PEROXISOME PROLIFERATOR-ACTIVATED RECEPTOR 2 AND ACYL-COA SYNTHETASE 5 POLYMORPHISMS INFLUENCE DIET RESPONSE

Page created by Louis Moody
 
CONTINUE READING
Brief Genetic Analyses

Peroxisome Proliferator-activated Receptor ␥ 2
and Acyl-CoA Synthetase 5 Polymorphisms
Influence Diet Response
Kristi B. Adamo,*† Robert Dent,‡ Carl D. Langefeld,§ Miranda Cox,§ Kathryn Williams,† Kevin M. Carrick,¶
Joan S. Stuart,¶ Scott S. Sundseth,¶ Mary-Ellen Harper,储 Ruth McPherson,†储 and Frédérique Tesson†储

Abstract                                                                                     0.03), and the rs2419621, located in the 5⬘untranslated
ADAMO, KRISTI B., ROBERT DENT, CARL D.                                                       region of the ACSL5 gene, displayed the strongest associa-
LANGEFELD,           MIRANDA          COX,        KATHRYN                                    tion with diet response (odds ratio ⫽ 3.45, 95% confidence
WILLIAMS, KEVIN M. CARRICK, JOAN S. STUART,                                                  interval ⫽ 1.61 to 7.69, p ⫽ 0.001). Skeletal muscle ACSL5
SCOTT S. SUNDSETH, MARY-ELLEN HARPER, RUTH                                                   mRNA expression was significantly lower in carriers of the
MCPHERSON,           AND       FRÉDÉRIQUE          TESSON.                                 wildtype compared with the variant rs2419621 allele (p ⫽
Peroxisome proliferator-activated receptor ␥ 2 and acyl-                                     0.03). Our results suggest a link between PPAR␥2 and
CoA synthetase 5 polymorphisms influence diet response.                                      ACSL5 genotype and diet responsiveness.
Obesity. 2007;15:1068 –1075.
Peroxisome proliferator-activated receptor ␥ (PPAR␥) and                                     Key words: peroxisome proliferator-activated receptor
its response gene, Acyl CoA synthetase 5 (ACSL5), which                                      ␥, Acyl CoA synthetase, diet response, obese women,
has an important role in fatty acid metabolism, may affect                                   haplotype
weight loss in response to caloric restriction. Therefore, we
aimed to determine whether these genes were involved in                                         Weight loss in response to caloric restriction shows sig-
the interindividual response to dietary treatment. Genotypic/                                nificant interindividual variability (1). Genes involved in
phenotypic comparisons were made between selected obese                                      fatty acid (FA)1 partitioning are strong candidates for
women from the quintiles losing the most (diet responsive,                                   weight gain and loss. We focused on the genes encoding the
n ⫽ 74) and the quintiles losing the least (diet-resistant, n ⫽                              peroxisome proliferator–activated receptor ␥ (PPAR␥), the
67) weight in the first 6 weeks of a 900-kcal formula diet.                                  master regulator of adipogenesis (2), and the acyl CoA
Two common PPAR␥ single nucleotide polymorphisms,                                            synthetase 5 (ACSL5), one of the putative PPAR␥ response
Pro12Ala and C1431T, and eight polymorphisms across the                                      genes, mapping to the obesity locus 10q25.1–2 (3,4).
ACSL5 gene were selected for single locus and haplotypic                                        The PPAR␥ response gene acyl CoA synthetases (5,6)
association analyses. The PPAR␥ Pro12Ala single nucleo-                                      (EC 6.2.1.3; ACSL) are involved in the supply of FAs by
tide polymorphism was associated with diet resistance (odds                                  catalyzing the activation step of FA metabolism. PPAR␥
ratio ⫽ 3.48, 95% confidence interval ⫽ 1.41 to 8.56, p ⫽                                    has been shown to regulate rat liver ACSL (5,7), and by
                                                                                             examining the 5⬘ flanking region of the human ACSL5 gene,
                                                                                             we identified a putative peroxisome proliferator response
Received for review July 19, 2005.                                                           element (GACTGTGGCACAGCTCA) located 826 bp up-
Accepted in final form November 17, 2006.                                                    stream of exon 1, suggesting that ACSL5 may be regulated
The costs of publication of this article were defrayed, in part, by the payment of page
charges. This article must, therefore, be hereby marked “advertisement” in accordance with   by PPAR␥. Because of its location on the outer mitochon-
18 U.S.C. Section 1734 solely to indicate this fact.                                         drial membrane in liver, the ACSL5 isoform has been
*Department of Cellular and Molecular Medicine, Faculty of Medicine, University of
Ottawa, Ontario, Canada; †University of Ottawa Heart Institute, Ontario, Canada; ‡Ottawa
                                                                                             proposed to provide acyl-CoA destined primarily for mito-
Hospital Weight Management Clinic, Ontario, Canada; §Department of Public Health             chondrial oxidation (8,9). In human skeletal muscle, the
Sciences, Wake Forest University School of Medicine, Winston-Salem, North Carolina;
¶GlaxoSmithKline, Research Triangle Park, North Carolina; and 储Department of Biochem-
istry and Immunology, Faculty of Medicine, University of Ottawa, Ontario, Canada.
                                                                                             1
Address correspondence to Frédérique Tesson, University of Ottawa Heart Institute, 40        Nonstandard abbreviations: FA, fatty acid; PPAR, peroxisome proliferator–activated re-
Ruskin Street, Ottawa, Ontario, Canada K1Y 4W7.                                              ceptor; ACSL5, acyl CoA synthetase 5; SNP, single nucleotide polymorphism; OR, odds
E-mail: ftesson@ottawaheart.ca                                                               ratio; CI, confidence interval; UTR, untranslated region; DHPLC, denaturing high-perfor-
Copyright © 2007 NAASO                                                                       mance liquid chromatography.

