Volume 3 Supplement 7

Genetic Analysis Workshop 16

Open Access

Genome-wide discovery of maternal effect variants

  • Jack W KentJr1Email author,
  • Charles P Peterson1,
  • Thomas D Dyer1,
  • Laura Almasy1 and
  • John Blangero1
BMC Proceedings20093(Suppl 7):S19

DOI: 10.1186/1753-6561-3-S7-S19

Published: 15 December 2009

Abstract

Many phenotypes may be influenced by the prenatal environment of the mother and/or maternal care, and these maternal effects may have a heritable component. We have implemented in the computer program SOLAR a variance components-based method for detecting indirect effects of maternal genotype on offspring phenotype. Of six phenotypes measured in three generations of the Framingham Heart Study, height showed the strongest evidence (P = 0.02) of maternal effect. We conducted a genome-wide association analysis for height, testing both the direct effect of the focal individual's genotype and the indirect effect of the maternal genotype. Offspring height showed suggestive evidence of association with maternal genotype for two single-nucleotide polymorphisms in the trafficking protein particle complex 9 gene TRAPPC9 (NIBP), which plays a role in neuronal NF-κB signalling. This work establishes a methodological framework for identifying genetic variants that may influence the contribution of the maternal environment to offspring phenotypes.

Background

Many phenotypes may be influenced by the prenatal environment of the mother and/or maternal care, and these maternal effects may have a heritable component. Much research has focused on the impact of measurable properties of the mother (e.g., adiposity, diabetes, alcohol, or tobacco use) on subsequent phenotypes in their children (e.g., birthweight [1], insulin resistance [2], cognitive function [3]). A more general question is: does the mother's measured genotype influence offspring phenotypes, whether or not the intermediate maternal phenotypes are known or measurable? This 'agnostic' (with respect to maternal phenotype) approach has the potential both to identify novel genetic variants of maternal effect and, via 'reverse epidemiology,' to identify novel maternal phenotypes for such effects.

For the purposes of this study, we accept the strict definition of a genetic maternal effect as the indirect effect of maternal genotype on offspring phenotype [4], as distinct from asymmetric transmission of parental alleles (e.g., mitochondrial inheritance [5]) or asymmetric expression of alleles in the offspring depending on parent of origin (e.g., imprinting [6]). Here we develop mixed variance-components models in the computer program SOLAR [7] to estimate maternal random effects on quantitative phenotypes, and use the best such models as null hypotheses for measured-genotype genome-wide association tests of single-nucleotide polymorphism (SNP) genotypes of individuals and their mothers.

Methods

Data

Data include adult quantitative phenotypes and Affymetrix SNP genotypes provided in the Genetic Analysis Workshop (GAW) 16 Framingham Heart Study (FHS) data release (Problem 2). All authors of this study are 'approved users' of these data per the NHLBI Data Use Certification of April 2008. Analysis of these data was approved by the Institutional Review Board of the University of Texas Health Science Center, San Antonio.

Outlying phenotype measurements (more than four standard deviations from the mean) were removed from the lipid measures (10 for total cholesterol, 9 for high-density lipoprotein (HDL), 43 for triglyceride (TG)) on the assumption that these represented assay errors. The data were normal-quantile-transformed before analysis using the SOLAR "inormal" option to meet the distributional assumptions of the variance components and regression methods. The normal quantile ("inverse normal") transformation is robust to a range of departures from normality and also removed scale effects by standardizing the data. Transformations of this type are convenient for batch processing of multiple phenotypes (e.g., Peng et al. [8]).

Individuals with incomplete genotype data were given imputed genotype scores for the missing markers using the -- infer option in the computer program Merlin [9, 10] Merlin imputes an expected genotype score based on the probability of each possible genotype at a locus given information on marker allele frequency, adjacent markers, and pedigree relationships. We chose not to exclude any SNPs or individuals on the basis of number of incomplete genotypes (unless no genotypes were available at all), given the robustness of imputation from family data [11]. Genotypes were similarly imputed for all SNPs for the non-genotyped implicit mothers of genotyped and phenotyped founders. These maternal genotypes entered the association analysis as properties of their offspring (see "Measured genotype analysis," below); the 'virtual' mothers did not enter the analysis otherwise.

