Telomeres are linked to susceptibility to cancer, the aging process, and the lifespan of individual cells [1]. Also, the dysfunction of a single enzyme involved in telomere maintenance has been shown to be sufficient to cause severe telomere damage, premature tissue aging, and development of neoplastic lesions [2]. Telomere length is controlled by the enzyme telomerase, a component of which is Telomerase Reverse Transcriptase (TERT). Therefore, variations in telomerase activity could result from coding polymorphisms affecting the enzymatic function of TERT, or from non-coding polymorphisms affecting the magnitude or kinetics of TERT expression. Interestingly, it has also been observed that while deficient telomerase activity results in a decreased rate of cancer in mice [3], over-expression and ubiquitious expression of telomerase result in an increased rate of tumors [4].

Domestic dogs provide an exceptional experimental model to assess the potential effect of such polymorphisms on organismal lifespan, as it is well recognized that many large breed dogs age faster and have shorter life spans than dogs of small breeds. Furthermore, cancer susceptibility varies widely by breed [5] making dogs excellent models for studying cancer susceptibility. It has been specifically shown that in dogs, as in humans, cancer cells rely on telomerase to maintain telomeres and extend their lifespan [6].

For this study, polymorphisms in the TERT sequence were observed in Shih Tzus, Dachshunds, Irish Wolfhounds and Newfoundlands: four breeds that show large differences in size, expected lifespan, cancer susceptibility, and geographic origins. Shih Tzus are small dogs that have an average lifespan of 13 years [7]; the breed is thought to be “ancient”, and it dates back at least to the year 624 A.D. when it was first depicted in Chinese art [8]. These dogs are thought to be at a decreased risk for cancer on the whole [5]. Dachshunds are small dogs that were first bred in Germany in the early 1600s [8] and also have an average lifespan of 13 years [7]. Evidence points to them having a decreased risk of cancer compared to the average, but they appear to have a predilection for mammary tumors [9]. Records of Irish Wolfhounds, large dogs first bred in Ireland, date back to 391 A.D. [8]. These dogs only live 7 years on average [7], and are particularly susceptible to osteosarcoma, even as compared to other large, fast-growing breeds that are typically prone to this cancer [10]. Finally, the large Newfoundland breed was developed in Newfoundland, Canda in the early 1600s [8] and live 9 years on average [7]. These dogs are thought to be no more or less susceptible to cancer than average [5].

Although reproductive isolation among breeds was not formally standardized until the mid-19th century with the formation of breed clubs and breed standards [11], geographic isolation of these breeds in their place of origin makes interbreeding among these breeds unlikely. Furthermore, Parker et al. showed that modern dog breeds are “distinct genetic units” and assigned the dog breeds used in this study to genetically distinct “clusters”.

Our objectives here were to assess conserved polymorphism in TERT, and to establish if they segregated according to the breeds or randomly. In order to do so, we sequenced the TERT gene of 10 dogs each from the aforementioned breeds and compared their sequences and any polymorphism that were found.

The DNA samples from individual dogs were prepared and PCR carried out as described in Supplementary Materials. PCR reactions were originally carried out for 74 primer pairs, and it was found that 69 of these worked on the sample dog (Additional file 1: Figure S1). It was verified that 1,245 of the 1,380 PCR samples yielded a clearly defined band of the expected size (Additional file 1: Table S1, Additional file 1: Figure S1).

Clustalx was used to analyze the consensus sequence of all twenty dogs and compared them with the published sequence. As hypothesized, we identified polymorphisms within both exons and introns (Figure 1a and b). Exonic sequences within the TERT gene were more conserved compared to the intronic sequences. Strikingly, for many observed SNPs we found two groups: one predominantly homozyougous for the reference sequence and the other heterozygous. However, there were specific loci where the variant homozygous allele was never seen when a heterozygous allele was present. For example, at bp 19298 of the TERT gene (Figure 1a), located within exon 17, dogs of all breeds surveyed here were either homozygous C or heterozygous C/G (S), but none were homozygous G. It is unlikely this was the result of read contamination, as the sequence trace chromatograms for this section were clean and the peaks were clear (Figure 1b). A further example of this variation pattern can be seen at bp 17273 of the TERT gene (Figure 1c and d), located within an intronic region: dogs of all breeds surveyed here were either homozygous A or heterozygous A/C (M), but none were homozygous C. This variation pattern was observed throughout the gene in at least 115 instances in both sets of 20 samples, with 14 of these instances occurring in exonic regions, and 7 of those exonic variants causing a coding change. The locations of the exonic changes and their effect on codons within the gene are displayed in Table 1. A complete analysis of our sequence and the SNPs obtained is shown in Additional file 1: Figure S2 and S3.