1068       OBESITY Vol. 15 No. 5 May 2007
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

presence of ACSL5 transcripts (10) and mitochondrial acyl           phisms across the 46-kb region containing the ACSL5 gene.
CoA synthetase activity (11) have been reported. Because            Polymorphisms were selected based on the following crite-
the ability to maintain or increase FA transport capacity           ria: 1) more than one database reported loci as polymorphic,
may be a determining factor in the success of weight reduc-         2) priority was given to polymorphisms with higher re-
tion, and ACSL5 has been shown to increase with food                ported frequencies (⬎10%), and 3) they resulted in an
deprivation in rats (8), we hypothesized that the ACSL5             amino acid substitution or were located in the 5⬘untranslated
genotype may influence the rate of weight loss in response          region (UTR). The G-778ex1 SNP, found during our SSCP
to energy restriction. We, therefore, tested for an association     screening for the rs3840746 I/D, has not been previously
between selected polymorphisms in both the PPAR␥ and                reported. All ACSL5 polymorphisms were consistent with
ACSL5 genes and weight loss in a population of obese                Hardy-Weinberg proportions. Pairwise linkage disequilib-
women enrolled in a weight management program (1).                  rium was generally strong (Table 2). Single marker associ-
   Women were split into three groups based on initial              ation analyses (Table 1) revealed that rs2419621, located in
weight (⬍90.9, 90.9 to 113.6, and 113.6 to 136.4 kg). After         the 5⬘UTR of the ACSL5 gene, displayed the highest level of
6 weeks of meal replacement, the mean weight change was             association with diet response (OR ⫽ 3.45, 95% CI ⫽ 1.61
significantly different among the three groups (see Appen-          to 7.69). Of the genetic models tested, the dominant model
dix, available online at the Obesity website, www.obesity-          was most significant (p ⫽ 0.001), even after adjusting for
research.org), with the heavier group losing more than the          age (p ⫽ 0.003). There were 50 carriers (67.6%) of the rare
lighter group. Hence, the lower two quintiles in each of the        allele in the responsive group vs. 27 (40.3%) in the resistant
three weight groups were summed and labeled diet resistant          group (p ⫽ 0.009). After adjusting for age, total cholesterol,
(n ⫽ 67), and the top two quintiles were labeled diet responsive    and triglycerides, the ACSL5 rs2419621 remained the only
(n ⫽ 74). Weight change was significantly different between         statistically significant predictor of diet responsiveness
the diet-responsive and diet-resistant individuals in each of the   (OR ⫽ 2.87, 95% CI ⫽ 1.4 to 5.9, p ⫽ 0.004). Although all
three groups (Group 1: 9.8 ⫾ 0.9 vs. 7.3 ⫾ 1.0 kg, p ⬍ 0.0001;      two and three tandem marker haplotypes were tested, as
Group 2: 12.4 ⫾ 1.3 vs. 8.0 ⫾ 2.0 kg, p ⬍ 0.0001; Group 3:          well as the eight marker haplotypes, these analyses did not
14.8 ⫾ 2.1 vs. 10.4 ⫾ 1.3 kg, p ⬍ 0.0001). Other than the           improve association between rs2419621 SNP and the rate of
diet-responsive group being slightly younger (p ⫽ 0.05), there      weight loss.
were no significant differences at baseline between these two          A logistic regression analysis, assuming a dominant
populations, independently or pooled, in terms of initial body      model for both the ACSL5 and the PPAR␥ SNPs, found
weight, BMI, waist circumference, fasting plasma glucose,           modest evidence for an interaction between these two loci
impaired glucose tolerance, diabetes status, total cholesterol,     (p ⫽ 0.058). Studies with a greater number of subjects are
low-density lipoproteins, high-density lipoproteins, and trig-      required to confirm this relationship. The next study will use
lycerides (data not shown).                                         a full cohort of subjects and weight loss adjusted for initial
   In the PPAR␥ gene, we analyzed two polymorphisms                 weight as a continuous trait to determine whether genotype
previously reported to be associated with obesity-related           predicts the amount of weight loss.
phenotypes (12), the Pro12Ala located in exon B of the                 We measured and compared ACSL5 mRNA expression
PPAR␥2 isoform and the silent C1431T single nucleotide              in a subset of women who provided skeletal muscle biopsies
polymorphism (SNP) located in exon 6. Both polymor-                 (n ⫽ 19; Figure 1). ANOVA indicated that individuals
phisms were consistent with Hardy-Weinberg proportions.             homozygous for the diet-responsive allele (T) of SNP
These SNPs showed linkage disequilibrium (D’ ⫽ 0.66) as             rs2419621 displayed a significantly greater expression of
previously reported (13). Although there was no difference          ACSL5 compared with the wildtype (p ⫽ 0.04). It should be
in the allele frequency for the silent C1431T SNP (p ⫽              noted that these results are based on a small number of
0.66), a ␹2 test for homogeneity showed a significant asso-         individuals. Interestingly, 1) the heterozygotes displayed an
ciation between diet resistance and the Pro12Ala allele vari-       intermediary phenotype (Figure 1), and 2) both wildtype
ant under a dominant model (p ⫽ 0.03; Table 1). Using               individuals, displaying the lowest mRNA levels, belonged
logistic regression, after adjustment for age, initial weight,      to the diet-resistant group, and all four homozygous mu-
initial waist circumference, total cholesterol, and triglycer-      tants, displaying the highest mRNA levels, belonged to the
ides, the Pro12Ala genotype remained the only statistically         diet-responsive group, suggesting the physiologic relevance
significant predictor of diet resistance [odds ratio (OR) ⫽         of ACSL5 mRNA levels in diet responsiveness.
2.99, 95% confidence interval (CI) ⫽ 1.15 to 7.78, p ⫽                 Our results showing an association between the PPAR␥
0.025]. Distribution of haplotypes was not significantly            Ala allele and resistance to diet-induced weight loss corrob-
different between weight loss groups.                               orate the findings of Nicklas et al. (14), who found that
   For ACSL5, we attempted to capture as much haplotype             postmenopausal women carrying the Ala allele experienced
diversity as possible by examining 8 of the 60 National             a reduction in resting fat oxidation after 6 months of a
Center for Biotechnology Information reported polymor-              hypocaloric diet. An inverse association between polyun-

