Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Aug 1.
Published in final edited form as: Biogerontology. 2016 Mar 10;17(4):725–736. doi: 10.1007/s10522-016-9643-y

Single nucleotide polymorphisms linked to mitochondrial uncoupling protein genes UCP2 and UCP3 affect mitochondrial metabolism and healthy aging in female nonagenarians

Sangkyu Kim 1, Leann Myers 2, Eric Ravussin 3, Katie E Cherry 4, S Michal Jazwinski 1
PMCID: PMC4935613  NIHMSID: NIHMS767739  PMID: 26965008

Abstract

Energy expenditure decreases with age, but in the oldest-old, energy demand for maintenance of body functions increases with declining health. Uncoupling proteins have profound impact on mitochondrial metabolic processes; therefore, we focused attention on mitochondrial uncoupling protein genes. Alongside resting metabolic rate (RMR), two SNPs in the promoter region of UCP2 were associated with healthy aging. These SNPs mark potential binding sites for several transcription factors; thus, they may affect expression of the gene. A third SNP in the 3′-UTR of UCP3 interacted with RMR. This UCP3 SNP is known to impact UCP3 expression in tissue culture cells, and it has been associated with body weight and mitochondrial energy metabolism. The significant main effects of the UCP2 SNPs and the interaction effect of the UCP3 SNP were also observed after controlling for fat-free mass (FFM) and physical-activity related energy consumption. The association of UCP2/3 with healthy aging was not found in males. Thus, our study provides evidence that the genetic risk factors for healthy aging differ in males and females, as expected from the differences in the phenotypes associated with healthy aging between the two sexes. It also has implications for how mitochondrial function changes during aging.

Keywords: healthy aging, frailty, energy metabolism, mitochondrial uncoupling, single nucleotide polymorphism

Introduction

RMR measures the energy demand for body functioning at rest. Normally, the amount of energy consumed for resting metabolism exceeds that for physical activity several-fold (Ruggiero and Ferrucci 2006). Both of these major forms of energy expenditure decline with increasing age (Kim and Jazwinski 2015). In nonagenarians, elevated RMR is associated with declining health, which is represented by increased FI34, a frailty index based on 34 health variables (Kim et al. 2013). FI34 is a reliable indicator of biological age, with a heritability estimate of 0.39 (Kim et al. 2013). Some factors show gender-specific associations with FI34. In male nonagenarians, circulating creatine kinase (CK) is positively associated with FI34; in female nonagenarians, fat mass (FM) and FFM are important contributors to healthy aging (Kim et al. 2014).

UCP2 and UCP3 (UCP2/3) are two of the five homologs of UCP1, the first UCP protein demonstrated to have a mitochondrial uncoupling function (Gimeno et al. 1997; Matthias et al. 2000). UCP2/3 share 57-59% amino-acid sequence identity with UCP1 (Krauss et al. 2005). UCP4 and UCP5 have much lower sequence identity with UCP1, and relatively less is known about the functions of these homologs (Ho et al. 2012; Kwok et al. 2010).

The original uncoupling function assigned to UCP1 involves dissipation of the proton motive force to thermogenesis (Rousset et al. 2004). Some protons in the mitochondrial intermembrane space are transferred to the matrix by proton carriers without participating in oxidative phosphorylation. The best known example of this proton ‘leak’ is dissipation of the proton motive force as heat in brown adipose tissue (BAT), where ATP synthase expression is relatively low but UCP1 expression is high (Klaus et al. 1991; Thomas and Palmiter 1997). Unlike UCP1 expression, UCP2/3 expression is not limited to BAT. UCP3 mRNA and protein are abundant in skeletal muscle and heart tissues (Boss et al. 1997; Rousset et al. 2004). Because of translational control, UCP2 protein is hardly detected in these tissues even though its mRNA is abundant (Velloso et al. 2009). Production of UCP2 protein can increase more than 10-fold under certain stress conditions, like fasting (Pecqueur et al. 2001). Thus, abundance of UCP2/3 mRNA or protein in tissues other than fat suggests that thermogenesis may not be the function of these proteins.

Indeed, UCP2/3 have multiple roles. They affect mitochondrial energy generation, key metabolite use, insulin secretion, and generation of reactive oxygen species (ROS). Thymocytes isolated from UCP2-knockout mice show lower proton leak, higher mitochondrial membrane potential, and higher cellular ATP concentrations, compared with those from wild-type mice (Krauss et al. 2002). These uncoupling proteins are carriers of such important metabolites as fatty acids and mitochondrial four-carbon metabolites (Jaburek et al. 1999; Vozza et al. 2014). UCP2 negatively regulates insulin secretion from pancreatic β cells by decreasing the ATP/ADP ratio (Zhang et al. 2001). UCP3 protein levels are lower in type 2 diabetic patients (Schrauwen et al. 2001). Besides their effects on metabolic processes, UCP2/3 can lower oxidative stress by reducing ROS generation from incomplete reduction of oxygen, and oxidative stress can be an important regulator of cell survival and aging (Mookerjee et al. 2010). Importantly, polymorphisms in UCP2/3 are linked to RMR in humans (Bouchard et al. 1997).

The functions of UCP2/3 in metabolism, bioenergetics, and ROS generation point to these genes as potential genetic risk factors for healthy aging. In this report, we examined association of the UCP2/3 genes with healthy aging by evaluating the relationship between single-nucleotide polymorphisms (SNP) in UCP2/3 and FI34 after adjustment for relevant covariates.

Materials and methods

Participants

Study participants were 67 Caucasian nonagenarians from the Louisiana Healthy Aging Study (LHAS) (Jazwinski et al. 2010; Kim et al. 2014), 37 females aged from 90 to 98 and 30 males aged from 90 to 97. Ages were based on demographic questionnaires and documentary evidence. All participants provided informed consent according to protocols approved by the respective Institutional Review Boards.

