The Molecular Genetics of Exercise Induced Collapse in Labrador Retrievers
ACVIM 2008
Edward (Ned) E. Patterson, DVM, PhD, DACVIM (SAIM); Katie M. Minor, BA, RN; James R. Mickelson, PhD
St. Paul, MN, USA


Exercise-induced collapse (EIC) is a recently recognized disorder of increasing significance in Labrador Retrievers, especially those dogs used for hunting and field trials. Dogs affected with EIC develop muscle weakness, incoordination and life-threatening collapse after just five to fifteen minutes of intense exercise or excitement, and cannot participate in many types of strenuous activities. The condition also unknowingly exists in dogs that are not routinely participating in such activities. Affected dogs can tolerate mild to moderate exercise, but after 5 to 15 minutes of strenuous exercise they develop weakness, apparent incoordination, and then collapse. Dogs do not seem to be painful during or after the collapse, and are usually anxious to continue participating in the trigger activity.1 After 10 to 30 minutes of rest, most dogs return to normal, but some dogs have died during an episode of EIC. The rectal temperature of dogs during the EIC episode typically reaches 41.7oC (107oF) or higher, a finding that originally lead to some confusion in the literature as to whether this syndrome is a form of exercise-induced malignant hyperthermia.2 It has been documented, however, the post-exercise rectal temperature is no higher, and the rate of cooling is no lower in EIC-affected dogs than unaffected Labrador Retrievers.1

A comprehensive investigation of EIC in Labrador Retrievers has been ongoing for nearly a decade, involving investigators from the University of Minnesota (EE Patterson, JR Mickelson, KM Minor), the University of Saskatchewan (SM Taylor, CL Shmon), and the Comparative Neuromuscular Unit at the University of California (GD Shelton). The objectives of this research have been to 1) describe the syndrome so that it can be recognized by dog owners, veterinarians and trainers, 2) to thoroughly evaluate affected dogs to try to establish an efficient means of diagnosis and to gain some insight into the cause of collapse and 3) to determine the mode of inheritance and the genetic basis for EIC, and 4) to develop a DNA test for the condition. Objectives 1 and 2 are described in detail in the talk immediately preceding this.1 Here-in are the results of objectives 3 and 4.


Multi-generation pedigrees of Labrador Retrievers affected with EIC were identified. Criteria for EIC were: Presumed EIC affected; Dogs with a well-documented history of more than one typical collapse episode in which the pelvic limbs became ataxic and then flaccid, consistent with the criteria of Taylor et al.,1 and failure to identify a systemic, metabolic, cardiac or neurologic reason for the collapse and Unaffected; Dogs never observed to collapse. DNA and medical information was obtained for 275 Labs (143 affected and 132 unaffected) with 8 extended family pedigrees identified.


All eight pedigrees could be made into one very large extended pedigree. Examination of the pedigrees established that many affected dogs were closely related. Within these 8 pedigrees, there were 16 matings that resulted in two or more affected offspring. Males and females were equally represented (51%:49%), excluding an X-linked mode of inheritance. Only six affected dogs were known to have an affected parent. From these affected parents there is one family with full phenotypic information in which one affected parent mated to an unaffected parent produced two affected and two unaffected offspring. There are three families in which an affected dog produced multiple (in one case 15) affected second and third generation offspring. The pedigree analysis was most consistent with an autosomal recessive mode of inheritance, although a dominant disorder with partial penetrance or a polygenic disorder could not be excluded.

Exclusion of the Malignant Hyperthermia Gene

Fourteen clearly affected dogs were genotyped for the V547A Malignant Hyperthermia (MH) ryanodine receptor (RYR1) mutation on canine chromosome 1 using a protocol described by Roberts.3 DNA from dogs with MH known to contain the RYR1 mutation served as a positive control. Six canine microsatellite genetic markers adjacent to the RYR1 locus on canine chromosome 1 were genotyped. Significant evidence for linkage is a Logarithm of Odds (LOD) score > 3.3 and exclusion would be indicated by a LOD score less than -2.0. The dogs with EIC did not have the known RYR1 MH mutation and no chromosome one markers near the RYR1 gene locus tested showed linkage to EIC. Marker FH2598 excluded linkage to EIC with a LOD score of -2.1 or less up to 20 centimorgans, which includes the region containing the RYR1 gene.

Whole Genome Scan

A whole genome scan with 444 microsatellite genetic markers was performed. A locus for the EIC gene on canine chromosome 9 (CFA09) was identified based on significant linkage to multiple markers, with a maximal LOD score for linkage of 12.2.

