Partial sequencing of ESR1 and CDK5RAP2 genes in dogs with mammary tumours 

Canine mammary tumours (CMT) are among the most common canine cancer types in female dogs. Although the development of CMTs is multifactorial, high breed susceptibility indicates the genetic basis of the disease (Babu et al., 2012; Borge et al., 2011). Studies on candidate genes, which have been suggested to be associated with CMT, are increasing in the last decades. The genes encoding estrogen receptor (ER) and progesterone receptor (PR), which have been reported to be present < 50% in mammary tumours in dogs and proliferative markers like P53 and C-erb B2 are important candidates for diagnose and prognosis of CMTs (Veena et al., 2014). The importance of these studies also originated from the similarities with human breast cancer. Dogs provide an adaptable model system for human breast cancer studies. Both species share complex and monogenic diseases, a similar gene set and same environment. Similar to women, increasing age is associated with development of mammary tumours (Melin et al., 2016).
        
It is important to identify the underlying genetic basis to improve early detection of cancer and dog breeding systems (Rivera and Euler, 2011). An important problem related with identification of predisposing genes in CMTs is the genetic heterogeneity between dog breeds. It is unclear whether the results of previous association studies can be applied to other dog breeds or even different populations from the same breed. (Borge et al., 2013; Melin et al., 2016).  Previously identified genetic risk factors associated with CMT in a particular breed need to be validated in other dog breeds with further research.
        
Estrogen receptor 1 (ESR1) gene is one of the candidate genes associated with CMTs (Borge et al., 2013;Melinet_al2016). The ESR1 gene encodes an estrogen receptor which acts as a ligand-activated transcription factor. It takes part in the sexual development and reproduction system and it is also involved in pathogenesis of breast cancer. Previous studies on humans suggested an association between genetic variations of ESR1 and breast cancer (Dahlman-Wright et al., 2006). Borge et al., (2013) have found a SNP in exon 4 of ESR1 in dogs associated with CMT, which supported ESR1 as a candidate gene also in dogs. This non-synonymous SNP is an A/G transition, which leads to an amino acid change from isoleucin to leucin. Risk allele is A for this SNP. Comparison of two SNPs both in human and canine ESR1 exon 4 revealed that ss244244343 in dogs and rs1801132 in human are located in similar positions, difference is only 1bp. The codon including this SNP encodes amino acids in the hormone binding domain of the human ESR1, which is related to receptor dimerization, chaperon binding and co-regulator elements (Anghel et al., 2010).
        
Another candidate gene in dogs with CMT is cyclin-dependent kinase 5 regulatory subunit-associated protein 2 (CDK5RAP2) gene. CDK5RAP2 protein expression is important in cell cycle checkpoint. The levels of this protein can also influence cancer treatment response. The knock-down mammary cancer cell lines are found to be more resistant to paclitaxel and doxorubicin (Zhang et al., 2009). Melin et al., (2016) identified a genome-wide significant peak on CFA11, which includes the CDK5RAP2 gene. According to the results of different dog cancer genome-wide association studies (GWAS) cyclin-dependent kinases have been suggested to have an important role in cancer development (Rao et al., 2008). The most associated SNP observed in the above mentioned study is a T/C transition downstream of CDK5RAP2 gene with risk allele C and protective allele T. This SNP is thought to play a role in the function of a photoreceptor cell-specific nuclear receptor, which is involved in mammary cancer pathogenesis due to its relationship with ESR1. In ER positive breast cancer cells, CDK5RAP2 has a role in tumour growth, cell migration and metastasis through regulating ESR1 (Park et al., 2012; Zhao et al., 2015).
        
The aim of the present study was to determine novel mammary cancer associated variations and also to check the validity of previously reported SNPs in ESR1 and CDK5RAP2 genes in dogs with mammary tumours.

Samples
 
DNA samples from 25 dogs with CMTs and 10 dogs with healthy mammary glands were obtained from DNA collection in Biochemistry Department, Faculty of Veterinary Medicine, Istanbul University (Istanbul University Animal Researches and Ethic Committee - protocol number 126).
        
Data regarding breeds, vaginal cytology examination results, CMT grouping made by histopathological evaluation were obtained from the databank established in previous studies (e.g. Enginler et al., 2014). Histopathological diagnosis was evaluated on haematoxylin and eosin (H&E) stained sections by the WHO’s classification for canine mammary tumours, dysplasias and normal mammary glands. Breeds and CMT groups (malignant/benign) of samples were given in Table 1.
 