Genotyping and SNPs in UCP2 and UCP3

The Illumina GoldenGate assay was used for SNP genotyping (Jazwinski et al. 2010). SNPs were selected according to the Illumina Assay Design Tool, and quality control measures were applied using Illumina GenomeStudio. For SNP and sample clustering, the 10th percentile GenCall score was set at 0.4. Samples with call rates below 90% were excluded. UCP2/3 are located in tandem on chromosome 11, and the three SNPs analyzed in this study are polymorphic and in Hardy-Weinberg equilibrium (Table 1). To examine the additive genetic effect of each SNP in the multiple linear regression analysis, its genotype was numerically recoded according to the copy number of the minor allele (0, 1, 2).

Table 1. Genotype frequencies of SNPs analyzed.

SNP A/A A/B B/B MAF HWE
rs1626521 20 (0.54) 14 (0.38) 3 (0.08) 0.27 > 0.05
rs591758 15 (0.41) 16 (0.43) 6 (0.16) 0.38 > 0.05
rs675547 13 (0.35) 16 (0.43) 8 (0.22) 0.43 > 0.05
Open in a new tab

MAF = minor allele frequency; HWE = Hardy-Weinberg equilibrium; The major and minor alleles (A and B, respectively) correspond to G/A for rs1626521, C/G for rs591758, and A/G for rs675547; The numbers in parentheses are the proportions of the respective genotype in the sample

Data management and analysis

Ethnic origin of each participant was inferred genetically using Alu genotypes, and Caucasians were included in data analysis to avoid population confounding (Jazwinski et al. 2010). The summary statistics of the study variables for the 67 subjects are shown in Table 1. Details of data collection and calculation and characterization of FI34 were described elsewhere (Kim et al. 2013). FI34 is normally distributed among the nonagenarians. Measurements of body and metabolic parameters used in this study were described previously (Frisard et al. 2007). FFM is the total body weight multiplied by the proportion of non-fat mass, measured using dual-emission X-ray absorptiometry (DXA). RMR was measured by indirect calorimetry during rest in the fasting state. The energy expenditure summary index (EESI) is one of the three indices of the Yale Physical Activity Survey (Dipietro et al. 1993). It is calculated by multiplying the time for an activity by an intensity code and summing the product over all the activities for a week. All the statistical analyses were performed using R (R_Core_Team 2014).

SNP annotation

Linkage disequilibrium (LD) blocks for UCP2/3 were examined using Haploview with the standard color scheme using D′/LOD values (Barrett 2009). Images summarizing genomic features around UCP2/3 were obtained using the UCSC Genome Browser (Rosenbloom et al. 2015) and RegulomeDB (Boyle et al. 2012), based on the Human Feb. 2009 (GRCh37/hg19) Assembly.

Results

Characteristics of the nonagenarian sample

Male and female nonagenarians in our study sample differed in FI34, RMR, and FFM (Table 2). The mean FI34 of the females was significantly higher than that of the males (p = 0.0022), indicating that females were overall less healthy than males of the same age group. On the other hand, mean RMR and FFM values were much higher in males (p < 0.001). EESI means didn't differ, probably because EESI sums up mostly light or moderate physical activities in both genders.

Table 2. Basic characteristics of the LHAS nonagenarian subjects (mean ± standard deviation).

Variable Male (n=30) Female (n=37) All (n=67)
Age (years) 92 ± 2 92 ± 2 92 ± 2
FI34** 0.183 ± 0.073 0.232 ± 0.069 0.210 ± 0.075
EESI (kcal/w) 3731.5 ± 3038.3 3632.8 ± 2862.8 3677.7 ± 2921.3
RMR (kcal/d)*** 1281.8 ± 147.3 1029.2 ± 122.0 1142.3 ± 183.5
FFM (kg)*** 52.1 ± 5.1 37.4 ± 5.3 44.0 ± 9.5
Open in a new tab

EESI = Energy Expenditure Summary Index of the Yale Physical Activity Survey; RMR = resting metabolic rate; FFM = fat-free mass;

**

p ≤ 0.01 and

***

p ≤ 0.001 by Wilcoxon test between female and male

Although RMR goes down with age overall (Kim and Jazwinski 2015), it goes up with FI34 among nonagenarians, which was observed after adjustment for FFM, FM, and age (Kim et al. 2014). This suggests increasing resting energy consumption with deteriorating health. Although this association of RMR with FI34 is common to both sexes, we found male-specific elevation of circulating CK levels and female-specific decline in FFM (Kim et al. 2014).

SNPs rs591758 and rs675547 are significant predictors of FI34

The lower FFM in unhealthy female nonagenarians coupled to higher RMR is surprising. It suggests that features of energy metabolism, especially in muscle, may change in female nonagenarians as their health declines. To gain an understanding of these changes, we searched for genetic risk factors associated with these changes. Mitochondrial uncoupling proteins affect muscle energy metabolism in multiple ways. Therefore, we examined association of UCP2/3 SNPs with FI34 with adjustment for RMR. SNP rs591758 is predicted to affect binding of transcription factors (Fig. 1). It is a variant in risk haplotypes associated with type 2 diabetes in Caucasian women (Hsu et al. 2008). SNP rs675547 is also predicted to affect transcription factor binding sites (Fig. 2). Both rs591758 and rs675547 are in an LD block (Block 1) covering UCP2 (Fig. 3). SNP rs1626521, located in the 3′ untranslated region of UCP3 (Fig. 4), is associated with body weight and mitochondrial energy metabolism and affects expression of a reporter gene in an allele-specific manner (Acosta et al. 2015). Therefore, we chose these three SNPs (Table 1) to test for association with FI34.

Fig. 1.

