John C. Fyfe, DVM, PhD
College of Veterinary Medicine, Michigan State University, East Lansing, MI, USA
Objectives of the Presentation
To discuss the rationale and interplay of motives directing investigations of genetic disease in companion animals.
To discuss 3 examples of genetic disease investigation and the genetic testing programs resulting from each.
Overview of the Issue
Investigation of genetic disease in dogs and cats benefits everyone involved and the animal and human communities at large. To learn what gene mutation causes a disorder allows the animal producer to detect the mutant gene in otherwise healthy animals and, thereby, avoid producing diseased animals in future matings. Reciprocally, to learn what goes wrong health-wise when a gene is mutated builds the community-held stock of biomedical knowledge and the potential to devise new medical therapies. Thus, genetic diseases, in any species, are health problems to be eliminated if one is producing animals, but for those in medical research they are windows that open onto the complex physiology underlying normal health.
New genetic disorders arise spontaneously and continually in all species and any breed. None are immune and no person needs to assume fault when a new disorder is observed. A previously unrecognized clinical disorder, or a recognized genetic disease occurrence in a breed where it was not previously seen, should evoke a high level of suspicion for genetic disease, particularly if it occurs in a young or a purebred animal. Most genetic diseases in veterinary medicine are inherited recessively; the carriers do not exhibit clinical abnormalities and may be allowed to produce successive generations of carriers unless some method of carrier detection is devised. Other disorders are inherited dominantly, but when they are late-onset, the signs of disease are not apparent until an age when affected animals may have already been used for breeding to produce the next generation. For the foreseeable future, prevention is the only practical and ethical solution to genetic disease in dogs and cats, and for that, mutation detection is crucial in order to make appropriate breeding decisions.
For recessive or late-onset disease, convenient and reliable mutation detection is best achieved with a DNA-based test that can be performed on samples obtained noninvasively. To develop such a test, one must determine the mutation underlying a disorder, and this in turn requires careful investigation of the clinical, pathologic, and genetic aspects of the disorder. Characterization of a genetic disease requires an integrated approach and the cooperative expertise of a variety of individuals including breeders, practicing veterinarians, geneticists, and molecular biologists. For this, there must be an open exchange of information in a non-accusatory, non-threatening manner.
Exacting clinical information that includes signalment, evaluation of all body systems, age of disease onset, progression of signs, clinical laboratory data, histopathology, metabolic screening, and appropriate provocative tests must be obtained from each animal that appears to be affected by the disorder. The patient history should include pedigree information, questions regarding normal littermates (e.g. growth rates), and the occurrence of similar disorders in related animals. Obtaining the latter information is traditionally difficult because related animals and their attending veterinarians are often geographically dispersed, and because it requires that information that may be negatively perceived is shared between breeders. It is particularly difficult in later-onset disorders because the breeder-owner relationship attenuates with the passage of time and potential environmental causes of the disorder become more plausible. Posting inquiries in breed-specific or veterinary medical-specialty chat rooms or list serves via the Internet can be fruitful. Of utmost importance is that the diagnostic details be sufficiently specific that one is assured that every animal grouped under the diagnosis actually has the same disorder. In our favor is that simply inherited diseases demonstrate remarkably consistent age of onset, signs, laboratory abnormalities, and progression. Therefore, it is wise to exclude animals from the analysis if they do not conform to strict diagnostic criteria or are insufficiently evaluated.
The crux of moving from the description of a disease to the causative mutation is to first eliminate all but the fewest possible genes that deserve detailed analysis such as expression assays and/or sequencing. This may be accomplished in one of two ways. In a functional gene approach, specific laboratory information about the disease may suggest one or a few candidate genes, genes that one can reasonably hypothesize to harbor mutations that cause the disorder. Such a hypothesis is typically made because the gene has a known function that is important in a metabolic pathway or structural complex that is abnormal in the affected animals. Enzyme and collagen defects are examples. Information about similar human or other species' genetic diseases can be quite helpful in a functional gene approach, and it is feasible to take this approach even if one has DNA samples from only one or a few affected animals and one (preferably both) parent(s).