Variance components estimation

We have implemented in SOLAR a general model for incorporating polygenic maternal effects [1214]. Briefly, in the absence of dominance and epistatic effects, the phenotypic covariance between individuals i, j may be decomposed into additive genetic and environmental components in the usual way:
https://static-content.springer.com/image/art%3A10.1186%2F1753-6561-3-S7-S19/MediaObjects/12919_2009_Article_2766_Equ1_HTML.gif
(1)
where I(i, j) is an identity term (1 if i = j, or 0 otherwise), ϕ(i, j) is a coefficient of coancestry, σ z (i, j) is the phenotypic covariance, and σ2 e and σ2 a are, respectively, environmental and additive genetic variances. The additive genetic covariance can be further decomposed to include maternal effects:
https://static-content.springer.com/image/art%3A10.1186%2F1753-6561-3-S7-S19/MediaObjects/12919_2009_Article_2766_Equ2_HTML.gif
(2)
where 2ϕ(i, mo_j) is the coancestry coefficient for i and the mother of j, and 2ϕ(mo_i, mo_j) is the coancestry of the mothers. https://static-content.springer.com/image/art%3A10.1186%2F1753-6561-3-S7-S19/MediaObjects/12919_2009_Article_2766_IEq1_HTML.gif is the additive genetic variance due to maternal effects, and ρa, amis the additive genetic correlation between direct and maternal effects. Decomposition of the environmental component of Eq. 1 is modified from Eq. 14 of Bijma [14]:
https://static-content.springer.com/image/art%3A10.1186%2F1753-6561-3-S7-S19/MediaObjects/12919_2009_Article_2766_Equ3_HTML.gif
(3)

with R = 1 if i = j (equivalent to the identity matrix in Eq. 1), ρ sib [0,1] if i, j are siblings or half-siblings, ρmo [-1,1] if i, j are mother and offspring, or 0 otherwise. Our modification from Bijma [14] was that twins were not treated differently than other siblings because dizygotic twins could not be distinguished in the de-identified FHS data. Our full mixed model also included the fixed effects of relevant covariates and the random effect of mitochondrial inheritance, σ2 mito ; the mitochondrial variance component is structured by a matrix whose elements are 1 if i, j belong to the same matriline or 0 otherwise, as described by Czerwinski et al. [15].

Measured genotype analysis

Measured genotype analysis was conducted for each polymorphic SNP by including its genotype score (the number of copies of the minor allele, range [0,2] with non-integral values for imputed genotypes) as a covariate in the mixed model [16]. Unlike standard association analysis, we included the indirect effect of the mother's genotype in addition to the direct effect of the focal subject's genotype. These effects were tested separately, with an additional test of the mother's genotype conditional on that of the focal subject. The latter test was intended to account for the non-independence of maternal and offspring genotypes: reduction of evidence of maternal association in the conditional test would suggest that the unconditioned maternal effect represented a 'bleed-through' of the direct effect, while an increase in evidence would suggest that the locus affects the trait both directly and indirectly.

Results

Screening for evidence of maternal effects

We tested our maternal random effects model on quantitative phenotypes (height, weight, body mass index (BMI), systolic blood pressure (SBP) and diastolic blood pressure (DBP), fasting total and HDL cholesterol and triglycerides) in individuals measured at exams when they were as similar as possible in mean age (Table 1). Sex, age, age2, and their interactions were included as covariates in all models, and use of antihypertensive medication was a covariate for SBP and DBP. The use of an indicator variable-type covariate for medication has been questioned, especially with regard to BP [17, 18]. In response to a reviewer's concern, we re-ran the BP analyses while correcting for medication by adding 10 mm Hg to BP measures in medicated individuals, as recommended [17]; this did not substantially change our results (data not shown). The impact of alternative corrections for medication may have been greater if we had proceeded to association analysis of the BP traits, because we would then be testing for a difference in means. HDL-C and TG measures were not available for the original cohort and did not give evidence of maternal effect (data not shown); they were not considered further. Results for the remaining phenotypes are given in Table 2.
Table 1