Fig. 1

Fig. 1

Open in a new tab

UCSC Genome Browser and RegulomeDB outputs for rs591758 at position 73,698,059 on chromosome 11 (based on dbSNP141 and hg19). (a) Genomic features around rs591758, where the vertical dotted line is located, are displayed by UCSC Genome Browser. The grey box in the DNase I Hypersensitivity Clusters track indicates a hypersensitive site to DNase I. The Transcription Factor ChIP-seq tracks show transcription factor binding sites from ChIP-seq experiments carried out by the ENCODE project. The DNA binding motifs are from the ENCODE Factorbook repository, which can be viewed as a matrix of all ENCODE transcription factor ChIP-seq datasets. The darkness is proportional to the signal strength, and the green segment highlights the highest scoring-site motif, with the arrows in the green segment denoting the matching strand (5′ to 3′) of the motif. (b) According to RegulomeDB, SNP rs591758 scores 2b (“likely to affect binding”) for its potential effects on transcription factor binding, a motif, DNase footprint, and DNase peak. The SNP falls in FOXA protein binding sites (HepG2 is a hepatocellular carcinoma cell line). (c) The SNP also overlaps the MyoD binding motif CANNTG. PWM is the position-weight matrix method for transcription factor binding. The base position in the motif that overlaps the SNP is shown in a box. HelaS3lfna4h is a cervical carcinoma cell line.

Fig. 2.

Fig. 2

Fig. 2

Open in a new tab

UCSC Genome Browser and RegulomeDB outputs for rs675547 at position 73,700,409 on chromosome 11 (based on dbSNP141 and hg19). (a) Genomic features around rs675547, where the vertical line is located, are displayed by the UCSC Genome Browser. The grey box in the DNase I Hypersensitivity Clusters track indicates a hypersensitive site to DNase I. The Transcription Factor ChIP-seq track shows a CTCF binding site from ChIP-seq experiments carried out by the ENCODE project. The DNA binding motif is from the ENCODE Factorbook repository, which can be viewed as a matrix of all ENCODE transcription factor ChIP-seq datasets. The darkness is proportional to the signal strength, and the green segment highlights the highest scoring-site motif. Note that rs675547 is only 15 bp away from rs67870053, which is a 7-bp indel variation (-/ACCAACA from 73,700,388 to 73,700,394; MAF=0.46 by the 1000 Genomes Project). The last four base pairs of rs67870053 overlap the first four base pairs of the CTCF binding motif. (b) According to RegulomeDB, SNP rs675547 scores 2a (“likely to affect binding”) for its potential effects on transcription factor binding, matched TF motif, matched DNase footprint, and DNase peak. PWM is the position-weight matrix method for transcription factor binding. The SNP position adjoining the motif is shown in a box. Mcf7 is a mammary gland, adenocarcinoma cell line, and H1hesc is an embryonic stem cell line.

Fig. 3.

Fig. 3

Open in a new tab

SNPs in LD in the UCP2/3 region on chromosome 11. The SNPs examined in this study are enclosed in red, dotted lines. The Green line represents UCP2 and the blue UCP3, with SNP locations indicated. The LD plot was created using Haploview with the standard color scheme: bright red for D′ = 1, LOD ≥ 2 and all other colors for D′ < 1 or LOD < 2 or both. The numbers shown inside the boxes are D′ values multiplied by 100 and empty boxes represent the D′ value of 1.

Fig. 4.

Fig. 4

Open in a new tab

Genomic features around rs1626521, where the vertical line is located, are shown by the UCSC Genome Browser. SNP names enclosed in red, dotted lines are those shown in LD Block 2 in Fig. 3. SNPs whose names are colored black are in introns, green are in coding (synonymous), red in coding (non-synonymous), and blue in untranslated regions. At the top, below the coordinate scale, two splice variants are shown, and rs1626521 falls in the 3′ untranslated region of the shorter transcript. Different colors in the histone modification tracks represent results from different cell lines, and peak levels show enrichment levels of the corresponding histone marks as determined by ChIP-seq assays of the ENCODE project. The grey boxes in the DNase I Hypersensitivity Clusters track indicate hypersensitive sites to DNase I with the darkness proportional to the sensitivity. The number to the left of each box shows the number of cell lines in which the region was hypersensitive. The Transcription Factor ChIP-seq tracks show transcription factor binding sites from ChIP-seq experiments carried out by the ENCODE project. The darkness is proportional to the signal strength, and the green highlights indicate the highest scoring-site motifs. The TargetScan Regulatory Sites track shows conserved microRNA target sites for conserved microRNA families in the 3′ UTR regions of genes. Seven miRNA target sites were included in the track, but shown are the top three sites with highest performance prediction scores.

The UCP2 SNP rs591758 was significantly associated with FI34 without affecting the significance and direction of association of RMR (Table 3). The additive genetic effect of the minor allele of rs591758 was positively associated with FI34, indicating that an increase in the copy number of the minor allele is associated with unhealthy aging. Similar results were obtained with rs675547. These associations were observed only in the female group. On the other hand, no association of the UCP3 SNP rs1626521 with FI34 was seen; however, its inclusion in the model eliminated the significant association of the covariate RMR (Table 3). These results suggest that the UCP2 variants have an effect on healthy aging independent of RMR, while the effect of the UCP3 variant is closely related to it.

Table 3. Association of UCP2 variants with FI34 in female nonagenarians.

rs591758 rs675547 rs1626521
b se b se b se
RMR 2.7e-4 ** 7.7e-5 2.8e-4 ** 8.5e-5 2.3e-4 8.6e-5
SNP 4.3e-2 ** 1.3e-2 3.1e-2 * 1.4e-2 -8.1e-3 1.6e-2
R2 0.34 0.24 0.13
P 0.00035 0.0035 0.032
Open in a new tab

b = regression coefficient

*

p ≤ 0.05 and

**

p ≤ 0.01);

se = standard error of b; RMR = resting metabolic rate; SNP = additive genetic model for each SNP; R2= adjusted R2; P = p value of the coefficient of determination (R2) obtained by application of the F-statistic in analysis of variance; sample size (n = 37)

A significant interaction between rs1626521 and RMR

To evaluate our conclusions regarding the effects of the UCP2/3 variants, we examined interactions of the SNPs with RMR (Table 4). There was no significant interaction of the two UCP2 SNPs with RMR found. For the UCP3 SNP, however, a significant interaction with RMR was detected (p = 0.018; Table 4). This indicates that the association of RMR with FI34 is inseparable from UCP3 genotypes (Fig. 5).