                                                                                      OBESITY Vol. 15 No. 5 May 2007        1069
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

Table 1. PPAR␥ and ACSL5 SNP tests for association with diet resistance
                                              Tests for association with diet resistance
                                              Genotype                                               P for genetic model
                           1/1               1/2                    2/2             General       Dominant         Additive       Recessive
PPAR␥ Pro12Ala
  Count                  107                 29                       3                             0.0271
  OR (95% CI)              1.00       3.48 (1.41,8.56)
PPAR␥ C1431T
  Count                  104                 34                       1                             0.6589
  OR (95% CI)              1.00       3.43 (1.58,7.46)
ACSL5 rs2419621
  Count                   64                 66                    11                0.0043         0.0010          0.0027          0.3436
  OR (95% CI)              1.00       0.29 (0.13,0.62)       0.27 (0.06,1.2)
ACSL5 G-778ex1
  Count                  130                 11                       0                             0.4614
  OR (95% CI)              1.00       1.64 (0.44,6.14)
ACSL5 rs2277208
  Count                   40                 75                     26               0.5641         0.3284          0.5619          0.8977
  OR (95% CI)              1.00       0.64 (0.28,1.47)       0.79 (0.27,2.31)
ACSL5 rs6144093
  Count                   46                 68                     25               0.7024         0.5633          0.4168          0.4387
  OR (95% CI)              1.00       0.87 (0.39,1.96)       0.65 (0.23,1.8)
ACSL5 rs4918747
  Count                  119                 22                       0                             0.1063
  OR (95% CI)              1.00       2.38 (0.83,6.84)
ACSL5 rs876873
  Count                  103                 34                      4               0.9306         0.7081
  OR (95% CI)              1.00       1.17 (0.51,2.66)       1.10 (0.15,8.15)
ACSL5 rs3736946
  Count                  103                 36                       2                             0.1131
  OR (95% CI)              1.00       1.74 (0.76,3.98)
ACSL5 rs3740142
  Count                   78                 54                      8               0.2413         0.2187          0.1071          0.1432
  OR (95% CI)              1.00       0.73 (0.34,1.54)       0.17 (0.02,1.56)

SNP, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval. For PPAR␥ SNPs and ACSL G-778ex1, rs4918747,
rs876873, and rs3736946 SNPs, the frequency of 2/2 homozygous is low. Inference should be based on the dominant model only. Most and
least frequent alleles are indicated in Table 2 for each SNP. In this table, 1 ⫽ most frequent allele, 2 ⫽ least frequent allele. An empty box
means that the value has not been calculated because of the small number of individuals carrying the genotype.

saturated fat intake and BMI among Ala carriers was found                 the C1431T SNP, common to all PPAR␥ isoforms, was not,
in one study (15) but not in others (16,17). Considering that             suggesting the involvement of the PPAR␥2 isoform, found
all of the women in this study were provided with an                      predominantly in adipose tissue, in influencing the rate of
identical meal replacement product, one would not expect                  diet-induced weight loss. The Ala variant, associated with
there to be differences in dietary fat content between diet-              reduced transcriptional activity (18,19) and reduced binding
responsive and -resistant individuals. The Pro12Ala SNP                   affinity to the cognate promoter element (18), was signifi-
was associated with a slower rate of weight loss, whereas                 cantly more frequent in the diet-resistant vs. diet-responsive

1070    OBESITY Vol. 15 No. 5 May 2007
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

                                                                                                        Pairwise linkage disequilibrium coefficients (D’) and numbers in parentheses are the correlation coefficients (r2) between ACSL5 gene polymorphisms in the combined population.
                                                                                                        Underlined coefficients indicate single nucleotide polymorphisms not in complete linkage disequilibrium. Bold coefficient is indicative of the weakest linkage disequilibrium
                                                                                      0.8478 (0.1920)
                                                                                      0.8923 (0.1962)
                                                                                      0.8601 (0.0563)
                                                                                      1.0000 (0.0126)