Unfortunately, much is not yet known about gene function(s), so in many cases even the best description of the disorder does not sufficiently reduce the number of candidate genes. Neurodegenerative disorders and epilepsy are examples. In these cases, a positional gene approach is needed, but this approach is 10-100 times more labor-intensive and requires much greater technical and financial resources. To map a disease gene by linkage analysis, one needs DNA samples from many members of a relatively large family segregating the disorder. One also needs a large number of mapping markers and high resolution, species-specific genetic and physical genome maps. Fortunately, the latter have been developed in the last decade through efforts of a worldwide community of dog and cat genome researchers and federal support of the dog and cat genome sequencing projects. Once the map position for a disorder is established, genes within the critical interval (the interval between markers whose alleles are found to have recombined with the disease alleles) become candidate genes by virtue of being near the mapped disease locus, even without knowledge of the gene function or how it relates to the disease abnormalities.
Another positional gene approach is to map the disease locus by association mapping, a procedure of looking for a segment of the genome exhibiting alleles of markers that are common in unrelated affected animals but different from unaffected control animals. This can be a very powerful technique, but as in linkage analysis, mapping resolution is directly related to both the number of animals available for analysis and the number of available markers. With the recent advent of single nucleotide polymorphism (SNP) arrays that collect allele data from thousands of markers simultaneously, disease loci can be mapped by either method with relatively few affected animals needed (Leegwater et al., 2007).
By whatever approach one gets there, the candidate genes are sequenced in affected animal DNA, and the sequences are compared to those of normal animals. Normal variation of DNA sequence exists in all genes, so while an observed sequence difference can be considered a putative mutation, it must be shown to conform to various standard criteria. These criteria include that the mutation segregates with the disease allele in all available animals of known disease locus genotype and that it is not seen as a normal polymorphism in a large population of unrelated animals. However, these criteria are not enough because, by definition, many sequence variations in the critical interval determined by disease gene mapping will satisfy them. A putative mutation must be either of a nature that it is obviously incompatible with gene function (e.g. a premature stop codon, a large deletion of coding sequence, or a frameshift mutation) or it must be shown experimentally to disrupt gene expression or function, before it should be accepted as the disease mutation. The latter is particularly challenging when the putative disease gene is one of unknown function. However, with the disease mutation in hand, it is a relatively trivial exercise to devise a reliable DNA-based test to detect the mutation in patient samples.
Example I: Congenital hypothyroidism with goiter (CHG) in the toy fox terrier (TFT) is an autosomal recessive trait that is lethal in the preweaning period. Unfortunately, some very popular sires were carriers, and the carrier state was quite widespread in TFT and had even been bred into rat terriers before the disorder was studied (Pettigrew et al., 2007). We initiated the mutation search at the behest of a TFT breeder whose tenacity when the problem occurred in her kennel marshaled the efforts of local veterinarians, veterinary endocrinologists, and molecular biologists. In this instance, specialized clinical testing of 2 affected puppies that responded partially to thyroid hormone replacement therapy provided detailed information that supported an efficient functional candidate gene approach. An inability to attach iodine to thyroglobulin was demonstrated in radioiodine uptake and perchlorate discharge tests performed by a veterinary radiologist in specialty practice. In current understanding, those clinical findings reflect a defect in the synthesis of thyroid hormone that is best explained by a mutation in one of only 4 genes. Laboratory studies further narrowed the field of candidates by demonstrating a deficiency of thyroid peroxidase (TPO) enzymatic activity in affected dog thyroid tissue. Sequencing of the affected dog TPO cDNA and confirmation in genomic DNA revealed a C>T mutation that predicted a premature stop codon at amino acid residue 111 of the protein coding sequence (p.R111X). The mutation also obliterated a Mwo I restriction endonuclease recognition sequence, thus providing a convenient way to discriminate the mutant and normal alleles in PCR products amplified from genomic DNA (Fyfe et al., 2003).
In the last 6 years, the Laboratory of Comparative Medical Genetics at Michigan State University has tested 808 TFT buccal brush samples and found 248 CHG carriers (30%). In addition, 5 pups have been tested as CHG suspects, and the test show each of them to be homozygous for the mutation sequence, thus confirming the CHG diagnosis. Clearly, these results are from a biased sample and do not represent a true carrier rate for the breed. As expected, initially many samples were submitted from dogs related to known carriers. Subsequently, we received many samples from offspring of known carriers as the TFT breeders began to use the recommended "test and replace" approach to eliminating the CHG problem in their kennels. What is a bit disappointing is that in the last year we detected 45 carriers (39%) among the 115 dogs tested, suggesting that that there is still a long way to go in eliminating this mutation. Carriers have been detected among dogs from the USA, Canada, New Zealand, and, just recently, Europe (pers. comm. Prof. R. Renaville).