Table 4. Interaction between RMR and the UCP3 variant explains FI34 in female nonagenarians.

rs591758 rs675547 rs1626521
b se b se b se
RMR 2.4e-4 1.4e-4 1.4e-4 1.6e-4 -7.1e-5 1.5e-4
SNP 1.6e-2 0.11 -9.4e-2 0.12 -0.40 * 0.16
RMRxSNP 2.6e-5 1.1e-4 1.2e-4 1.2e-4 3.9e-4 * 1.6e-4
R2 0.32 0.24 0.25
P 0.0013 0.0066 0.0057
Open in a new tab

b = regression coefficient

*

p ≤ 0.05 and

**

p ≤ 0.01);

se = standard error of b; RMR = resting metabolic rate; SNP = additive genetic model for each SNP; R2= adjusted R2; P = p value of the coefficient of determination (R2) obtained by application of the F-statistic in analysis of variance; sample size (n = 37)

Fig. 5.

Fig. 5

Open in a new tab

Interaction plot for RMR and rs1626521 in prediction of FI34 in female nonagenarians (N = 37) based on Table 4. There are only three individuals homozygous for the minor allele A (Table 1); therefore, these homozygotes (closed red circles each with an asterisk) were plotted together with heterozygotes as carriers of A (closed red circles; A/A or A/G) compared to non-carriers (open blue circles; G/G).

Association of the UCP2/3 SNPs in multiple covariate models

The models involving RMR alone did not adjust for other factors that may be related to the association of RMR with FI34 and thus affect the relationship between FI34 and the SNPs. FFM and EESI are two such likely factors (FM is the body mass remaining after subtraction of FFM, which is obtained by multiplying total body weight by percent non-fat body mass determined by DEXA. Thus, they are complementary measures.) UCP2/3 are expressed predominantly in tissues other than fat, especially in skeletal muscle, and FFM is associated with FI34 in the females, as stated earlier. Physical activity energy expenditure is related to skeletal muscle mass and function, and our studies have shown that it is related to FI34 (Kim and Jazwinski 2015). Therefore, we included FFM and EESI in multiple regression analyses (Table 5). In the models containing each of the individual UCP2 SNPs, the association of the SNP and RMR with FI34 remained, while FFM and EESI showed the expected, significant association with FI34. In the model with the UCP3 SNP, the same relationships were apparent, except that no association of the SNP with FI34 was found, as before. Thus, both FFM and EESI were inversely associated with FI34, indicating that muscle mass and physical activity are associated with healthy aging and that UCP2 contributes as well.

Table 5. Association of UCP2/3 SNPs with FI34 in female nonagenarians after adjustment for muscle mass and physical activity.

rs591758 rs675547 rs1626521
b se b se b se
FFM -6.0e-3 ** 2.1e-3 -6.3e-3 ** 2.3e-3 -5.1e-3 * 2.5e-3
EESI -8.6e-6 ** 2.8e-6 -9.1e-6 ** 3.0e-6 -1.1e-5 ** 3.3e-6
RMR 4.4e-4 *** 9.2e-5 4.6e-4 *** 1.0e-4 3.7e-4 ** 1.1e-4
SNP 4.0e-2 ** 1.1e-2 3.1e-2 * 1.2e-2 1.2e-3 1.6e-2
R2 0.56 0.49 0.38
P < 0.0001 < 0.0001 0.00066
Open in a new tab

b = regression coefficient

*

p ≤ 0.05,

**

p ≤ 0.01,

***

p ≤ 0.001);

se = standard error of b; FFM = fat-free mass; EESI= energy expenditure summary index in the Yale Physical Activity Survey; RMR = resting metabolic rate; SNP = additive genetic model for each SNP; R2= adjusted R2; P = p value of the coefficient of determination (R2) obtained by application of the F-statistic in analysis of variance; sample size (n = 37)

We next examined the interactions between RMR and the UCP2/3 variants in the presence of FFM and EESI, to determine whether they were affected by muscle mass and physical activity energy expenditure. The interaction between RMR and the UCP3 SNP remained significant, while no interaction was detected for the UCP2 SNPs as before (Table 6). Therefore, the models that take into account FFM and EESI confirm the association of UCP2/3 SNPs with FI34 seen when only RMR is included.

Table 6. Interaction between RMR and the UCP3 variant in female nonagenarians explains FI34 after adjustment for muscle mass and physical activity.

rs591758 rs675547 rs1626521
b se b se b se
FFM -5.7e-3 ** 2.1e-3 -5.4e-3 * 2.3e-3 -3.6e-3 2.3e-3
EESI -9.7e-6 ** 3.0e-6 -1.1e-5 ** 3.1e-6 -1.2e-3 *** 2.9e-6
RMR 3.1e-4 * 1.5e-4 2.1e-4 1.7e-4 5.1e-6 1.5e-4
SNP -6.5e-2 9.7e-2 -0.15 0.11 -0.41 ** 0.13
RMRxSNP 1.0e-4 9.5e-5 1.8e-4 1.0e-4 4.1e-4 ** 1.3e-4
R2 0.56 0.52 0.52
P < 0.0001 < 0.0001 < 0.0001
Open in a new tab

b = regression coefficient

*

p ≤ 0.05,

**

p ≤ 0.01,

***

p ≤ 0.001);

se = standard error of b; FFM = fat-free mass; EESI= energy expenditure summary index in the Yale Physical Activity Survey; RMR = resting metabolic rate; SNP = additive genetic model for each SNP; R2= adjusted R2; P = p value of the coefficient of determination (R2) obtained by application of the F-statistic in analysis of variance; sample size (n = 37)

Discussion

We examined three UCP2/3 SNPs that are predicted or known to be functional for their association with FI34. SNPs rs591758 and rs675547, which are linked to UCP2, were significantly associated with FI34, whereas the UCP3 SNP rs162651 was not (Tables 3 and 5). The UCP2 SNPs showed the same allele specificity. An increase in the copy number of the minor allele was associated with an increase in FI34 of about 0.03 to 0.04. Based on the exponential accumulation of health deficits (FI34 = 0.03 • e(0.02•age)) (Kim et al. 2013), an increase in age from 90 to 91 is expected to increase FI34 on average by 0.0037 in the population studied. Thus, the additive genetic effect of the UCP2 SNPs is substantially greater than the annual chronological effect at age 90. However, this must be placed in context. The activity of UCP2 occurs over a lifetime and not only for one year at advanced ages. In contrast to the UCP2 SNPs, the UCP3 SNP moderated the association of RMR with FI34 (Tables 4 and 6, Fig. 5), without an effect on its own. Thus, this differentiates the UCP3 SNP from the UCP2 SNPs, which likely reflects functional differences of the gene products. All the associations were specific only to female nonagenarians.