                                                                                      0.9996 (0.0289)
                                                                                      1.0000 (0.0566)
                                                                                      1.0000 (0.0566)
                                                                          rs3740142

                                                                                             .
                                                                                      0.7215 (0.0385)
                                                                                      0.9340 (0.0059)
                                                                                      1.0000 (0.1344)
                                                                                      1.0000 (0.1229)
                                                                                      0.9977 (0.0141)
                                                                                      0.9999 (0.0286)
                                                                          rs3736946

                                                                                             .
                                                                                             .
                                                                                      0.9788 (0.0067)
                                                                                      1.0000 (0.2127)
                                                                                      0.8741 (0.1734)
                                                                                      1.0000 (0.0146)
                                                                                      0.8601(0.0563)
                                                                          rs876873

                                                                                                                                                                                                                                                                                          Figure 1: ANOVA comparing skeletal muscle ACSL5 mRNA
                                                                                                                                                                                                                                                                                          levels split by ACSL5 rs2419621 genotype. Wildtype, n ⫽ 2
                                                                                             .
                                                                                             .
                                                                                             .

                                                                                                                                                                                                                                                                                          (white); heterozygotes, n ⫽ 13 (black); homozygous mutant n ⫽ 4
                                                                                                                                                                                                                                                                                          (striped). Bars not sharing a common letter are significantly dif-
                                                                                                                                                                                                                                                                                          ferent at p ⬍ 0.05.
                                                                                      0.9999 (0.0378)
                                                                                      0.0307 (0.0005)
                                                                                      1.0000 (0.0688)
                                                                                      1.0000 (0.0628)
                                                                          rs4918747

                                                                                                                                                                                                                                                                                          women. Thus, it is plausible that the rate of weight loss was
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .

                                                                                                                                                                                                                                                                                          attenuated because of less activation of genes involved in
                                                                                                                                                                                                                                                                                          adipose tissue lipolysis, such as perilipin (20) and caveo-
                                                                                                                                                                                                                                                                                          lin-1 (21).
                                                                                                                                                                                                                                                                                             Given its location in liver mitochondrial membranes,
                                                                                      0.7474 (0.1833)
                                                                                      1.0000 (0.0519)
                                                                                      0.9524 (0.8425)

                                                                                                                                                                                                                                                                                          large increase in protein expression in response to food
                                                                          rs6144093
      Table 2. Marker-marker multi-allelic D’ (bounded between 0 and 1)

                                                                                                                                                                                                                                                                                          deprivation (8), and the lack of change in mRNA expression
                                                                                                                                                                                                                                                                                          during adipocyte differentiation (22), the ACSL5 isoform is
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .

                                                                                                                                                                                                                                                                                          suggested to provide acyl-CoA destined primarily for ␤-ox-
                                                                                                                                                                                                                                                                                          idation rather than triacylglycerol synthesis (8,9). FA oxi-
                                                                                                                                                                                                                                                                                          dation was shown to be reduced in skeletal muscle from
                                                                                                                                                                                                                                                                                          obese black women compared with obese white women, and
                                                                                      0.8088 (0.2339)
                                                                                      1.0000 (0.0507)

                                                                                                                                                                                                                                                                                          this was associated with lower ACSL activity in the mito-
                                                                          rs2277208

                                                                                                                                                                                                                                                                                          chondrial and microsomal fractions (11). Considering that
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .

                                                                                                                                                                                                                                                                                          the prevalence of obesity is greater in black than in white
                                                                                                                                                                                                                                                                                          women, Privette et al. (11) suggested that the decrease in
                                                                                                                                                                                                                                                                                          ACSL activity might be caused by reduced expression of
                                                                                                                                                                                                                                                                                          the protein. Skeletal muscle, which comprises about one
                                                                                      1.0000 (0.0181)

                                                                                                                                                                                                                                                                                          half of human body mass, is highly metabolically active and
                                                                          G-778ex1

                                                                                                                                                                                                                                                                                          oxidizes a substantial amount of FAs at rest. Consequently,
                                                                                                                                                                                                                                                                                          a variant modifying FA metabolism in skeletal muscle may
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .
                                                                                             .

                                                                                                                                                                                                                                                                                          have considerable influence in determining the rate of
                                                                                                        between G-778ex1 and rs4918747.

                                                                                                                                                                                                                                                                                          weight loss, and the rs2419621 ACSL5 SNP, located in the
                                                                                                                                                                                                                                                                                          enhancer/promoter region shortly upstream of a peroxisome
                                                                                                                                                                                                                                                                                          proliferator response element motif, is associated with pos-
                                                                          rs2419621

                                                                                                                                                                                                                                                                                          itive diet response. Based on our finding of differential
                                                                                           .
                                                                                           .
                                                                                           .
                                                                                           .
                                                                                           .