        
The CMT group consisted of 15 intact and 10 previously spayed dogs and all the dogs in control group were intact. According to their vaginal cytology, only DNA samples from dogs in anoestrus were included in this study. Dogs which have been spayed 1 year before the examination were accepted as in anoestrus.
 
Polymerase Chain Reaction and Sequencing
 
Exon 4 of ESR1 and an intergenic region downstream of CDK5RAP2 were amplified. Primers used for amplification, their anne- aling temperatures and product sizes were given in Table 2.
 

        
PCR amplifications were performed in a reaction volume of 25 μl using 1 U Taq polymerase, 2 μl 10XPCR buffer (100 mMKCl, 20 mM Tris HCl (pH 8.0), 0.1 mM EDTA, 0.5 mM PMSF, 1 mM DTT, 50 % glycerol), 2.5 mM MgCl2, 50-100 ng genomic DNA, 100μ MdNTP and 10 pmol of each primer. Amplification was carried out for 94°C for 2 min; 30 cycles of 94°C for 60 s, annealing temperature for 60 s, 72°C for 60 s and a final extension at 72°C for 10 min. Sequencing was performed by using an ABI-3100 sequencer (PE Biosystems, Germany) and the BigDyeTM terminator cycle sequencing kit, after purification of the PCR products.
 
Single nucleotide polymorphism genotyping and asso- ciation analysis
 
Nucleotide sequences of the amplified regions were aligned by using Clustal W program in the MEGA 6 software program (Tamura et al., 2013). The previously identified SNPs were obtained from GenBank for comparison. All positions given in the study are from CanFam3.1.
        
Association analysis was performed in order to examine if the observed SNPs are related to CMT by comparison allele frequencies in cases and controls. The significance of the association was calculated by using Pearson’s two-sided chi-square test in SPSS 13.0 software. Odds ratios (OR) with 95% confidence intervals (CI) were estimated by using unconditional logistic regression. Due to the small number of benign cases (n=2), no statistical analysis was performed to compare allele frequencies between malignant and benign cases.

Naturally occurring tumours in animals are excellent models of human cancer due to their genetic and pathological similarities. As a part of the One Health Initiative, comparative oncology is a research area important for new diagnostic and therapeutical discoveries for both humans and animals (Davis and Ostrander, 2014). Dog is a prominent model for especially studies on breast cancer. Unravelling of the dog genome in 2005 accelerated researches on genetic structure of dogs associated with CMTs. Nevertheless, to date there are limited data obtained from studies on inherited genetic risk factors of CMT (Rivera and von Euler, 2011).
        
In this research we studied SNPs in partial regions in/near two cancer associated genes, namely ESR1 and CDK5RAP2 in a cohort comprised of different dog breeds.
        
Results of the association analysis including minor allele frequency (MAF), odd ratios (OR) and p-values of the SNPs observed in this study were given in Table 3.
 
@tablw3
 
Association of ESR1 polymorphisms
 
Sequencing of exon 4 of ESR1 revealed two previously identified SNPs. No novel polymorphism was observed. The A/G substitution (rs21970417) at position 42,208,601 is a synonymous polymorphism. rs21970417 was found to be monomorphic in our sample set with only GG homozygous individuals, therefore no further statistical analysis was performed for this locus.
        
The second SNP (ss244244343) at position 42,208,686, an A/G transition, is a mis-sense polymorphism, which leads to an amino acid substitution from isoleucine to leucine. According to previous studies, A is identified as the risk allele and T is protective allele for CMT. No TT homozygous individuals were observed in cases and controls. The risk allele had higher frequencies both in cases and controls, but its frequency is higher in control group (0.91), than the frequency in dogs with CMT (0.83). But the difference between cases and controls was found not to be statistically significant.
        
There are several CMT-associated SNPs observed in intronic and exonic SNPs in ESR1 gene. But one of them in exon4 (ss244244344), which we also identified, was found to be significant after Bonferroni correction. According to the association study on English Springer Spaniel (ESS) samples from Sweden, A allele of the SNP ss244244344 was found as the risk allele for CMT. The risk allele A of ss244244343 is observed to be the major allele in the study on ESS. The same study had also another dataset comprised of 450 individuals from nine different dog breeds. These breeds, excluding ESS, varied in allele frequencies of risk and protective alleles. In our study, the frequency of risk allele A was found to be lower in cases than the frequency in the control group contrary to ESS dataset from Sweden (Borge et al., 2013). Although no significant correlation was found in our study, allele frequencies point out to the G allele as a possible risk allele for our dataset. Comparison of the high and low risk breeds also supported A allele as a risk factor with a higher frequency in all of the samples from high risk breeds. Nevertheless, there are also large allele frequency differences between breeds in the same risk group. For example, Bernese Mountain Dog in the low risk group was monomorphic for A allele, just like Dachshund from the high risk group, whereas two high risk breeds Collie and Shetland Sheepdog had frequencies of protective G allele 0.791 and 0.766, respectively (Borge et al., 2013). Our results also supported a genetic heterogeneity in terms of associated risk alleles.
 