Functions of the SNPs and healthy aging

UCP2 variants have been associated with various traits related to body mass and composition. UCP2 negatively regulates insulin secretion by lowering the ATP/ADP ratio, implying a role in obesity and type 2 diabetes (Krauss et al. 2005; Rousset et al. 2004). A recent study highlights UCP2 as a transporter of important metabolites of the Krebs cycle rather than as an uncoupling protein associated with thermogenesis (Vozza et al. 2014). This study shows that UCP2 negatively controls oxidation of acetyl-CoA by transporting four-carbon intermediates, such as oxaloacetate, out of mitochondria. This export of metabolic intermediates, assisted by proton gradients, lowers the redox potential, ATP/ADP ratio, and possibly ROS generation (Vozza et al. 2014). Besides limiting glucose oxidation, UCP2 can promote mitochondrial processing of other energy sources such as fatty acids and glutamine. In fact, UCP2 can enhance fatty acid oxidation (Pecqueur et al. 2008) and possibly glutaminolysis, because its translation from mRNA is activated by glutamine (Hurtaud et al. 2007). Furthermore, overexpressed in many actively proliferating tumor cells, UCP2 can promote energy production from aerobic breakdown of glucose to lactate (Ayyasamy et al. 2011), a phenomenon termed the Warbug effect (Warburg 1956). Thus, increased UCP2 expression can alter cellular profiles of energy metabolism under certain physiological conditions.

Included in the same LD block, rs591758 and rs675547 are predicted to affect binding of transcription factors such as MyoD and the CCCTC-binding factor CTCF, respectively. Along with Myf6, MyoD belongs to a group of myogenic regulatory transcription factors (Anand et al. 1994; Cao et al. 2010). Indeed, MyoD can activate expression of UCP2/3 in several cell lines, including differentiating myoblasts, with the expression level depending on the availability of additional cofactors (Kim et al. 2007). CTCF is a highly conserved, zinc-finger transcription factor with various effects on gene expression (Filippova et al. 1996). CTCF enhances the myogenic activity of MyoD (Delgado-Olguin et al. 2011), and these two proteins interact physically through chromatin looping (Battistelli et al. 2014). In this study, we found that the additive genetic effects of the minor alleles of these UCP2 SNPs were positively associated with FI34. The minor allele (G) of rs591758 corresponds to the last sequence position of the MyoD-binding motif (CANNTG; Fig. 3c). Therefore, assuming an active role of this MyoD motif in UCP2 expression, the minor allele would promote binding of MyoD, activating UCP2 expression. We speculate that increased expression of UCP2 in the oldest-old could be a sign of altered cellular energy metabolism associated with unhealthy aging, similar to the Warburg effect in cancer cells. In other words, energy metabolism profiles may vary depending on the genotypes of the UCP2 SNPs. A better understanding of the relationship between UCP2 expression and healthy aging awaits experimental data on the functions of the SNPs and further characterization of the uncoupling protein.

The UCP3 SNP rs1626521 is functional and associated with body weight and mitochondrial energy metabolism in skeletal muscle (Acosta et al. 2015). Other SNPs in UCP3 are associated with childhood obesity and hand-grip strength (Crocco et al. 2011; Dato et al. 2012; Musa et al. 2012). The LD block 2 in which rs1626521 belongs contains seven miRNA target sites as well as a missense variant rs8179179 (Fig. 4), increasing the likelihood of the functionality of the SNP. Expression of a reporter gene in the presence of the minor allele is significantly higher than in the presence of the major allele (Acosta et al. 2015). The interaction of the SNP with RMR indicates that the positive association of RMR with FI34 depends on the minor allele of the SNP (Fig. 5). Specifically, an increase in RMR among the frail nonagenarian females accompanies an increase in UCP3 expression. It is unclear how exactly UCP3 expression relates to RMR in frail female nonagenarians. We speculate, however, that increased expression of UCP3 in the oldest-old with deteriorating health leads directly to elevated RMR without altering the essential characteristics of energy metabolism. Other studies have shown that an allele of a SNP (rs1800849) in the promoter region that drives higher UCP3 expression (Schrauwen et al. 1999) is associated with higher hand-grip strength (Crocco et al. 2011). However, this association was not significant in the nonagenarian sample. Furthermore, the female nonagenarians in this study displayed lower hand-grip strength. Thus, we suggest that increased UCP3 expression and associated increase in RMR compensates for the decline in health and related decline in muscle mass in female nonagenarians.

Oxidative damage

Another possibility is that the association of UCP2/3 with FI34 could be through their effects on mitochondrial ROS generation. The uncoupling proteins can protect cells from oxidative damage by reducing ROS production (Brand et al. 2004; Krauss et al. 2005). Superoxide even enhances proton conductance of skeletal muscle mitochondria by activating UCP2 and UCP3 (Echtay et al. 2002). This indicates operation of a coordinated feedback mechanism to regulate the mitochondrial proton leak and membrane potential. Mitochondrial membrane potential and ROS generation are highly correlated, and ROS generation is sensitive to changes in membrane potential; therefore, mild uncoupling through increased expression of UCP2/3 can be an efficient way of controlling mitochondrial oxidative damage (Echtay 2007).

