Analysis of Randomly Amplified Polymorphic DNA (RAPD) for Identifying Genetic Markers Associated with Canine Hip Dysplasia

Analysis of Randomly Amplified Polymorphic DNA (RAPD) for Identifying Genetic Markers Associated with Canine Hip Dysplasia X. Wang, A. B. Miller, A. J. Lepine, J. D. Scott, and K. E. Murphy Canine hip dysplasia is a heritable developmental disease resulting, in part, from increased laxity in hip joints and is a precursor to degenerative joint disease. Identification of genetic markers linked to joint laxity would foster development of more accurate diagnostic methods, facilitate identification of the disease gene(s), and supplement efforts to establish physical/genetic maps of the canine genome. Work presented here describes analysis of randomly amplified polymorphic DNA in the search for markers which cosegregate with increased joint laxity in Canis familiaris, the domestic dog. The Boykin spaniel, a highly inbred breed afflicted with an extremely high incidence of hip dysplasia, served as a model for study of canine hip dysplasia. Only 5% of 200 random primers revealed significant polymorphisms within this breed. However, polymorphisms were detected in seemingly nonpolymorphic amplification products when digested with restriction enzymes. Restriction digestion revealed polymorphisms in 15% of the monomorphic amplification products. Among the primers that revealed polymorphisms, one primer correctly identified 9 of 12 dogs with regard to joint laxity. However, extensive evaluation is required before any assertion can be made regarding linkage of this marker to joint laxity. Of interest, another primer amplified a genomic segment unique to the canine Y chromosome. From the Department of Microbiology and Molecular Cell Sciences, University of Memphis, Memphis, TN 38152-6041 (Wang, Miller, Scott, and Murphy), and The Iams Company, Lewisburg, Ohio (Lepine). We thank Drs. Robert Pernell and Paul Shealy for clinical examinations of the dogs used in this study and thank the many owners/breeders for donating dog blood for the research. Advice from Drs. Gail Smith, George Lust, Rory Todhunter, and Michael Olivier is gratefully acknowledged. We also thank Dr. Gustavo Aguirre for assistance in preparation of the manuscript. This work is supported by grants from The Iams Company and the Boykin Spaniel Society. Address correspondence to K. E. Murphy at the address above or e-mail: kemurphy@cc.memphis.edu. This paper was delivered at the International Workshop on Canine Genetics at the College of Veterinary Medicine, Cornell University, Ithaca, New York, July 12 13, 1997. 1999 The American Genetic Association 90:99 103 Canine hip dysplasia (CHD) is a developmental disease that can result in malformation of the hip joint and eventual degenerative joint disease ( Lust 1997). It affects all breeds of Canis familiaris, the domestic dog, but prevalence of the disease varies drastically from breed to breed. According to the Orthopedic Foundation for Animals (OFA; http://www.offa.org), the five breeds with the highest incidences are the bulldog (70.8%), otterhound (53.9%), Clumber spaniel (53.6%), St. Bernard (47.8%), and the Boykin spaniel (46.5%). Many popular breeds such as the Golden retriever, rottweiler, and German shepherd have about 20 30% incidences. The breeds with the lowest incidences include the greyhound (4.1%) and Belgian sheepdog (2.8%). It is probable that these incidences are artificially low because many cases of CHD may not be reported. Diagnosis of CHD is made by physical and radiographic examinations. The two radiographic methods are termed OFA and PennHip (Smith et al. 1990). Both have contributed significantly to the control of canine hip dysplasia during the past three decades. The PennHip method provides a biomechanical assessment of joint laxity as measured by the degree of distraction. The higher the distraction index ( DI), the looser the hip joint. This work employs the PennHip method and therefore is directed toward understanding the genetic basis of increased joint laxity, a major factor in development of CHD (Smith 1997). In the PennHip registry (with more than 12,000 dogs), the Boykin spaniel was listed among the three