Skip to content

Advertisement

  • Proceedings
  • Open Access

Mitochondrial genetic effects on latent class variables associated with susceptibility to alcoholism

  • 1,
  • 2,
  • 2, 3,
  • 2,
  • 2 and
  • 2, 3Email author
BMC Genetics20056 (Suppl 1) :S158

https://doi.org/10.1186/1471-2156-6-S1-S158

  • Published:

Abstract

We report the results of statistical genetic analyses of data from the Collaborative Study on the Genetics of Alcoholism prepared for the Genetic Analysis Workshop 14 to detect and characterize maternally inherited mitochondrial genetic effects on variation in latent class psychiatric/behavioral variables employed in the diagnosis of alcoholism. Using published extensions to variance decomposition methods for statistical genetic analysis of continuous and discrete traits we: 1) estimated the proportion of the variance in each trait due to the effects of mitochondrial DNA (mtDNA), 2) tested for pleiotropy, both mitochondrial genetic and residual additive genetic, between trait pairs, and 3) evaluated whether the simultaneous estimation of mitochondrial genetic effects on these traits improves our ability to detect and localize quantitative trait loci (QTL) in the nuclear genome. After correction for multiple testing, we find significant (p < 0.009) mitochondrial genetic contributions to the variance for two latent class variables. Although we do detect significant residual additive genetic correlations between the two traits, there is no evidence of a residual mitochondrial genetic correlation between them. Evidence for autosomal QTL for these traits is improved when linkage screens are conditioned on significant mitochondrial genetic effects. We conclude that mitochondrial genes may contribute to variation in some latent class psychiatric/behavioral variables associated with alcoholism.

Keywords

  • Quantitative Trait Locus
  • Genetic Correlation
  • Latent Class
  • Additive Genetic Variance
  • Variance Decomposition

Background

A number of studies have report associations between mtDNA variation and chronic alcoholism. However, the deletions, insertions, modifications, and other rearrangements described in these studies represent damage done to mtDNA in the liver, white blood cells, and elsewhere as a consequence of oxidative stress associated with chronic ethanol consumption [1, 2]. Only a small number of studies have detected associations between mtDNA polymorphisms and psychiatric disorders and syndromes that overlap those of chronic alcoholism [3, 4]. These observations, plus the possibility that heritable mtDNA mutations could influence susceptibility to alcohol-induced oxidative stress damage, motivate our current study in which we analyze data from the Collaborative Study of the Genetics of Alcoholism (COGA) [5] to detect and characterize mitochondrial genetic effects on variation in latent class psychiatric/behavioral variables.

Methods

Data

The COGA dataset for the Genetic Analysis Workshop 14 includes 1,614 individuals, 788 females and 826 males, aged 17–91 years (mean = 40.09 ± 15.2 years), in 143 pedigrees of varying size and complexity. Families were recruited into COGA on the basis of a positive diagnosis of alcoholism (DSM-III-R (later, DSM-IV) and Feighner criteria) for a focal proband and 2 first degree relatives.

We analyzed data on the 14 latent class variables assessed in 1,181 to 1,388 of the COGA family members. The majority of these latent class variables were measured by COGA researchers on either a dichotomous (no = 1 and yes = 2; recoded as no = 0 and yes = 1 in our analyses) or ordinal scale (0, 1, 2, ..., n). These variables include: persistent desire to stop drinking (DESIRE), morning drinking (MORNING), craving for a drink (CRAVING), binge drinking (BINGE), narrowing of drinking repertoire (NARROW), gave up other activities to drink (GAVE UP), experience three or more blackouts (BLACKOUT), experienced withdrawal symptoms (WITHDRAWAL), experience physical health problems as a result of drinking (HEALTH), and experienced emotional/psychological problems as a result of drinking (PSYCHOL). By combining the 2 non-zero categories for the ordinal scale variable, spent so much time drinking there was little time for anything else (SPENT), we converted it to a dichotomous trait. We also analyzed the 2 continuous scale traits: maximum number of drinks in a 24 hour period (MAX DRINK) and the number of cigarette packs per day for one year (CIGPKYRS). In linkage analyses reported here we used genotype data at 315 microsatellite marker loci distributed across the 23 chromosomes in 1,376 individuals.