UCP genes are associated with human longevity (Rose et al. 2011). The importance of oxidative damage as a determinant of longevity is found in rodent studies. For example, longevity in rats is correlated with ROS production (Mookerjee et al. 2010). Mice lacking UCP2 have shorter lifespans compared with the wild type, and mice lacking the superoxide dismutase 2 gene tend to live longer as the copy number of UCP2 increases (Andrews and Horvath 2009). However, other studies show no effect of sod2- on murine life span (Gallagher et al. 2000; Zhang et al. 2009). According to our current study, the minor alleles of the UCP2/3 SNPs examined were correlated with increased gene expression, which is evident from the gene reporter assay with the UCP3 SNP (Acosta et al. 2015). Because increased expression of UCP2/3 corresponds to increased uncoupling, and increased uncoupling is expected to reduce ROS production, the minor alleles should be associated with healthier aging. On the contrary, we found the opposite: all three minor alleles of the UCP2/3 SNPS were positively associated with FI34. Therefore, we think that reduction in ROS generation, if any, through increased uncoupling may have a minor effect on healthy aging. It is possible that the effect of the UCP2/3 gene variants on longevity comes from combinations of several functions of these genes rather than the sole effect on oxidative stress (Rose et al. 2011).

Gender specificity

The association of UCP2/3 with FI34 was specific to the female nonagenarians. The population study of rs1626521 by Acosta et al. (2015) involved both males and females, but mostly in their 30s. In our study, the largest gender difference was in FFM. The mean FFM of the male nonagenarians was about 40% higher than that of the female nonagenarians (p < 0.0001, Table 2). Because UCP2/3 expression is most abundant in FFM such as heart and skeletal muscle, one would expect greater associations of these uncoupling genes with the healthy aging of males. Yet, we didn't find such associations in males. However, we noticed that RMR per unit FFM is significantly higher in the female nonagenarians than in the males (27.5 vs 24.6, p < 0.0001). Therefore, any change in FFM would have a greater effect on energy metabolism in the females than in the males. Thus, the genetic effect of UCP2/3 on energy metabolism would be greater in the females.

A caveat to our conclusions is the sample size in this analysis. Although it begins with 67 nonagenarians, it then focuses on an examination of the females in this group who show a gender-specific association of FFM with healthy aging, by investigation of the impact of UCP2/3 on this association. Despite the relatively small sample, post-hoc power analysis showed that the 1-β values of all the models described here exceed 0.8 at n = 37 and α = 0.05.

In sum, we investigated the effect of UCP2/3 on healthy aging in nonagenarians. We used two predicted functional SNPs in UCP2 and an experimentally-validated functional SNP in UCP3. For each of the UCP2 SNPs, the minor allele copy number was positively associated with an increase in FI34. Unlike the UCP2 SNPs, the UCP3 SNP moderated the relationship between RMR and FI34. The net effect of the UCP3 SNP was the same as that of the UCP2 SNPs: the minor allele was positively associated with FI34. All these associations were specific to female nonagenarians. Our results provide evidence for the function of UCP2/3 in healthy aging in nonagenarian females.

Acknowledgments

We thank the people of Louisiana for participation in our studies. We also thank Katie Kwong, Natalie Harold, and Tiffany Kaul in graduate programs for their help with data management. This study was supported by grants from the National Institute on Aging of the National Institutes of Health (P01AG022064 to S.M.J.), the National Institute of General Medical Sciences of the National Institutes of Health (P20GM103629) to S.M.J. and S.K., the Louisiana Board of Regents through the Millennium Trust Health Excellence Fund [HEF(2001–06)-02] to S.M.J, Support from Pennington NORC P30DK072476 (E.R.), and by the Louisiana Board of Regents RC/EEP Fund through the Tulane-LSU CTRC at LSU Interim University Hospital.

Footnotes

Conflicts of interests: None declared.

