 and 2 in Belgian Blue cattle and ichthyosis fetalis in centers to detect emerging defects and centralize samples of affected
Italian Chianina cattle. Identification of these causative animals that can then be used to map and identify causal mutations.
mutations has an immediate translation into breeding practice, The ensuing diagnostic markers could then be used to avoid
allowing marker assisted selection against the defects through at-risk matings.
avoidance of at-risk matings. Toward that goal, we collected samples from affected animals for
five emerging defects in four cattle breeds: congenital muscular
Livestock productivity has improved remarkably over the last 50 years, dystonia 1 (CMD1), congenital muscular dystonia 2 (CMD2) and
to a large extent because of efficient breeding programs rooted in crooked tail syndrome (CTS) in Belgian Blue cattle, renal lipofusci-
quantitative genetics. However, although stringent selection increases nosis (RL) in Holstein-Friesian and Danish Red cattle and ichthyosis
profitability, it may cause drastic reductions in effective population fetalis (IF) in Italian Chianina cattle. All calves with CMD have
size, particularly when artificial insemination is heavily relied upon. episodes of generalized muscle contractures, but careful clinical
For example, although the Holstein-Friesian dairy cattle population examination suggested two distinct phenotypes (CMD1 and
numbers several million individuals worldwide, its effective popula- CMD2). Calves affected by CMD1 show impaired swallowing, fatigue
tion size is in the low hundreds1,2. Because of the accompanying upon stimulation or exercise, and muscle myotonia resulting in an
increase in coancestry and inbreeding, the industry faces regular inability to flex limbs and injurious falling. CMD1 calves usually die
outbreaks of recessive defects. For instance, in the 1990s, B14% of within a few weeks as a result of respiratory complications. CMD2
Holstein-Friesian bulls were shown to carry a missense mutation in calves suffer severe episodes of myoclonus upon acoustic or tactile
gene encoding the leukocyte b2 integrin subunit CD18, causing stimulation and typically die within a few hours of birth. CMD2 is
bovine leukocyte adhesion deficiency (BLAD). This lethal immuno- reminiscent of congenital myoclonus in Hereford cattle caused by a

1Unit of Animal Genomics, GIGA-Research and Department of Animal Sciences, Faculty of Veterinary Medicine University of Liège (B34), 1 Avenue de l’Hôpital,
4000 Liège, Belgium. 2Department of Large Animal Internal Medicine and 3Unit of Pathology, Department of Morphology and Pathology, Faculty of Veterinary
Medicine, University of Liège, 20 Boulevard de Colonster, 4000 Liège, Belgium. 4Department of Veterinary Pathobiology, Faculty of Life Sciences, University of
Copenhagen, 15 Groennegaardsvej, 1870 Frederiksberg C, Denmark. 5Department of Pharmacology, The School of Pharmacy, 29-39 Brunswick Square, London WC1N
1AX, UK. 6Department of Animal Science, Faculty of Veterinary Medicine, University of Milan, 10 Via Celoria, 20133 Milan, Italy. 7Recherche & Développement,
Association Wallonne de l’Elevage, 4 Rue des Champs Elysées, 5590 Ciney, Belgium. 8Division of Genetics and Bioinformatics, Department of Animal and Veterinary
Basic Sciences, Faculty of Life Sciences, University of Copenhagen, 3 Groennegaardsvej, 1870 Frederiksberg C, Denmark. 9Centre for Integrative Genetics,
Department of Animal and Aquacultural Sciences, Norwegian University of Life Sciences, PO Box 5003, 1432 Aas, Norway. 10Present address: Animal Breeding and
Genetics Group, Wageningen University, Wageningen, The Netherlands. 11Deceased. Correspondence should be addressed to M.G. ([Link]@[Link]).

Received 5 December 2007; accepted 10 January 2008; published online 16 March 2008; doi:10.1038/ng.96

NATURE GENETICS VOLUME 40 [ NUMBER 4 [ APRIL 2008 449

 LETTERS

Table 1 Overview of mapping studies

Population Mapping

Defect Breed Casesa Controlsa Log(1/p)b Chrom. Interval Gene

Congenital muscular dystonia 1 (CMD1) Belgian Blue 12 (81) 14 (2,000) 44c 25 2.12 Mb ATPA2A1
Congenital muscular dystonia 2 (CMD2) Belgian Blue 7 (21) 24 (2,000) 44c 29 3.61 Mb SLC6A5
Ichthyosis fetalis (IF) Chianina 3 (3) 9 (96) 3.30d 2 11.78 Mb ABCA12
Crooked tail syndrome (CTS) Belgian Blue 8 (36) 14 (2,000) 44c 19 2.42 Mb –
Renal lipofuscinosis (RL) Holstein Friesian Danish Red 6 (16) 6 (27) 24 (141) 14 44c 17 0.87 Mb –
© 2008 Nature Publishing Group [Link]

aNumbers correspond to sample sizes used to perform the genome-wide scan, whereas the numbers in brackets correspond to the total number of samples available. bHighest genome-wide log(1/p)
value obtained using either ASSHOMc or ASSISTd. –, gene not known.

