Canine Cytogenetics - From band to basepair

Abstract

Humans and dogs have coexisted for thousands of years, during which time we have developed a unique bond, centered on companionship. Along the way, we have developed purebred dog breeds in a manner that has resulted unfortunately in many of them being affected by serious genetic disorders, including cancers. With serendipity and irony the unique genetic architecture of the 21^st Century genome of Man's best friend may ultimately provide many of the keys to unlock some of nature's most intriguing biological puzzles. Canine cytogenetics has advanced significantly over the past 10 years, spurred on largely by the surge of interest in the dog as a biomedical model for genetic disease and the availability of advanced genomics resources. As such the role of canine cytogenetics has moved rapidly from one that served initially to define the gross genomic organization of the canine genome and provide a reliable means to determine the chromosomal location of individual genes, to one that enabled the assembled sequence of the canine genome to be anchored to the karyotype. Canine cytogenetics now presents the biomedical research community with a means to assist in our search for a greater understanding of how genome architectures altered during speciation and in our search for genes associated with cancers that affect both dogs and humans. The cytogenetics ‘toolbox’ for the dog is now loaded. This review aims to provide a summary of some of the recent advancements in canine cytogenetics.

Introduction

The description of the domestic dog (Canis familiaris) as Man's best friend is a concept that is derived from millennia of co-existence and companionship. The bond between these two species is perhaps unlike that shared by any other two mammalian species. The availability of high quality genome resources for both human and dog has taken the relationship to a new level and comparative studies of their genomes are providing a plethora of new insights into the underlying biomedical similarities between these two species. Prerequisite to an understanding of normal genome organization of a species is the ability to define gross genome architecture. The 2.4Gb of DNA comprising the haploid nuclear genome of the (female) domestic dog are partitioned into 38 pairs of acrocentric autosomes and two metacentric sex chromosomes (Figure 1). The gross morphological similarity shared by many of the autosomes of the dog continues to present significant challenges to chromosome identification using conventional cytogenetics. The development of reagents and resources for molecular cytogenetic analyses of the canine genome has played a pivotal role in the generation of comprehensive and effective genome and comparative maps for the dog. Development and use of this molecular cytogenetics ‘toolbox’ allowed the development of a standardized chromosome nomenclature that led subsequently to the generation of integrated genome maps and played a key role in the anchoring of the canine genome assembly. In addition, molecular cytogenetics has allowed the gross genome architecture of the dog to be compared to other mammalian species and has also facilitated the development of means to assess genome changes associated with a variety of canine cancers. This article will review the development and application of these cytogenetic reagents and approaches.

Figure 1. The dog karyotype.

a) DAPI stained and b) enhanced DAPI banded metaphase preparation of the domestic dog, with 38 pairs of acrocentric autosomes and a pair of metacentric sex chromosomes.

Development of a standard karyotype for the dog

The presence of 78 chromosomes in cells of the dog was determined by studies of meiotic cells (Minouchi 1928) and later confirmed using cultured peripheral lymphocytes (Gustavsson 1964). The karyotype comprises 38 pairs of acrocentric autosomes, a large sub-metacentric X chromosome and a small metacentric Y chromosome (Figure 1). Based on genome sequence data, the largest autosome, dog chromosome 1 (CFA 1) is ∼ 125Mb in size (Lindblad-Toh et al. 2005) and is thus smaller than human chromosome 12, with all but the five largest dog chromosomes (CFA 1-5) being smaller in Mb size than human chromosome 18 (Table 1). Evaluation of unbanded chromosome preparations of the dog allows precise identification only for the sex chromosomes (due to their size and morphology) and chromosomes 1 and 38 (by virtue of their size). The gradual decrease in size of the remainder of the karyotype renders reliable recognition of conventionally stained homologous pairs almost impossible.

Table 1.

Physical size of the chromosomes comprising the canine genome, derived from the earlier study of flow sorted chromosomes (Langford et al. 1996) and by the more recent canine genome sequence (Lindblad-Toh et al. 2005).

Chromosome	∼Mb size derived from flow sorting	∼ Mb size derived from the genome assembly
CFA01	135	125
CFA02	100	88
CFA03	104	94
CFA04	102	91
CFA05	99	91
CFA06	88	80
CFA07	95	83
CFA08	87	77
CFA09	79	64
CFA10	81	72
CFA11	87	77
CFA12	86	75
CFA13	76	66
CFA14	73	63
CFA15	76	67
CFA16	74	62
CFA17	81	67
CFA18	67	58
CFA19	67	56
CFA20	68	61
CFA21	63	54
CFA22	72	64
CFA23	63	55
CFA24	55	50
CFA25	60	54
CFA26	48	42
CFA27	57	48
CFA28	55	44
CFA29	51	44
CFA30	47	43
CFA31	50	42
CFA32	51	41
CFA33	41	34
CFA34	50	45
CFA35	38	29
CFA36	41	33
CFA37	40	33
CFA38	38	26
CFA X	137	126
CFA Y	27	--
TOTAL	2782*	2424

^*Total size of haploid female genome indicated.