References

  1. Acosta A, et al. Association of UCP-3 rs1626521 with obesity and stomach functions in humans Obesity. Silver Spring; 2015. pp. 23pp. 898–906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anand G, Shapiro DN, Dickman PS, Prochownik EV. Rhabdomyosarcomas do not contain mutations in the DNA binding domains of myogenic transcription factors. J Clin Invest. 1994;93:5–9. doi: 10.1172/jci116982. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Andrews ZB, Horvath TL. Uncoupling protein-2 regulates lifespan in mice. Am J Physiol Endocrinol Metab. 2009;296:E621–627. doi: 10.1152/ajpendo.90903.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Ayyasamy V, et al. Cellular model of Warburg effect identifies tumor promoting function of UCP2 in breast cancer and its suppression by genipin. PLoS One. 2011;6:e24792. doi: 10.1371/journal.pone.0024792. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barrett JC. Haploview: Visualization and analysis of SNP genotype data. Cold Spring Harb Protoc. 2009;2009:pdb ip71. doi: 10.1101/pdb.ip71. [DOI] [PubMed] [Google Scholar]
  6. Battistelli C, Busanello A, Maione R. Functional interplay between MyoD and CTCF in regulating long-range chromatin interactions during differentiation. J Cell Sci. 2014;127:3757–3767. doi: 10.1242/jcs.149427. [DOI] [PubMed] [Google Scholar]
  7. Boss O, et al. Uncoupling protein-3: a new member of the mitochondrial carrier family with tissue-specific expression. FEBS Lett. 1997;408:39–42. doi: 10.1016/s0014-5793(97)00384-0. [DOI] [PubMed] [Google Scholar]
  8. Bouchard C, Perusse L, Chagnon YC, Warden C, Ricquier D. Linkage between markers in the vicinity of the uncoupling protein 2 gene and resting metabolic rate in humans. Hum Mol Genet. 1997;6:1887–1889. doi: 10.1093/hmg/6.11.1887. [DOI] [PubMed] [Google Scholar]
  9. Boyle AP, et al. Annotation of functional variation in personal genomes using. RegulomeDB Genome Res. 2012;22:1790–1797. doi: 10.1101/gr.137323.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Brand MD, et al. Mitochondrial superoxide: production, biological effects, and activation of uncoupling proteins. Free Radic Biol Med. 2004;37:755–767. doi: 10.1016/j.freeradbiomed.2004.05.034. [DOI] [PubMed] [Google Scholar]
  11. Cao Y, et al. Genome-wide MyoD binding in skeletal muscle cells: a potential for broad cellular reprogramming. Dev Cell. 2010;18:662–674. doi: 10.1016/j.devcel.2010.02.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Crocco P, Montesanto A, Passarino G, Rose G. A common polymorphism in the UCP3 promoter influences hand grip strength in elderly people. Biogerontology. 2011;12:265–271. doi: 10.1007/s10522-011-9321-z. [DOI] [PubMed] [Google Scholar]
  13. Dato S, Soerensen M, Montesanto A, Lagani V, Passarino G, Christensen K, Christiansen L. UCP3 polymorphisms, hand grip performance and survival at old age: association analysis in two Danish middle aged and elderly cohorts. Mech Ageing Dev. 2012;133:530–537. doi: 10.1016/j.mad.2012.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Delgado-Olguin P, Brand-Arzamendi K, Scott IC, Jungblut B, Stainier DY, Bruneau BG, Recillas-Targa F. CTCF promotes muscle differentiation by modulating the activity of myogenic regulatory factors. J Biol Chem. 2011;286:12483–12494. doi: 10.1074/jbc.M110.164574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Dipietro L, Caspersen CJ, Ostfeld AM, Nadel ER. A survey for assessing physical activity among older adults. Med Sci Sports Exerc. 1993;25:628–642. [PubMed] [Google Scholar]
  16. Echtay KS. Mitochondrial uncoupling proteins--what is their physiological role? Free Radic Biol Med. 2007;43:1351–1371. doi: 10.1016/j.freeradbiomed.2007.08.011. [DOI] [PubMed] [Google Scholar]
  17. Echtay KS, et al. Superoxide activates mitochondrial uncoupling proteins. Nature. 2002;415:96–99. doi: 10.1038/415096a. [DOI] [PubMed] [Google Scholar]
  18. Filippova GN, et al. An exceptionally conserved transcriptional repressor, CTCF, employs different combinations of zinc fingers to bind diverged promoter sequences of avian and mammalian c-myc oncogenes. Mol Cell Biol. 1996;16:2802–2813. doi: 10.1128/mcb.16.6.2802. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Frisard MI, et al. Aging, resting metabolic rate, and oxidative damage: results from the Louisiana Healthy Aging Study. J Gerontol A Biol Sci Med Sci. 2007;62:752–759. doi: 10.1093/gerona/62.7.752. doi:62/7/752 [pii] [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gallagher IM, Jenner P, Glover V, Clow A. CuZn-superoxide dismutase transgenic mice: no effect on longevity, locomotor activity and 3H-mazindol and 3H-spiperone binding over 19 months. Neurosci Lett. 2000;289:221–223. doi: 10.1016/s0304-3940(00)01279-9. [DOI] [PubMed] [Google Scholar]
  21. Gimeno RE, et al. Cloning and characterization of an uncoupling protein homolog: a potential molecular mediator of human thermogenesis. Diabetes. 1997;46:900–906. doi: 10.2337/diab.46.5.900. [DOI] [PubMed] [Google Scholar]
  22. Ho PW, et al. Uncoupling protein-4 (UCP4) increases ATP supply by interacting with mitochondrial Complex II in neuroblastoma cells. PLoS One. 2012;7:e32810. doi: 10.1371/journal.pone.0032810. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hsu YH, Niu T, Song Y, Tinker L, Kuller LH, Liu S. Genetic variants in the UCP2-UCP3 gene cluster and risk of diabetes in the Women's Health Initiative Observational Study. Diabetes. 2008;57:1101–1107. doi: 10.2337/db07-1269. [DOI] [PubMed] [Google Scholar]
  24. Hurtaud C, Gelly C, Chen Z, Levi-Meyrueis C, Bouillaud F. Glutamine stimulates translation of uncoupling protein 2mRNA. Cell Mol Life Sci. 2007;64:1853–1860. doi: 10.1007/s00018-007-7039-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Jaburek M, et al. Transport function and regulation of mitochondrial uncoupling proteins 2 and 3. J Biol Chem. 1999;274:26003–26007. doi: 10.1074/jbc.274.37.26003. [DOI] [PubMed] [Google Scholar]
  26. Jazwinski SM, et al. HRAS1 and LASS1 with APOE are associated with human longevity and healthy aging. Aging Cell. 2010;9:698–708. doi: 10.1111/j.1474-9726.2010.00600.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kim D, Jitrapakdee S, Thompson M. Differential regulation of the promoter activity of the mouse UCP2 and UCP3 genes by MyoD and myogenin. Journal of biochemistry and molecular biology. 2007;40:921–927. doi: 10.5483/bmbrep.2007.40.6.921. [DOI] [PubMed] [Google Scholar]