nonsense mutation in the gene encoding the a1 subunit of the animals7,8. The incidence of RL was estimated at 0.44% in Holstein-
inhibitory glycine receptor5. Despite the high incidence of CMD in Friesian and 2.5% in Danish Red breeds. Genealogical analysis
Belgian Blue cattle (0.1–0.2%), the distinction between the two forms, suggests autosomal recessive inheritance8. Finally, severe forms of IF
as well as their inherited nature, has remained controversial6. CTS is have been reported in Chianina cattle. Affected calves show cutaneous
an emerging syndrome in Belgian Blue cattle. Affected animals have a lesions, with deep fissures separating hyperkeratotic skin plaques and
crooked tail and shortened head, growth retardation, extreme muscu- eversion of mucocutaneous junctions (ectropion and eclabium),
larity and spastic paresia, although some characteristics show variable reminiscent of harlequin ichthyosis in humans9.
penetrance. CTS is not lethal per se, but
causes substantial economic losses due to
growth retardation and treatment. RL is
characterized by the accumulation of lipofus-
a 4
cine granules in secondary lysosomes of renal
proximal tubular epithelial cells, as well as in 3 27
macrophages and in the reticular and

Chrom. no.
Log(1/p)

endothelial cells of kidney-associated lymph 2 18

nodes. Affected animals are identified at
slaughter by their discolored kidneys, justify-
1 9
ing the more common designation of ‘black
kidney disease’. RL is not accompanied by
gross renal dysfunction, yet it may cause
more insidious deleterious effects, as evi-
denced by a reduced longevity of affected b CTRLS BTA25 – 1,380 SNPs

Figure 1 Congenital muscular dystonia 1

(CMD1). (a) Results of genome-wide association
mapping of the CMD1 mutation with 12 affected CASES
animals and 14 controls using ASSHOM (red)
and ASSIST (blue). The black horizontal bars
mark the limits between the 29 autosomes.
Evidence for linkage (y axis) is measured as
log(1/p), with p being determined by 10,000 2.12 Mb
locus (ASSHOM) or phenotype (ASSIST) c
permutations (see Methods). (b) Genotypes of
XPO6

SBK1

SPNS1

ATP2A1
ATXN2L

CLN3

QPRT

SLC5A2

cases and controls for SNPs on chromosome 25.

Homozygous genotypes are shown in yellow or
white, and heterozygous genotypes in red.
Fetal form
Overlapping blocks of extended homozygosity in
the cases are marked in black and gray. The Adult form
limits and size of the haplotype that is shared ATP2A1
Exon 14
homozygous and identical by state among all
Cow MUT GAPEGVIDRCNYVRVGTTRVPMTGPVKEKILSVIKEWGTGRDTLCCLALATRDTPPKREEMVLDDSTKFMEYE
cases is highlighted in red. (c) Above, partial Cow WT GAPEGVIDRCNYVRVGTTRVPMTGPVKEKILSVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDSTKFMEYE
gene content of the shared chromosome segment, Human GAPEGVIDRCNYVRVGTTRVPLTGPVKEKIMAVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDSARFLEYE
Rat GAPEGVIDRCNYVRVGTTRVPLTGPVKEKIMSVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDSAKFMEYE
focusing on the studied candidate genes involved Mouse GAPEGVIDRCNYVRVGTTRVPLTGPVKEKIMSVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDSAKFMEYE
Rabbit GAPEGVIDRCNYVRVGTTRVPMTGPVKEKILSVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDSSRFMEYE
in neuromuscular physiology. Middle, exonic Dog GAPEGVIDRCNYVRVGTTRVPMTGPVKDKILSVIKEWGTGRDTLRCLALATRDTPPKREEMILDDSARFMEYE
Pig GAPEGVIDRCTHIRVGSTKVPMTPGVKQKIMSVIREWGSGSDTLRCLALATHDNPMRREEMNLEDSANFIKYE
Q
composition of the fetal and adult forms of ********** *** ** * ** * ** *** * ********** * * **** * ** * **
ATP2A1 highlighting exon 14, which harbors the Zebrafish GAPEGVIDRCAYVRVGTTRVPLTGPVKDKIMAVIKEWGTGRDTLRCLALATRDNPLRPEEMNLEDSTKFAEYE
Fugu GAPEGVIDRCAYIRVGTARVPLTGPVKDHIMGVIKEWGTGRDTLRCLALATRDTPLRKEEMNLEDSTRFAEYE
mutation encoding R559C. Below, bovine Xenopus GAPEGVIDRCNYVRVGTTRVPLTPAIKDKILTVIKEWGTGRDTLRCLALATRDTPPKREEMVLDDATKFVDYE
Chicken GAPEGVIDRCNYVRVGTTRVPLTPAVKEKILAVIKEWGTGRDTLRCLALATRDTPPKMEDMMLVDSTKFAEYE
sequence and interspecies alignments of the Fly GAPEGVLERCTHARVGTTKVPLTSALKAKILALTGQYGTGRDTLRCLALAVADSPMKPDEMDLGDSTKFYQYE
Worm GAPEGVLGRCTHVRVNGQKVPLTSAMTQKIVDQCVQYGTGRDTLRCLALGTIDTPVSVSNMNLEDSTQFVKYE
region flanking R559C, showing the extreme ****** ** ** ** * * * * ******** * * * * * **
conservation of the Arg559 residue. R559C