Variance decomposition

Pedigree, phenotype, and genotype data were managed using PEDSYS [6] routines. We used a maximum likelihood based variance decomposition approach and extensions to this approach, implemented in SOLAR [7], to conduct all statistical genetic analyses. For basic statistical genetic analyses of continuously distributed traits, this approach models the phenotypic covariance as the sum of the additive genetic and random environmental covariances. We simultaneously estimated for each trait its mean, the mean effects of age and sex, a residual heritability (i.e., h2 = ), and residual environmental standard deviation. We extended this basic approach to discrete traits using a multivariate normal threshold model in which the discrete trait is assumed to be determined by a threshold process acting on an underlying, continuous liability distribution [8]. To detect and estimate the effects of a maternally inherited mitochondrial effect on these traits, we employed the approach in Czerwinski et al. [9] in which the within-pedigree covariance matrix is modified by introducing an additional variance component, , and its structuring matrix M. M specifies the mitochondrial relationships in the pedigree that share a maternal line of descent [9]; thus matrix element m ij = 1 if individuals i,j are members of the same maternal lineage and m ij = 0 otherwise [10, 11]. Aspects of mitochondrial inheritance such as heteroplasmy, threshold effects, phenotypic heterogeneity, and mtDNA/nuclear DNA interactions were not specifically modeled [11].

Bivariate variance decomposition

We tested for possible mitochondrial genetic pleiotropy between any pairs of traits for which we detected a significant mitochondrial genetic component using a bivariate extension to the univariate variance components approach. This extension allows for the simultaneous estimation of all parameters estimated in the univariate analyses, plus the estimation of a residual additive genetic correlation, a mitochondrial genetic correlation, and a random environmental correlation between pairs of traits. For bivariate analyses of continuous and discrete trait pairings, we employed the method introduced by Williams et al. [12] to allow the liability of the discrete trait to be correlated with the quantitative trait.

Adjustment for multiple tests

We used genetic correlations between all 78 latent trait pairs in a modified Bonferroni correction procedure to control for false-positive detection of mitochondrial genetic components to the variance in these 14 traits. To obtain an adjusted p-value consonant with α = 0.05, we divided 0.05 by 1 + (N - 1)(1 - |ρG|), where N is the number of phenotypes, and |ρG| is the mean of the absolute value of the genetic correlations.

Linkage

We conducted a series of multipoint, whole genome linkage screens for each latent class variable exhibiting a significant (corrected p < 0.009) mitochondrial effect to detect evidence for quantitative trait loci (QTL) in nuclear chromosomal regions (only marker-specific linkage analyses were conducted with X-chromosome markers). Second, we performed locus-specific linkage analyses at the multipoint peaks providing the best evidence for a QTL for each trait to obtain an estimate of the improvement in the LOD score conditional on the mitochondrial genetic component.

Likelihood ratio tests

Significance of all parameters was determined by likelihood ratio tests in which the likelihood of a more general model in which all parameters are estimated was compared to a nested restricted model in which some parameter values are constrained. The test statistic, twice the difference in the log likelihoods of the two models, was asymptotically distributed approximately as a χ2 variate with degrees of freedom equal to the difference in the numbers of estimated parameters in the models. However, when a parameter value was constrained to the boundary of the parameter space (e.g., h2 = 0), the test statistic was distributed as a 1/2:1/2 mixture of χ2 and a point mass of zero [13]. To control for the overall false-positive rate given the finite marker locus density in the COGA microsatellite genome linkage map, we estimated genome-wide p-values by means of a method suggested by Feingold et al. [14]

Results

Table 1 provides maximum likelihood estimates (MLEs) for components of the variance from a restricted model in which the proportion of the residual phenotypic variance due to the mitochondrial genetic component is constrained to zero (left side) and the more general model in which the effect of mitochondrial genetic component is estimated. A nominally significant mitochondrial genetic component (p(h2mt = 0) ≤ 0.05) is detected for three traits: DESIRE, CIGPKYRS, and CRAVING but, after correction for multiple testing, only the CIGPKYRS and CRAVING remain significant (p < 0.009; 78 pair-wise genetic correlations |mean ρG| = 0.64). The mitochondrial genetic component accounts for 14% and 19% of the residual phenotypic variance and 77% and 65% of the additive genetic variance in these two traits, respectively.
Table 1

Proportions of the residual phenotypic variance in latent class variables

Trait

Restricted (h2mt = 0)

Unrestricted (h2mt ≠ 0)

 

h2

e2

c2a

e2

h2

h2mt

c2

p(h2mt = 0)