In analyzing the phylogenetic tree obtained from the two separate sets of dogs at the TERT locus, it was found there was no instance that placed all individuals of the same breed in one branch. Figures 2a and b illustrate this: even when as many as three members of the same breed were grouped together there was at least one other dog from a different breed in that branch. This was not an artifact of the phylogenetic inference method used, as nearly all clades within the phylogeny having high confidence (i.e., with posterior probabilities > 50%) were preserved by diverse phylogenetic methods (MrBayes, parsimony and Neighbor-joining) (data not shown).

We found that polymorphisms exist within both exons and introns of the TERT gene. A salient and unexpected observation was that in some locations, individuals were either homozygous to the reference sequence, or had polymorphisms that were heterozygous SNPs. The variant homozygote SNPs at these loci were never observed. This is likely due to the fact that dog breeds as populations are not subject to Hardy-Weinberg Equilibrium: matings are not random, and traits are deliberately selected for or against by human breeders. However, this phenomenon may also be explained by a deleterious effect of the alternative homozygous variant: it could be lethal for a conceptus. This is a definite possibility, given that inbred populations tend to have a higher concentration of deleterious recessive traits than less inbred populations [20]. Alternatively, the small sample size analyzed could be the reason for the lack of variants. In 7 of the 14 exonic occurrences, the variant caused a coding change and thus could prove deleterious. However, it is unknown if the alleles have dominant negative effects.

We expected to observe breed-specific SNPs shared amongst the majority of individuals in each breed. We also hypothesized the specific SNPs shared in each breed would be distinct from other breeds, possibly in correlation to breed size and lifespan. However, it was found that SNPs occurred in individuals without breed specificity, and there was no correlation between dog size and the observed polymorphisms; most strikingly this was seen in the close relation of the Shi Tzu and Irish Wolfhound in Figure 2b (SH2 and IW5, respectively). This lack of distinct breed association in addition to the geographic and genetic isolation of the breeds’ places of origin implies that polymorphic diversification of TERT preceded derivation of the four breeds studied. Despite the founder effects, population bottlenecks, and careful human selection for or against traits involved in breed creation, the TERT gene has remained diverse within and among dog breeds. The gene’s mutation rate and the prevalence of mutations do not seem to have been drastically altered by breed creation.

Access to breeds was limited in this study. Thus, future studies with expanded the sample sizes (individuals and breeds) and where the TERT genes in wolves and indigenous dogs from South East Asia were sequenced would further enhance our understanding of how and if the TERT gene has changed with the derivation of modern dog breeds. This also could help answer whether TERT is resistant to high levels of inbreeding, clarifying the actual distribution of the homozygous variants and their potential effects on viability.

Our results suggest that the TERT gene is not wholly responsible for differences in lifespan and cancer susceptibility seen in dog breeds. Furthermore, the diversity and surprisingly higher number of SNPs in TERT makes us infer that mutations in this sequence are well tolerated. However, given the heterozygous SNPs that were found, we believe that the gene is not immune to the depressive effects of inbreeding. Furthermore, previous work has identified quantitative trait loci in canine chromosome 7 (CFA7) and 15 (CFA15) that were associated with size and longevity [21]. The IGF1 gene in CFA15 was proposed as one candidate having a major influence on these traits. However, the density of SNPs used in this study would not conclusively eliminate TERT as an additional, related or independent modulator.

Mutations in the human TERT gene have been shown to be important in dyskeratosis congenita, aplastic anemia, and idiopathic pulmonary fibrosis, among other disorders [22]. In addition, heterozygous mutations in the TERT gene have been shown to impair telomerase activity through haploinsufficiency [22]. Despite the fact that many of these disorders are heritable, it is not clear as to what extent mutations of TERT contribute to these diseases. Given our findings, we propose that these diseases do not arise solely due to polymorphic variants of the TERT gene, especially since the canine TERT protein is the closest homologue to the human protein known thus far [23].

The data set(s) supporting the results of this article are included within the article (and its additional files).