- Email: [email protected]
Association of the Serotonin Transporter Gene with Serum Cholesterol Levels and Heart Disease
Molecular Genetics and Metabolism 67, 248 –253 (1999) Article ID mgme.1999.2870, available online at http://www.idealibrary.com on
Association of the Serotonin Transporter Gene with Serum Cholesterol Levels and Heart Disease David E. Comings,* ,1 James P. MacMurray,† Nancy Gonzalez,* Linda Ferry,‡ and Warren R. Peters‡ *Department of Medical Genetics, City of Hope National Medical Center, Duarte, California 91910; †Department of Psychiatry, Loma Linda University School of Medicine, Loma Linda, California; and ‡Loma Linda University, Loma Linda, California Received March 10, 1999, and in revised form April 27, 1999
We had a unique opportunity to examine the role of genetic factors in the health of an elderly population of male athletes. This was a group of men over 55 years of age who compete each year in a series of athletic competitions called the Golden Age Games. Cholesterol levels and a history of the presence or absence of heart disease, angina, and heart attack were obtained on these subjects. Since low cholesterol levels have been reported to be associated with the development of depression and violent death by suicide, especially in the elderly (1– 4), and since the serotonin transporter gene has been implicated in affective disorders (5), we were particularly interested in the potential interaction between this gene and cholesterol levels and heart disease. As a replication group we utilized a group of subjects in good general health from the Loma Linda University Center for Health Promotion (the CHP group) since cholesterol and triglyceride levels were available on these subjects. As an additional replication we obtained blood samples for a third group of subjects from Jerry L. Pettis VA Medical Center, Loma Linda, California, on whom a history of heart attacks (the LLHosp group) was available. A well-characterized insertion (L)/deletion (S) polymorphism (HTTLPR) at the promoter of the HTT gene was utilized since it is known to be associated with variations in HTT gene expression (6). Expression studies with a human choriocarcinoma cell line showed that the nondeletion or long or L allele was associated with three times the rate of expression of the serotonin transporter compared to the S allele. A further relevant aspect of this report is the finding of a molecular heterosis effect at the HTT gene. Heterosis or overdominance refers to the presence of a greater (or
In a study of a group of elderly athletes we observed an unexpected association between serum cholesterol levels and the HTTLPR insertion/deletion polymorphism of the promoter region of the serotonin transporter gene (HTT, SLC6A4). As a follow-up we examined the potential association of this polymorphism with cholesterol and triglyceride levels, or heart disease, in two other groups of subjects. We examined the possible association between cholesterol levels and heart disease and genotypes of the HTTLPR insertion/deletion polymorphism of the promoter region of the HTT gene, in three independent study populations ranging from 42 to 90 years of age. For subjects 55 to 70 years of age in Group 1, cholesterol levels were significantly greater in the LS heterozygotes than either LL or SS homozygotes, indicating a heterosis effect (P < 0.0001). This was replicated in Group 2 (P < 0.015). Triglyceride levels were also significantly elevated in the LS subjects (P < 0.001). In Groups 1 and 3 there was a significant association between LS heterozygosity and heart disease, angina, and heart attacks in subjects 70 years of age or less. All of these associations were absent in subjects >70 years of age. While these studies are preliminary and exploratory, they are consistent with a relationship of the HTT gene in cholesterol levels and a risk for heart disease. Replication of these findings in independent, epidemiologically based studies is required. © 1999 Academic Press
Key Words: serotonin; transporter; cholesterol; triglycerides; age; coronary heart disease.
1
To whom correspondence should be addressed. Fax: (626) 301-8980. E-mail: [email protected]. 248 1096-7192/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL
lesser) phenotypic effect in heterozygotes or hybrids than in homozygotes or parental strains. It is well studied in corn where hybrids of two strains show consistently higher yields in bushels per acre than either parent strain (7,8). Molecular heterosis refers to a situation in which heterozygotes for a polymorphic gene marker show a greater or lesser phenotypic effect than either homozygote (9). We have found molecular heterosis to be a common phenomena and have observed it in a wide range of genes (10). Little et al. observed molecular heterosis at the HTT gene (11). They examined levels of [ 125I]b-CIT (citalopram) binding (fmol/mg) to the serotonin transporter in the dorsal and median raphe nuclei and substantia nigra of human controls and subjects with chronic cocaine use genotyped at the HTTLPR polymorphism. This showed that [ 125I]b-CIT binding was lower in the LS heterozygotes than either the SS or LL homozygotes in all three regions. A twoway ANOVA was significant for genotype and region and genotype main effect (P , .001). Several recent studies have shown that elevated cholesterol levels are less associated with cardiovascular disease in elderly individuals than in younger subjects (12,13). The Golden Age Games sample allowed us to investigate this by separately examining participants who were younger or older than 70 years of age. The age 70 was chosen to be past the age of the greatest risk for death due to coronary artery disease and yet not so great that it would exclude all the older subjects, but instead the subjects were divided into two approximately equal groups. Thus, 54% were in the ,70 group and 46% were in the $70-year-old group. If the cut had been made at 65 years of age, only 34% would have been in the ,65year-old group. Thus, the above three groups allowed us to do the following: (1) Determine if the HTTLPR polymorphism of the HTT gene predicted cholesterol levels. (2) Retest this association in an independent group of subjects. (3) Determine if the HTT gene was also associated with heart attacks. (4) Retest this association in an independent group of subjects. (5) Determine if the risks were different in the subjects 70 years of age or less versus those over age 70.
249
for DNA and cholesterol testing. Of the 100 subjects, 74% were non-Hispanic Caucasian, 18% were African-American, and 8% were Hispanic or other. They ranged in age from 55 to 91 years. The mean age of the 58 subjects in the #70 age group was 63.9 years (SD 4.1 years). The mean age of the 42 subjects in the .70 age group was 74.8 (SD 3.1). Subjects were coded for a history of different cardiovascular disease problems (angina, heart attacks, or hypertension). The Center for Health Promotion group (CHP). The subjects from the CHP study consisted of 102 non-Hispanic Caucasians from the Loma Linda University Center for Health Promotion. The age, sex, weight, height, and waist-hip ratio were determined on each subject. The subjects ranged in age from 42 to 70 years of age with a mean age of 55.4 years (SD 7.5 years). A fasting blood sample was obtained for determination of cholesterol and triglycerides. The recruitment particularly targeted staff members of the Loma Linda University and Loma Linda University Medical Center. The Loma Linda V.A. Hospital group (LLHosp). The subjects from the LLHosp group consisted of a random sample of 83 non-Hispanic Caucasian male inpatients from the medical wards of the Jerry L. Pettis Memorial Veterans Administration Medical Center, Loma Linda, California. They consisted of individuals 42 to 74 years of age with a mean age of 57.0 years (SD 8.87 years). There were too few subjects in the .70 age group to analyze separately. Data on the presence or absence of heart attacks were available. In all three studies coded samples of blood were sent to the Department of Medical Genetics at the City of Hope National Medical Center where the genetic studies were performed blind to clinical data. All studies were approved by the IRBs of both institutions.
METHODS
Laboratory tests. Cholesterol and triglyceride levels were determined using the REP Ultra-30 HDL, VLDL/LDL cholesterol system of Helena Laboratories (Beaumont, TX). Polymerase chain reaction genotyping of the HTTLPR polymorphism of the HTT gene was performed using the primers and conditions reported by Heils et al. (6).
The Golden Age Games group (GAG). Each year veterans over 55 years of age volunteer for a national athletic competition called the Golden Age Games. In 1996 the Golden Age Games were held in Southern California and we obtained blood samples
Statistics. The mean cholesterol and triglyceride levels for the different HTT genotypes, SS versus LS versus LL, and LS versus LL 1 SS, were compared by ANOVA. The significance levels were determined on the basis of the F ratio. For the three-way com-
250
COMINGS ET AL.
parisons, a post hoc Tukey test indicated any individual means that were significantly different from other individual means at a 5 0.05. Linear regression analysis of the association between cholesterol and triglyceride levels and genotype was performed scoring the HTT gene as LL or SS 5 0 and LS 5 1. To examine the percentage of the variance for heart attack, heart disease, and angina, their presence was scored as 1 and their absence as 0. This allowed the determination of r and r 2 or the percentage of the variance attributable to the HTT gene. x 2 analysis was used to compare the number of subjects with a history of heart disease, angina, or heart attack in the GAG group, and heart attacks in the LLHosp group, versus the LS and LL or SS HTT genotype groups. All statistical analysis was by the SPSS statistical package (SPSS, Inc., Chicago, IL). RESULTS The results for the 100 Golden Age Games subjects are shown in Table 1. We first examined all races together. The cholesterol levels for the entire GAG group ranged from 100 to 312 mg/dl indicating there were no extreme outliers. The mean cholesterol was 231.41 mg/dl for the LS heterozygotes versus 197.00 mg/dl for the LL and 206.36 mg/dl for the SS groups (P # 0.0056). When tested for heterosis by comparing LS heterozygotes to LL 1 SS homozygotes, the mean cholesterol for the homozygotes was 200.65 mg/dl (P # 0.0017). When restricted to Caucasians the LS heterozygotes again had the higher mean cholesterol levels but the results were not significant. When the subjects in the 55- to 70-year-old group were examined for all races the mean cholesterol for the LS heterozygotes was 240.00 mg/dl, compared to 190.76 mg/dl for the LL homozygotes and 201.00 mg/dl for the SS homozygotes (P # 0.0005). When tested for heterosis by comparing LS heterozygotes to LL 1 SS homozygotes, the mean cholesterol for the homozygotes was 194.00 mg/dl (P # 0.0001). When restricted to Caucasians the results were still significant for all three genotypes (P # 0.011) and for heterozygotes versus homozygotes (P # 0.0026). By comparison, none of these associations were significant for subjects over age 70 (Table 1). To determine if these findings could be replicated, we examined the CHP subjects. In addition to cholesterol levels these subjects also had triglyceride levels. We examined individuals in the 42- to 70year age group (Table 2). There were too few subjects in the .70-year group for statistical analysis.
TABLE 1 Association between the Genotypes of the HTT Gene and Serum Cholesterol Levels in Golden Age Games Males in Two Age Groups Mean cholesterol Genotype
n
(mg/dl)
SD
F
P
41.45 48.30 49.10 48.30 48.30
5.48
#0.0056
10.36
#0.0017
1.89
#0.158
3.79
#0.055
8.72
#0.0005
17.07
#0.0001
Age all races (HTTLPR) (n 5 100) LL LS SS LL 1 SS LS
39 36 25 64 36
197.00 231.41* 206.36 200.65 231.41
Age Caucasians only (HTTLPR) (n 5 72) LL LS SS LL 1 SS LS
25 27 20 45 27
205.20 227.63* 208.05 206.46 227.62
41.52 40.82 53.55 46.69 47.12
Age 55 to 70 all races (HTTLPR) (n 5 58) LL LS SS LL 1 SS LS
26 20 12 38 20
190.76 240.00# 201.00 194.00 240.00
42.11 44.05 28.41 38.22 44.05
Age 55 to 70 Caucasians only (HTTLPR) (n 5 38) LL LS SS LL 1 SS LS
16 14 8 24 14
199.37 243.21* 202.00 200.25 243.21
39.95 44.93 28.71 35.96 44.93
5.11
#0.011
10.49
#0.0026
Age .70 all races (HTTLPR) (n 5 42) LL LS SS LL 1 SS LS
13 16 13 26 16
209.46 220.68 211.31 210.38 220.68
38.67 52.58 63.45 51.51 52.58
0.194
#0.82
0.390
#0.53
Age .70 Caucasians only (HTTLPR) (n 5 34) LL LS SS LL 1 SS LS
9 13 12 21 13
215.55 210.84 212.08 213.57 210.84
44.60 28.91 66.21 28.91 28.91
0.025
#0.97
0.025
#0.87
* Significantly different from LL by Tukey test at a 5 0.05. # Significantly different from LL and SS by Tukey test at a 5 0.05.