  28. Kim S, Jazwinski SM. Quantitative measures of healthy aging and biological age. Healthy Aging Res. 2015;4 doi: 10.12715/har.2015.4.26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kim S, Welsh DA, Cherry KE, Myers L, Jazwinski SM. Association of healthy aging with parental longevity. Age (Dordr) 2013;35:1975–1982. doi: 10.1007/s11357-012-9472-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kim S, Welsh DA, Ravussin E, Welsch MA, Cherry KE, Myers L, Jazwinski SM. An elevation of resting metabolic rate with declining health in nonagenarians may be associated with decreased muscle mass and function in women and men, respectively. J Gerontol A Biol Sci Med Sci. 2014;69:650–656. doi: 10.1093/gerona/glt150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Klaus S, Casteilla L, Bouillaud F, Ricquier D. The uncoupling protein UCP: a membraneous mitochondrial ion carrier exclusively expressed in brown adipose tissue. The International journal of biochemistry. 1991;23:791–801. doi: 10.1016/0020-711x(91)90062-r. [DOI] [PubMed] [Google Scholar]
  32. Krauss S, Zhang CY, Lowell BB. A significant portion of mitochondrial proton leak in intact thymocytes depends on expression of UCP2. Proc Natl Acad Sci U S A. 2002;99:118–122. doi: 10.1073/pnas.012410699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Krauss S, Zhang CY, Lowell BB. The mitochondrial uncoupling-protein homologues. Nat Rev Mol Cell Biol. 2005;6:248–261. doi: 10.1038/nrm1572. [DOI] [PubMed] [Google Scholar]
  34. Kwok KH, et al. Mitochondrial UCP5 is neuroprotective by preserving mitochondrial membrane potential, ATP levels, and reducing oxidative stress in MPP+ and dopamine toxicity. Free Radic Biol Med. 2010;49:1023–1035. doi: 10.1016/j.freeradbiomed.2010.06.017. [DOI] [PubMed] [Google Scholar]
  35. Matthias A, Ohlson KBE, Fredriksson JM, Jacobsson A, Nedergaard J, Cannon B. Thermogenic Responses in Brown Fat Cells Are Fully UCP1-dependent: UCP2 OR UCP3 DO NOT SUBSTITUTE FOR UCP1 IN ADRENERGICALLY OR FATTY ACID-INDUCED THERMOGENESIS. Journal of Biological Chemistry. 2000;275:25073–25081. doi: 10.1074/jbc.M000547200. [DOI] [PubMed] [Google Scholar]
  36. Mookerjee SA, Divakaruni AS, Jastroch M, Brand MD. Mitochondrial uncoupling and lifespan. Mech Ageing Dev. 2010;131:463–472. doi: 10.1016/j.mad.2010.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Musa CV, et al. Four novel UCP3 gene variants associated with childhood obesity: effect on fatty acid oxidation and on prevention of triglyceride storage. Int J Obes (Lond) 2012;36:207–217. doi: 10.1038/ijo.2011.81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Pecqueur C, et al. Uncoupling protein 2, in vivo distribution, induction upon oxidative stress, and evidence for translational regulation. J Biol Chem. 2001;276:8705–8712. doi: 10.1074/jbc.M006938200. [DOI] [PubMed] [Google Scholar]
  39. Pecqueur C, et al. Uncoupling protein-2 controls proliferation by promoting fatty acid oxidation and limiting glycolysis-derived pyruvate utilization. Faseb j. 2008;22:9–18. doi: 10.1096/fj.07-8945com. [DOI] [PubMed] [Google Scholar]
  40. R_Core_Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing; Vienna, Austria: 2014. [Google Scholar]
  41. Rose G, Crocco P, De Rango F, Montesanto A, Passarino G. Further support to the uncoupling-to-survive theory: the genetic variation of human UCP genes is associated with longevity. PLoS One. 2011;6:e29650. doi: 10.1371/journal.pone.0029650. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Rosenbloom KR, et al. The UCSC Genome Browser database: 2015 update. Nucleic Acids Res. 2015;43:D670–681. doi: 10.1093/nar/gku1177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Rousset S, Alves-Guerra MC, Mozo J, Miroux B, Cassard-Doulcier AM, Bouillaud F, Ricquier D. The biology of mitochondrial uncoupling proteins. Diabetes. 2004;53(Suppl 1):S130–135. doi: 10.2337/diabetes.53.2007.s130. [DOI] [PubMed] [Google Scholar]
  44. Ruggiero C, Ferrucci L. The endeavor of high maintenance homeostasis: resting metabolic rate and the legacy of longevity. J Gerontol A Biol Sci Med Sci. 2006;61:466–471. doi: 10.1093/gerona/61.5.466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Schrauwen P, Hesselink MK, Blaak EE, Borghouts LB, Schaart G, Saris WH, Keizer HA. Uncoupling protein 3 content is decreased in skeletal muscle of patients with type 2 diabetes. Diabetes. 2001;50:2870–2873. doi: 10.2337/diabetes.50.12.2870. [DOI] [PubMed] [Google Scholar]
  46. Schrauwen P, Xia J, Walder K, Snitker S, Ravussin E. A novel polymorphism in the proximal UCP3 promoter region: effect on skeletal muscle UCP3 mRNA expression and obesity in male non-diabetic. Pima Indians Int J Obes Relat Metab Disord. 1999;23:1242–1245. doi: 10.1038/sj.ijo.0801057. [DOI] [PubMed] [Google Scholar]
  47. Thomas SA, Palmiter RD. Thermoregulatory and metabolic phenotypes of mice lacking noradrenaline and adrenaline. Nature. 1997;387:94–97. doi: 10.1038/387094a0. [DOI] [PubMed] [Google Scholar]
  48. Velloso LA, Degasperi GR, Vercesi AE, Saad MA. Methods for assessing and modulating UCP2 expression and function. Methods Enzymol. 2009;457:395–404. doi: 10.1016/s0076-6879(09)05022-8. [DOI] [PubMed] [Google Scholar]
  49. Vozza A, et al. UCP2 transports C4 metabolites out of mitochondria, regulating glucose and glutamine oxidation. Proc Natl Acad Sci U S A. 2014;111:960–965. doi: 10.1073/pnas.1317400111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Warburg O. On respiratory impairment in cancer cells. Science. 1956;124:269–270. [PubMed] [Google Scholar]
  51. Zhang CY, et al. Uncoupling protein-2 negatively regulates insulin secretion and is a major link between obesity, beta cell dysfunction, and type 2 diabetes. Cell. 2001;105:745–755. doi: 10.1016/s0092-8674(01)00378-6. [DOI] [PubMed] [Google Scholar]
  52. Zhang Y, et al. Mice deficient in both Mn superoxide dismutase and glutathione peroxidase-1 have increased oxidative damage and a greater incidence of pathology but no reduction in longevity. J Gerontol A Biol Sci Med Sci. 2009;64:1212–1220. doi: 10.1093/gerona/glp132. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES