Background

The uncoupling proteins (UCPs) in the mitochondrial inner membrane are members of the mitochondrial anion carrier protein family [1, 2]. Mammals have five UCP homologs, of which UCP1, UCP2, and UCP3 are closely related, while UCP4 and UCP5 are more divergent from the other UCPs [3].

Based on genetic association studies, UCP2, UCP3, or both are reportedly associated with obesity, insulin resistance, type 2 diabetes mellitus, and metabolic syndrome in humans [4–11]. For example, a SNP in the 5′ untranslated region in human UCP3, the UCP3 -55CT SNP, is known to be a genetic marker associated with mRNA expression [12], elevated high density lipoprotein cholesterol levels, a reduced body mass index (BMI), weight, waist circumference, waist to hip ratio, fat mass, low density lipoprotein (LDL) cholesterol, and total cholesterol (T-Cho) [13–15].

The treatment and prevention of obesity and metabolic-related diseases are also clinically important in dogs [16–25]. Our previous report showed that the nucleotide sequences, predicted amino acid sequences and the genomic structures of human UCP2 and UCP3 are highly homologous to the canine orthologs [26, 27]. In this study, we investigate whether the dog UCP2 and UCP3 genes are associated with alterations in metabolism.

Results and discussion

Figure 1 shows a schematic representation of the canine UCP2 and UCP3 genes and the identified DNA polymorphisms from 119 animals from 11 breeds. For analysis of the dog UCP2 gene, six regions were individually amplified from genomic DNA and sequenced. We then identified 10 SNPs (9 intronic and 1 exonic) and 4 indels (intronic) in UCP2 (Figure 1, Additional file 1). In the dog UCP3 gene, 13 SNPs (11 intronic and 2 exonic) and 1 indel (exonic) were revealed by sequencing nine regions of this gene (Figure 1, Additional file 1).

Figure 1
figure 1

Schematic representation of the DNA polymorphisms detected in the UCP2 and UCP3 genes in dog. □: Exon(UTR) ■: Exon(CDS) ▬: Intron │: SNP or INDEL : PCR Fragment. The position of identified DNA polymorphism was numbered from the A of the initiator methionine ATG codon as the +1 revealed in exon. In case of intron, a positive number indicates the number of nucleotides away from the previous exon, while a negative number indicates the number of nucleotide away from the next exon.

Full size image

To test the association between the dog UCP2 and UCP3 genes and metabolic data, we determined the genotype of 50 Labrador Retrievers for each of 14 polymorphic sites (10 SNPs and 4 indels) in the UCP2 gene, and examined whether any of the genotypes were associated with biochemical measurements of glucose (GLU), total cholesterol (T-Cho), lactate dehydrogenase (LDH), or triglyceride (TG). To exclude any contamination by disease of the animals, we selected Labrador Retrievers that had undergone a health examination for breeding for guide dogs by the Kyushu Guide Dog Association.

The average of measurements was calculated with respect to the genotype group. Nine of the 14 loci in the UCP2 gene were polymorphic in this population of Labrador Retrievers. None of these DNA polymorphisms in the UCP2 gene were significantly associated with any of the biochemical parameters in this study (Additional file 2). We also subjected the 14 polymorphic sites (13 SNPs and 1 indel) in the UCP3 gene to this association analysis. Ten of the 14 sites were polymorphic in this population of Labrador Retrievers. There were no significant differences between genotype and GLU, LDH, or TG measurements for any polymorphic site. However, the T-Cho levels differed significantly among the genotype groups at four sites: -4399C/T, -4339T/C, -930T/C and -803C/T in intron 1 of the UCP3 gene (UCP 3 intron1). The average T-Cho levels in dogs carrying CC or CT at -4399 C/T were 273.5 ± 49.0 and 237.2 ± 53.3, respectively. The average T-Cho levels for the TT, TC, or CC genotypes at -4339T/C and -930T/C were 264.3 ± 49.6, 276.9 ± 49.5, and 233.5 ± 51.2, respectively. Those for CC or CT at -803C/T were 271.6 ± 49.5 and 239.1 ± 54.5, respectively (Table 1). The genotype distributions were in a Hardy–Weinberg equilibrium.