                                                                                                                                                                                                                                                                                          skeletal muscle ACSL5 mRNA levels between genotypes,
                                                                                                                                                                                                                                                                                          we propose that the rs2419621 SNP acts as a cis-acting
                                                                                                                                                                                                                                                                                          regulatory variant affecting ACSL5 expression levels or is in
                                                                                      rs2419621

                                                                                      rs2277208
                                                                                      rs6144093
                                                                                      rs4918747

                                                                                      rs3736946
                                                                                      rs3740142

                                                                                                                                                                                                                                                                                          linkage disequilibrium with the causative regulatory variant.
                                                                                      G-778ex1

                                                                                      rs876873

                                                                                                                                                                                                                                                                                          Because diet was not controlled before the study, the puta-
                                                                                                                                                                                                                                                                                          tive effect of the rs2419621 SNP on ACSL5 expression
                                                                                                                                                                                                                                                                                          levels and on the rate of weight loss was not reflected in the

                                                                                                                                                                                                                                                                                                              OBESITY Vol. 15 No. 5 May 2007          1071
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

anthropometric measurements at baseline. These findings          tected by restriction fragment length polymorphism with
and the fact that ACSL5 rs2419621 rare allele carriers have      BstU-1 using primers designed by Yen et al. (26). Polymer-
lower total cholesterol and tend to have lower low-density       ase chain reaction was performed in a total volume of 25 ␮L
lipoproteins and triglycerides support that this ACSL5 SNP       using 100 ng of genomic DNA. The mix included 200 ␮M
may influence the rate of weight loss by increasing ACSL5        dNTPs, 1.2 ␮M primers, 1⫻ Qiagen buffer, 1⫻ Q solution
levels and promoting the FA ␤-oxidation pathway.                 (Qiagen GmbH, Hilden, Germany), 1.5 mM MgCl2, and
   In summary, our results suggest that polymorphisms in         1.25 U of taq polymerase. The polymerase chain reaction
both the PPAR␥2 and ACSL5 genes are significantly asso-          conditions were optimized as follows: initial denaturation at
ciated with diet response in a population of obese women.        95 °C for 5 minutes, followed by 30 cycles at 95 °C for 1
However, it is still possible that other SNPs, which might be    minute, 59 °C for 1 minute, and 72 °C for 1 minute, with a
functional or in linkage disequilibrium with a functional        final extension of 10 minutes at 72 °C. The products were
SNP, have been missed in our study.                              digested with BstU-1 at 60 °C for 1 hour and electropho-
                                                                 resed on a 3% agarose gel stained with ethidium bromide.
                                                                 This digestion produces fragments of 270 bp for Pro12Pro
       Research Methods and Procedures                           wildtype and 227 and 43 bp for Ala12Ala homozygous
Subjects and Dietary Intervention                                mutants.
   Subjects were selected from a group of ⬎1000 obese               To detect the silent exon 6 C1431T SNP, which is com-
women who completed the weight management program at             mon to all PPAR␥ isoforms, we first used denaturing high-
the Ottawa Hospital between September 1992 and February          performance liquid chromatography (DHPLC; Helix; Var-
2003 and gave their informed consent. This program con-          ian Medical Systems, Palo Alto, CA) analysis. DHPLC runs
sists of a year-long course in lifestyle change and a total      were performed as recommended by Varian using buffer A
meal replacement (Optifast 900; Novartis, Cambridge, MA)         and buffer B (Varian) and a flow rate of 0.45 mL/min. One
for the first 6 or 12 weeks (23). Participants were asked to     half of each polymerase chain reaction product was mixed
refrain from engaging in structured physical activity during     with an equal amount of a previously sequenced and con-
the initial 6-week meal replacement component of the             firmed wildtype DNA. Mixed and unmixed samples were
weight loss program. White women were selected based on          denatured at 95 °C for 3 minutes and reannealed by decreas-
their initial weight and compliance to the program (1).          ing the temperature from 95 °C to 64 °C at a rate of 1 °C/
Women with medical conditions or taking medication that          min. Polymerase chain reaction samples displaying aberrant
might alter rate of weight loss were excluded from the           DHPLC profiles compared with wildtype controls were
study. Because initial weight is a predictor of rate of weight   double-stranded sequenced (ABI Prism Big Dye; Applied
loss, subjects were divided into four groups [Group 1:           Biosystems, Foster City, CA). Using this methodology, we
⬍90.9 kg (200 lbs, n ⫽ 35), Group 2: 90.9 to 113.6 kg (200       detected and genotyped a new SNP (C1442T) but were
to 250 lbs, n ⫽ 72), Group 3: 113.6 to 136.4 kg (250 to 300      unable to correctly genotype C1431T. Hence, C1431T was
lbs, n ⫽ 34), and Group 4: ⬎136.4 kg (⬎300 lbs, n ⫽ 25)          genotyped using the SNaPshot methodology.
and ranked within each group according to rate of weight            Primers for the ACSL5 gene were based on sequences
loss in the first 6 weeks of meal replacement. To select an      generated from the International Human Genome Sequenc-
obese population as homogenous as possible, data from the        ing Consortium-UCSC version hg16 (www.genome.uc-
25 women from Group 4 (⬎136.4 kg) were a priori ex-              sc.edu). Sets of primers flanking the selected polymor-
cluded from the analysis because, although medical man-          phisms (Figure 1) were designed using the Vector NTI
agement remains the mainstay of treatment for obesity, it        program producing amplicons of distinguishable lengths
has been shown to be of little long-term benefit in patients     (see Appendix, available online at the Obesity website,
with severe obesity (24). Moreover, low percentage change        www.obesityresearch.org). Primers, except those for I/D
in body weight after dietary intervention was shown to be        polymorphisms, were used in a multiplex polymerase chain
significantly associated with BMI (25). The genotypic and        reaction. Multiplex polymerase chain reaction was per-
phenotypic comparisons were made between diet-respon-            formed in a total volume of 30 ␮L using 200 ␮g of genomic
sive (top two quintiles) and diet-resistant (lower two quin-     DNA, individual primer concentration range 0.4 to 1.0 ␮M,
tiles) groups. Measurements of glucose and lipids were           2⫻ polymerase chain reaction buffer with 2.5 mmol MgCl2
performed by the Ottawa Hospital Central laboratory using        (Qiagen), 1⫻ Qiagen solution, 0.4 mM dNTPs, and 2.5 U
standard techniques. All clinical and laboratory data were       Taq with the following polymerase chain reaction condi-
handled by the weight management clinic software (23).           tions: 1 cycle at 95 °C for 5 minutes, 35 cycles at 95 °C for
                                                                 1 minute, 62 °C for 1 minute, and 72 °C for 1 minute, and
PPAR␥ and ACSL5 Polymorphism Genotyping                          a final extension at 72 °C for 10 minutes. Genotyping was
  Total genomic DNA was prepared from leukocytes. The            done with the ABI-3100 sequencer using the SNaPshot
Pro12Ala variant, unique to the PPAR␥2 isoform, was de-          technique and specifically designed primers for a SNaPshot