450 VOLUME 40 [ NUMBER 4 [ APRIL 2008 NATURE GENETICS

 LETTERS

We performed genome-wide SNP scans on DNA samples from ATP2A1, which encodes a fast-twitch skeletal-muscle Ca2+ ATPase.
affected individuals and controls (Table 1), using either the 25K ATP2A1 (also known as SERCA1) is responsible for pumping Ca2+
Affymetrix SNP panel or a custom-made 60K Illumina panel from the cytosol back into the sarcoplasmic reticulum, thereby
(Supplementary Table 1 and Supplementary Fig. 1 online). The latter inducing muscle relaxation12. Nonsense and splice-site mutations in
was assembled from SNPs in the public domain obtained from the the human ATP2A1 gene cause Brody myopathy (OMIM 601003), a
Bovine Genome Project of the Human Genome Sequencing Center at rare autosomal recessive disorder characterized by exercise-induced
Baylor College of Medicine and from comparing publicly available muscle cramps and impaired muscle relaxation13,14. Arg559 in bovine
BAC end sequences with the bovine reference sequence (W. Coppieters ATP2A1 is located within the nucleotide-binding domain and is a
et al., unpublished data). The assumed order of the SNP markers was highly conserved, functionally significant residue that directly interacts
based on build 4.0 of the Bos taurus genome. (See Methods for URLs with the b-phosphate moiety of ATP (Fig. 1)15. Site-directed muta-
© 2008 Nature Publishing Group [Link]

for these data sources.) To map the corresponding disease loci, we genesis of this residue abolished ATP2A1 function15. Genotyping
developed two heuristic programs detecting overlapping, unusually 2,000 unaffected Belgian Blue animals for the R559C variant did not
long, homozygous chromosome segments among affected cattle. The reveal a single individual homozygous for the Cys559-encoding
first program (ASSHOM) performs homozygosity mapping by search- mutation despite a frequency of the Cys559 allele of 0.075 (P ¼
ing for shared segments of homozygosity that need not be identical by (1 – 0.0752)2,000 ¼ 1.26
105), whereas all 81 confirmed CMD1
state (IBS)10. The second program (ASSIST) performs autozygosity cases were homozygous for the mutation. Taken together, these results
mapping by searching for shared homozygous and IBS chromosome strongly suggest that the R559C substitution is causative for CMD1,
segments among affected cattle. Unique, genome-wide significant disrupting ATP2A1 function by impairing ATP binding.
signals (P r 0.0005) were readily obtained with one or both The 3.61-Mb segment containing the CMD2 locus on bovine
approaches for CMD1 (BTA25), CMD2 (BTA29), CTS (BTA19), RL chromosome 29 encompasses 13 genes, including one strong candi-
(BTA17) and IF (BTA2). These results validated the genetic etiology of date: SLC6A5, encoding the Na+/Cl–-dependent glycine transporter
the five defects and confirmed that CMD1 and CMD2 have distinct (GlyT2), which is responsible for maintaining an abundant pre-
determinants. Closer examination of the autozygous segments defined synaptic pool of neurotransmitter at glycinergic synapses (Supple-
critical regions of 2.12, 3.61, 2.42, 0.87 and 11.78 Mb for CMD1, mentary Fig. 2)16. In humans, missense, nonsense and frameshift
CMD2, CTS, RL and IF, respectively (Table 1, Fig. 1 and Supple- mutations in SLC6A5 underlie autosomal recessive hyperekplexia
mentary Figs. 2–5 online). The 0.87-Mb interval for RL corresponds (OMIM 149400), which is characterized by an exaggerated startle
to a chromosomal segment at which all affected cattle (Holstein- response to tactile or acoustic stimuli resulting in muscle hypertonia
Friesian and Red Danish) are homozygous for the same haplotype. and life-threatening neonatal apnea episodes17, symptoms highly
This observation suggests that RL is allelically homogeneous across reminiscent of CMD2 in Belgian Blue cattle. Sequencing SLC6A5 in
these breeds, and illustrates the potential gain in mapping resolution CMD2 cases revealed a missense mutation (T809C) in exon 4,
from studying the same trait across breeds, as was recently shown resulting in a L270P substitution in the third membrane-spanning
in the dog11. domain of GlyT2. Although the Leu270 residue is not predicted to be
The 2.12-Mb segment encompassing the CMD1 locus on bovine involved in glycine or Na+ binding17, a leucine residue is conserved at
chromosome 25 encompasses 95 transcriptional units, of which 6 were this position in all vertebrates studied, indicative of structural and
plausible candidates because their protein products are known to have functional importance (Fig. 2a,b). We did not find a single individual
functions in neuromuscular physiology (Fig. 1). RT-PCR analyses in homozygous for the Pro270-encoding mutation among 2,000 unaf-
brain, spinal cord and skeletal muscle of affected individuals (cases) fected Belgian Blue animals, despite a frequency of 0.023 for the
and controls did not reveal any striking alterations in gene expression Pro270 allele, whereas all 21 confirmed CMD2 cases were homozygous
for these genes. However, DNA sequencing revealed a missense for the Pro270-encoding allele (P ¼ (1 – 0.0232)2,000 ¼ 0.35). We
mutation (C1676T) leading to a R559C substitution in exon 14 of introduced the corresponding mutation (affecting Leu269 in human

a b c d
MDCSAPKEMNKQPANSLEAAVPSGHHDGPCAPRTSPEQELPAATAAPPPR 50
VPRSASTGAQTFHAADARACEAERPGGAGACKLSSLRSPAASAALRDWSE 100
AHGAQTAPPSGGAGPGNALPCKITALRGLEGDANVSVGKGTLERNNTPAV 150 Cow MUT PAPQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV 1.25 *
GWVNMSQSTVVLGTDGITSVLPGSVATAATQEDEQGDENKARGNWSSKLD 200 Cow WT PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV
(nmol/min per mg protein)

TM1 TM2 Human PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV

FILSMVGYAVGLGNVWRFPYLAFQNGGGAFLIPYLMMLALAGLPIFFLEV 250 1.00
[3H]Glycine uptake

L270P TM3 Chimp PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV

SLGQFASQGPVSVWKAIPALQGCGIAMLIISVLIAIYYNVIICYTLFYLF 300 Macaque PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV
ASFVSVLPWGSCNNPWNTPECKDKTKLLLDSCVINDHPKIQIKNSTFCMT 350 Mouse PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV 0.75
TM4
Rat PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV
mycGlyT2
AYPNLTMVNFTSQANKTFVSGSEEYFKYFVLKISAGIEYPGEIRWPLAFC 400
TM5
LFLAWVIVYASLAKGIKTSGKVVYFTATFPYVVLVILLIRGVTLPGAGAG 450 Dog PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV
TM6 Opossum PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV 0.50
IWYFITPKWEKLTDATVWKDAATQIFFSLSAAWGGLITLSSYNKFHNNCY 500 Chicken PALQGCGIAMLIISVLIAIYYNVIICYTLFYLFASFV
TM7
RDTLIVTCTNSATSIFAGFVIFSVIGFMANERKVNIENVADQGPGIAFVV 550 Xenopus PALQGCGIAMLIISVLIAIYYNIIMCYTIFYLFASLV
TM8
YPEALTRLPLSPFWAIIFFLMLLTLGLDTMFATIETIVTSISDEFPKYLR 600 Zebrafish PALQGCGIAMLIISVLIAIYYNIIMCWTLYYLFASLK 0.25
TM9 TM10
THKPVFTLGCCVCFFIMGFPMITQGGIYMFQLVDTYAASYALVIIAIFEL 650 Fugu PALQGCGIAMLIISVLIAIYYNIIMCWTLYYLFASLK
TM11 ********************** * * * ***** mycGlyT2L269P 0
VGISYVYGLQRFCEDIEMMIGFQPNIFWKVCWAFVTPTILTFILCFSFYQ 700
r
T2

TM12
o

750
69

69

WEPMTYGSYRYPNWSMVLGWLMLACSVIWIPIMFVIKMHLAPGRFIERLK
ct

ly
Ve

L2

L2
G

LVCSPQPDWGPFLAQHRGERYKNMIDPLGTSSLGLKLPVKDLELGTQC 798

Figure 2 Congenital muscular dystonia 2 (CMD2). (a) Sequence of bovine GlyT2 protein highlighting the 12 membrane-spanning domains (TM1–12), the
predicted Na+ (blue triangles) and glycine (black circles) binding residues and the L270P mutation (black shading). (b) Sequence alignment of TM3 showing
its remarkable conservation down to amphibians and fish, including the Leu270 residue. MUT, L270P mutant; WT, wild type. (c) Confocal images of
transfected HEK293 cells showing membrane localization of myc-hGlyT2 and the myc-hGlyT2L269P mutant counterstained with the nuclear marker
TOPRO3. (d) Glycine uptake in HEK293 cells transiently transfected with pRKmyc-hGlyT2 and the pRKmyc-hGlyT2L269P mutant after 5 min incubation
with 200 mM [3H]glycine. Data are means ± s.e.m. (n ¼ 6–20). Statistical comparisons were made using an unpaired Student’s t-test. *, significantly
different from vector control.