Table 1 Association analysis of UCP3 DNA polymorphisms with biochemical parameters among healthy Labrador Retrievers
Full size table

Shetland Sheepdogs are considered to have a predisposition to primary hyperlipidemia as determined by the levels of cholesterol, triglycerides, and free fatty acids [28, 29]. Therefore, we investigated the distribution of genotypes for SNPs and indels of the UCP2 and UCP3 genes in a population of Shetland Sheepdogs (n = 30). Shiba (n = 30) were also tested as a comparative contrast breed in this study. Statistically significant differences in allele frequency between the two breeds were found in five of the 14 polymorphic sites in UCP2 (-3629C/G, -2931A/T, -748G/A, -636A/G and IVS6-133delTCTCCCC, Additional file 3). Four SNPs (-4339T/C, -930T/C, 143A/C and IVS3+121T/C) of the 14 UCP3 polymorphic sites were significantly different in allele frequency between the two breeds (Table 2). Despite the different genetic background in each of the dog breeds [30–32], the different allele frequencies in the UCP2 and UCP3 polymorphic site between the two breeds may result from the susceptibility of Shetland Sheepdogs to hypercholesterolemia in a limited number of individuals.

Table 2 Genotyping data and interbreed analysis of DNA polymorphisms in UCP3
Full size table

The T allele at -4339T/C and -930T/C located in the UCP3 intron 1 is associated with higher T-Cho levels, as shown by two different experiments: the association between polymorphisms and metabolic data (Table 1), and the distribution of allele of genotype in the breed that is susceptible to hypercholesterolemia (Table 2). These results suggest that the dog UCP3 gene might be associated with T-Cho levels in a limited number of individuals.

It is known that the peroxisome proliferator activated receptors (PPAR) ligands activate UCP3 expression [33, 34]. The UCP3 intron 1 contains that the putative binding elements of MyoG/MyoD, PPARγ/RXRα and SP1/SP3 that enhanced the UCP3 gene transcription mainly regulated by PPARs in hamster, rat, and mouse [33]. Recently, we find the similar nucleotide sequences of the PPARγ/RXRα element in the dog UCP3 intron 1 (Canine Genome Draft, NC_006603.3). These findings imply that the dog UCP3 intron 1 may be associated with regulation of UCP3 gene expression. Further studies will be needed to demonstrate whether PPAR ligands bind or not this intronic region in dog.

With each genetic study, a different sample size is used to identify the candidate gene associating with genotypes and phenotypes in common diseases (multifactorial diseases) and/or single gene disorders. For instance, genome-wide association studies (GWAS) have reported the candidate gene associated with a mild form of disproportionate dwarfism using 23 cases and 37 controls [35], atopic dermatitis using 91 cases and 88 controls [36], and the chromosomal region of Patellar Luxation using 45 cases and 40 controls [37]. Some of the candidate genes were also tested using more than a hundred samples. The examination of larger sample sizes and further analysis will lead to increased precision of our results. In addition, because the association analysis in this study was performed using only polymorphisms within the UCP2 and UCP3 genes, we cannot exclude the possibility that a gene that is closely linked to UCP3 is causal.

Conclusions

A genetic association study between polymorphisms in the dog uncoupling protein 2 and 3 genes and metabolic data showed that the SNPs of the UCP3 intron 1 were associated with T-Cho levels in Labrador Retrievers. Alleles associated with high T-Cho levels of these polymorphisms were also present at higher frequencies in a breed that is susceptible to hypercholesterolemia (Shetland Sheepdogs), than in the control group (Shiba). The results obtained from a limited number of individuals suggest that the UCP3 gene in dogs may be associated with total cholesterol levels. Therefore, the UCP3 gene could be an interesting target, not only for lipid metabolism, but also for the treatment and prevention of obesity and metabolic-related diseases in dogs.

Methods

Animals and DNA

All animal experiments were approved by The Experimental Animal Ethics Committee in Nippon Veterinary and Life Science University. The blood samples were originally collected at the Veterinary Medical Teaching Hospital at NVLU with the written consent of each owner or the Kyushu Guide Dog Association. The collection of samples was handled by licensed veterinarians only.