CFA09 Genetic Markers

57 single nucleotide polymorphism (SNP) markers within the 56-61 Mb region of CFA09 were genotyped and the allele frequencies of presumed-affected dogs was compared to that of unaffected dogs. This analysis revealed 18 SNPs in the region with chi-square p-values < 0.001, with the lowest p-value of 1.17 x10-11. These results confirmed that this CFA09 locus contains the EIC gene.

Gene Sequencing and Mutation Identification

Four positional candidate genes were analyzed, and a missense mutation in the Dynamin 1 gene (DNM1) that is very strongly associated (chi-square p-value < 10-15) with EIC was identified. This Arginine256Leucuine (R256L) amino acid substitution was deemed a compelling candidate causal mutation due to the critical central nervous system function and remarkable evolutionary conservation of the DNM1 amino acid sequence (99.5% identity between human and canine). Thirteen amino acids surrounding both sides of this mutation are totally conserved in all species that we looked at including: other mammals (human, cow, mouse), other vertebrates (chicken and zebrafish), and invertebrates (Drosophila, and C. elegans). The DNM1 mutation produced a LOD score of 16.39 for linkage on the EIC pedigrees, and a p-value for association of 1.07 x 10-16. This mutation involves a single DNA nucleotide base pair change that can be detected by a restriction fragment length polymorphism (RFLP) assay.

Ninety-seven percent of all dogs that fulfilled the study criteria for presumed EIC affected were homozygous for the DNM1 mutation. Of 132 dogs for which the owners reported no episodes of collapse, 9% were homozygous the DNM1 mutation, 49% were heterozygous, and 42% were homozygous for the DNM1 normal allele. All 20 parents of affected dogs were heterozygous or homozygous for the mutant allele, which is consistent with EIC being autosomal recessive. A significant false negative phenotyping rate, in which genetically susceptible dogs have not been exposed to conditions sufficient to initiate collapse, as well as the possibility of genetic and environmental modifying factors, may explain why 9% of dogs without a history of collapse are homozygous for the mutant allele. Twelve heterozygotes were reported to have single collapse episodes or collapse episodes that did not fit the more stringent criteria for presumed affected. This could be consistent with a less severe phenotype in carriers than for the homozygotes, and indicate the possibility of a partially penetrant dominant trait. There were, however, 65 heterozygotes with no known episodes of collapse, and the high frequency of heterozygotes in the population makes conclusions concerning genotype-phenotype relationships in heterozygotes ambiguous at present. SNP haplotype analysis on 23 dogs with the strongest evidence of being EIC affected revealed a small minimally-conserved 137 kilobase (Kb) haplotype block shared by affected dogs that is only three times longer than the length of the DNM1 gene. Collectively, this genotype data is consistent with the DNM1 mutation causing a highly penetrant autosomal recessive condition.

Dynamin 1 Mutation Frequency in Labrador Sub-Populations

We have genotyped over 400 Labradors from 5 different field trials conducted in the upper Midwest in the summer of 2007 that included dogs from 20 different states and three Canadian provinces. This data revealed a carrier frequency of > 30%, and homozygous affected frequency of 3%, with 11/15 of the homozygous affected dogs having had repeated episodes of collapse. Though not a random sample, we also tested 153 conformation Labradors and found >30% to be carriers, and 19 dogs to be homozygous for the mutant DNM1 allele. Many of these conformation dogs did not perform exercise to the level of field trial and hunting dogs, but some of the homozygous affected conformation dogs were reported to have collapse episodes. In addition we genotyped 68 Labrador Retrievers from a veterinary teaching hospital storage bank and found 29% were carriers, and 1.5% were homozygous affected. In limited testing to date, we have also found homozygous affected Labrador Retrievers with collapse consistent with EIC from Australia, New Zealand, Israel, and Germany.


The dynamin gene family (DNM1, DNM2, AND DNM3) encodes proteins that are essential for synaptic transmission by virtue of their role in the formation of new synaptic vesicles at the presynaptic nerve terminal. The DNM1 gene encodes the dynamin 1 protein that is also a member of the subfamily of GTP-binding proteins that regulate clathrin-mediated endocytic vesicle formation. DNM1 appears to be expressed exclusively in the brain and spinal cord4, where it plays a key role in synaptic vesicle fission by assembling into collar-like structures around coated pits on the pre-synaptic terminal. These structures are severed to release coated vesicles, thereby re-forming synaptic vesicles to contain neurotransmitter and enabling continuous synaptic communication. No other naturally occurring mutation in DNM1 has been identified to date in any mammalian species. DNM2 mutations have been associated with centronuclear myopathy5 and Charcot-Marie-Tooth disease6 in people.