Association of CDK5RAP2 polymorphisms
 
The SNP (BICF2G630310626) was observed in a region downstream of CDK5RAP2. The polymorphism is a C/T transition, while C is risk allele and T is protective allele. All animals in the control group were found to be TT homozygous. Only 2 CC homozygous individuals were observed in the group with CMT. The frequency of protective allele is higher in total. Two novel SNPs were identified in this region. One of these is an A/G transition and the other one is an A/T transversion. The latter two SNPs are inherited together and can be identified as a haplotype. Minor allele frequencies of both novel identified SNPs were found to be lower in cases than controls. But according to the association analysis performed in cases and controls, no statistically significant association was found between these SNPs and CMTs.
        
Many key cell cyclins take part in tumour pathogenesis. It has been reported that cyclin-activating cyclins were overexpressed, whereas cell cycle inhibitors were underexpressed in canine mammary tumour cell lines. (Rao et al., 2008).
        
Melin et al., (2016) identified a genome-wide significant locus on chromosome 11, which overlaps the CDK5RAP2 gene. Researchers also reported that the results indicating significant CMT association of these regions, need to be validated in other dog breeds and for other geographical locations with further studies. In this study, we performed SNP genotyping of  partial sequence of the same region to check the relationship with CMT.
        
Although all the animals in the control group are TT homozygote, which is consistent with the protective status of the T allele, risk allele had a very low frequency in cases. Therefore, no significant association was detected between minor allele C and CMT. The limited number of dogs with CMT analysed in the study and different breeds found in the study cohort, may have revealed these results. Another possibility to consider is that  the relationship between C allele and CMTs are not valid for our dataset, but sample numbers per breed are not enough for confirmation.
        
According to the results of a study on 336 ESS samples, the risk allele C is accumulated in ESS population in Sweden. Consistent with the results of our study, no significant association was observed in ESS population from United Kingdom and Norway, with sample numbers 40 and 15, respectively (Melin et al., 2016). Effects of the artificial selection on genes of different dog breeds or sub-populations from different countries may cause the difference (Ostrander and Franklin, 2012). The small sample size in three studies including our study does not allow us to make a definitive conclusion. Further studies on different dog breeds, as well as from different geographical regions with larger sample numbers would give important information on differences of genes and pathways involved in CMTs.
        
Most of the studies on the genetic basis of canine mammary tumours were conducted on ESS from Sweden, which is accepted as a high risk breed. This breed may be an adaptable model for human studies, but because of the fact that spaying of dogs is uncommon in Sweden (Jitpean et al., 2012), hormonal factor should be taken into consideration during genetic analysis of this breed. Although number of intact dogs in our dataset is higher than the spayed samples, results of this study differ from those of studies on ESS samples.
        
The advantage of dog studies in which the dog can be accepted as a model animal, originated from limited intra breed heterogeneity. But high genetic heterogeneity between breeds is complicating for evaluation of the results from different breeds from the same risk group (Borge et al., 2013).
        
The prevalence of CMTs exhibit differences among breeds. There are different high risk breeds in different geographical locations. Moreover, for a specific geographical region, various studies report different breeds as high risk breeds (Dobson, 2013; Gupta et al., 2014). These controversies arise from geographic and breed popularity contributions and also from the fact that insufficient data were obtained up to date (Goebel and Merner, 2017).
        
Especially studies on pure-bred dogs could provide important insights into the genetic aetiology of different forms of cancer, because of their widely homogenous genetic structure shaped by artificial selection.

Cancer is a very complex disease, with a pathogenesis involved by many genes. Both the ESR1 and CDK5RAP2 genes analyzed in this study seem to have heterogeneity between different dog breeds. To overcome this issue, larger datasets of different breeds, even subpopulations, should be analyzed in further studies. Case and control studies within a single breed is important to compare the results from different breeds. Whether the risk alleles are the same for all of the high risk breeds or different loci playing roles in tumour incidence should be investigated in future studies.

This study was supported by Istanbul University, Faculty of Veterinary Medicine with project number VET-TB-16-6 and by the Scientific Research Projects Coordination Unit of Istanbul University with project number BEK-2017-24017.