NATURE GENETICS VOLUME 40 [ NUMBER 4 [ APRIL 2008 451

 LETTERS

ABCA12 ortholog of all sequenced vertebrates (including amphibians

a and fish) as well as in 7 of the 11 human ABCA transporters,
underlining its key structural and functional importance (Fig. 3).
Genotyping 96 healthy Chianina cattle identified two individuals
that were heterozygous for the His1935- and the Arg1935-encoding
alleles, whereas the three IF cases were homozygous for the Arg1935-
encoding allele. The Arg1935 allele was not found in other breeds
(n ¼ 60). Again, these findings support the disease causality of the
H1935R mutation.
b ABCA12 We have shown that using SNP arrays with a marker density on the
© 2008 Nature Publishing Group [Link]

H2N EC1 NBD1 IC3 EC4 NBD2 -COOH order of one SNP per 55 kb allows highly effective localization of
mutations causing inherited defects in livestock. It is noteworthy that
neither CMD2, RL nor IF could be mapped by linkage analysis or
Cow MUT
Cow WT association studies using a panel of 400 microsatellite markers and the
Mouse
Horse
Dog
available pedigree material (data not shown). Moreover, we show that,
Macaque
Human after having determined the map location, it is possible to rapidly
Armadillo
Orthologs
Opossum
identify the causal mutations in a substantial proportion of cases.
Platypus
Lizard Indeed, we have located the genes and mutations responsible for
Chicken
Fish
Xenopus
CMD1 (ATP2A1) and CMD2 (SLC6A5) in Belgian Blue cattle, and for
ichthyosis fetalis (ABCA12) in Italian Chianina, defining new missense
HS-ABCA12
HS-ABCA13
mutations that disrupt protein function and identifying the corre-
HS-ABCA1
HS-ABCA7 sponding human disorders (Brody myopathy, hyperekplexia and
HS-ABCA4
HS-ABCA2
HS-ABCA3
harlequin ichthyosis). The 2.42- and 0-87 Mb intervals encompassing
Paralogs HS-ABCA6
HS-ABCA10 the mutations causing CTS and RL are presently being screened
HS-ABCA8
HS-ABCA9
HS-ABCA5
to identify the corresponding causative mutations (Supplementary
Figs. 4 and 5).
Figure 3 Ichthyosis fetalis (IF). (a) Cutaneous lesions and eversion These advances have important practical consequences, as they
of mucocutaneous junctions in a Chianina IF case. (b) Schematic allow the rapid control of emerging recessive defects with otherwise
representation of the structure of the ABCA12 protein showing the position major economic and animal welfare implications. This is well illu-
of the transmembrane helices (green) and the nucleotide-binding domains
strated by the fact that the diseases CMD1 and CMD2 were essentially
(blue). The position of the H1935R substitution associated with IF in
Chianina cattle is marked by the red star. The largest extracellular (EC) and
eliminated from the Belgian Blue breed within a matter of months
intracellular (IC) loops are numbered. Coordinates are according to ref. 25. after the discovery of the causal mutations. Virtually all artificial
(c) Clustal-W alignments of the amino acid sequences flanking the H1935R insemination sires were genotyped for both defects (revealing that
mutation among sequenced vertebrate ABCA12 orthologs and human ABCA B11.4% were carriers of CMD1, 4.6% of CMD2 and 0.6% of both, in
paralogs. MUT, H1935R mutant; WT, wild type. good agreement with reported CMD incidence) and the genotypes
published. The test was offered to breeders, allowing them to test
their cows and private bulls and combine the information to avoid
GlyT2) into the expression construct pRK5myc-hGlyT2 (ref. 17) by at-risk matings. Carrier animals are not systematically culled, as
site-directed mutagenesis, and analyzed two independent mutant their overall breeding worth may outweigh the economic penalty of
constructs for subcellular localization and [3H]glycine uptake in carrier status.
HEK293 cells. Consistent with the effect of most missense mutations Domestic animal populations are a treasure trove of inherited
in human GlyT2 (ref. 17), this substitution did not affect membrane phenotypic variability as the result of centuries of selection for
trafficking, but abolished [3H]glycine uptake (Fig. 2c,d). Taken useful or fashionable attributes. Our work, as well as comparable
together, these results demonstrate that the L270P mutation is results obtained in dogs11 and undoubtedly soon to be obtained
causative for CMD2 and disrupts presynaptic uptake of glycine in other species, shows that the analysis of domestic animal
by GlyT2. populations has the potential to make an extremely valuable con-
The three IF cases available were closely related; hence, the shared tribution toward unraveling genotype-phenotype relationships and
segment on bovine chromosome 2 containing the IF mutation was genome function.
large, 11.78 Mb (Supplementary Fig. 3). However, we noted that the
orthologous region on human chromosome 2 contains ABCA12, METHODS
which encodes an ATP-binding cassette (ABC) transporter found in Genotyping. Genotyping using the custom-made 60K iSelect panel as well as
the lamellar granules of keratinocytes, where it is thought to have an the 25K Affymetrix array was performed using standard procedures as
important role in the regulation of lipid trafficking18. Missense recommended by the respective manufacturers. CMD1, CTS, RL and IF cases
mutations in the first ATP-binding domain of human ABCA12 and controls were genotyped with the Illumina 60K panel, whereas CMD2 cases
cause lamellar ichthyosis type 2 (ref. 19), whereas nonsense, frameshift were genotyped with the Affymetrix 25K panel.
and in rare cases missense mutations have been shown to cause the
Association analysis. ASSHOM first scans the genome of each of the N studied
more severe harlequin ichthyosis (OMIM 242500)20–24, which mark- cases for blocks of adjacent homozygous markers. Markers within such blocks
edly resembles IF. We therefore sequenced ABCA12 in the affected receive a score sij corresponding to the sum of –log(p2i ) over all markers in the
cases and controls and identified a missense mutation in exon 39 block. In this, pi is the frequency (estimated in controls) of the allele for which
(A5804G) resulting in an H1935R substitution in the fourth extra- case i is homozygous and j is the marker number (out of M markers). Thus, the
cellular loop. The corresponding His1935 residue is conserved in the longer and rarer the homozygous haplotype, the higher the score. All markers