The cholesterol levels ranged from 109 to 352 mg/dl with only two individuals with a level greater than 275. The F values for all three genotypes were sig-
251
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL
TABLE 2 Association between the Genotypes of the HTT Gene and Serum Cholesterol and Triglyceride Levels in CHP Males and Females 42 to 70 Years of Age (n 5 99) Genotype
n
Mean (mg/dl)
SD
F
P
32.71 40.67 44.62 36.02 40.67
3.29
#0.042
6.09
#0.015
Cholesterol LL LS SS LL 1 SS LS
35 51 13 48 51
196.80 218.60* 205.46 199.14 218.25
TABLE 3 HTT Genotype and History of Angina, Heart Disease, or Heart Attack (% Yes or No, All 1 df)
Triglycerides (log transformed) LL LS SS LL 1 SS LS
35 51 13 48 51
1.99 2.15* 2.02 1.99 2.15
0.23 0.22 0.20 0.22 0.22
9.6% of the variance of the triglyceride levels (P # 0.0066). We examined the association between the presence of the HTT genotypes and history of heart disease, angina, or heart attack for the two age ranges in the GAG group (Table 3). For the GAG subjects less than 70 years of age, the frequencies for those with the respective conditions in those with the LS versus the LL or SS genotypes were are follows: heart disease 68.4% versus 35.1%, angina 42.1% versus 16.2%, and heart attack 42.1% versus
5.76
#0.0043
11.46
#0.001
HTT Genotype Condition
* Significantly different from LL by Tukey test at a 5 0.05.
LS
LL 1 SS
x2
P
5.59
#0.018
Golden Age Games A. Heart disease
nificant at P # 0.045. The Tukey test showed that the LS values were significantly higher than the LL values at a 5 0.05, and the test for heterosis by comparison of heterozygotes versus homozygotes was significant (P # 0.015). The triglyceride levels ranged from 40 to 496 mg/dl with skewing of the higher levels. Because of this wide range the values were -log transformed. The log-transformed values showed a normal distribution. The results of ANOVA for the triglyceride levels were significant for all three genotypes (P # 0.0056) and for the comparison of heterozygotes versus homozygotes (P # 0.002). When sex was used as a covariant it was not significant for either cholesterol or triglyceride levels. When BMI was used as a covariant the association between LS subjects and elevated cholesterol remained significant for both GAG and CHP groups. When non-log-transformed triglyceride levels were used ANOVAs for the three genotype groups and the LL 1 SS versus LS groups were still significant (P , 0.01 and #0.0052, respectively). For the Golden Age Games subjects 70 years of age or less, the HTT gene accounted for 23.3% of the variance of cholesterol levels, P # .0001; 10.2% of the variance for heart attack, P # 0.016; 9.9% of the variance for heart disease, P # 0.018; and 8.0% of the variance for angina, P # 0.034. For the CHP subjects, the HTT gene accounted for 5.3% of the variance of the cholesterol levels (P # 0.019), and
Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
13 (68.4) 6 (31.6)
13 (35.1) 24 (64.9)
Age .70 all races (HTTLPR) (n 5 42) Yes No
8 (50.0) 8 (50.0)
13 (50.0) 13 (50.0)
.000
#1.00
B. Angina Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
8 (42.1) 11 (57.9)
6 (16.2) 31 (83.8)
4.48
#0.034
Age .70 All races (HTTLPR) (n 5 42) Yes No
5 (31.3) 11 (68.8)
8 (30.8) 18 (69.2)
.001
#0.97
C. Heart attack Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
8 (42.1) 11 (57.9)
5 (13.5) 32 (86.5)
5.75
#0.016
1.00
#0.31
4.91
#0.03
Age .70 all races (HTTLPR) (n 5 42) Yes No
6 (37.5) 10 (62.5)
6 (23.1) 20 (76.9)
LLHosp group Heart attack (n 5 83) Yes No
11 (23.4) 36 (76.6)
2 (5.6) 34 (94.4)
252
COMINGS ET AL.
13.5%. The increase in the frequency for those with the LS genotype was significant for all three diagnostic groups (P # 0.016 to # 0.034). By comparison, for those over 70 years of age there was no significant difference in the frequency for any of these heart problems in those with the LS genotypes versus the LL or SS genotypes (P # 0.31 to 1.0). To determine if we could replicate the association between the HTT alleles and heart attacks, we examined the LLHosp group (Table 3). Here 23.4% of the LS subjects had a history of a heart attack versus 5.6% of the LL or SS subjects (P , 0.03). DISCUSSION While the present results are consistent with a role of the HTT gene in cholesterol levels and risk of heart disease, we clearly recognize a number of limitations of the present study. Foremost is the unique nature of the GAG group in that these were elderly athletes. As such, they were not typical of the general population. However, we specifically chose this group because we felt they might provide some unique insights into genetic factors associated with good health rather than disease. Thus, they may have magnified an effect that would be less dramatic in a comparable aged group in the general population. This is supported by the observation that in the GAG group that were 70 years of age or less, the HTT gene accounted for 23% of the variance of cholesterol, while in the CHP group it accounted for only 5.3%. While the CHP individuals also took an interest in their health they were not necessarily healthier than subjects in the general population. Finally, the LLHosp group was atypical in that they are drawn from the general inpatient medical wards. Despite these limitations, the results for cholesterol and/or risk of heart disease were consistent across all groups. Statistically it could be argued that the comparison of LS heterozygotes versus SS 1 LL homozygotes was a post hoc analysis of the data. However, as discussed in the introduction this finding was not unexpected since we have observed molecular heterosis for phenotypes associated with a wide range of other genes (ADRA1C, ADRA2C, AR, CD4, CD8, CHRNA4, CNR1, DBH, DRD4, DRD5, DAT1, ESRA, FRAXA, GABRA1, GABRA3, GABRA5, GABBR1, GABRB3, GABBR, GAD2, HTR1A, INFG, INS, MAOA, MAOB, MC3R, NOS1A, NOS2A, OB1875, OB1871, PENK, TH,TDO2) (10). In addition, while the molecular heterosis effect was post hoc for the GAG group, this was not the case for the other two groups.