FHS cohorts/examination periods used in this study

FHS Cohort/Exam Period

N

Age in years [mean (SE)]

Original/Exam 4 (1954-1958)

356

40.9 (0.20)

Offspring/Exam 3 (1983-1987)

2,422

46.3 (0.19)

Generation 3/Exam 1 (2002-2005)

3,997

40.2 (0.14)

Table 2

Parameter estimates [mean (SE)] for saturated random-effects model

Trait

σ2a

σ2am

ρa, am

σ2e

ρsib

ρmo

σ2mito

Height

0.64 (0.02)

0.19 (0.07)

-0.21 (0.16)

0.23 (0.08)

0.03 (0.25)

0.05 (0.29)

0.05 (0.12)

Weight

0.68 (0.03)

0.22 (0.08)

-0.38 (0.16)

0.50 (0.04)

0.00a

0.02 (0.12)

0.00 (0.09)

BMI

0.76 (0.04)

0.24 (0.10)

-0.51 (0.14)

0.60 (0.04)

0.00a

-0.02 (0.12)

0.00a

SBP

0.53 (0.04)

0.04 (0.04)

-1.00a

0.74 (0.04)

0.04 (0.03)

0.00 (0.05)

0.00a

DBP

0.48 (0.04)

0.24 (0.09)

-0.33 (0.30)

0.80 (0.03)

0.00a

0.07 (0.05)

0.00 (0.08)

Cholest.

0.58 (0.05)

0.12 (0.32)

-0.11 (0.75)

0.75 (0.06)

0.05 (0.06)

0.06 (0.07)

0.13 (0.09)

aEstimate on boundary; no SE computed.

Table 3 gives the log likelihoods for the minimal polygenic (PG) model (Eq. 1), a PG model with mitochondrial effect, and the saturated model of Table 2. Height showed the strongest evidence of a maternal effect (compared with the PG-mito model, P = 0.02 at 4 degrees of freedom; 4 df is probably over-conservative [19]). Interestingly, this trait initially showed a significant mitochondrial effect compared with the PG model (P = 0.008, 1 df), evidently capturing some of the maternal effects when these were not explicitly modeled.
Table 3

Comparative evidence (log likelihoods), variance-components models

Model

Height

Weight

BMI

SBP

DBP

Cholest.

Polygenic

-112.7

-1850.0

-2788.8

-2460.2

-2724.4

-2615.8

Polygenic + mitochondrial

-109.1

-1850.0

-2788.8

-2460.2

-2724.4

-2613.9

Saturated

-103.3

-1846.9

-2783.5

-2458.4

-2722.7

-2612.8

Measured genotype analysis

We performed measured genotype (MG) tests of association for own genotype (OMG), maternal measured genotype (MMG), and conditional maternal measured genotype (CMMG), for 476,987 autosomal SNPs from the Affymetrix 500k panel. The saturated maternal effects model was used as the null for all analyses. No SNP gave significant evidence of own- or maternal-genotype association with height when corrected for multiple testing using a Bonferroni test (critical test statistic Λ = 28.374 for genome-wide α = 0.05 and 1 df). We did not attempt to account for any linkage disequilibrium among the SNPs in our sample. The SNPs with strongest evidence for OMG, MMG, and CMMG are listed in Table 4.
Table 4

Suggestive associations with height

SNPa

Gene

Chrom.