MAX DRINK

0.16

0.84

0.21

0.85

0.12

0.03

0.21

0.10

SMOKER

0.43

0.57

n/a

0.75

0.10

0.14

n/a

0.15

DESIRE

0.47

0.53

n/a

0.55

0.35

0.10

n/a

0.05

MORNING

0.18

0.82

n/a

0.84

0.12

0.04

n/a

0.29

CIGPKYRS

0.17

0.83

0.21

0.82

0.04

0.14

0.21

0.002

CRAVING

0.29

0.71

n/a

0.71

0.10

0.19

n/a

0.004

BINGE

0.06

0.94

n/a

0.96

0.00

0.03

n/a

0.28

SPENT

0.07

0.93

0.01

0.94

0.05

0.01

0.01

0.31

NARROW

0.34

0.66

n/a

0.68

0.26

0.06

n/a

0.23

GAVE UP

0.26

0.74

n/a

0.82

0.07

0.11

n/a

0.07

BLACKOUTS

0.38

0.62

n/a

0.62

0.36

0.02

n/a

0.38

WITHDRAWAL

0.20

0.80

n/a

0.84

0.11

0.05

n/a

0.28

HEALTH

0.26

0.74

n/a

0.76

0.23

0.01

n/a

1.00

PSYCHO

0.36

0.64

n/a

0.64

0.34

0.01

n/a

0.42

aProportion of total phenotypic variance due to covariates (c2), additive genetic effects (h2), mitochondrial genetic effects (h2mt), and random environmental factors (e2).

Bivariate genetic analysis of the two traits identified above yielded the following correlation estimates when no mitochondrial genetic component was included in the model: ρG = 0.06 ± 0.17, ρE = 0.24 ± 0.07, and ρP = 0.06. When a mitochondrial genetic component was included in the bivariate model for the traits, we detected a significant residual additive genetic correlation (ρG = 1.00 ± 0.11), but the estimated mitochondrial genetic correlation between these two traits (ρM= -0.36 ± 0.34) was not significantly different from zero (p > 0.05). The MLE for the residual additive genetic correlation between these two latent class variables was indicative of complete, or nearly complete, pleiotropy. Given the standard errors around the estimate, 60% to 100% of the residual additive genetic variance in each of these traits was due to the effects of genes that also influence the other trait. The residual correlation due to the shared effects of unmeasured environmental factors (i.e., due to the effects of covariates not included in the models, measurement errors, and/or non-additive genetic components) was significantly different from zero (ρE = 0.20 ± 0.05, p < 0.05).

Our univariate multipoint linkage screens detected no significant evidence for a QTL influencing these two latent class variables on any of the autosomes. The best evidence for an autosomal QTL for each of these three traits is presented in Table 2. Subsequent maximization of these linkage models at the locations of their multipoint LOD peaks, conditional on the mitochondrial genetic component, raised statistical support for each QTL. None achieved genome-wide significance at the α = 0.05 level (LOD = 2.77).
Table 2

LOD scores from linkage screens with and without mitochondrial effects

Trait

h2mt = 0

h2mt ≠ 0

QTL Location

 

LOD

Genome-wide p

LOD

Genome-wide p

Locus-specific p

cM from pter

pter STR locus

qter STR locus

CIGPKYRS

1.68

0.76

2.67

0.06

0.00045

158

D2S1328

D2S1334

CRAVING

1.41

1.49

2.21

0.20

0.00014

258

D3S1311

3-qter

Discussion

The magnitude of the mitochondrial genetic effect on the phenotypic variance of the two latent class traits is likely small. However, mitochondrial genetic effects do account for a substantive proportion of the additive genetic variance in each trait and this proportion may be of biological, as well as statistical, significance.

Possible confounding of mitochondrial genetic and non-mitochondrial maternal effects (both genetic and environmental) is a valid concern; the range of pedigree size and complexity in the COGA families may mitigate this concern. We also believe the following observations provide circumstantial evidence that such confounding has not occurred. 1) Addition of the mitochondrial genetic component results in diminution of the residual additive genetic, rather than the environmental, component of the variance. 2) Re-analyses of these data incorporating the following surrogates for maternal effects as covariates, do not eliminate the mitochondrial genetic effect: maternal age at birth and household effects (unpublished data).

Our analyses of these data cannot identify the genes, gene products that may be involved, or the mechanisms by which they operate. However, the lack of a significant mitochondrial genetic correlation between the two traits tentatively suggests that multiple mtDNA variants, either in the same gene or in different genes, are responsible for the mitochondrial effects.

Lastly, our analyses offer tentative support for inclusion of significant mitochondrial genetic components in variance components models used in linkage screens. In each case, the LOD scores for non-significant multipoint peaks increased, although not quite to the level of genome-wide significance given the finite marker density of the linkage map and the mean recombination frequency in the COGA pedigrees.

Conclusion

We detect evidence of mitochondrial genetic variation in two latent class phenotypes associated with susceptibility to alcoholism in data from the COGA families. To our knowledge, this is the first report of such an effect on any alcoholism-related traits.

Abbreviations

COGA: 

Collaborative Study on the Genetics of Alcoholism

MLE: 

Maximum likelihood estimate

mtDNA: 

Mitochondrial DNA

QTL: 

Quantitative trait locus

Declarations

Acknowledgements

This work was supported in part with research grants from the National Institutes of Health: P01 HL045522 and P01 HL028922. Development of statistical genetic methods used in this work and their implementation in SOLAR software is supported by a MERIT award to J. Blangero from the National Institute of Mental Health: R01 MH059490.