The observation that elderly subjects with elevated cholesterol levels are at less risk for cardiovascular disease stimulated us to examine the GAG subjects less than 70 years of age versus those greater than 70 years of age. This could only be tested in the Golden Age Games subjects since they were the only group with a sufficient number of subjects over 70 to have statistical validity. The pattern for cholesterol levels, heart disease, angina, and heart attack were all similar. Thus, the effect on these variables was greater for LS heterozygotes only in the less than 70 age group. It disappeared or was negative (less effect in the LS heterozygotes) in the over 70 age group. These findings tend to parallel the observations that elevated cholesterol levels are less of a risk factor for cardiovascular disease in older individuals. Much of the association between the HTT gene and cardiovascular disease may be through the well known role of serotonin on vascular constriction, essential and pulmonary hypertension, platelet aggregation, thrombosis, and atheromata formation (14 –18). At present, we have no explanations of why the HTT gene should be correlated with cholesterol levels, no explanation for the molecular heterosis effect, and no explanation for the age effect. Since these are preliminary, exploratory studies, before struggling with a explanation for these observations they need to be replicated in an epidemiologically based random sample of the population. REFERENCES 1. 2.
3.
4.
5.
6.
7.
Golomb BA. Cholesterol and violence: Is there a connection? Ann Int Med 128:478 – 487, 1998. Lindberg G, Rastam L, Guilberg B, Eklund GA. Low serum cholesterol concentration and short term mortality from injuries in men and women. Br Med J 305:277–279, 1992. Brown SL, Dalive ME, Harris TB, Simonsick EM, Guralnik JM, Kohout FJ. Low cholesterol concentrations and severe depressive symptoms in elderly people. Br Med J 308:1328 – 1332, 1994. Rozzini R, Bertozzi B, Barbisoni P, Trabucchi M. Low serum cholesterol and sertonin metabolism. Risk of depression is higher in elderly patients with lowest serum cholesterol values. Br Med J 312:1298 –1299, 1966. Ogilvie AD, Battersby S, Bubb VJ, FInk G, Hamaar AJ, Goodwin GM, Smith CAD. Polymorphism in the serotonin transporter gene associated with susceptability to major depression. Lancet 347:731–733, 1996. Heils A, Teufel A, Petri S, Sto¨ber G, Riederer P, Bengel D, Lesch KP. Allelic variation of human serotonin transporter gene expression. J Neurochem 66:2621–2624, 1996. Strickberger MW. Genetics. New York, Macmillian, pp 757– 763, 1968.
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL 8. 9.
10.
11.
12.
13.
Gardner EJ, Snustad DP. Principles of Genetics, 6th ed. New York, Wiley, pp 544 –547, 1981. Comings DE, MacMurray JM. Molecular heterosis: Implications for complex inheritance. Am J Hum Genet 61:A196, 1997. Comings DE. Molecular heterosis as the explanation for the controversy about the effect of the DRD2 gene on dopamine D2 receptor density. Mol Psychiatry, 1999, in press. Little KY, McLaughlin DP, Zhang L, Livermore CS, Dalack GW, McFinton PR, Delproposta ZS, Hill E, Cassin BJ, Watson SJ, Cook EH. Cocaine, ethanol, and genotype effects on human midbrain serotonin transporter binding sites and mRNA levels. Am J Psychiatry 155:207–213, 1998. Curb JD, Schatz I, Grove J, Sharp DS, Burchfiel CM, Rodriquez B, Chen R, Huang B, Yano K. Association of short term mortality with plasma cholesterol in the elderly. Can J Cardiol (Suppl)13:A0101, 1997. Weverling-Rijnsburger AW, Blauw GJ, Lagaay AM, Knook
14. 15. 16.
17.
18.
253
DL, Meinders ARE, Westendorp RG. Total cholesterol and risk of mortality in the oldest old. Lancet 350:1119 –1123, 1997. Yildiz O, Smith JR, Purdy RE. Serotonin and vasoconstrictor synergism. Life Sci 62:1723–1732, 1998. De Clerck F. The role of serotonin in thrombogenesis. Clin Physiol Biochem 8(Suppl 3):40 – 49, 1990. Vanhoutte P, Amery A, Birkenhager W, Breckenridge A, Buhler F, Distler A, Dormandy J, Doyle A, Frohlich E, Hansson L, et al. Serotoninergic mechanisms in hypertension. Focus on the effects of ketanserin. Hypertension 11: 111–133, 1988. Amstein R, Fetkovska N, Luscher TF, Kiowski W, Buhler FR. Age and the platelet serotonin vasoconstrictor axis in essential hypertension. J Cardiovasc Pharmacol 11(Suppl 1):S35–S40, 1988. Vanhoutte PM. Serotonin, hypertension and vascular disease. Neth J Med 38:35– 42, 1991.