For many mammalian species, the development of chromosome banding techniques in the early 1970's provided the ability to identify homologous chromosome pairs and led to the subsequent development of internationally recognized chromosome numbering and banding nomenclatures. A variety of banded karyotypes for the dog have been published over the past 30 years and these are summarized in Table 2. It became clear that, at least for many of the smaller dog autosomes, chromosome banding patterns alone did not offer sufficient discrimination to describe the complete karyotype of the dog with confidence. Consequently, to ensure consistency in chromosome description, a Committee for the Standardization of the Karyotype of the Domestic Dog was established in the early 1990's, under the auspices of the DogMap Workshop. This Committee was charged with proposing a chromosome nomenclature of the dog karyotype that would be accepted on the international arena. Using conventional cytogenetics the Committee was able to reach a consensus on the identification of dog chromosomes 1-21 (CFA 1-21), but concluded that a complete standardized karyotype would require the use of molecular cytogenetic reagents (Switonski et al. 1996). Simultaneously, Langford et al (Langford et al. 1996) had used high-resolution bivariate flow sorting of dog chromosomes to generate a panel of whole chromosome paint probes (WCPP) for fluorescence in situ hybridization (FISH) analysis of the dog. These WCPP provided the essential common resource to members of the Committee, and the joint efforts of many researchers led subsequently to the development of a consensus chromosome nomenclature for CFA 21-38 (Breen et al. 1998). The combined chromosome numbering system of all chromosomes of the dog that proposed by the Committee (Breen et al. 1999; Breen et al. 1998; Switonski et al. 1996) was later endorsed by the International Society of Animal Genetics (ISAG) DogMap workshop held in Minneapolis, July 2000. At that time, the physical size of each dog chromosomes was estimated by inference from bivariate-chromosome sorting with mixtures of human and dog chromosomes (Breen et al. 1999; Langford et al. 1996) and suggested that the haploid genome of a female dog was in the region of 2,800Mb (Table 1). It is interesting to note that while the dog karyotype was initially assembled based on conventional microscopic assessment of chromosome size, we know now that the actual physical sizes of the dog chromosomes bear little resemblance to their order in the established karyotype. Despite the introduction of a more sophisticated means of identification, the initial ordering of the karyotype was left as was and so the continued use of more sophisticated tools is still essential for reliable dog chromosome identification.

Table 2.

Published banded karyotypes of the domestic dog.

Author (year)	Banding
Selden et al (1975)	GTG
Manolache et al (1976)	GTG
Howard-Peebles and Pryor (1980)	R
Fujinaga et al (1989)	GTG
Stone et al (1989)	GTG
Poulsen et al (1990)	R
Stone et al (1991)	GTG
Moreno-Millan et al (1991)	R
Graphodatsky et al (1995)	GTG
Langford et al (1996)	DAPI
Reimann et al (1996)	GTG
*Switonski et al (1996)	GTG
Christian et al (1998)	digital
**Breen et al (1998)	DAPI/GTG
***Breen et al (1999)	DAPI
Yang et al (1999)	DAPI
Graphodatsky et al (2000)	DAPI

^*CFA 1-21 were defined by the use of whole chromosomes paint probes generated by Langford et al (1996), with the chromosome numbering based on Selden et al (1975).

^**CFA 22-38 were defined by the whole chromosomes paint probes generated by Langford et al (1996) and ordered according the recommendation of the Committee for the Standardization of Dog Karyotype.

^***all chromosomes were defined by the use of whole chromosomes paint probes generated by Langford et al (1996) and ordered according the recommendation of the Committee for the Standardization of Dog Karyotype.

The introduction of canine molecular cytogenetics

The generation of whole chromosome paint probes and larger insert genomic libraries (bacteriophage and cosmid) saw the introduction of canine molecular cytogenetics. Simultaneously, the development of sophisticated cooled charged coupled device cameras, refinement of fluorescence filter band widths and the introduction of advanced multi-plane imaging software, facilitated the development of multicolor FISH analysis. As a consequence more laboratories began to adopt fluorochrome based banding techniques (e.g. DAPI banding) to allow concurrent chromosome identification during FISH analyses, rather than the need to perform pre- or post hybridization chromosome banding. Sequential chromosome painting analysis with fluorescently labeled chromosome paint probes (Langford et al. 1996) onto single metaphases produced a series of complete DAPI banded karyotypes of the dog in which the chromosome numbering followed that adopted by the standardization committee (Breen et al. 1999). To allow accurate assignment of FISH mapped single locus probes (SLPs) to DAPI banded dog chromosomes, a 460-band ideogram of the dog karyotype with five grey levels was also produced (Breen et al. 1999).