Example II: Spinal muscular atrophy (SMA) in Maine coon cats is an autosomal recessive trait that causes skeletal muscle weakness and neurogenic atrophy. The disorder becomes clinically apparent as gait abnormalities, tremors, and jumping difficulty beginning at 12-14 weeks of age but is compatible with many years of high quality indoor life (He et al., 2005). The primary lesion is loss of spinal lower motor neurons. The homologous disorder in humans is the leading genetic cause of infant mortality, most cases of which are due to mutation of the survival of motor neuron gene (SMN1). The SMN function that is crucial to motor neuron survival remains obscure, despite intense investigation by many laboratories in the last decade.
Of utmost importance to this investigation were, once again, the efforts of a breeder of Maine coons for whom the problem was a reality. She collected information regarding other similar cases, used her contacts to collect DNA samples, and donated the initial affected cats for study. While the histopathology documented the primary tissue involved in the disorder, motor neuron degeneration has many causes, and there was no reasonable functional candidate gene approach to discovering the disease gene. It was timely that the tools for gene mapping in cats, including the first generation of domestic cat whole genome sequencing, were just becoming available, so we collaborated with the Laboratory of Genomic Diversity at the National Cancer Institute to map the cat SMA gene. As indicated above, the effort did require that we establish a breeding colony to enlarge the SMA kindred that, when combined with the pedigree of privately owned Maine coon cats we were able to discover, provided sufficient statistical power to expect success from a whole genome scan for marker linkage. The analysis led to discovery of a large (~140 kilobase) deletion that abrogates expression of 2 contiguous genes, LIX1 and LNPEP in the affected cats (Fyfe et al., 2006). Studies are continuing to determine what loss of function or combination of losses caused by the deletion actually lead to motor neuron degeneration in cats.
Even though the disease mechanism remains to be elucidated, the deletion itself defines the disease allele and is easily detected by PCR amplification of genomic DNA isolated from buccal brushes. In the two years we have offered this test, we've detected 13 carriers (6%) among 225 cats tested. Interestingly, we've tested 8 Maine coon cats that had very similar presentation and progression of clinical signs, but which did not have pedigrees that went back to the cats of the original SMA kindred and which responded positively to steroid administration. None of the steroid-responsive cats had the deletion allele. On the other hand, carrier cats were recently and unexpectedly detected in a French line of Maine coon cats that have none of the known or suspected carriers in the extended pedigree. This event serves as a caution to Maine coon cat breeders of the danger of limiting genetic testing to cats that descend from a "known" disease ancestry. Rather, new potential breeding cats should be tested regardless of the pedigree "hung" on the cat. It is a lesson that is applicable to all genetic disease testing programs.
Example III: Glycogen storage disease type IIIa (GSD IIIa) in curly-coated retrievers (CCR) is a newly described autosomal recessive trait that causes episodic hypoglycemia and liver and skeletal muscle disease Gregory et al., 2007). In this instance, a functional candidate gene approach was the obvious choice because the glycogen storage disorders are well described inborn errors of metabolism in humans and the relevant metabolic pathways are highly conserved across mammalian species. To date, we are aware of only 2 affected dogs, probably because the clinical signs are subtle in the young dogs and not very specific. The initial findings in a single pair of seemingly normal littermates were elevations of liver and muscle enzyme activities observed in routine presurgical serum chemistry screens. Liver biopsy of one dog during elective ovariohysterectomy revealed hepatocyte swelling and excessive glycogen storage. Analysis of frozen portions of liver and muscle biopsies by the Glycogen Disease Laboratory of the Duke University Medical Center revealed deficiency of the glycogen debranching enzyme (GDE) and a compatible abnormality of the glycogen structure. Attempts to amplify the GDE cDNA from affected dog tissue failed, presumably due to the nature of the mutation, but the available dog genome sequence allowed us to amplify and sequence all of the 33 coding exons. We found a single base deletion in exon 32 that predicts a frameshift and premature termination of translation that would truncate the protein by 126 amino acids. Genomic DNA flanking the deletion site is readily amplified by PCR from buccal brush samples, but there is no convenient indirect way to discriminate the co-amplified alleles. Therefore, in this test the amplification products are purified and sequenced directly.
In the single year that this test has been available, 37 carriers (23%) were detected among 161 dogs tested. Carriers have been found among dogs originating in USA, Canada, New Zealand, Australia, and Finland. The Curly-Coated Retriever Club of America officially embraced this testing program and has donated funds to offset the cost of testing for its members. Members that receive financial support for testing their dogs are required to have the results posted in a public database, but many of these breeders are already posting their results, both clear and carrier, on the OFA and CHIC sites as well as on their own websites. Given this degree of concern and openness among the breeders, we expect the GSD IIIa mutation to rapidly drop in frequency among CCR.