Association of the Serotonin Transporter Gene with Serum Cholesterol Levels and Heart Disease David E. Comings,* ,1 James P. MacMurray,† Nancy Gonzalez,* Linda Ferry,‡ and Warren R. Peters‡ *Department of Medical Genetics, City of Hope National Medical Center, Duarte, California 91910; †Department of Psychiatry, Loma Linda University School of Medicine, Loma Linda, California; and ‡Loma Linda University, Loma Linda, California Received March 10, 1999, and in revised form April 27, 1999
We had a unique opportunity to examine the role of genetic factors in the health of an elderly population of male athletes. This was a group of men over 55 years of age who compete each year in a series of athletic competitions called the Golden Age Games. Cholesterol levels and a history of the presence or absence of heart disease, angina, and heart attack were obtained on these subjects. Since low cholesterol levels have been reported to be associated with the development of depression and violent death by suicide, especially in the elderly (1– 4), and since the serotonin transporter gene has been implicated in affective disorders (5), we were particularly interested in the potential interaction between this gene and cholesterol levels and heart disease. As a replication group we utilized a group of subjects in good general health from the Loma Linda University Center for Health Promotion (the CHP group) since cholesterol and triglyceride levels were available on these subjects. As an additional replication we obtained blood samples for a third group of subjects from Jerry L. Pettis VA Medical Center, Loma Linda, California, on whom a history of heart attacks (the LLHosp group) was available. A well-characterized insertion (L)/deletion (S) polymorphism (HTTLPR) at the promoter of the HTT gene was utilized since it is known to be associated with variations in HTT gene expression (6). Expression studies with a human choriocarcinoma cell line showed that the nondeletion or long or L allele was associated with three times the rate of expression of the serotonin transporter compared to the S allele. A further relevant aspect of this report is the finding of a molecular heterosis effect at the HTT gene. Heterosis or overdominance refers to the presence of a greater (or
In a study of a group of elderly athletes we observed an unexpected association between serum cholesterol levels and the HTTLPR insertion/deletion polymorphism of the promoter region of the serotonin transporter gene (HTT, SLC6A4). As a follow-up we examined the potential association of this polymorphism with cholesterol and triglyceride levels, or heart disease, in two other groups of subjects. We examined the possible association between cholesterol levels and heart disease and genotypes of the HTTLPR insertion/deletion polymorphism of the promoter region of the HTT gene, in three independent study populations ranging from 42 to 90 years of age. For subjects 55 to 70 years of age in Group 1, cholesterol levels were significantly greater in the LS heterozygotes than either LL or SS homozygotes, indicating a heterosis effect (P < 0.0001). This was replicated in Group 2 (P < 0.015). Triglyceride levels were also significantly elevated in the LS subjects (P < 0.001). In Groups 1 and 3 there was a significant association between LS heterozygosity and heart disease, angina, and heart attacks in subjects 70 years of age or less. All of these associations were absent in subjects >70 years of age. While these studies are preliminary and exploratory, they are consistent with a relationship of the HTT gene in cholesterol levels and a risk for heart disease. Replication of these findings in independent, epidemiologically based studies is required. © 1999 Academic Press
Key Words: serotonin; transporter; cholesterol; triglycerides; age; coronary heart disease.
1
To whom correspondence should be addressed. Fax: (626) 301-8980. E-mail: [email protected]. 248 1096-7192/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL
lesser) phenotypic effect in heterozygotes or hybrids than in homozygotes or parental strains. It is well studied in corn where hybrids of two strains show consistently higher yields in bushels per acre than either parent strain (7,8). Molecular heterosis refers to a situation in which heterozygotes for a polymorphic gene marker show a greater or lesser phenotypic effect than either homozygote (9). We have found molecular heterosis to be a common phenomena and have observed it in a wide range of genes (10). Little et al. observed molecular heterosis at the HTT gene (11). They examined levels of [ 125I]b-CIT (citalopram) binding (fmol/mg) to the serotonin transporter in the dorsal and median raphe nuclei and substantia nigra of human controls and subjects with chronic cocaine use genotyped at the HTTLPR polymorphism. This showed that [ 125I]b-CIT binding was lower in the LS heterozygotes than either the SS or LL homozygotes in all three regions. A twoway ANOVA was significant for genotype and region and genotype main effect (P , .001). Several recent studies have shown that elevated cholesterol levels are less associated with cardiovascular disease in elderly individuals than in younger subjects (12,13). The Golden Age Games sample allowed us to investigate this by separately examining participants who were younger or older than 70 years of age. The age 70 was chosen to be past the age of the greatest risk for death due to coronary artery disease and yet not so great that it would exclude all the older subjects, but instead the subjects were divided into two approximately equal groups. Thus, 54% were in the ,70 group and 46% were in the $70-year-old group. If the cut had been made at 65 years of age, only 34% would have been in the ,65year-old group. Thus, the above three groups allowed us to do the following: (1) Determine if the HTTLPR polymorphism of the HTT gene predicted cholesterol levels. (2) Retest this association in an independent group of subjects. (3) Determine if the HTT gene was also associated with heart attacks. (4) Retest this association in an independent group of subjects. (5) Determine if the risks were different in the subjects 70 years of age or less versus those over age 70.