Coordinate (bp)

Λb, OMG

Λb, MMG

Λb, CMMG

MAFc

N, typedd

Sorted by OMG

rs2213883

MPZL1

1

138,961,675

23.8587

0.8336

0.6778

0.201

6836

rs12129308

MCOLN2

1

83,533,561

20.9235

10.6998

3.0661

0.324

6827

rs536609

MCOLN2

1

83,540,695

20.8421

11.3493

3.4800

0.325

6786

rs597630

MCOLN2

1

83,529,929

20.7087

10.5223

2.9965

0.325

6852

rs600924

MCOLN2

1

83,536,001

20.6884

10.5223

3.0128

0.325

6850

Sorted by MMG

rs11166947

TRAPPC9

8

136,376,532

2.3202

24.9813

22.8014

0.322

6271

rs1426022

 

18

16,635,128

3.7962

22.6984

18.9815

0.468

6675

rs12709669

 

18

16,653,893

3.3445

22.6099

19.2853

0.467

6838

rs756228

TRAPPC9

8

141,138,313

2.5298

21.1932

18.6637

0.31

6841

rs7096364

 

10

107,344,433

1.8847

21.1904

19.3774

0.112

6849

Sorted by CMMG

rs11166947

TRAPPC9

8

136,376,532

2.3202

24.9813

22.8014

0.322

6271

rs7607015

 

2

81,548,367

0.3632

21.1517

22.1813

0.077

6670

rs17198973

 

4

178,164,796

0.3836

16.228

20.2233

0.401

6848

rs11048399

RASSF8

12

25,992,998

0.9614

20.2601

19.5277

0.042

6787

rs1195768

 

10

107,348,297

1.6268

20.201

18.6879

0.118

6553

aLikelihood ratio test statistic (Λ = 2 times the difference in log likelihood of test model and its null)

bAll MG tests have 1 df

cMAF = minor allele frequency.

dN, typed = number of genotyped individuals (before imputation).

Discussion

Several recent studies have undertaken genome-wide association analysis of human height [2022]. These studies have typically examined very large numbers of individuals (~10,000-25,000, with multi-stage designs), larger than the 6,775 individuals in the FHS cohort available for this study. These studies agree in finding numerous loci associated with height, as may be expected for a trait long assumed to be polygenic. Under these circumstances, it is not surprising that we did not replicate specific SNPs or locations identified in these larger studies. It should be noted, however, that we did find suggestive evidence of association (OMG) with a broad genomic region identified by Gudbjartsson et al. [21]: 1q24-25 (Table 4). Our candidate genes in this region is MPZL1 (OMIM #604376), a protein tyrosine phosphatase involved in cell proliferation and differentiation. Our next four highest 'hits' were in the mucolipin2 gene MCOLN2 on 1p22.

Because the published genome-wide association study on height used unrelated individuals, none reported maternal effects. Interestingly, among our strongest maternal associations are repeated hits in two regions: the TRAPPC9 gene on chromosome 8 and an intergenic region on chromosome 18. The trafficking protein particle complex 9 gene TRAPPC9 (NIBP) plays a role in neuronal NF-κB signalling [23] but has not, to our knowledge, been associated with stature in any published study. The existence of repeated (albeit suggestive) associations in this gene makes it a candidate for further investigation of the effects of maternal genotype on height.

Conclusion

We have implemented combined random-effects, measured-genotype fixed effects approach for discovery of genetic variants contributing to the indirect effect of maternal genotype on offspring phenotype. We have identified two regions on chromosomes 2 and 8 - with suggestive association at two SNPs in each region - that may contribute to maternal effects on human height. The tools developed here should be of use for a variety of phenotypes and diseases for which an effect of maternal environment is known or suspected, including height, hypertension, birthweight, and the metabolic syndrome.

List of abbreviations used

BMI: 

Body mass index

CMMG: 

Conditional maternal measured genotype

DBP: 

Diastolic blood pressure

FHS: 

Framingham Heart Study

GAW16: 

Genetic Analysis Workshop 16

HDL: 

High-density lipoprotein

MG: 

Measured genotype

MMG: 

Maternal measured genotype

OMG: 

Own measured genotype

PG: 

Polygenic SBP: Systolic blood pressure

SNP: 

Single-nucleotide polymorphism

TG: 

Triglyceride.

Declarations

Acknowledgements

The Genetic Analysis Workshops are supported by NIH grant R01 GM031575 from the National Institute of General Medical Sciences. We thank the NHLBI and the Framingham Heart Study participants and investigators for making these data available. Our work in this study was supported by NIH grant R01 MH59490 (John Blangero, PI) and was performed in computing facilities funded in part by the AT&T Foundation.