A quirk of testing for this disorder is that one of the published PCR primer sequences that we used initially in testing fell by coincidence over a single nucleotide polymorphism (C/T) seen in some CCR. The GSD IIIa mutation apparently occurred on a C allele chromosome, so there are both C and T alleles on normal chromosomes in the CCR population. The T allele creates a primer mismatch in the penultimate 3' position that prevents a GSD IIIa normal allele from amplifying, so dogs that are GSD IIIa carriers and are heterozygous for the SNP may be misidentified as affected because their normal allele (T at the SNP) will not be represented in the amplification products. Our current testing protocol avoids the problem by using a PCR primer that is upstream of the SNP.
It is essential that a DNA-testing laboratory provide genetic counseling to clients. When carriers of recessive disease or animals affected with late-onset dominant disease can be reliably and conveniently detected, it is shortsighted to advocate that all carriers of genetic disease be eliminated from future breeding programs. Such absolute selection against a disease allele may seriously deplete the pool of genetic diversity in a breed. To preserve genetic diversity but prevent production of animals affected with a recessive disorder, one can counsel that disease mutation carriers that exhibit other desirable traits can be paired safely with animals that test clear of the mutation (safety implying that no affected offspring will be produced). Offspring of such a mating that have inherited the desirable traits can then be tested, and if clear, used to replace breeding stock that carry the disease mutation. Non-tested or carrier offspring should not be used for breeding. Thus, in one or two generations, a breeding concern can be free of the disease mutation.
In dominantly inherited disease, there is no such thing as a clinically normal "carrier", even if the disease is of late onset. Therefore, there is no "safe" use in breeding of an animal that has been demonstrated to harbor the disease mutation. On average, half of the offspring will themselves be affected. One simply has to search for animals that test clear to stock ones breeding program.
In each of the example investigations above, there were obvious benefits to the breeders and owners of animals, to future generations of the animals themselves, and to veterinary practice. Less obvious perhaps, but just as important to generate their involvement, is the benefit to the laboratory researcher. The scholarship involved in genetic disease investigation leads to publications and grant proposals, training opportunities for future generations of veterinary professionals and research scientists, and the simple joy of discovery and new knowledge.
These types of investigations are expensive, but funding for them is often available from animal health foundations when one can be plausibly predict that results of the investigation will reduce animal suffering or promote animal health. Similarly, funds are available from government or private, disease-specific agencies when one can plausibly predict outcomes of the research that will contribute in a comparative way to understanding of the homologous human disease or new therapeutics.
Investigation of genetic disease in dogs and cats provides benefit to the breeders and owners of such animals by developing tools for mutation detection in carriers of recessive disease and in affected animals of dominant disease before the age of onset of clinical signs. Such mutation tests are breed specific and can aid practitioners in confirming a presumptive diagnosis. The spontaneously-occurring genetic disorders of companion animals are an abundant biomedical research resource, the use of which can lead to improved understanding of normal physiology and health in those species. The comparative aspect of modern biomedical research is founded on the fact that findings in one species lead to solutions for health problem in other species, including humans.
1. Fyfe JC, Kampschmidt K, Dang V, et al. Congenital hypothyroidism with goiter in toy fox terriers. J Vet Intern Med, 2003;17:50-57.
2. Fyfe JC, Menotti-Raymond M, David VA,, et al. An ~140 kb deletion associated with feline spinal muscular atrophy implies an essential LIX1 function for motor neuron survival. Genome Res, 2006;16:1084-1090.
3. Gregory BL, Shelton GD, Bali D, et al. Glycogen storage disease type IIIa in curly-coated retrievers. J Vet Intern Med, 2007;21:40-46.
4. He Q, Lowrie C, Shelton GD, Castellani RJ, et al. Inherited motor neuron disease in domestic cats: a model of spinal muscular atrophy. Pediatric Res 2005;57:324-330.
5. Leegwater PA, van Hagen MA, van Oost BA. Localization of white spotting locus in boxer dogs on CFA20 by genome-wide linkage analysis with 1500 SNPs. J Hered 2007; [Jun 4; Epub ahead of print].
6. Pettigrew R, Fyfe JC, Gregory BL, et al. CNS hypomyelination in rat terriers with congenital goiter and a mutation in the thyroid peroxidase gene. Vet Path, 2007;44:50-56.