249
for DNA and cholesterol testing. Of the 100 subjects, 74% were non-Hispanic Caucasian, 18% were African-American, and 8% were Hispanic or other. They ranged in age from 55 to 91 years. The mean age of the 58 subjects in the #70 age group was 63.9 years (SD 4.1 years). The mean age of the 42 subjects in the .70 age group was 74.8 (SD 3.1). Subjects were coded for a history of different cardiovascular disease problems (angina, heart attacks, or hypertension). The Center for Health Promotion group (CHP). The subjects from the CHP study consisted of 102 non-Hispanic Caucasians from the Loma Linda University Center for Health Promotion. The age, sex, weight, height, and waist-hip ratio were determined on each subject. The subjects ranged in age from 42 to 70 years of age with a mean age of 55.4 years (SD 7.5 years). A fasting blood sample was obtained for determination of cholesterol and triglycerides. The recruitment particularly targeted staff members of the Loma Linda University and Loma Linda University Medical Center. The Loma Linda V.A. Hospital group (LLHosp). The subjects from the LLHosp group consisted of a random sample of 83 non-Hispanic Caucasian male inpatients from the medical wards of the Jerry L. Pettis Memorial Veterans Administration Medical Center, Loma Linda, California. They consisted of individuals 42 to 74 years of age with a mean age of 57.0 years (SD 8.87 years). There were too few subjects in the .70 age group to analyze separately. Data on the presence or absence of heart attacks were available. In all three studies coded samples of blood were sent to the Department of Medical Genetics at the City of Hope National Medical Center where the genetic studies were performed blind to clinical data. All studies were approved by the IRBs of both institutions.
METHODS
Laboratory tests. Cholesterol and triglyceride levels were determined using the REP Ultra-30 HDL, VLDL/LDL cholesterol system of Helena Laboratories (Beaumont, TX). Polymerase chain reaction genotyping of the HTTLPR polymorphism of the HTT gene was performed using the primers and conditions reported by Heils et al. (6).
The Golden Age Games group (GAG). Each year veterans over 55 years of age volunteer for a national athletic competition called the Golden Age Games. In 1996 the Golden Age Games were held in Southern California and we obtained blood samples
Statistics. The mean cholesterol and triglyceride levels for the different HTT genotypes, SS versus LS versus LL, and LS versus LL 1 SS, were compared by ANOVA. The significance levels were determined on the basis of the F ratio. For the three-way com-
250
COMINGS ET AL.
parisons, a post hoc Tukey test indicated any individual means that were significantly different from other individual means at a 5 0.05. Linear regression analysis of the association between cholesterol and triglyceride levels and genotype was performed scoring the HTT gene as LL or SS 5 0 and LS 5 1. To examine the percentage of the variance for heart attack, heart disease, and angina, their presence was scored as 1 and their absence as 0. This allowed the determination of r and r 2 or the percentage of the variance attributable to the HTT gene. x 2 analysis was used to compare the number of subjects with a history of heart disease, angina, or heart attack in the GAG group, and heart attacks in the LLHosp group, versus the LS and LL or SS HTT genotype groups. All statistical analysis was by the SPSS statistical package (SPSS, Inc., Chicago, IL). RESULTS The results for the 100 Golden Age Games subjects are shown in Table 1. We first examined all races together. The cholesterol levels for the entire GAG group ranged from 100 to 312 mg/dl indicating there were no extreme outliers. The mean cholesterol was 231.41 mg/dl for the LS heterozygotes versus 197.00 mg/dl for the LL and 206.36 mg/dl for the SS groups (P # 0.0056). When tested for heterosis by comparing LS heterozygotes to LL 1 SS homozygotes, the mean cholesterol for the homozygotes was 200.65 mg/dl (P # 0.0017). When restricted to Caucasians the LS heterozygotes again had the higher mean cholesterol levels but the results were not significant. When the subjects in the 55- to 70-year-old group were examined for all races the mean cholesterol for the LS heterozygotes was 240.00 mg/dl, compared to 190.76 mg/dl for the LL homozygotes and 201.00 mg/dl for the SS homozygotes (P # 0.0005). When tested for heterosis by comparing LS heterozygotes to LL 1 SS homozygotes, the mean cholesterol for the homozygotes was 194.00 mg/dl (P # 0.0001). When restricted to Caucasians the results were still significant for all three genotypes (P # 0.011) and for heterozygotes versus homozygotes (P # 0.0026). By comparison, none of these associations were significant for subjects over age 70 (Table 1). To determine if these findings could be replicated, we examined the CHP subjects. In addition to cholesterol levels these subjects also had triglyceride levels. We examined individuals in the 42- to 70year age group (Table 2). There were too few subjects in the .70-year group for statistical analysis.
TABLE 1 Association between the Genotypes of the HTT Gene and Serum Cholesterol Levels in Golden Age Games Males in Two Age Groups Mean cholesterol Genotype
n
(mg/dl)
SD
F
P
41.45 48.30 49.10 48.30 48.30
5.48
#0.0056
10.36
#0.0017
1.89
#0.158
3.79
#0.055
8.72
#0.0005
17.07
#0.0001
Age all races (HTTLPR) (n 5 100) LL LS SS LL 1 SS LS
39 36 25 64 36
197.00 231.41* 206.36 200.65 231.41
Age Caucasians only (HTTLPR) (n 5 72) LL LS SS LL 1 SS LS
25 27 20 45 27
205.20 227.63* 208.05 206.46 227.62
41.52 40.82 53.55 46.69 47.12
Age 55 to 70 all races (HTTLPR) (n 5 58) LL LS SS LL 1 SS LS
26 20 12 38 20
190.76 240.00# 201.00 194.00 240.00
42.11 44.05 28.41 38.22 44.05
Age 55 to 70 Caucasians only (HTTLPR) (n 5 38) LL LS SS LL 1 SS LS
16 14 8 24 14
199.37 243.21* 202.00 200.25 243.21
39.95 44.93 28.71 35.96 44.93
5.11
#0.011
10.49
#0.0026
Age .70 all races (HTTLPR) (n 5 42) LL LS SS LL 1 SS LS
13 16 13 26 16
209.46 220.68 211.31 210.38 220.68
38.67 52.58 63.45 51.51 52.58
0.194
#0.82
0.390
#0.53
Age .70 Caucasians only (HTTLPR) (n 5 34) LL LS SS LL 1 SS LS
9 13 12 21 13
215.55 210.84 212.08 213.57 210.84
44.60 28.91 66.21 28.91 28.91
0.025
#0.97
0.025
#0.87
* Significantly different from LL by Tukey test at a 5 0.05. # Significantly different from LL and SS by Tukey test at a 5 0.05.