1072   OBESITY Vol. 15 No. 5 May 2007
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

Table 3. Between-genotype univariate comparison for ACSL5 rs2419621 and PPAR␥2 Pro12Ala single nucle-
otide polymorphism
                                   ACSL5 rs2419621 genotype                                PPAR␥ Pro12Ala genotype
                              Wildtype                 Carrier                      Wildtype                  Carrier
      Baseline                (n ⴝ 64)                (n ⴝ 77)            p         (n ⴝ 107)                (n ⴝ 32)             p
Age (yrs)            46.1 ⫾ 9.7 (47.6)    43.7 ⫾ 10.3 (43.5)            0.15 44.7 ⫾ 10.6 (46.9)         45.5 ⫾ 8.5 (44.2)   0.70
Weight (kg)         101.6 ⫾ 14.0 (100.7) 102.8 ⫾ 14.5 (101.6)           0.62 100.0 ⫾ 13.8 (96.8)       108.9 ⫾ 13.2 (108.5) 0.002
BMI (kg/m2)          37.9 ⫾ 4.6 (37.4)    38.9 ⫾ 5.8 (37.7)             0.23 38.0 ⫾ 5.2 (36.8)          39.8 ⫾ 5.5 (38.6)   0.09
Waist circumference
  (cm)              106.4 ⫾ 12.0 (107.3) 105.6 ⫾ 11.1 (104.1)           0.68 104.7 ⫾ 11.3 (104.1) 109.9 ⫾ 11.3 (113.7) 0.03
Glucose (mM)          5.4 ⫾ 1.2 (5.3)      5.7 ⫾ 1.5 (5.2)              0.29   5.6 ⫾ 1.5 (5.3)      5.4 ⫾ 1.1 (5.2)    0.49
Total cholesterol
  (mM)                5.5 ⫾ 0.9 (5.3)      5.1 ⫾ 0.9 (5.0)              0.02     5.3 ⫾ 1.0 (5.2)          5.1 ⫾ 0.7 (5.0)      0.17
Low-density
  lipoproteins
  (mM)                3.4 ⫾ 0.8 (3.3)      3.1 ⫾ 0.8 (3.0)              0.07     3.2 ⫾ 0.9 (3.2)          3.2 ⫾ 0.6 (3.2)      0.99
High-density
  lipoproteins
  (mM)               1.29 ⫾ 0.3 (1.3)     1.31 ⫾ 0.3 (1.3)              0.72     1.3 ⫾ 0.3 (1.3)        1.28 ⫾ 0.3 (1.2)       0.65
Triglycerides
  (mM)               1.73 ⫾ 0.83 (1.6)    1.48 ⫾ 0.69 (1.4)             0.06    1.69 ⫾ 0.8 (1.6)        1.29 ⫾ 0.44 (1.3)      0.009

Values are given as mean ⫾ standard deviation (median). Wildtype indicates wildtype genotype, whereas carrier corresponds to those who
carry at least one rare allele, most and least frequent alleles being indicated in Table 2.