452 VOLUME 40 [ NUMBER 4 [ APRIL 2008 NATURE GENETICS

 LETTERS

for which individual i is heterozygous receive the same 105 score. ASSHOM Functional analysis of GlyT2. [3H]Glycine uptake assays were performed as
then computes a summary score at each marker position, Sj, corresponding to described17 in HEK293 cells (ATCC CRL1573) transfected with 0.25 mg total
the harmonic mean of the N corresponding sij scores. If one or more of the pRK5myc (vector control), pRK5myc-hGlyT2 or the L269P mutant using
cases are heterozygous at the corresponding marker position, their 105 score Lipofectamine LTX (Invitrogen). After 24 h, cells were washed with a buffer
will drastically penalize Sj. The statistical significance of each Sj is estimated containing 116 mM NaCl, 1 mM NaH2PO4, 26 mM NaHCO3, 1.5 mM MgSO4,
from its rank with respect to a list of ‘best out of M’ Pj scores calculated by 5 mM KCl, 1.3 mM CaCl2 and 5 mM glucose pregassed with 5% CO2/95% air
‘locus permutation’. Locus permutation scores Pj are computed as the harmonic and then incubated for 5 min in 1 mCi/ml [3H]glycine (60 Ci/mmol, NEN) at a
mean of N randomly selected sij scores (one per case). The corresponding final concentration of 200 mM. Uptake was terminated by two additions of ice-
P values are thus corrected for the testing of M marker loci and account for the cold buffer followed by aspiration. Cells were digested in 0.1 M NaOH and used
level of inbreeding of the cases. ASSIST scans the genome for core markers for scintillation counting and determination of protein concentration using the
defined as polymorphic in controls yet monomorphic in cases. For each Bradford reagent (Bio-Rad). [3H]Glycine uptake was calculated as nmol/min
© 2008 Nature Publishing Group [Link]