The cholesterol levels ranged from 109 to 352 mg/dl with only two individuals with a level greater than 275. The F values for all three genotypes were sig-
251
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL
TABLE 2 Association between the Genotypes of the HTT Gene and Serum Cholesterol and Triglyceride Levels in CHP Males and Females 42 to 70 Years of Age (n 5 99) Genotype
n
Mean (mg/dl)
SD
F
P
32.71 40.67 44.62 36.02 40.67
3.29
#0.042
6.09
#0.015
Cholesterol LL LS SS LL 1 SS LS
35 51 13 48 51
196.80 218.60* 205.46 199.14 218.25
TABLE 3 HTT Genotype and History of Angina, Heart Disease, or Heart Attack (% Yes or No, All 1 df)
Triglycerides (log transformed) LL LS SS LL 1 SS LS
35 51 13 48 51
1.99 2.15* 2.02 1.99 2.15
0.23 0.22 0.20 0.22 0.22
9.6% of the variance of the triglyceride levels (P # 0.0066). We examined the association between the presence of the HTT genotypes and history of heart disease, angina, or heart attack for the two age ranges in the GAG group (Table 3). For the GAG subjects less than 70 years of age, the frequencies for those with the respective conditions in those with the LS versus the LL or SS genotypes were are follows: heart disease 68.4% versus 35.1%, angina 42.1% versus 16.2%, and heart attack 42.1% versus
5.76
#0.0043
11.46
#0.001
HTT Genotype Condition
* Significantly different from LL by Tukey test at a 5 0.05.
LS
LL 1 SS
x2
P
5.59
#0.018
Golden Age Games A. Heart disease
nificant at P # 0.045. The Tukey test showed that the LS values were significantly higher than the LL values at a 5 0.05, and the test for heterosis by comparison of heterozygotes versus homozygotes was significant (P # 0.015). The triglyceride levels ranged from 40 to 496 mg/dl with skewing of the higher levels. Because of this wide range the values were -log transformed. The log-transformed values showed a normal distribution. The results of ANOVA for the triglyceride levels were significant for all three genotypes (P # 0.0056) and for the comparison of heterozygotes versus homozygotes (P # 0.002). When sex was used as a covariant it was not significant for either cholesterol or triglyceride levels. When BMI was used as a covariant the association between LS subjects and elevated cholesterol remained significant for both GAG and CHP groups. When non-log-transformed triglyceride levels were used ANOVAs for the three genotype groups and the LL 1 SS versus LS groups were still significant (P , 0.01 and #0.0052, respectively). For the Golden Age Games subjects 70 years of age or less, the HTT gene accounted for 23.3% of the variance of cholesterol levels, P # .0001; 10.2% of the variance for heart attack, P # 0.016; 9.9% of the variance for heart disease, P # 0.018; and 8.0% of the variance for angina, P # 0.034. For the CHP subjects, the HTT gene accounted for 5.3% of the variance of the cholesterol levels (P # 0.019), and
Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
13 (68.4) 6 (31.6)
13 (35.1) 24 (64.9)
Age .70 all races (HTTLPR) (n 5 42) Yes No
8 (50.0) 8 (50.0)
13 (50.0) 13 (50.0)
.000
#1.00
B. Angina Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
8 (42.1) 11 (57.9)
6 (16.2) 31 (83.8)
4.48
#0.034
Age .70 All races (HTTLPR) (n 5 42) Yes No
5 (31.3) 11 (68.8)
8 (30.8) 18 (69.2)
.001
#0.97
C. Heart attack Age 55 to 70 all races (HTTLPR) (n 5 56) Yes No
8 (42.1) 11 (57.9)
5 (13.5) 32 (86.5)
5.75
#0.016
1.00
#0.31
4.91
#0.03
Age .70 all races (HTTLPR) (n 5 42) Yes No
6 (37.5) 10 (62.5)
6 (23.1) 20 (76.9)
LLHosp group Heart attack (n 5 83) Yes No
11 (23.4) 36 (76.6)
2 (5.6) 34 (94.4)
252
COMINGS ET AL.
13.5%. The increase in the frequency for those with the LS genotype was significant for all three diagnostic groups (P # 0.016 to # 0.034). By comparison, for those over 70 years of age there was no significant difference in the frequency for any of these heart problems in those with the LS genotypes versus the LL or SS genotypes (P # 0.31 to 1.0). To determine if we could replicate the association between the HTT alleles and heart attacks, we examined the LLHosp group (Table 3). Here 23.4% of the LS subjects had a history of a heart attack versus 5.6% of the LL or SS subjects (P , 0.03). DISCUSSION While the present results are consistent with a role of the HTT gene in cholesterol levels and risk of heart disease, we clearly recognize a number of limitations of the present study. Foremost is the unique nature of the GAG group in that these were elderly athletes. As such, they were not typical of the general population. However, we specifically chose this group because we felt they might provide some unique insights into genetic factors associated with good health rather than disease. Thus, they may have magnified an effect that would be less dramatic in a comparable aged group in the general population. This is supported by the observation that in the GAG group that were 70 years of age or less, the HTT gene accounted for 23% of the variance of cholesterol, while in the CHP group it accounted for only 5.3%. While the CHP individuals also took an interest in their health they were not necessarily healthier than subjects in the general population. Finally, the LLHosp group was atypical in that they are drawn from the general inpatient medical wards. Despite these limitations, the results for cholesterol and/or risk of heart disease were consistent across all groups. Statistically it could be argued that the comparison of LS heterozygotes versus SS 1 LL homozygotes was a post hoc analysis of the data. However, as discussed in the introduction this finding was not unexpected since we have observed molecular heterosis for phenotypes associated with a wide range of other genes (ADRA1C, ADRA2C, AR, CD4, CD8, CHRNA4, CNR1, DBH, DRD4, DRD5, DAT1, ESRA, FRAXA, GABRA1, GABRA3, GABRA5, GABBR1, GABRB3, GABBR, GAD2, HTR1A, INFG, INS, MAOA, MAOB, MC3R, NOS1A, NOS2A, OB1875, OB1871, PENK, TH,TDO2) (10). In addition, while the molecular heterosis effect was post hoc for the GAG group, this was not the case for the other two groups.