Panel 1, for the first SNP discovery, was collected from 11 dogs that represented 11 different breeds: Miniature Dachshund, Welsh Corgi, Labrador Retriever, Shetland Sheepdog, Beagle, Yorkshire Terrier, Dobermann, Whippet, Weimaraner, Papillon, and Shiba. Panel 2 was used for SNP discovery and a study of associations between SNP variants and biochemical parameters; these samples were collected from 50 Labrador Retrievers. Panel 3 was used for SNP discovery and an interbreed analysis was collected from 30 Shetland Sheepdogs and 30 Shibas containing each one animals from Panel 1. A list of breeds and number of individuals are presented in Table 3. Genomic DNA was extracted from whole blood with the Puregene kit (Qiagen, Valencia CA, USA).

Table 3 List of 119 DNA samples from 11 breeds
Full size table

PCR

We used sequences of UCP2 and UCP3 (Canine Genome Draft, NC_006603.3), to design 15 pairs of primers for amplification of each exon of the UCP2 and UCP3 genes (Table 4). Each PCR using TaKaRa Ex Taq was performed in a total volume of 25 μl and contained 20 ng genomic DNA, 2.5 μl 10× Ex Taq Buffer (including 20 mM Tris–HCl, 100 mM KCl, 0.1 mM EDTA, 1 mM DTT, 0.5% Tween 20, 0.5% Nonidet P-40, 50% Glycerol, 20 mM Mg2+), 0.4 mM of each primer, 200 μM dNTP (dATP, dTTP, dCTP and dGTP), and 1U TaKaRa Ex Taq (TaKaRa, Shiga, Japan). Each PCR using FastStart Taq DNA polymerase (Roche, Basel, Switzerland)) was performed in a total volume of 25 μl and contained 20 ng genomic DNA, 2.5 μl 10× reaction Buffer (including 500 mM Tris–HCl, 100 mM KCl, 50 mM (NH4)2SO4, 20 mM MgCl2), 0.4 mM of each primer (F12: 0.2 mM of each primer), 200 μM dNTP (dATP, dTTP, dCTP and dGTP), and 1U FastStart Taq DNA polymerase. If necessary, we used FastStart Taq for primer pairs that did not work with TaKaRa Ex Taq. The PCR reactions were performed on TaKaRa PCR Thermal Cycler Dice TP600 (TaKaRa). The conditions for PCR are shown in Table 5.

Table 4 Sequences of primers for PCR
Full size table
Table 5 Conditions for PCR
Full size table

Sequencing and SNP detection

The PCR products were purified with High Pure PCR Product Purification Kit (Roche). Cycle sequencing was then performed with the Big Dye Terminator v3.1 kit (Applied Biosystems, Foster City CA, USA); each reaction was run in a 10 μl reaction volume containing 1 μl purified PCR amplification product, 1 μl Ready Reaction Premix, 1.5 μl 5× Big Dye Sequence Buffer, 1 μl primer (1.6 pmol/μl), and 5.5 μl sterile water. Cycle sequencing reactions were performed with the following conditions: 60 s at 96°C followed by 25 cycles of 10 s at 96°C, 5 s at 50°C and 4 min at 60°C. BigDye Xterminator Purification kits were used according to the manufacturer’s instructions (Applied Biosystems) to purify dye-labeled fragments. Samples were analyzed on an ABI PRISM 310 genetic analyzer (Applied Biosystems). We identified DNA polymorphisms by comparing each sequence with the reference sequence (Canine Genome Draft. NC_006603.3) by BLAST in NCBI (National Center for Biotechnology Information) and GENETYX program Ver. 11(GENETYX Corporation, Tokyo, Japan). The position of identified DNA polymorphism was numbered from the A of the initiator methionine ATG codon as the +1 revealed in exon. In case of intron, a positive number indicates the number of nucleotides away from the previous exon, while a negative number indicates the number of nucleotide away from the next exon.

Measurement of biochemical parameters

Blood samples were collected into heparinized plastic tubes at least 12 h postprandial. Plasma was separated by centrifugation at 1500× g for 10 min. Glucose (GLU), triglyceride (TG), total cholesterol (T-Cho), and lactate dehydrogenase (LDH) were measured using a Spotchem EM SP-4430 (Arkray, Kyoto, Japan) with the manufacturer’s reagents.

Statistical analysis

Deviation from the Hardy–Weinberg equilibrium was assessed by the Chi-squared test. SNPAlyze (Dynacom, Chiba, Japan) was used to estimate haplotype frequencies. Genotype frequencies were compared using the Fisher’s exact test. Differences of p < 0.05 were considered statistically significant. Associations between genotype frequencies and metabolic data were analyzed by one-way analysis of variance (ANOVA).