multiplex reaction (see Appendix, available online at the             TTGGGAAAGAAAGTGGCCTT; reverse, TGGAAA-
Obesity website, www.obesityresearch.org). Insertion/dele-            GCTCTCCTCGCTTTG; and probe, TCCATTGAAAAT-
tion polymorphism rs6144093 was detected using agarose                GGGCTCTTGACACCA.
gel electrophoresis and ethidium bromide staining, while we
screened for the reported 5⬘UTR I/D rs3840746 using sin-              Statistical Analyses
gle-strand confirmation polymorphism.                                    For each polymorphism, we used a ␹2 goodness of fit test
                                                                      to determine whether the observed allele frequencies de-
TaqMan Quantitative Gene Expression Assay                             parted from Hardy-Weinberg proportions. The degree of
   Nineteen of the obese women consented to having a                  marker-marker linkage disequilibrium was estimated by the
rectus femoris muscle biopsy taken. The biopsies were                 statistic D’. Individual allele frequencies were computed,
taken after 4 weeks of weight stabilization following com-            and differences in these frequencies were tested using a
pletion of the meal replacement intervention. Of these bi-            permutation test (10,000 permutations) of the likelihood
opsies, 12 were taken from the diet-responsive group and 7            ratio statistic. Haplotype frequencies and tests for differ-
from the diet-resistant group. Isolation of total RNA was             ences in haplotype frequencies were completed using the
done as previously described (1). RNA was reverse tran-               expectation-maximization algorithm-based software Dande-
scribed, and aliquots of cDNA were amplified for ACSL5                lion (27). Statistical significance was assessed using a per-
and for four control housekeeping genes (18S, cyclophilin,            mutation test of the likelihood ratio statistic. To span the
glyceraldehyde-3-phosphate dehydrogenase, and ␤-actin).               region, all two and three tandem marker haplotypes and the
All amplified products were measured by real-time quanti-             entire eight marker haplotypes were computed and tested
tative polymerase chain reaction using the Taqman Reverse             using a permutation test; inferences are based on haplotypes
Transcription Reagent Kit (Applied Biosystems) and the                with overall frequencies ⬎0.10. To compute haplotype
Applied Biosystems PRISM 7700 Sequence Detection sys-                 analyses adjusting for covariates, we used the expectation-
tem. The sequences of the primers and fluorogenic probe,              maximization algorithm in Dandelion and determined all of
specific to the ACSL5 isoform, were as follows: forward,              the haplo-genotype (haplotype pairs) probabilities for

                                                                                         OBESITY Vol. 15 No. 5 May 2007         1073
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

haplo-genotypes consistent with the genotypic data. These                6. Martin G, Schoonjans K, Lefebvre AM, Staels B, Auwerx
probabilities were used as weights in a weighted logistic                   J. Coordinate regulation of the expression of the fatty acid
regression analysis. Inferences are focused on the more                     transport protein and acyl-CoA synthetase genes by PPARal-
frequent haplotypes. For both the unadjusted and covariate                  pha and PPARgamma activators. J Biol Chem. 1997;272:
                                                                            28210 –7.
adjusted analyses, ORs and corresponding 95% CIs were
                                                                         7. Martin G, Schoonjans K, Staels B, Auwerx J. PPARgamma
computed for each haplotype relative to all other haplo-
                                                                            activators improve glucose homeostasis by stimulating fatty
types. All single SNP genotypic association analyses were                   acid uptake in the adipocytes. Atherosclerosis. 1998;137:S75–
computed using 2 ⫻ 3 contingency tables and ␹2 statistics.                  80.
Multiple logistic regression was used to evaluate whether                8. Lewin TM, Kim J-H, Granger DA, Vance JE, Coleman
the association between genotype and weight loss was still                  RA. Acyl-CoA synthetase isoforms 1, 4, and 5 are present in
present after adjusting for variables with a p ⬍ 0.2 in the                 different subcellular membranes in rat liver and can be inhib-
diet-responsive vs. -resistant univariate analyses (see Ap-                 ited independently. J Biol Chem. 2001;276:24674 –9.
pendix, available online at the Obesity website, www.obe-                9. Coleman RA, Lewin TM, Van Horn CG, Gonzalez-Baro
sityresearch.org) or in the genotype-specific univariate                    MR. Do long-chain Acyl-CoA synthetases regulate fatty acid
comparison (Table 3). To test whether the proportion of the                 entry into synthetic versus degradative pathways? J Nutr.
ACSL5 rs2419621 rare allele carriers differed between the                   2002;132:2123– 6.
                                                                        10. Yamashita Y, Kumabe T, Cho YY, et al. Fatty acid induced
diet-responsive and diet-resistant groups, we used the one
                                                                            glioma cell growth is mediated by the acyl-CoA synthetase 5
sample test of a proportion. Interaction between the PPAR␥                  gene located on chromosome 10q25.1-q25.2, a region fre-
Pro12Ala and the rs2419621 ACSL5 polymorphisms was                          quently deleted in malignant gliomas. Oncogene. 2000;19:
tested using logistic regression analyses. Finally, an                      5919 –25.
ANOVA model was performed to test whether the level of                  11. Privette JD, Hickner RC, Macdonald KG, Pories WJ,
ACSL5 mRNA varied between rs2419621 genotypes.                              Barakat HA. Fatty acid oxidation by skeletal muscle homog-
                                                                            enates from morbidly obese black and white American
                                                                            women. Metabolism. 2003;52:735– 8.
                    Acknowledgments                                     12. Lohmueller K, Pearce C, Pike M, Lander E, Hirschlorn J.
   These studies would not have been possible without the                   Meta-analysis of genetic association studies supports a con-
invaluable assistance of the women participating in the                     tribution of common variants to susceptibility to common
study. We thank Andrej Teren for assistance in designing                    disease. Nat Genetics. 2003;33:177– 82.
the ACSL5 multiplex and SNaPshot primers and Andrej                     13. Masud S, Ye S. Effect of the peroxisome proliferator acti-
Teren and Diana Maalouf for aid establishing the multiplex                  vated receptor-gamma gene Pro12Ala variant on body mass
and SNaPshot protocol for the ACSL5 SNPs. This study was                    index: a meta-analysis. J Med Genet. 2003;40:773– 80.
supported by grants from the University of Ottawa (to F.T.),            14. Nicklas BJ, van Rossum EF, Berman DM, Ryan AS, Den-
the Canadian Foundation for Innovation (F.T.), the Cana-                    nis KE, Shuldiner AR. Genetic variation in the peroxisome
dian Institutes for Health Research (CIHR) Doctoral Schol-                  proliferator-activated receptor-gamma2 gene (Pro12Ala) af-
                                                                            fects metabolic responses to weight loss and subsequent
arship (K.B.A.), and Wyeth-Ayerst/CIHR Chair in Cardio-
                                                                            weight regain. Diabetes. 2001;50:2172– 6.
vascular Disease (R.M.).                                                15. Luan J, Browne PO, Harding AH, et al. Evidence for
                                                                            gene-nutrient interaction at the PPARgamma locus. Diabetes.
                            References                                      2001;50:686 –9.
 1. Harper M, Dent R, Monemdjou S, et al. Decreased mito-               16. Memisoglu A, Hu FB, Hankinson SE, et al. Interaction
    chondrial proton leak and reduced expression of uncoupling              between a peroxisome proliferator-activated receptor gamma
    protein 3 in skeletal muscle of obese and diet-resistant women.         gene polymorphism and dietary fat intake in relation to body
    Diabetes. 2002;51:2459 – 66.                                            mass. Hum Mol Genet. 2003;12:2923–9.
 2. Spiegelman B, Flier J. Adipogenesis and obesity: rounding           17. Robitaille J, Despres JP, Perusse L, Vohl MC. The PPAR-
    out the big picture. Cell. 1996;87:377– 89.                             gamma P12A polymorphism modulates the relationship be-
 3. Dong C, Wang S, Li WD, Li D, Zhao H, Price RA. Inter-                   tween dietary fat intake and components of the metabolic
    acting genetic loci on chromosomes 20 and 10 influence                  syndrome: results from the Quebec Family Study. Clin Genet.
    extreme human obesity. Am J Hum Genet. 2003;72:115–24.                  2003;63:109 –16.
 4. van der Kallen CJ, Cantor RM, van Greevenbroek MM, et               18. Deeb SS, Fajas L, Nemoto M, et al. A Pro12Ala substitution
    al. Genome scan for adiposity in Dutch dyslipidemic families            in PPARgamma2 associated with decreased receptor activity,
    reveals novel quantitative trait loci for leptin, body mass index       lower body mass index and improved insulin sensitivity. Nat
    and soluble tumor necrosis factor receptor superfamily 1A. Int          Genet. 1998;20:284 –7.
    J Obes Relat Metab Disord. 2000;24:1381–91.                         19. Masugi J, Tamori Y, Mori Y, Koike T, Kasuga M. Inhib-
 5. Schoonjans K, Watanabe M, Suzuki H, et al. Induction of                 itory effect of a proline-to-alanine substitution at codon 12 of
    the acyl-coenzyme A synthetase gene by fibrates and fatty               peroxisome proliferator-activated receptor gamma 2 on thia-
    acids is mediated by a peroxisome proliferator response ele-            zolidinedione-induced adipogenesis. Biochem Biophys Res
    ment in the C promoter. J Biol Chem. 1995;270:19269 –76.                Commun. 2000;268:178 – 82.