breeds with the highest distraction indices. However, the fact that these methods are not infallible has induced workers to search for genetic markers of this disease. While environmental factors play a role in the development of CHD, the disease is known to be heritable, possessing a polygenic mode of transmission (Brass 1989; Hedhammar et al. 1979). Information concerning the presence or absence of the disease-causing genes as suggested by markers would provide increased accuracy in predicting the development of CHD. Markers would also facilitate the search for the gene(s) responsible for CHD due to linkage of markers and genes. Finally, in the search for genetic markers of diseases, it is important to catalog all polymorphisms because these are useful for general mapping of genes. The 99

importance of mapping the canine genome is underscored by recent work to identify linkage groups ( Lingass et al. 1997; Mellersh et al. 1997) and establishment of hybrid cell lines for study of specific canine chromosomes ( Langston et al. 1997). DNA fingerprinting techniques rely on detection of polymorphisms. Polymorphisms have been detected by analyses of randomly amplified polymorphic DNA (RAPD) (Gu et al. 1997a; Rothuizen and Van Wolferen 1994; Williams et al. 1990) and of microsatellite DNA (Ostrander et al. 1993). RAPD is especially useful since allelic polymorphisms can be quickly identified. The utility of the RAPD approach is exemplified by its use in identification of markers in insect (Garner and Slavicek 1996), barley (Poulsen et al. 1995), canine (Gu et al. 1997a), and sheep (Cushwa et al. 1996) genomes. The real advantages of using RAPD techniques in any genome poor species are (1) the set of RAPD primers may be purchased commercially and can be used in any organism, (2) genome screening may be faster since RAPD, unlike microsatellite markers, simultaneously screens several loci in the genome. As our interests are in canine genetics, it is important to note that RAPD has been used to identify a marker linked to canine progressive rod-cone degeneration (Gu et al. 1997b). Reported here are some interesting polymorphisms revealed by RAPD analysis of the highly inbred Boykin spaniel, although it is not clear at present whether any of these polymorphisms are related to the CHD phenotype. Materials and Methods Dogs and Genomic DNA The Boykin spaniel was used as a model for study of clinical CHD because this breed has a high incidence of hip dysplasia and high DI. The average DI for this breed is 0.66, which is significantly higher than 0.38, the average of all dogs (more than 12,000) examined using the PennHip method. The Boykin spaniel has a small breeding population, which certainly contributed to its high incidence of CHD. The dogs reported here represented a family of 12 members spanning three generations. Blood was collected from dogs into buffered sodium citrate tubes. Genomic DNA was isolated using the Puregene DNA Isolation Kit (Gentra Systems Inc., NC). RAPD-PCR Random decanucleotide primers were obtained from Genosys ( TX) and Operon Figure 1. The family of dogs that were used in the arbitrary-primed PCR. Male dogs are represented by squares; female by circles. Joint laxity is indicated by different levels of shading: no shading tight hips (DI 0.4); dotted shading mild laxity (DI: 0.4 0.6); gray shading loose hips (DI 0.6). Letter C indicates that the dog is clinically confirmed with hip dysplasia. Technologies (CA). The GC content of these primers varies from 50 to 70%. Other PCR reagents were from Perkin-Elmer (CT). PCRs were carried out in a total volume of 25 l and contained the following: 100 ng of genomic DNA, 1 mm MgCl 2,1 PCR buffer, 0.1 mm of each dntp, 0.5 units of AmpliTaq DNA polymerase, and 0.4 M of random primer. The reaction mixtures were incubated in a PCR thermal cycler (Perkin-Elmer 2400) using the following parameters: a holding step at 94 C for 3 min; 3 cycles at 94 C for 2 min, 33 C for 1 min, and 72 C for 2 min; and 42 cycles at 94 C for 1 min, 33 C for 1 min, and 72 C for 2 min. After cycling, the elongation reaction was extended 7 min at 72 C. After PCR, 5 l of the reaction mixtures along with 2 l of loading dye were loaded on a 1.5% agarose gel prestained with ethidium bromide (0.5 g/ml). Electrophoresis proceeded at 60 V for 2.5 h in 1 TAE buffer. Sometimes restriction digestion was carried out on nonpolymorphic amplification products. Briefly, 5 l of the PCR product was digested with 1 unit of restriction enzyme (HaeIII, HinP1I, AluI, Sau3AI, MspI, or MseI) in a total volume of 10 l at37 C for 1 2 h. The digestion products were then resolved by electrophoresis. Results In the Boykin spaniel family tested ( Figure 1), the parent dog 11 has loose hips (DIs of 0.48 and 0.60) and dog 12 has tight hips (DI 0.32 and 0.37). All offspring of these dogs have loose hips, with DIs ranging from 0.50 to 0.71, but at least one hip presents with a DI greater than 0.60. When one of the offspring was bred to a tighthipped dog, dog 5 (DI 0.24 and 0.24), the resultant offspring had loose hips ( DI 0.65 and 0.75). Therefore it seems that the high distraction index was found in all dogs related to dog 11. Most offspring of dog 11 have been clinically confirmed to have CHD. A total of 200 primers have been used to amplify genomic DNA from selected Boykin spaniels. About 95% of these primers amplified nonpolymorphic DNA fragments. The lack of polymorphisms within the family is a result of the breed s high level of inbreeding, as suggested by the breed s narrow national distribution (mainly in South Carolina). The failure of the primers to generate a significant number of polymorphisms led to incorporation of the restriction fragment length polymorphism (RFLP) technique. RFLPs arise from sequence differences at restriction sites. While a powerful and widely used method, the need for large amounts of DNA for analysis has limited its use in DNA fingerprinting. However, the amount of DNA generated from PCR amplification is sufficient for RFLP analysis. Thus it was theorized that restriction digestion of amplification products would reveal polymorphisms within the study family. Among 39 restriction digestion experiments performed on nonpolymorphic PCR products, 25 experiments showed restriction on these products, but only 6 revealed internal polymorphisms. Hence the chance for restriction digestion to reveal internal polymorphisms was about 15% (6/39), much higher than the RAPD tech- 100 The Journal of Heredity 1999:90(1)

Figure 3. Primer OPW9 (GTGACCGAGT) revealed polymorphism associated with gender in Boykin spaniel. The 800 bp band is present in all male dogs (lanes 2, 3, 6, 7, and 11) but absent in all female dogs. Lanes 1 12: the dog numbers are the same as lane numbers. Figure 2. Restriction digestion revealed internal polymorphism among seemingly nonpolymorphic fragments obtained by arbitrary primer PCR with primer r37 (GAGTCACTCG). (A) The primer amplified a single band for all 12 Boykin spaniels tested. These bands have the same size. An example of these bands is shown in lane 13. Lanes 1 12: the dog numbers are the same as lane numbers. The amplification products of these dogs were digested with the restriction enzyme HaeIII. Lane 14: DNA marker. (B) Comparison of DNA sequences of dogs 1 and 2 at the HaeIII restriction site (GGCC). The restriction sites are missing in some regions in the DNA of dog 1, but present in the corresponding regions of dog 2. nique alone (5%). For example, primer r37 (GAGTCACTCG) amplified a single identical DNA fragment from all 12 dogs. Digestion with HaeIII revealed that there exist internal polymorphisms in the individual fragments ( Figure 2A). The observed polymorphisms follow the Mendelian pattern of inheritance. Sequencing results confirmed the internal polymorphisms, for example, dogs 1 and 2 have different numbers of HaeIII restriction sites ( Figure 2B). While certainly the majority of random primers are not informative, about 5% of random primers have revealed polymorphisms within this Boykin spaniel family. For example, amplification using primer r55 (CGCATTCCGC) yielded a 1,600 bp band in dogs 6 10 and 12, but not in the other dogs (data not shown). Apparently the DNA region corresponding to this band is only present in dog 12 and its offspring. Interesting polymorphisms were observed with primer OPW9 (GTGA- CCGAGT). An extra band (800 bp) was present in the amplification products of dogs 2, 3, 6, 7, and 11 (Figure 3). All of these dogs are male. However, this band was not present in other dogs, all of which are female. Further testing of this primer on additional Boykin spaniels and other breeds such as the golden retriever, Labrador retriever, and samoyed confirmed that the band is always associated with male dogs. The primer identified the gender of dogs with 100% accuracy in a total of 52 dogs tested (23 males and 29 females). Thus the 800 bp fragment may be used as a gender-specific marker in the dog. Of interest, another RAPD primer that amplifies a region specific for the canine Y chromosome has previously been described (Olivier et al. 1997). Marker sequence comparison (not shown) indicated that OPW9 and the primer used by Olivier et al. amplified distinct regions of the canine Y chromosome with no sequence overlap. While polymorphisms are most often seen as the presence/absence of specific fragments, polymorphisms are also evident as differences in band intensity of similarly sized fragments. As shown in Figure 4, primer r105 (GCACCGAACG) generated bands of about 1,000 bp in all family members, but two distinct intensities existed, suggesting that bands of different intensities are unique. These obvious amplification differences allowed segregation of family members into two groups. One group includes dogs 2, 3, 6 9, and 11. The high-intensity band was present in amplification products of these dogs. All dogs except dogs 3 and 6 in this group were confirmed to have hip dysplasia. In the other group, a band of identical size but of lower intensity was present. Dogs 4 and 10 in this group were diagnosed with hip dysplasia. Overall the correctness of this primer in predicting the presence and absence of CHD in this family is 67% (8/12). Similarly the correctness of this primer in predicting high joint laxity (DI 0.6) in this family is 75% (9/12). The average DIs for the first and second groups are 0.60 and 0.44, respectively. The DNA fragments in these bands were sequenced. The sequences of the bands of dogs 11 and 12 exhibit very low similarity (data not shown). This indicates that although the bands of dogs 11 and 12 are the same size, they represent different regions of the genome or radically different sequences from the same loci. It must be noted that this analysis is preliminary and that further testing of this primer on additional Boykin spaniels and other breeds is needed to find out whether this 1,000 bp fragment has potential as a marker for elevated joint laxity. Discussion CHD occurs in all breeds, with drastic differences in prevalence varying from 3 to 70% among breeds. The working assumption is that the same genes are responsible for CHD in all breeds. The Boykin spaniel was chosen as a model for studying molecular genetics of CHD based on two considerations. First, for genetic study of any heritable disease it is imperative that a small population with welldefined pedigrees and a higher than normal incidence of the disease be used. For example, in small or isolated populations, consanguineous mating increases the incidences of recessive diseases. In certain of these small populations, the founder effect is a primary factor responsible for abnormally high incidences of various genetic diseases. For example, the high incidence of Ellis van Creveld syndrome (a form of dwarfism) (Brueton et al. 1990) Figure 4. RAPD generated by using primer r105(gcaccgaacg). Polymorphism is indicated by two intensities for the 1000 bp band. Lanes 1 12: the dog numbers are the same as lane numbers. Wang et al Genetic Markers for Canine Hip Dysplasia 101

and several other diseases has been described within the Amish community of Lancaster County, Pennsylvania. It follows then that efforts to identify the genetic cause(s) of canine hip dysplasia will have a higher probability of success if research focuses on a breed that has (1) a high incidence of CHD and (2) a relatively small population in which breeding practices and events have resulted in an increased frequency of disease alleles as compared with open-breeding populations. The Boykin spaniel meets these criteria as it is among the breeds with the highest incidence of hip dysplasia and/or joint laxity. Of importance, its population is quite small (restricted mostly to South Carolina) and there exist extensive pedigree data. Second, since identification of genetic markers using the RAPD technique is a random search and there are likely copious polymorphisms among individual dogs, we tried to minimize the noise polymorphisms by selecting the highly inbred Boykin spaniel. Specifically, initial molecular analyses were carried out on dogs within one Boykin spaniel family in order to test the possibility that identification of markers in such a pedigree would be enhanced as compared with other pedigrees with greater diversity. Unfortunately this approach can actually work against marker identification since too great a reduction in availability of polymorphic markers may hinder identification of a marker associated with CHD. Within the selected family of 12 Boykin spaniels, 7 dogs were confirmed by veterinary radiographic and physical examinations to suffer from hip dysplasia. This family showed a great degree of genetic homogeneity since only 5% of arbitrary primers tested in arbitrarily primed-pcr revealed DNA polymorphisms. Among the primers able to reveal polymorphism, OPW9 distinguished male dogs from female dogs with 100% accuracy; r105 identified five of seven dogs that were diagnosed with hip dysplasia. While this latter polymorphism is intriguing, any conclusion regarding its linkage to increased joint laxity would be premature. That being understood, it is important to realize that based on the work of Gu et al. (1997a,b), it is possible that testing of 200 primers may be insufficient to draw definite conclusions regarding the utility of the RAPD approach for identification of markers linked to CHD. Analysis of dinucleotide microsatellites has also been performed as a tool for identifying markers linked to increased joint laxity but not shown to be useful since there were hardly any polymorphisms detected. However, it is likely that even in such an inbred population many polymorphisms are present and will be more readily found through examination of tetranucleotide microsatellites (Ostrander E, personal communication). While the high level of genetic homogeneity reduced the number of polymorphisms irrelevant to CHD, and thus facilitated the search progress, it is clear from these results that the utilization of additional DNA fingerprinting techniques in the search for markers for CHD is warranted. These techniques include restriction digestion of RAPD and analyses of amplified fragment length polymorphisms (AFLP). Tests of RFLP on seemingly nonpolymorphic DNA fragments generated by RAPD-PCR using primer r37 indicated that RFLP, when combined with RAPD-PCR, can reveal polymorphisms that can not be revealed by RAPD-PCR alone. AFLP (Vos et al. 1995) is based on the selective amplification of fragments from complete restriction digestion of genomic DNA. In this technique, genomic DNA is first digested by two restriction enzymes followed by selective amplification of the digests using selective primers. The AFLP approach is a variation of the RFLP approach, but requires less genomic DNA and allows specific amplification of a large number (50 100) of restriction fragments, whereas RAPD-PCR generates 3 10 amplified fragments. Finally, the search for markers of canine hip dysplasia is not restricted to the Boykin spaniel, but has been extended to other breeds such as the golden retriever, the Labrador retriever, and the samoyed. Furthermore, humans have been incorporated into this work because developmental dislocation of the hip ( DDH), a disease that has a very similar pathology to canine hip dysplasia, exists in humans. The incidence of DDH is 1 1.5 per 1000 live births (Kim and Weinstein 1998) and 75% of all cases of secondary osteoarthritis of the hip result from uncorrected DDH (Michaeli et al. 1997). The rationale for studying DDH is that more is known about the human genome with respect to chromosome markers and specific genes. Therefore it may be advantageous to try to identify markers for DDH and use these to probe for their canine homologues. Should polymorphisms tightly linked to increased joint laxity be identified, fragments containing these will be cloned and sequenced. A pair of primers of more than 20 bases each could be designed so that they selectively amplify the DNA region corresponding to the polymorphic fragment. Diagnostic tests can then be developed using this information. In addition, work can be directed toward identification of the genes responsible for the disease. References Brass W, 1989. Hip dysplasia in dogs. J Small Anim Pract 30:166 170. Brueton LA, Dillon MJ, and Winter RM, 1990. Ellis-van Creveld syndrome, Jeune syndrome, and renal-hepaticpancreatic dysplasia: separate entities or disease spectrum? J Med Genet 27:252 255. Cushwa WT, Dodds KG, Crawford AM, and Medrano JF, 1996. Identification and genetic mapping of random amplified polymorphic DNA (RAPD) markers to the sheep genome. Mamm Genome 7:580 585. Garner KJ and Slavicek JM, 1996. Identification and characterization of a RAPD-PCR marker for distinguishing Asian and North American gypsy moths. Insect Mol Biol 5:81 91. Gu WK, Acland GM, Aguirre GD, and Ray K, 1997a. Evaluation of RAPD analysis for identification of polymorphisms in canine genomic DNA. Anim Biotechnol 8: 207 219. Gu WK, Acland GM, Ray K, and Aguirre GD, 1997b. Identification of a RAPD marker linked to progressive rodcone degeneration (prcd). In: International Workshop: Canine Genetics: The Map, The Genes, The Diseases, Cornell University, July 12 13; 40. Hedhammar A, Olsson SE, Anderson SA, Persson L, Pettersson L, Olausson A, and Sundgren PE, 1979. Canine hip dysplasia: Study of heritability in 401 litters of German shepherd dogs. J Am Vet Med Assoc 9:1012 1016. Kim HW and Weinstein SL, 1998. Intervening early in developmental hip dysplasia. J. Musculoskeletal Med February:70 81. Langston AA, Mellersh CS, Neal CL, Ray K, Acland GM, Gibbs M, Aguirre GD, Fournier RE, Ostrander EA, 1997. Construction of a panel of canine-rodent hybrid cell lines for use in partitioning of the canine genome. Genomics 46:317 325. Lingass F, Sørensen A, Junia RK, Johansson S, Fredholm M, Winterø AK, Sampson J, Mellersh C, Curzon A, Holmes NG, Binns MM, Dickens HF, Ryder EJ, Gerlach J, Baumle E, and Dolf G, 1997. Towards construction of a canine linkage map: establishment of 16 linkage maps. Mamm Genome 8:218 221. Lust G, 1997. An overview of the pathogenesis of canine hip dysplasia. J Am Vet Med Assoc 210:1443 1445. Mellersh CS, Langston AA, Acland GM, Fleming MA, Ray K, Wiegand NA, Francisco LV, Gibbs M, Aguirre GD, and Ostrander EA, 1997. A linkage map of the canine genome. Genomics 46:326 336. Michaeli DA, Murphy SB, and Hipp JA, 1997. Comparison of predicted and measured contact pressures in normal and dysplastic hips. Med Eng Phys 19:180 186. Olivier M, Lust G, Breen M, and Binns M, 1997. A new DNA marker specific for the canine Y chromosome. In: Molecular Genetics and Canine Genetic Health Conference, Canine Health Foundation, Oct. 31 Nov. 1; 129. Ostrander EA, Sprague GE Jr, and Rine J, 1993. Identification and characterization of dinucleotide repeat (CA)n markers for genetic mapping in dog. Genomics 16:207 213. Poulsen DME, Henry RJ, and Rees RG, 1995. The use of bulk segregant analysis to identify a RAPD marker 102 The Journal of Heredity 1999:90(1)

linked to leaf rust resistance in barley. Theor Appl Genetics 91:270. Rothuizen J and Van Wolferen M, 1994. Randomly amplified DNA polymorphisms in dogs are reproducible and display Mendelian transmission. Anim Genet 25: 13 18. Smith G, 1997. Advances in diagnosing canine hip dysplasia. J Am Vet Med Assoc 210:1451 1457. Smith GK, Biery DN, and Gregor TP, 1990. New concepts of coxofemoral joint stability and the development of a clinical stress-radiographic method for quantitating hip joint laxity in the dog. J Am Vet Med Assoc 196:59 70. Vos P, Hogers R, Blecker M, Reijans M, Van de Lee T, Hornes M, Frijters A, Pot J, Peleman J, Kuiper M, and Zabeau M, 1995. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res 23:4407 4414. Williams JGK, Kubelik AR, Livak KJ, Rafalski JA, and Tingy AV, 1990. DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Res 18:6531 6535. Corresponding Editor: Gustavo Aguirre Wang et al Genetic Markers for Canine Hip Dysplasia 103