Independently, a second set of dog chromosome paint probes was generated (Yang et al. 1999) and a detailed comparative chromosome painting approach (dog vs red fox vs human chromosomes) was applied in order to identify the dog chromosomes indirectly. The authors were able to integrate their chromosome nomenclature with that of the standardization committee for all but 10 of the smaller autosomes (Graphodatsky et al. 2000). These 10 discrepancies are shown in Table 3. A second DAPI banding pattern for the whole karyotype of the dog was generated in the form of a black and white ideogram. The nomenclature of the standardization committee was subsequently used to define the chromosomes in the 7.6x canine genome assembly (Lindblad-Toh et al. 2005) and so use of the nomenclature proposed by Yang et al (Yang et al. 1999), which has been used subsequently by Graphodatsky et al (Graphodatsky et al. 2000), requires numerical conversion of these 10 autosomes to be compliant with the genome assembly data.

Table 3.

Dog chromosome nomenclature differences between the studies of Breen et al (Breen et al. 1999) and Yang et al (Yang et al. 1999).

Breen et al 1999b	Yang et al 1999
25	28
27	29
28	31
29	34
30	25
31	27
33	30
34	35
35	37
37	33

The role of canine molecular cytogenetics in genome mapping

Following the establishment of standardized chromosome nomenclature, the use of multicolor FISH technology in canine genomics opened the door to expedite the physical mapping of numerous clones that contained key elements of emerging genome maps. In the first integrated genome map of the dog (Breen et al. 2001), 266 cosmid clones, each containing a polymorphic microsatellite marker, were used as chromosome-specific singe locus probes (SLPs) to define their cytogenetic location. Distributed across all chromosomes these clones thus served as key cytogenetic points and so enabled the 1,800 marker integrated meiotic/radiation hybrid (RH) map of the canine genome (Breen et al. 2001) as well as the subsequent 3,279 marker (Guyon et al. 2003) and 4,249 marker (Breen et al. 2004) RH maps of the dog to be firmly anchored to the karyotype. At the cytogenetic level, this 4,249 RH map reported the precise location of 1,000 canine bacterial artificial chromosome (BAC) clones (see below), each of which had been ordered along the length of each chromosome using stepwise multicolor SLP analysis. Of these 1,000 BAC clones, 804 were in common with the RH-map. The use of a common set of markers for both the cytogenetic and the RH-maps ensured that the integrated map was firmly anchored at multiple locations along the length of each chromosome, providing a very high level of confidence in marker order for this and the subsequent 10,000 marker RH map (Hitte et al. 2005). Molecular cytogenetics played a valuable role in anchoring the 7.6x genome assembly; indirectly via the use of cytogenetically-anchored RH map data as a means to provide useful ordering information; and directly as a tool to determine the precise location of sequences directly on chromosomes.

The development of additional reagents to enhance the growing canine molecular cytogenetics ‘toolbox’ became available with the development of two BAC libraries. The first, RPCI-81 (Li et al. 1999) was constructed from a male Doberman Pinscher and represents 8.1x genome coverage with an average insert size of 155kb (http://bacpac.chori.org/mcanine81.htm). Numerous groups have used this library as a source of large insert genomic clones that have been used as SLPs to determine the precise cytogenetic location of many loci within the canine genome. In addition to the ∼1,000 clones mapped as part of the development of the canine RH-map (Breen et al. 2004) numerous BACs were selected by screening the library specifically to obtain clones containing genes of interests. Many of these clones, which were FISH mapped to determine their cytogenetic location, contained genes associated with disease (eg (Debenham et al. 2005; Klukowska et al. 2004a; Klukowska et al. 2004b; Lozier et al. 2002; Philipp et al. 2005; Rak et al. 2002; Rak et al. 2003; Stabej et al. 2004; Stabej et al. 2005; Thomas et al. 2001a; van de Sluis et al. 2001; van den Berg et al. 2004; van den Berg et al. 2003; van_de_Sluis et al. 1999; Wagner 2003; Wagner et al. 1999); also see http://faculty.vetmed.ucdavis.edu/Faculty/lalyons/Sites/CompGenBACsDOG.htm) and cancer (Thomas et al. 2003a; Thomas et al. 2003b). Additional clones from the RPCI-81 library have been used as an aid to visualize comparative genome organization e.g. (Szczerbal et al. 2003; Yudkin et al. 2007).

