Evolutionary selection underlying the genetic architecture of - - PowerPoint PPT Presentation

Evolutionary selection underlying the genetic architecture of complex traits

Carolina Medina-Gomez Oscar Lao SNPs and Diseases Molecular School of Medicine Thursday, November 15th, 2018

Ethnic differences in bone mineral density and fracture risk have been described

• Finkelstein et al. 2002. JCEM. African American, Chinese, Japanese

premenopausal women (SWAN, N~2,000). African Americans and Asians (after adjustment) have higher BMD than Caucasians

• Marshall et al. 2008 JBMR. Black, Asian, Hispanic, Caucasians. Men
• ver 65 years. (MrOS, N~3,300). Self reported ethnicity. Greater

cortical thickness and trabecular vBMD in Blacks and Asians.

• Kalkwarf et al. 2012. JBMR. Children 1-36 months. (CHOPs, N~307).

No differences in LS BMD between Blacks and Whites. Blacks higher BMD at age >5 years.

1. Are the reported ethnic differences in BMD variation already present at early ages?

• 2. If so, can we attribute these differences to genetic factors?

Research questions (to start)

Generation R Study is a prospective multiethnic birth cohort

• f children followed to early adulthood
• Ongoing population-based

longitudinal study including 9,778 mothers followed since pregnancy (04/2002-01/2006)

• Parents coming from over 100

different countries

• N~6,500. DXA measurements

average age of 6 years

• N~5,733 with GWAS
• N~4,000 with GWAS & DXA

Rotterdam 2010 Generation R sample

Advantages of the Generation R multiethnic design

Possible advantages of the Generation R setting:

• Restricted geographical area
• > Similar light exposition
• < variation in diet and physical activity
• = Health Care quality

• The classification is based on the mother’s country of birth. If she is

also born in the Netherlands, the background is determined by the father’s country of birth.

Working Ethnicity Definition from questionnaires based on the Dutch Central Bureau of Statistics

Country of birth might not be the best surrogate for ancestry in genetic studies

Using genetic data to assess ancestry is a well-known technique

Using genetic data (instead of questionnaire ) to assess ancestry

GENERATION R GWAS POPULATION

Generation R ethnic groups show heterogeneous / scattered clustering when using genetic data

In this Graph the pink corresponds to the Yoruba panel in the Hapmap while dark and light blue correspond to JPN and NE panels respectively. Dutch Dutch Antilles Surinamense Indonesian African Capo Verdians American Western American non Western Morrocan Asians non Western Asians Western Turkish Europeans Oceanic

Applying stringent criteria for ancestry definition is not possible ( ±4SD in PC1, PC2 HapMap cluster)

Africans 31 Asians 36 Europeans 2921 Mix 2745

Genetic ancestry clustering algorithms should help

• vercoming biases generated by ethnicity definitions

Admixture analysis: ancestry proportions and population allele frequencies

Caucasian African Asian Mix

23&me most popular feature: Ancestry Analysis

Using admixture one can define 4 main clusters for a more powerful setting for analysis

GENERATION R GWAS POPULATION

Differences in BMD according to ethnic background based

• n questionnaire data

6,126 children DXA and ethnicity information (15 ethnicities)

Africans Africans Suri_Creole Antillans Europeans Europeans Oceanic Dutch Americans Turkish NorthAfricans* Asians Asians Suri_Hindu

African Caucasian Asian

EthnicGroup LSmean Difference Pval Asian 0.559 0.022 <2E-16 Caucasian 0.552 0.016 <2E-16 African 0.575

• After adjustment for age, gender, fat

mass, lean mass and height

*Additional cofounders: Maternal smoking, Maternal homocysteine level, Maternal marital status, Maternal weight, Maternal height, Infant birth weight, Child Breastfeeding, Child protein intake, Child sports. Diet questionnaires?

Differences in BMD according to ethnic background based

• n genetic data

Within the Caucasians the BMD increases as the % of African ancestry increases

SNP A1 A2 Ind #1 SCORE Ind #2 SCORE

rs17482952 G A GG 2 GG 2 rs12407028 C T TT CC 2 rs7521902 A C CA 1 CA 1 rs1346004 A G GG GA 1 rs6426749 C G CG 1 CC 2 rs479336 G T GG 2 GG 2 rs4233949 C A AC 1 AC 1

7 11

• 1. Score
• 2. Score

61/63 SNPs in GenR data

The allele score is positively associated with BMD in children of the Generation R cohort

• Score explains ~5% of phenotypic variance. 2.4% when corrected by PCs

By mendelian randomization similar distribution of score- bins for a particular covariate is expected accross bins

Distribution of gender across the score bins

• Male
• Female

BMD-increasing alleles were more likely to have higher frequencies in African than in Caucasian participants

Estrada et al. 2012. Genome-wide meta-analysis identifies 56 bone mineral density loci and reveals 14 loci associated with risk of fracture. Nature Genetics. 44, 491–501