This article has been published as part of BMC Proceedings Volume 3 Supplement 7, 2009: Genetic Analysis Workshop 16. The full contents of the supplement are available online at http://​www.​biomedcentral.​com/​1753-6561/​3?​issue=​S7.

Authors’ Affiliations

(1)
Department of Genetics, Southwest Foundation for Biomedical Research

References

  1. Kazumi T, Kawaguchi A, Yoshino G: Associations of middle-aged mother's but not father's body mass index with 18-year-old son's waist circumferences, birth weight, and serum hepatic enzyme levels. Metabolism. 2005, 54: 466-470. 10.1016/j.metabol.2004.10.014.View ArticlePubMed
  2. Gonzalez-Ortiz M, Martinez-Abundis E: Maternal effect of type 2 diabetes mellitus on insulin sensitivity and metabolic profile in healthy young Mexicans. Diabetes Nutr Metab. 1999, 12: 32-36.PubMed
  3. Julvez J, Ribas-Fitó N, Torrent M, Forns M, Garcia-Esteban R, Sunyer J: Maternal smoking habits and cognitive development of children at age 4 years in a population-based birth cohort. Int J Epidemiol. 2007, 36: 825-832. 10.1093/ije/dym107.View ArticlePubMed
  4. Hager R, Cheverud JM, Wolf JB: Maternal effects as the cause of parent-of-origin effects that mimic genomic imprinting. Genetics. 2008, 178: 1755-1762. 10.1534/genetics.107.080697.PubMed CentralView ArticlePubMed
  5. Yang Q, Kim SK, Sun F, Cui J, Larson MG, Vasan RS, Levy D, Schwartz F: Maternal influence on blood pressure suggests involvement of mitochondrial DNA in the pathogenesis of hypertension: the Framingham Heart Study. J Hypertens. 2007, 25: 2067-2073. 10.1097/HJH.0b013e328285a36e.View ArticlePubMed
  6. Xiong DH, Wang JT, Wang W, Guo YF, Xiao P, Shen H, Jiang H, Chen Y, Deng H, Drees B, Recker RR, Deng HW: Genetic determinants of osteoporosis: lessons learned from a large genome-wide linkage study. Hum Biol. 2007, 79: 593-608.PubMed
  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 ArticlePubMed
  8. Peng B, Yu RK, DeHoff KL, Amos CI: Normalizing a large number of traits using empirical normal quantile transformation. BMC Proc. 2007, 1 (Suppl 1): S156-10.1186/1753-6561-1-s1-s156.PubMed CentralView ArticlePubMed
  9. Abecasis G, Cherny SS, Cookson WO, Cardon LR: Merlin - rapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet. 2002, 30: 97-101. 10.1038/ng786.View ArticlePubMed
  10. Burdick JT, Chen W-M, Abecasis GR, Cheung VG: In silico method for inferring genotypes in pedigrees. Nat Genet. 2006, 38: 1002-1004. 10.1038/ng1863.PubMed CentralView ArticlePubMed
  11. Chen W-M, Abecasis GR: Family-based association tests for genomewide association scans. Am J Hum Genet. 2007, 81: 913-926. 10.1086/521580.PubMed CentralView ArticlePubMed
  12. Willham RL: The covariance between relatives for characters composed of components contributed by related individuals. Biometrics. 1963, 19: 18-27. 10.2307/2527570.View Article
  13. Lynch M, Walsh B: Maternal effects. Genetics and Analysis of Quantitative Traits. 1998, Sunderland, MA; Sinauer Associates, Inc
  14. Bijma P: Estimating maternal genetic effects in livestock. J Anim Sci. 2006, 84: 800-806.PubMed
  15. 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.PubMed