The observation that elderly subjects with elevated cholesterol levels are at less risk for cardiovascular disease stimulated us to examine the GAG subjects less than 70 years of age versus those greater than 70 years of age. This could only be tested in the Golden Age Games subjects since they were the only group with a sufficient number of subjects over 70 to have statistical validity. The pattern for cholesterol levels, heart disease, angina, and heart attack were all similar. Thus, the effect on these variables was greater for LS heterozygotes only in the less than 70 age group. It disappeared or was negative (less effect in the LS heterozygotes) in the over 70 age group. These findings tend to parallel the observations that elevated cholesterol levels are less of a risk factor for cardiovascular disease in older individuals. Much of the association between the HTT gene and cardiovascular disease may be through the well known role of serotonin on vascular constriction, essential and pulmonary hypertension, platelet aggregation, thrombosis, and atheromata formation (14 –18). At present, we have no explanations of why the HTT gene should be correlated with cholesterol levels, no explanation for the molecular heterosis effect, and no explanation for the age effect. Since these are preliminary, exploratory studies, before struggling with a explanation for these observations they need to be replicated in an epidemiologically based random sample of the population. REFERENCES 1. 2.
3.
4.
5.
6.
7.
Golomb BA. Cholesterol and violence: Is there a connection? Ann Int Med 128:478 – 487, 1998. Lindberg G, Rastam L, Guilberg B, Eklund GA. Low serum cholesterol concentration and short term mortality from injuries in men and women. Br Med J 305:277–279, 1992. Brown SL, Dalive ME, Harris TB, Simonsick EM, Guralnik JM, Kohout FJ. Low cholesterol concentrations and severe depressive symptoms in elderly people. Br Med J 308:1328 – 1332, 1994. Rozzini R, Bertozzi B, Barbisoni P, Trabucchi M. Low serum cholesterol and sertonin metabolism. Risk of depression is higher in elderly patients with lowest serum cholesterol values. Br Med J 312:1298 –1299, 1966. Ogilvie AD, Battersby S, Bubb VJ, FInk G, Hamaar AJ, Goodwin GM, Smith CAD. Polymorphism in the serotonin transporter gene associated with susceptability to major depression. Lancet 347:731–733, 1996. Heils A, Teufel A, Petri S, Sto¨ber G, Riederer P, Bengel D, Lesch KP. Allelic variation of human serotonin transporter gene expression. J Neurochem 66:2621–2624, 1996. Strickberger MW. Genetics. New York, Macmillian, pp 757– 763, 1968.
SEROTONIN TRANSPORTER GENE AND CHOLESTEROL 8. 9.
10.
11.
12.
13.
Gardner EJ, Snustad DP. Principles of Genetics, 6th ed. New York, Wiley, pp 544 –547, 1981. Comings DE, MacMurray JM. Molecular heterosis: Implications for complex inheritance. Am J Hum Genet 61:A196, 1997. Comings DE. Molecular heterosis as the explanation for the controversy about the effect of the DRD2 gene on dopamine D2 receptor density. Mol Psychiatry, 1999, in press. Little KY, McLaughlin DP, Zhang L, Livermore CS, Dalack GW, McFinton PR, Delproposta ZS, Hill E, Cassin BJ, Watson SJ, Cook EH. Cocaine, ethanol, and genotype effects on human midbrain serotonin transporter binding sites and mRNA levels. Am J Psychiatry 155:207–213, 1998. Curb JD, Schatz I, Grove J, Sharp DS, Burchfiel CM, Rodriquez B, Chen R, Huang B, Yano K. Association of short term mortality with plasma cholesterol in the elderly. Can J Cardiol (Suppl)13:A0101, 1997. Weverling-Rijnsburger AW, Blauw GJ, Lagaay AM, Knook
14. 15. 16.
17.
18.
253
DL, Meinders ARE, Westendorp RG. Total cholesterol and risk of mortality in the oldest old. Lancet 350:1119 –1123, 1997. Yildiz O, Smith JR, Purdy RE. Serotonin and vasoconstrictor synergism. Life Sci 62:1723–1732, 1998. De Clerck F. The role of serotonin in thrombogenesis. Clin Physiol Biochem 8(Suppl 3):40 – 49, 1990. Vanhoutte P, Amery A, Birkenhager W, Breckenridge A, Buhler F, Distler A, Dormandy J, Doyle A, Frohlich E, Hansson L, et al. Serotoninergic mechanisms in hypertension. Focus on the effects of ketanserin. Hypertension 11: 111–133, 1988. Amstein R, Fetkovska N, Luscher TF, Kiowski W, Buhler FR. Age and the platelet serotonin vasoconstrictor axis in essential hypertension. J Cardiovasc Pharmacol 11(Suppl 1):S35–S40, 1988. Vanhoutte PM. Serotonin, hypertension and vascular disease. Neth J Med 38:35– 42, 1991.