Specifically for molecular cytogenetic studies, clones from the RPCI-81 BAC library have been used to generate panels of chromosome-specific FISH probes. A set of 41 clones was described by Thomas et al (Thomas et al. 2003c) that may be used to identify conclusively each dog chromosome according to the position of each cone on the corresponding chromosome. Also from this library, a panel of 80 BAC clones was assembled in groups of 1-3 clones per dog autosome that, when labeled/detected with one of five fluorochromes in a single hybridization, is reported to identify simultaneously all dog chromosomes within a metaphase spread (Courtay-Cahen et al. 2007). Both these chromosome-specific panels are useful for identifying chromosomes that are considered structurally normal. As such, these panels may be used for chromosome identification following metaphase-base comparative genomic hybridization (CGH) analysis. As part of the cytogenetic anchoring of dog RH-maps, over 1,000 BAC clones from the RPCI-81 library were cytogenetically assigned using multicolor FISH analysis and details of each assignment reported (Breen et al. 2004) (http://www.cvm.ncsu.edu/mbs/breen/dog_map.htm). Using these clones, Thomas et al (Thomas et al. 2005) generated a genome-wide BAC microarray, with approx 2Mb resolution, and demonstrated how this array may be used for detecting copy number changes in canine cancers using array comparative genomic hybridization (see below). Use of a cytogenetically characterized array overcomes the need for identification of individual dog chromosomes and allows for a much greater throughput of cases.

Genome integrated canine cytogenetics – band to base pair

A second dog BAC library, CHORI-82 was constructed from a female Boxer, ‘Tasha’, the same female whose DNA was used to generate the 7.6x canine genome assembly (http://bacpac.chori.org/library.php?id=253). This library represents 10x coverage of the canine genome and contains ∼198,000 clones with an average insert size of 172Kb. As part of the development of the 7.6x canine genome sequence, BAC end sequences were generated from the clones comprising this library and so the library represent a valuable cytogenetics resource of genome assembly-integrated clones. Using web-based resources such as the UCSC Genome Browser (http://genome.ucsc.edu/cgi-bin/hgGateway) CHORI-82 clones may be selected, based either on their position within the genome or their specific gene/sequence content, and used as SLPs in FISH analysis. Using this process chromosome tiling sets of arrays have been generated with clone spacing at 10Mb resolution (Thomas et al. 2007) and 1 Mb resolution (Thomas et al, submitted).

In molecular cytogenetics of any species the accepted resolution of metaphase FISH is in the region of 2-4Mb, assuming that the chromosome spreads are elongated and the probes signals are discrete. In general the more common use of mid-metaphase preparations reduces this resolution to approx 4-6Mb and so precise ordering of clones that are spaced by smaller distances requires the use of interphase FISH analysis (Figure 2). In developing the genome wide, assembly-integrated BAC panels, Thomas et al (2007, submitted) were able to determine that over 95% of the 2,122 clones mapped by FISH, did so with a unique cytogenetic location and to the region of the chromosome indicated by the genome assembly. This figure lends high confidence to the structural integrity of the assembled genome and confidence to selecting clones from the genome browsers.

Figure 2. Application of multicolor FISH to determining genome assembly location.

Using the canine genome assembly five canine BAC clones were selected at 1Mb intervals from the telomeric end of CFA 37. a) Metaphase spread of a dog showing co-localization of the five differentially labeled canine BAC clones to the telomeric end of CFA 37. At this resolution while all five clones may be shown to map to CFA 37, their order is unable to be resolved, even when display as separate color planes as shown in b). c) Interphase FISH of the same five BACs showing clear and consistent order of the five clones on both homologues

As with other species, the availability of BAC clones that are integrated into the genome assembly launched canine molecular cytogenetics into a new arena – observations of microscopically visualized chromosome aberrations may be rapidly translated to aberrations of specific regions of the genome sequence. With this in mind, genome integrated canine BAC arrays have been developed with the panels of genome-anchored BACs described above. The first is a genome-wide BAC array comprising 275 clones spaced at intervals of approximately 10Mb and also contains clones representing the dog orthologues of 31 genes implicated in human cancers (Thomas et al. 2007). This array has been shown to be of use for first pass assessment of gross genome-wide DNA copy number variation in canine tumors and provides a similar resolution to that produced by metaphase based CGH analysis, albeit without the need to be able to identify the dog chromosomes. A higher resolution genome-wide array has also been generated which contains 2,122 clones and represents a typical resolution of ∼1Mb (Thomas et al, submitted). In both cases, each clone on the array has been used as a SLP to verify that it a) has a unique cytogenetic location and b) maps to the chromosome region indicated by the assembly. The results of this study indicated that, with a few exceptions, the 7.6x canine genome assembly is able to act as a robust source of CHORI-82 clones that may be used for canine cytogenetics.

Canine Cytogenetics and Evolution

The domestic dog belongs to the Canidae, a family that is believed to have diverged from other carnivore families ∼50-60 million years ago. Within the extant Canidae, divergence from a common ancestor is reported to have commenced ∼ 7-10 million years ago (Wayne 1993). Previous studies indicated that the family is divided into two major groupings, the ‘dog-like’ and ‘fox-like’ canids (Bininda-Emonds et al. 1999; Graphodatsky et al. 2001). More recent genetic data, including data generated as part of the dog genome sequence project (Lindblad-Toh et al. 2005), have suggested that the family may be refined into four major phylogenetic groups represented by the fox-like canids (including the racoon dog), the gray and island fox species, the South American canids and the wolf-like canids (including the domestic dog)(Ostrander 2007). Cytogenetic studies of the 34 extant species comprising the Canidae have revealed considerable variation in chromosome number and morphology. The karyotype architecture of the Canidae ranges from 2n= 34 in the red fox (Vulpes vulpes) to 2n=78 in the wolf-like canids, including the domestic dog (Canis familiaris) (Wayne 1993). Canid karyotypes with lower chromosome numbers contain numerous bi-armed autosomes, while the autosomes of the wolf-like canids are all acrocentric. The family has thus undergone a relatively high rate of karyotype evolution and so offers an exciting opportunity to use detailed cytogenetic evaluation, with both conventional and molecular approaches, to assess chromosomal evolution.

Whole chromosome reciprocal chromosome painting and comparative chromosome banding has been performed on numerous species within the Canidae. In addition to the domestic dog (2n=78) these studies have primarily involved the fox-like canids such as the Red Fox (Vulpes vulpes, 2n=34+B [0-8]), Chinese Raccoon Dog (Nectereutes procyonoides procyonoides, 2n=54+B [0-4]), Japanese Raccoon Dog (Nectereutes procyonoides procyonoides, 2n=38 +B [0-8]), and Arctic Fox (Alopex lagopus, 2n=48-50). These studies suggest an apparent gross pattern of whole-arm rearrangements during speciation within the Canidae (Graphodatsky et al. 2001; Graphodatsky et al. 2000; Nie et al. 2003; Yang et al. 1999). Notable exceptions to the presence of single segment shuffling was evident for dog chromosomes 1, 13 and 19, each of which correspond to two chromosome segments of the Red Fox, Raccoon dog and Arctic Fox, and while dog chromosome 18 is represented by a single conserved segments in the Raccoon Dog, it also is split across two segments in the Red Fox and Arctic Fox (Graphodatsky et al. 2001; Graphodatsky et al. 2000).

An interesting feature of the karyotypes of several species within the Canidae is the presence of B chromosomes, supernumerary chromosomes within a karyotype that may vary in their number and have been described in a wide range of species, primarily plants and insects (Jones 1975; Jones and Diez 2004). The role of B chromosomes and their variability in number is still unknown, though in some non-mammalian species it has been suggested that B chromosome numbers may contribute to the formation of aberrant meiotic products (Camacho et al. 2004; Camacho et al. 2000). Early studies of B chromosomes in the Raccoon dogs indicated that they were composed of telomeric-like sequences (Wurster-Hill et al. 1988), while more recent studies indicated the presence of nucelolar organizer region (NOR)-like sequences (Szczerbal and Switonski 2003). B chromosomes of the red fox, however, have been shown to be rich in centromeric-like sequences (Yang et al. 1999). FISH analysis, using chromosome paints and/or single locus probes has suggested that Canid B chromosomes may share ancestry with small (<1Mb) autosomal DNA segments and thus contain active genes (Graphodatsky et al. 2005; Trifonov et al. 2002; Yudkin et al. 2007). A more detailed assessment of the genome organization of these supernumerary chromosomes will be needed to further evaluate their potential significance. The dog genome sequence will undoubtedly pay a key role in furthering these studies by providing a common reference point for the other canids.

Comparative maps from a wide range of mammalian species are providing key information about the dynamics of chromosome evolution (Murphy et al. 2005). For example, cytogenetic resources developed for the dog have also played a role in extending comparative cytogenetics beyond the Canidae by expanding cross-species painting studies to include members of the Felidae and Ursidae, both of which also belong to the Order Carnivora (Tian et al. 2004).

Canine Clinical Cytogenetics and Cancer

The presence of gross cytogenetic changes in non-neoplastic cells has been reported only rarely in the domestic dog. The most common chromosome abnormality reported in non-cancer specimens involves the sex chromosomes. Almost two decades ago, conventional cytogenetics was used to investigate dogs with sexual development disorders and revealed that several possessed X chromosome aneuploidy (79,XXY) or were chimeras (78,XX/78,XY) (reviewed by (Mellink and Bosma 1989). In addition infertile bitches have been shown, using both conventional cytogenetic and molecular cytogenetics, to present with X chromosome aneuploidy (Schelling et al. 2001; Switonski et al. 2003a; Switonski et al. 2003b). Molecular cytogenetic analysis of dogs presenting Duchenne muscular dystrophy (DMD) revealed that, as in humans, such cases presented with a small deletion of a region of Xp21, containing the canine DMD gene (Schatzberg et al. 1999). The availability of high-resolution cytogenetic resources for the dog will allow a more detailed evaluation of more subtle changes in dogs with unusual syndromes.

Clonal chromosome aberrations have been described in almost 27,000 human neoplasms that collectively represent approximately 75 different types of cancer (http://cgap.nci.nih.gov/Chromosomes/Mitelman). There is broad acceptance that the accurate identification of recurring chromosome abnormalities in malignant cells provides opportunities to increase the sophistication of diagnosis, sub-classification and prognosis of neoplastic disorders. In human medicine, the identity of cytogenetic aberrations has been shown to also assist in the localization of cancer-associated genes and even selection of the most appropriate therapeutic approach. The application of molecular cytogenetics to the analysis of human neoplasia has revolutionized the way in which we interrogate tumor cells for cytogenetic changes, whether they are numerical or structural in nature.

Veterinary medicine has provided a wealth of information about the clinical and pathological presentation of numerous canine cancers. Despite millions of years of divergent evolution, the high degree of similarity between human and dog also extends to their genome sequences (Lindblad-Toh et al. 2005). As pets, dogs (and thus their genomes) are exposed to the same environmental influences as humans. It is not surprising therefore that there are considerable pathophysiological similarities shared by many forms of human and canine cancer (Withrow and Vail 2006). In addition, the demographic history and genetic structure of many purebred domestic dogs has resulted in the development of distinct genetic isolates, or breeds, which have been shown to have highly reduced levels of genetic variation and specific association with genetic disease e.g.(Ostrander 2007; Ostrander and Giniger 1997; Ostrander and Kruglyak 2000; Parker et al. 2004; Sutter et al. 2004; Sutter and Ostrander 2004). In terms of cancer incidence, it is well known that that purebred dog breeds are associated with differing susceptibility to specific malignancies e.g.(Modiano et al. 2005; Patterson 2000; Priester and McKay 1980). Such breed-associated predisposition suggests that selected breeds of dog are inheriting ‘at risk’ alleles for very few genes, perhaps even a single gene, with a profound effect. These features, combined with the sophisticated genomic resources now available for the dog, have placed the dog in a position of high visibility as a model system for cancer research.

The combination of pathophysiological and genetic similarity shared between human and dog led to the hypothesis that canine tumors may contain the natural variety of chromosome aberrations that are observed in human cancers, a feature that is not evident with induced rodent tumors. Molecular cytogenetic studies of canine cancers have demonstrated that this is indeed the case. For example, in studying canine hematopoetic malignancies it has been shown that cells isolated from human and dog tumors of comparable malignancies share several evolutionarily conserved cytogenetic changes (Breen and Modiano, 2008; Thomas et al. 2003c). As is the case in human cancers, a detailed cytogenetic investigation of canine tumors may therefore reveal the identity of recurrent changes that are of diagnostic and/or prognostic significance. To further this concept, evaluation of cytogenetic changes in canine tumors has been performed using a combination of direct cytogenetic analysis of tumor karyotypes from individual cells, using both conventional and molecular cytogenetics, and indirect analysis of DNA from the tumor cell population, using comparative genomic hybridization.

Direct analysis – chromosome paints and single locus probes

The difficulties associated with identification of normal dog chromosomes (see above) are exacerbated when such chromosomes undergo structural changes and so evaluation of complex canine tumor karyotypes using G-banding or DAPI banding alone is not a realistic option. While there were numerous early studies aimed at evaluating chromosome changes in canine cancers, the limitations of using chromosome banding techniques was very evident. Consequently, early conventional cytogenetic analyses of canine cancer karyotypes were generally restricted to very few cases. Further, such while such reports were able to describe the presence of addition chromosomes and/or aberrant chromosome structures, they were unable to identify conclusively which chromosome were involved in the aberration (eg Basrur and Gilman 1966; Grindem and Buoen 1986; Idowu 1976; Mayr et al. 1992; Mayr et al. 1990; Miles et al. 1970; Stone et al. 1991; Whang-Peng 1969). The development of molecular cytogenetics reagents for the dog, however, permits detailed direct analysis of canine tumor genomes by FISH analysis. The use of chromosome paint probes and/or cytogenetically characterized single locus probes allows a rapid assessment of the gross numerical and structural characteristics of chromosomes within individual tumor cells. Chromosome painting analysis is widely used to unravel the complex nature of chromosome rearrangements and to determine chromosomal organization in the karyotypes of human cancers and syndromes (see for example, Langer et al. 2004, Blennow 2004). The availability of high quality chromosome paints for application to canine tumor cytogenetics (Langford et al. 1996; Yang et al. 1999) has enabled this approach to be used similarly in the investigation of karyotypic abnormalities in canine tumors including mammary carcinoma (Tap et al. 1998), lymphoma (Thomas et al. 2001b) and soft tissue sarcoma (Milne et al. 2004). The development of multiplex-FISH (M-FISH) (Speicher et al. 1996) and spectral karyotyping (Schrock et al. 1996) for human cytogenetics has led to many new cytogenetic discoveries over the past decade due to their ability to increase the amount of genomic data that can be generated in a single experiment. To identify all 40 chromosome types (38 autosomes + X + Y) of the dog in a single FISH reaction with different colors, i.e. 40 color M-FISH, would require the use of six spectrally resolvable fluorochromes, which is beyond the scope of many laboratories. To overcome this limitation, Milne et al (2004) used a robust seven-color system in their study of canine sarcomas and established karyotype coverage over a series of six different separate FISH reactions.

Though of broad use, chromosome paint probes generated for the dog have not been widely available. To overcome this, panels of cytogenetically verified chromosome-specific single locus probes have been generated as described above (Courtay-Cahen et al. 2007; Thomas et al. 2007; Thomas et al. 2003c)(Thomas et al, submitted). Since these SLP panels are generated from commercially available canine BAC libraries, and the identity of each clone used is publicly available, it is a simple process for any cytogenetics laboratory to purchase these clones and so generate their own SLP panels. The most recent of these chromosome-specific SLP panels was generated by selection of clones from the canine genome assembly (see above) and so offer an opportunity to tile aberrant chromosomes in tumor karyotypes and refine the precise region of chromosome breakpoints in tumor karyotypes beyond the resolution possible with chromosome paint probes (Breen and Modiano, 2008).

Indirect analysis – comparative genomic hybridization

Comparative genomic hybridization (CGH) is a molecular cytogenetic approach to screen an entire genome for regions of copy number aberration (CNAs) in a single experiment (Kallioniemi et al. 1992; Kallioniemi et al. 1994). Metaphase-based CGH was used to identify CNAs in a range of human malignancies including cancers of the brain (Petersen et al. 2000; Scheil et al. 2001; Wiltshire et al. 2000), prostate (Cher et al. 1994), breast (Rodriguez et al. 2000; Tirkkonen et al. 1998), bladder (Obermann et al. 2003), thyroid (Wreesmann et al. 2002), skin (Bastian et al. 1998) and hematopoetic system (Allen et al. 2002; Berglund et al. 2002; Franke et al. 2002). Chromosome-based CGH was developed for application to canine tumors (Dunn et al. 2000; Thomas et al. 2001b) and has been used to demonstrate that recurrent chromosome aberrations are present in canine lymphoma (Thomas et al. 2003c). However, as with other species, canine metaphase-based CGH presents two key limitations to high throughput; 1) the approach requires conclusive identification of dog metaphase chromosomes and so requires a secondary analysis with chromosome specific SLPs (Courtay-Cahen et al. 2007; Dunn et al. 2000; Thomas et al. 2003c); 2) the use of metaphase chromosomes as the target limits the ability to detect copy number changes to those that are generally greater than 5-10Mb (Lichter et al. 2000; Pinkel et al. 1998).

The recent development of microarray-based CGH (aCGH) analysis overcomes many of the limitations of conventional metaphase-based CGH, offering increased resolution that is dependent solely on clone spacing. Array CGH has been used to describe a wide variety of DNA copy number changes in human cancers, providing new and valuable insight into the underlying gene involvement of tumors (e.g. Bejjani and Shaffer 2006; Gunn et al. 2007; Shaffer and Bejjani 2006). Array-based CGH analysis has been developed for application to the dog, based on the use of degenerate-oligonucleotide-primed PCR products from canine BAC clones, each of which has been assigned to a precise cytogenetic location using FISH analysis (see above). Preliminary use of these arrays, as shown in Figure 3, has demonstrated that for canine cytogenetics this approach offers an exciting opportunity for significant improvement in resolution as well as throughput of canine cancer specimens (Thomas et al. 2007; Thomas et al, submitted). The use of integrated BAC arrays allows rapid translation from cytogenetic aberrations to genome sequence aberrations. In combination with reliable clinical and pathological data of canine cancer patients, canine cytogenetics should therefore emerge an approach that will provide help to establish more accurate classification of canine tumors and so advance our aim to enhance the clinical management of our pet dogs.

Figure 3. Whole genome aCGH profile of a canine tumor.

These data were obtained using a 1Mb assembly-integrated dog BAC array (Thomas et al, submitted). Data are plotted as the median, block-normalized and background-subtracted log₂ ratio of replicate spots for each of 2,122 BAC clones on the 1Mb array. Log₂ ratios representing genomic gain and loss are indicated by horizontal bars above (thin green line) and below (thin red line) the midline representing normal copy number. Using the aCGH Smooth algorithm (Jong et al. 2004), the chromosome copy-number status line appears as orange where no major genomic imbalances were apparent and as either red or green in regions where genomic imbalances were apparent (red=loss, green=gain). This profile indicates whole chromosome loss for CFA 17 and CFA 34. In addition there are several additional copy number aberration of isolated clones. Validation of these inferred copy number changes is verified by using the BAC clones on the array to perform single locus probe analysis of the tumor cells. This process provides a means to assess that distribution of locus copy number within the actual tumor cell population (as shown in Figure 4).

The availability of a high density genome wide, genome integrated, cytogenetically characterized panel of canine BAC clones (Thomas et al submitted) also provides the basis for the selection of clones for use in the characterization of both numerical and structural chromosomal changes associated with cancers. Individual clones may be used to determine DNA copy number variations within tumor cell populations and pooled clones may be used to visualize karyotype reassembly presenting in such cells (Figure 4). The identity of recurrent cytogenetic changes in canine cancers should lead to the discovery of those changes that are associated with diagnosis and/prognosis and thus open the door for clinical cytogenetics to become a routine component of the clinical management of dogs presenting with cancers.

Figure 4. SLP analysis of a canine tumor.

These data revealed dual monosomy and indication of a translocation event. In Figure 3, the aCGH data indicated a DNA copy number loss for CFA 17 and 34. Five BAC clones were used to generate multicolor single locus probe data for this case including a clone each of CFA 17 (purple) and CFA 34 (green) and two clones spaced at 1Mb on CFA 20. a) co-hybridzation of these five clones to an interphase nucleus from a normal canine cell, demonstrating the presence of a normal copy number (n=2) for each of the five clones and also the close association of the two clone from CFA 20 (shown in aqua and yellow). b) co-hybridization of the same five BAC clones to an interphase nucleus of a cell from the tumor profiles in Figure 3, reveled that while three of the five probes had normal (n=2) copy number (red, yellow and aqua), the two probes representing CFA 17 (purple) and CFA 34 (green) were present as one copy, confirming the copy number loss of these two chromosomes indicated by the aCGH profile in a). In addition, the two clones on CFA 20 (yellow and aqua) are now no longer closely associated in this tumor cell, indicating the presence of chromosome break in the 1Mb interval between these two clones.

Summary

The use of genome-integrated cytogenetic resources such as those described above have advanced canine cytogenetics to a stage that will allow a much more detailed evaluation of mammalian chromosomal evolution. Simultaneously the field has reached a point where aberrant chromosomal regions may be translated to specific regions of the canine genome and so may identify genes of significance in canine cancers. As with human patients, we anticipate that new therapies offered for our pet dogs will become available. The ability to characterize response-associated cytogenetic changes will be a valuable aide to the refinement of prognosis and thus the clinical management of patients.

The pathophysiological and genome similarities between dog and human cancers may also lead to genes being discovered in studies of canine cancers that are also of major significance in human cancers. It seems that the unique bond shared between human and dog extends now beyond one with an emotional basis; it is now a bond that is at least in part fortified by the biomedical relationship that we share. For thousands of years the bark of our dogs has provided an early warning system for our safety. In the 21^st Century that role may extend, with studies of the canine genome aiding our development of early warning systems for our health. The domestic dog and now its genome, quite rightly, continue to be Man's best friend.