  16. Blangero J, Göring HHH, Kent JW Jr, Williams JT, Peterson CP, Almasy L, Dyer TD: Quantitative trait nucleotide analysis using Bayesian model selection. Hum Biol. 2005, 77: 541-559. 10.1353/hub.2006.0003.View ArticlePubMed
  17. Cui JS, Hopper JL, Harrap SB: Antihypertensive treatments obscure familial contributions to blood pressure variation. Hypertension. 2003, 41: 207-210. 10.1161/01.HYP.0000044938.94050.E3.View ArticlePubMed
  18. Tobin MD, Sheehan NA, Scurrah KJ, Burton PR: Adjusting for treatment effects in studies of quantitative traits: antihypertensive therapy and systolic blood pressure. Statist Med. 2005, 24: 2911-2935. 10.1002/sim.2165.View Article
  19. Amos C, de Andrade M, Zhu D: Comparison of multivariate tests for genetic linkage. Hum Hered. 2001, 51: 133-144. 10.1159/000053334.View ArticlePubMed
  20. Lettre G, Jackson AU, Gieger C, Schumacher FR, Berndt SI, Sanna S, Eyheramendy S, Voight BF, Butler JL, Guiducci C, Illig T, Hackett R, Heid IM, Jacobs KB, Lyssenko V, Uda M, Diabetes Genetics Initiative; FUSION; KORA; Prostate, Lung Colorectal and Ovarian Cancer Screening Trial; Nurses' Health Study; SardiNIA, Boehnke M, Chanock SJ, Groop LC, Hu FB, Isomaa B, Kraft P, Peltonen L, Salomaa V, Schlessinger D, Hunter DJ, Hayes RB, Abecasis GR, Wichmann HE, Mohlke KL, Hirschhorn JN: Identification of ten loci associated with height highlights new biological pathways in human growth. Nat Genet. 2008, 40: 584-591. 10.1038/ng.125.PubMed CentralView ArticlePubMed
  21. Gudbjartsson DF, Walters GB, Thorleifsson G, Stefansson H, Halldorsson BV, Zusmanovich P, Sulem P, Thorlacius S, Gylfason A, Steinberg S, Helgadottir A, Ingason A, Steinthorsdottir V, Olafsdottir EJ, Olafsdottir GH, Jonsson T, Borch-Johnsen K, Hansen T, Andersen G, Jorgensen T, Pedersen O, Aben KK, Witjes JA, Swinkels DW, den Heijer M, Franke B, Verbeek AL, Becker DM, Yanek LR, Becker LC, Tryggvadottir L, Rafnar T, Gulcher J, Kiemeney LA, Kong A, Thorsteinsdottir U, Stefansson K: Many sequence variants affecting diversity of adult human height. Nat Genet. 2008, 40: 609-615. 10.1038/ng.122.View ArticlePubMed
  22. Weedon MN, Lango H, Lindgren CM, Wallace C, Evans DM, Mangino M, Freathy RM, Perry JR, Stevens S, Hall AS, Samani NJ, Shields B, Prokopenko I, Farrall M, Dominiczak A, Diabetes Genetics Initiative; Wellcome Trust Case Control Consortium, Johnson T, Bergmann S, Beckmann JS, Vollenweider P, Waterworth DM, Mooser V, Palmer CN, Morris AD, Ouwehand WH, Cambridge GEM Consortium, Zhao JH, Li S, Loos RJ, Barroso I, Deloukas P, Sandhu MS, Wheeler E, Soranzo N, Inouye M, Wareham NJ, Caulfield M, Munroe PB, Hattersley AT, McCarthy MI, Frayling TM: Genome-wide association analysis identifies 20 loci that influence adult height. Nat Genet. 2008, 40: 575-583. 10.1038/ng.121.PubMed CentralView ArticlePubMed
  23. Hu W-H, Pendergast JS, Mo X-M, Brambilla R, Bracchi-Ricard V, Li F, Walters WM, Blits B, He L, Schaal S, Bethea JR: NIBP, a novel NIK and IKK-beta-binding protein that enhances NF-kappa-B activation. J Biol Chem. 2005, 280: 29233-29241. 10.1074/jbc.M501670200.PubMed CentralView ArticlePubMed

Copyright

© Kent et al; licensee BioMed Central Ltd. 2009

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