Authors’ Affiliations

(1)
Youngstown State University, Department of Sociology & Anthropology, One University Plaza, Youngstown, OH 44555, USA
(2)
Department of Genetics, Southwest Foundation for Biomedical Research, San Antonio, TX, USA
(3)
Southwest National Primate Research Center, Southwest Foundation for Biomedical Research, San Antonio, TX 78245, USA

References

  1. Cahill A, Cunningham CC, Adachi M, Ishii H, Bailey SM, Fromenty B, Davies A: Effects of alcohol and oxidative stress on liver pathology: the role of the mitochondrion. Alcohol Clin Exp Res. 2002, 26: 907-915. 10.1097/00000374-200206000-00022.PubMed CentralView ArticlePubMedGoogle Scholar
  2. Tsuchishima M, Tsutsumi M, Shiroeda H, Yano H, Ueshima Y, Shimanaka K, Takase S: Study of mitochondrial DNA deletion in alcoholics. Alcohol Clin Exp Res. 2000, 24 (4 Suppl): 12S-15S.PubMedGoogle Scholar
  3. Inagaki T, Ishino H, Seno H, Ohguni S, Tanaka J, Kato Y: Psychiatric symptoms in a patient with diabetes mellitus associated with point mutation in mitochondrial DNA. Biol Psychiatry. 1997, 42: 1067-1069. 10.1016/S0006-3223(97)00351-X.View ArticlePubMedGoogle Scholar
  4. Ramoz N, Reichert JG, Smith CJ, Silverman JM, Bespalova IN, Davis KL, Buxbaum JD: Linkage and association of the mitochondrial aspartate/glutamate carrier SLC25A12 gene with autism. Am J Psychiatry. 2004, 161: 662-669. 10.1176/appi.ajp.161.4.662.View ArticlePubMedGoogle Scholar
  5. Begleiter H, Reich T, Nurnberger J, Li TK, Conneally PM, Edenberg H, Crowe R, Kuperman S, Schuckit M, Bloom F, Hesselbrock V, Porjesz B, Cloninger CR, Rice J, Goate A: Description of the Genetic Analysis Workshop 11 Collaborative Study on the Genetics of Alcoholism. Genet Epidemiol. 1999, 17 (Suppl 1): S25-S30.View ArticlePubMedGoogle Scholar
  6. Dyke B: PEDSYS: A Pedigree Database Management System Users Manual. 1996, San Antonio, TX: Population Genetics Laboratory, Department of Genetics, Southwest Foundation for Biomedical ResearchGoogle Scholar
  7. Almasy L, Blangero J: Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet. 1998, 62: 1198-1211. 10.1086/301844.PubMed CentralView ArticlePubMedGoogle Scholar
  8. Duggirala R, Williams JT, Williams-Blangero S, Blangero J: A variance component approach to dichotomous trait linkage analysis using a threshold model. Genet Epidemiol. 1997, 14: 987-992. 10.1002/(SICI)1098-2272(1997)14:6<987::AID-GEPI71>3.0.CO;2-G.View ArticlePubMedGoogle Scholar
  9. Czerwinski SA, Williams JT, Demerath EW, Towne B, Siervogel RM, Blangero J: Does accounting for mitochondrial genetic variation improve the fit of genetic models?. Genet Epidemiol. 2001, 21 (Suppl 1): S779-S782.PubMedGoogle Scholar
  10. Beavis WD, Pollak E, Frey KJ: A theoretical model for quantitatively inherited traits influenced by nuclear-cytoplasmic interactions. Theor Appl Genet. 1987, 74: 571-578. 10.1007/BF00288854.View ArticlePubMedGoogle Scholar
  11. Schork NJ, Guo SW: Pedigree models for complex human traits involving the mitochondrial genome. Am J Hum Genet. 1993, 53: 1320-1337.PubMed CentralPubMedGoogle Scholar
  12. Williams JT, Van Eerdewegh P, Almasy L, Blangero J: Joint multipoint linkage analysis of multivariate qualitative and quantitative traits. I. Likelihood formulation and simulation results. Am J Hum Genet. 1999, 65: 1134-1147. 10.1086/302570.PubMed CentralView ArticlePubMedGoogle Scholar
  13. Hopper JL, Mathews JD: Extensions to multivariate normal models for pedigree analysis. Ann Hum Genet. 1982, 46: 373-383.View ArticlePubMedGoogle Scholar
  14. Feingold E, Brown PO, Siegmund D: Gaussian models for genetic linkage analysis using complete high-resolution maps of identity by descent. Am J Hum Genet. 1993, 53: 234-251.PubMed CentralPubMedGoogle Scholar

Copyright

© Lease et al; licensee BioMed Central Ltd 2005

This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Advertisement