Increasing Alleles Caucasian: 43 to 78 (60) Asian: 48 to 74 (59) African 50 to 78 (66) Quintiles BMD Highest (88) 0.72 SDs Lowest (53) 0.53 SDs

The 61% of the African children are in the 2 highest quintiles (p < 1x10-16)

BMD-increasing alleles were more likely to have higher frequencies in African than in Caucasian participants

Estrada et al. 2012. Genome-wide meta-analysis identifies 56 bone mineral density loci and reveals 14 loci associated with risk of fracture. Nature Genetics. 44, 491–501

Two pediatric studies confirmed that genetic variants associated with higher BMD are more common in Africans

Higher BMD genetic variants Higher BMD genetic variants

Behavior of the BMD-increasing alleles support no stratification influencing the results

• 30/61 is Minor Allele (based in the CEU freq). For 23 of them the difference is

even higher than 10% favoring the African populations.

The bar plots in the background represent the frequency of BMD-increasing alleles in HapMap CEU panel. Dots represent the difference in frequency between S. African and European

The Human Genome Diversity Panel as a replication source of our findings

1,063 cultured lymphoblastoid cell lines (LCLs) from 1,050 individuals in 52 world populations Genotype 650,000 SNPs

Carolina Medina-Gómez et al, 2015 (MBE)

CEPH-HGDP panel

939 samples 51 human populations of global distribution

Li et al 2008 (Science)

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Genetic differentiation

Significant Differences in proportion of BMD increasing alleles is seen only in populations of African origin

Conclusions (1)

• Genetic variants influencing BMD variation in NE-adults are predictive

in children from different ethnicities.

• Children of African descent have a higher frequency of BMD-increasing

alleles than children of Asian and Caucasian descent, and these results cannot be attributed to stratification.

• Analysis throughout worldwide populations show a similar spatial

distribution of the associated variants as observed in the Generation R cohort.

• Can BMD distributions today be explained by mechanisms of polygenic

evolution (subtle allele frequencies shifts at many loci)?

Nielsen et al 2017

HUMAN EVOLUTION

Adapted from Vattathil & Akey, 2015. Cell

HUMAN EVOLUTION

Sankararaman et al 2 Simonti et al 2016

CONSEQUENCES OF HUMAN BIOLOGICAL EVOLUTION (1)

HUMAN CULTURAL EVOLUTION

https://es.pinterest.com/

Fan et al 2016

CONSEQUENCES OF HUMAN BIOLOGICAL EVOLUTION (2)

CONSEQUENCES OF HUMAN BIOLOGICAL EVOLUTION (3) DISUSE

Chirchir et al 2014 PNAS

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Ruff et al 2015 PNAS

73% BMD increasing alleles actually constitute ancestral alleles

Population Mean MAF P value compared to

• ther GWAS

SNPs CEU 0.3 8.3*10-3 CHB/JPT 0.26 0.13 YRI 0.21 0.83

Carolina Medina-Gómez et al, 2015 (MBE)

EVOLUTION BMD IS A CASE OF DISUSE?

Hypothesis one High BMD is the ancestral state Polygenic selection out of Africa towards lower BMD Hypothesis two Relaxation of selection in Africa so BMD alleles fluctuate at random compared to non-African populations Hypothesis three BMD decreasing alleles are introgressed from a non-Homo sapiens species

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Mathieson et al 2015 Nature Pvalue 0.0006, R2 0.109

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Trend in ancient samples

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Archaic samples

Population/species Mean BMD-GS Neanderthal 1.344 Denisova 1.345 Pan troglodytes 1.356 SubAfrica 1.131 NorthAfrica 1.027 MiddleEast 0.972 SouthAsia 0.987 Europe 0.984 EastAsia 1.029 Oceania 0.886 NativeAmerican 1.077

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Trend in ancient samples

Fu et al 2016 Nature R2 = 0.31; Pvalue = 0.023

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

GS/Neanderthal trend

Sankararaman et al 2014 Na

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Archaic introgression

Excluding rs10416218, for which Neanderthal shows the decreasing BMD allele

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Archaic introgression

Hypothesis one High BMD is the ancestral state Polygenic selection out of Africa towards lower BMD Hypothesis two Relaxation of selection in Africa so BMD alleles fluctuate at random compared to non-African populations Hypothesis three BMD decreasing alleles are introgressed from a non-Homo sapiens species

Systematic decrease of BMD increasing alleles in EUROPE with time BMD alleles from Neanderthals systematically depleted except for one SNP

CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Green et al. 2010 Reich et al. 2010

BMD BMD BMD BMD BMD CONSEQUENCES OF CULTURAL EVOLUTION BMD IS A CASE OF DISUSE?

Thank you very much!

• scar.lao@cnag.crg.eu

Acknowledgements

• Dr. Oscar Lao

CNAG Barcelona

Denise Heppe Claudia Kruithof

• Dr. Vincent Jaddoe