1074    OBESITY Vol. 15 No. 5 May 2007
Diet Responsiveness, PPAR␥2, and ACSL5, Adamo et al.

20. Nagai S, Shimizu C, Umetsu M, et al. Identification of a         24. Brownell KD, Rodin J. Medical, metabolic, and psychological
    functional peroxisome proliferator-activated receptor respon-        effects of weight cycling. Arch Intern Med. 1994;154:1325–30.
    sive element within the murine perilipin gene. Endocrinology.    25. Carels RA, Cacciapaglia HM, Douglass OM, Rydin S,
    2004;145:2346 –56.                                                   O’Brien WH. The early identification of poor treatment out-
21. Llaverias G, Vazquez-Carrera M, Sanchez RM, et al. Ros-              come in a women’s weight loss program. Eat Behav. 2003;4:
    iglitazone upregulates caveolin-1 expression in THP-1 cells          265– 82.
    through a PPAR-dependent mechanism. J Lipid Res. 2004;45:        26. Yen CJ, Beamer BA, Negri C, et al. Molecular scanning of
    2015–24.                                                             the human peroxisome proliferator activated receptor gamma
22. Oikawa E, Iijima H, Suzuki T, et al. A novel acyl-CoA                (hPPAR gamma) gene in diabetic Caucasians: identification of
    synthetase, ACS5, expressed in intestinal epithelial cells and
                                                                         a Pro12Ala PPAR gamma 2 missense mutation. Biochem
    proliferating preadipocytes. J Biochem (Tokyo). 1998;124:
                                                                         Biophys Res Commun. 1997;241:270 – 4.
    679 – 85.
                                                                     27. Green L, Lange E and Langefeld C. Power comparison of
23. Dent RM, Penwarden RM, Harris N, Hotz SB. Develop-
                                                                         phase-known versus phase-unknown haplotype analyses for
    ment and evaluation of patient-centered software for a weight-
    management clinic. Obes Res. 2002;10:651– 6.                         case-control designs. Am J Hum Genet. 2001;69:1948.

                                                                                        OBESITY Vol. 15 No. 5 May 2007          1075
You can also read