Dynamin Mutants in Other Species

Temperature dependent mutations in the homologous gene have been found in Drosophila melanogaster7 and C. elegans8 in which affected individuals have normal movement at ambient temperatures, but when the environmental temperature is elevated they develop locomotion defects that are reversible once the temperature is returned to ambient levels. These locomotor defects can become fatal if they are exposed to the elevated temperatures for a prolonged time period. Similar to Labrador Retrievers, the Drosophila mutants appear to have their rear end affected initially with a potential explanation that the longest nerves might be the first to be affected.

A recent study created knock-out mice with no functional DNM1 gene These knock-out mice are born alive, but postnatal viability is brief, due to inability to tolerate the neurological stimulation of everyday life.9 DNM2 and DNM3 appear to be able to handle low frequency neurological stimulation, and DNM1 expression becomes essential when a heightened stimuluscreates a heavy load on endocytosis and only as long as thestimulus persists. We believe it is likely that insufficient activity of the L256 dynamin 1 protein in homozygous affected dogs during times of high excitement, such as anticipation and performance of high intensity exercise, leads to a lack of sufficient vesicles for sustained synaptic transmission, resulting in a reversible loss of motor function. Further studies will be needed to determine if the insufficient activity is also temperature dependent for the DNM1 R256L mutation.

The Dnm1 Mutation in Other Dog Breeds

We have observed the identical 137 Kb haplotype in 3 Chesapeake Bay Retrievers homozygous or heterozygous for the DNM1 mutation. In addition we have found 6 Chesapeakes and 9 Curly-Coated Retrievers who where homozygous for the DNM1 mutation. In a non-random sample of 44 Duck Tolling Retrievers we did not detect the mutant allele in any individuals. We have ongoing studies to determine the prevalence of the mutant canine DNM1 allele in additional closely and distantly related breeds.

Genetic Testing

A simple genotyping assay can now help Labrador breeders avoid producing affected puppies in future generations. We are planning to offer an EDTA blood sample based genetic test to determine normal, carrier, and affected status for this DNM1 mutation as a screening test for EIC susceptibility through the University of Minnesota Veterinary Diagnostic Lab (VDL) in the spring to early summer of 2008.


Based on the function of DNM1, the short shared genetic haplotype of affected dogs, the very strong evolutionary conservation of DNM1 amino acid sequence, and the phenotype of the temperature dependent mutants and the knock-out mice, this R256L substitution is very likely to be true causative mutation for susceptibility to the syndrome of exercise induced collapse in dogs. Future functional studies will be needed to conclusively prove if this is the case. There is a small chance, however, that this "mutation" is actually only a single base pair polymorphism that does not cause a function change in the DNM1 protein. In this case it is possible that the actual causative mutation is in promoter regulatory region of the DNM1 gene, or in another very nearby gene that has not yet been identified or sequenced. Even if this is the case, this DNM1 DNA change we have identified is very highly associated with the EIC phenotype and still could be used as a very good genetic screening test for breeding decisions.

In conclusion, a recessive DNM1 gene mutation that is very highly associated with EIC in the Labrador Retriever dog was identified. This finding comes in close succession to the discovery of a SINE insertion mutation in the PTPLA gene responsible for centronuclear myopathy in this breed10, and further demonstrates the utility of gene mapping for inherited canine neurological diseases that do not have obvious candidate genes. While the physiology of neurotransmitters and their respective receptors has been extensively detailed for decades, the biology of neurotransmitter synaptic vesicles and associated proteins is just starting to be elucidated.11 This is the first naturally occurring mammalian DNM1 mutation to be identified and should provide critical insight into synaptic vesicle biology across many species.


1.  Taylor, Proc ACVIM 2008,

2.  O'Brien, et al. Res. Vet. Sci. 1990;48:124,

3.  Roberts, et al. Anesthesiology. 2001 95: 716,

4.  Noakes, et al. J. Comp. Neurol. 1999;410: 531,

5.  Echaniz-Laguna, et al. Neuromusc. Disord. 2007;17:955,

6.  Züchner, et al. Nature Genet. 2005;37: 289,

7.  van der Bliek, Meyerowitz. Nature. 1991;351:411,

8.  Clark, et al. Proc. Natl. Acad. Sci. U. S. A. 1997; 94:10438,

9.  Ferguson, et al. Science. 2007; 316:570,

10. Pele, et al. Hum. Mol. Gen. 2005;14:1417,

11. Burre, et al. J. Neurochem. 2007;101 :1448.

Speaker Information
(click the speaker's name to view other papers and abstracts submitted by this speaker)

Edward (Ned) Patterson, DVM, PhD, DACVIM (SAIM)
University of Minnesota
St. Paul, MN

MAIN : ACVIM Neurology Specialist : Exercise-Induced Collapse
Powered By VIN