core marker, ASSIST identifies a proximal reference case with the longest per mg protein and expressed as a percentage of the pRK5myc-transfected
contiguous stretch of homozygous markers proximal from the core, and a distal control. All statistical comparisons used an unpaired Student’s t-test. For GlyT2
reference case with the longest contiguous stretch of homozygous markers immunostaining, transfected cells were washed twice in PBS, fixed for 5 min in
distal from the core. Proximal (or, respectively, distal) scores are then computed 4% (w/v) paraformaldehyde in PBS and quenched in 50 mM NH4Cl for 10 min
as the sum of –log(p2i ) over contiguous (starting from the core) marker before being permeabilized in 0.01% (w/v) Triton X-100, further washed in
stretches shared IBS with the proximal (respectively, distal) reference. In this, PBS and stained with an anti-Myc primary antibody and Alexa Fluor 488–
pi is the frequency in controls of the allele for which the reference case is conjugated secondary antibodies (Invitrogen). Confocal microscopy was per-
homozygous. Proximal and distal scores are summed to yield an overall score formed using a Zeiss Axioscop LSM 510 Meta confocal microscope with
40
Sj. Their statistical significances are determined by ‘phenotype permutation’: objective (NA 1.3) and
3 digital zoom, capturing images at 1,024
1,024–
disease status is randomly redistributed amongst cases and controls, genome- pixel resolution.
wide association conducted with ASSIST and the ‘best out of M’ phenotype
permutation scores Pj scores stored. The P value of real Sjs are then estimated URLs. 25K Affymetrix bovine SNP panel, [Link]
from their rank with respect to a list of such ‘best out of M’ Pj permutation products/arrays/specific/[Link]; Bovine Genome Project of the Human
scores. ASSIST thus accounts for multiple-marker testing and allelic frequencies Genome Sequencing Center at Baylor College of Medicine, including build 4.0
in controls. of the Bos taurus genome, [Link]
Both programs assume that the analyzed trait has simple, monogenic
Note: Supplementary information is available on the Nature Genetics website.
recessive inheritance (incomplete penetrance is not accommodated). When
samples originating from distinct breeds are analyzed together, ASSHOM ACKNOWLEDGMENTS
assumes allelic homogeneity within but not necessarily between breeds, whereas The 60K genome-wide SNP panel was constructed with the financial support of
ASSIST assumes allelic homogeneity both within and between breeds. Holland Genetics (Arnhem, The Netherlands). This work was funded by grants
ASSHOM and ASSIST are available from the authors on request. from the Walloon Ministry of Agriculture, from the Belgian Science Policy
organization (SSTC Genefunc PAI), from the Communauté Française de Belgique
Resequencing. For each disease, at least one affected animal and an unrelated (Game & Biomod ARC), from the University of Liège and from the Medical
wild-type counterpart, matched for age and breed, were chosen after Research Council (G0601585 to K.H. and R.J.H.). C.C. is Chercheur Qualifié
from the Fonds National de la Recherche Scientifique. This work is dedicated to
careful phenotyping. They were slaughtered following ethical procedures.
the memory of Professor Roger Hanset.
Total blood was withdrawn for DNA extraction and tissues samples
were collected directly after the animal’s death, rapidly frozen in liquid nitrogen
AUTHOR CONTRIBUTIONS
and stored at –80 1C until RNA extraction. The tissue panel was disease C.C. and M.G. designed the experiments, analyzed data, wrote the manuscript
specific: skeletal muscle, spinal cord and brain for CMD1 and CMD2; cartilage, and supervised the project. W.C. designed the Illumina iSelect panel and
skeletal muscle and spinal cord for CTS; kidney and skeletal muscle for RL; and supervised genotyping. F.R., D.D., J.S.A., M.D., J.-C.F., R.H., C.J., S.V., M.L. and
skin for IF. M.F. confirmed diagnosis and provided samples of affected individuals. S.L. and
Total RNA was extracted from tissue samples after homogenization in M.K. performed genotyping on the Affymetrix panel. N.C., S.D., C.F., X.H., L.K.,
TRIzol Reagent (Invitrogen) following the manufacturer’s instructions. Total P.S., N.T. and H.N. performed genotyping and mutation scanning. X.H. analyzed
RNA was ethanol precipitated, resuspended in DEPC-treated water and diluted data and mapped the CMD1 mutation. C.J. and M.F. performed mutation
at 1 mg/ml. First-strand cDNA was synthesized with the Superscript First-Strand scanning of RL. E.C., K.H., B.R.P. and R.J.H. performed functional test of the
CMD2 mutation. R.J.H. participated in writing the manuscript.
Synthesis System for RT-PCR (Life Technologies) using a mixture of random
hexamers and oligo(dT) primers. The cDNA was directly amplified by PCR
Published online at [Link]
with gene-specific primers designed to allow full-length ORF amplification.
Reprints and permissions information is available online at [Link]
Primer pairs and PCR conditions for all the positional candidate genes screened reprintsandpermissions
are listed in Supplementary Table 2 online. PCR products were loaded on 1%
agarose gels to compare band intensities and sizes between affected and control
1. Young, C.W. & Seykora, A.J. Estimates of inbreeding and relationship among registered
animals. DNA fragments were cut and purified with the GENECLEAN Kit Holstein females in the United States. J. Dairy Sci. 79, 502–505 (1996).
(Qbiogene). Purified fragments were sequenced using the Big Dye terminator 2. Hayes, B.J., Visscher, P.M., McPartlan, H.C. & Goddard, M.E. Novel multilocus
cycle-sequencing kit (Applied Biosystems). Electrophoresis of purified sequen- measure of linkage disequilibrium to estimate past effective population size. Genome
cing reactions was performed on an ABI PRISM 3730 DNA analyzer Res. 13, 635–643 (2003).
3. Shuster, D.E., Kehrli, M.E. Jr., Ackermann, M.R. & Gilbert, R.O. Identification and
(PE Applied Biosystems). Multiple sequence traces from affected and wild- prevalence of a genetic defect that causes leukocyte adhesion deficiency in Holstein
type animals were aligned and compared using the phredPhrap program from cattle. Proc. Natl. Acad. Sci. USA 89, 9225–9229 (1992).
the Consed package. Pathogenic relevance of each mutation was established by 4. Thomsen, B. et al. A missense mutation in the bovine SLC35A3 gene, encoding a UDP-
aligning the protein sequence from available species. N-acetylglucosamine transporter, causes complex vertebral malformation. Genome
Res. 16, 97–105 (2006).
Genomic DNA was extracted from total blood through a standard phenol- 5. Pierce, K.D. et al. A nonsense mutation in the alpha 1 subunit of the inhibitory
chloroform extraction procedure. Each animal enrolled for the resequencing glycine receptor associated with bovine myoclonus. Mol. Cell. Neurosci. 17, 354–363
study was first genotyped for SNP markers from the minimum genetic interval (2001).
to asses its IBD (affected) or non- IBD (wild-type counterpart) status. Finally, 6. Vandeputte, S. et al. Congenital muscular dystonia (CMD): a new congenital pathology
in Belgian Blue calves. in Proceedings of the 24th World Buiatrics Congress, Nice,
all the available individuals (Table 1) were genotyped for the respective disease 15–19 October 2006 (ref. OS05–2).
mutations by PCR amplification of the culprit exon followed by resequencing. 7. Agerholm, J.S. Inherited disorders in Danish cattle. APMIS Suppl. 122, 1–76
Primer pairs for each mutation are listed in Supplementary Table 2. (2007).

NATURE GENETICS VOLUME 40 [ NUMBER 4 [ APRIL 2008 453

 LETTERS

8. Rude, H., Agerholm, J.S., Maddox-Hyttel, P., Christensen, K. & Flagstad, P. Renal 17. Rees, M.I. et al. Mutations in the human GlyT2 gene define a presynaptic component
lipofuscinosis in Danish slaughter cattle. J. Comp. Pathol. 132, 303–312 (2005). of human startle disease. Nat. Genet. 38, 801–806 (2006).
9. Molteni, L. et al. Ichthyosis in Chianina cattle. Vet. Rec. 158, 412–414 (2006). 18. Akiyama, M. Pathomechanisms of harlequin ichthyosis and ABCA transporters in
10. Lander, E.S. & Botstein, D. Homozygosity mapping: a way to map human recessive human diseases. Arch. Dermatol. 142, 914–918 (2006).
traits with the DNA of inbred children. Science 236, 1567–1570 (1987). 19. Lefévre, C. et al. Mutations in the transporter ABCA12 are associated with lamellar
11. Karlsson, E.K. et al. Efficient mapping of mendelian traits in dogs through genome- ichthyosis type 2. Hum. Mol. Genet. 12, 2369–2378 (2003).
wide association. Nat. Genet. 39, 1321–1328 (2007). 20. Kelsell, D.P. et al. Mutations in ABCA12 underlie the severe congenital skin disease
12. Toyoshima, C. & Inesi, G. Structural basis of ion pumping by Ca2+-ATPase of the harlequin ichthyosis. Am. J. Hum. Genet. 76, 794–803 (2005).
sarcoplasmic reticulum. Annu. Rev. Biochem. 73, 269–292 (2004). 21. Akiyama, M. et al. Mutations in lipid transporter ABCA12 in harlequin ichthyosis and
13. Odermatt, A. et al. Mutations in the gene encoding SERCA1, the fast-twitch skeletal functional recovery by corrective gene transfer. J. Clin. Invest. 115, 1777–1784
muscle sarcoplasmic reticulum Ca2+ ATPase, are associated with Brody disease. Nat. (2005).
Genet. 14, 191–194 (1996). 22. Thomas, A.C. et al. ABCA12 is the major harlequin ichthyosis gene. J. Invest.
14. Berchtold, M.W., Brinkmeier, H. & Müntener, M. Calcium ion in skeletal muscle: its Dermatol. 126, 2408–2413 (2006).
crucial role for muscle function, plasticity, and disease. Physiol. Rev. 80, 1215–1265 23. Akiyama, M. et al. Compound heterozygous ABCA12 mutations including a novel
© 2008 Nature Publishing Group [Link]

(2000). nonsense mutation underlie harlequin ichthyosis. Dermatology 215, 155–159 (2007).
15. Ma, H., Lewis, D., Xu, C., Inesi, G. & Toyoshima, C. Functional and structural roles of 24. Thomas, A.C. et al. Novel and recurring ABCA12 mutations associated with harlequin
critical amino acids within the ‘‘N’’, ‘‘P’’, and ‘‘A’’ domains of the Ca2+ ATPase (SERCA) ichthyosis: implications for prenatal diagnosis. Br. J. Dermatol. 158, 611–613.
headpiece. Biochemistry 44, 8090–8100 (2005). 25. Peelman, F. et al. Characterization of the ABCA transporter subfamily: identification of
16. Gomeza, J. et al. Deletion of the mouse glycine transporter 2 results in a hyperekplexia prokaryotic and eukaryotic members, phylogeny and topology. J. Mol. Biol. 325,
phenotype and postnatal lethality. Neuron 40, 797–806 (2003). 259–274 (2003).

454 VOLUME 40 [ NUMBER 4 [ APRIL 2008 NATURE GENETICS