Transcript CCMB colloquium 10-5-16 Samoan obesity gene

Identification of a Novel Missense Variant in
CREBRF Associated with Obesity and Diabetes in
Samoans
Stephen T McGarvey, PhD, MPH
Department of Epidemiology and International Health Institute
School of Public Health & Department of Anthropology
Brown University
Center for Computational Molecular Biology, Brown University
October 5, 2016
Supported by NIH Grants AG09375, HL52611, DK55406,
DK59642, DK075371 & HL093093
Map showing the direction of Austronesian expansion from Taiwan and likely timing of
expansion into the Pacific.
Matisoo-Smith PNAS 2015;112:13432-13433
©2015 by National Academy of Sciences
Samoa, N=~180,000
American Samoa
N=~56,000
$8,000 per capita income (2007)
58% of families incomes
below the U. S. poverty level
$4,250 per capita income (2012)
46% of working age men primarily in
subsistence related farming and fishing.
Prevalence Samoan Adult Adiposity 1975-2010: Males
Based on Polynesian Standards normal <26, overweight 26-32, obese > 32 kg/m2
70
Normal
60
Overwt
Obese
50
40
30
20
10
0
Am
Samoa
1976-78
Am
Samoa
1990
Am
Samoa
2002
Samoa
1979-82
Samoa
1991
Samoa
2003
Samoa
2010
Prevalence Samoan Adult Adiposity 1975-2010: Females
Based on Polynesian Standards normal <26, overweight 26-32, obese > 32 kg/m2
80
Normal
Overwt
Obese
70
60
50
40
30
20
10
0
Am
Samoa
1976-78
Am
Samoa
1990
Am
Samoa
2002
Samoa
1979-82
Samoa
1991
Samoa
2003
Samoa
2010
Lin et al Diab Med 2016; DOI: 10.1111/dme.13197 .
Why do Polynesians have high levels of adiposity?
• Environmental Perspective
• Energy balance changes associated with rapid changes in way of life =
modernization/development
• Positive energy balance – excess calories in diet & too few in physical activity
• Evolutionary biology perspective
• Polynesian demographic and ecological history – discovery & settlement of far
flung islands, fragility of island adaptations, small & fluctuating population size
• Putative advantage of “thrifty genes” for efficient metabolism to store energy
connected to population history of food shortages
• Gene by environment interaction perspective
• Population susceptibility not deterministic
• Former adaptive traits may increase risk of obesity & cardiometabolic
conditions in contemporary obesogenic environment
• Key roles of G x E interactions
• Epigenetic processes on gene expression at critical life periods
Diet
and
BMI
in
Samoa
19612010
Seiden, N Hawley, D
Schulz, S Raifman, ST
McGarvey. Am Jl
Human Biology, 24:
286-295, 2012
Genome Wide Association Studies
233,000 Participants,97 Loci, ~ 2.7% variance in BMI
Locke et al. Nat. Genet. 2015; Speliotes et al. Nat. Genet. 2010
Samoa GWAS 2010 Sampling
1,837 women & 1,235 men – all 4 grandparents
are Samoan
Hawley et al. Am. J. Hum. Biol. 2014
Discovery Sample Phenotypes
BMI
BMI & Adiposity in Discovery Sample
BMI, Adiposity & Body
Composition in Polynesians
Overweight - BMI ≥ 27
Obese - BMI ≥ 32 Severely
Obese - BMI ≥ 40
Genotypes
• Affymetrix 6.0
• 909,622 markers
⬇
• 659,492 markers
Statistical Model
BMI = age + age² + sex + age × sex + genotype
also adjusted for kinship
GWAS for BMI in Samoans
Identified 5q35.1, SNP rs12513649
“LocusZoom”
rs12513649
rs12513649 Allele Frequencies
• Samoa
0.258
•
•
•
•
•
0.063
0.059
0.003
0.000
0.001
East Asians
Admixed Americans
South Asians
Europeans
Africans
1000 Genomes
Replication Samples
• 1990 American Samoa & 1991 Samoa
- 541 women | 479 men
• 2002 American Samoa & 2003 Samoa
– 584 women | 499 men
– 220 girls | 189 boys
rs12513649 / BMI Association
• Discovery: 5.3 × 10−14
• Replication: 1.2 × 10−9
• Joint: 4.0 × 10−22
Initial Imputation
• Before targeted sequencing, imputed in our region of
interest from 1000 Genomes Project reference panel.
• Only one associated variant rs141606089 (predicted
frequency of 0.075), but when genotyped this variant in a
Samoan pilot sample was monomorphic.
Targeted sequencing, Imputation & Association
Interpretation of targeted sequencing, imputation &
association analyses
•
Sequence highly informative N=96 => impute genotypes into remaining 2,976
•
Analyses of imputed data found two significantly associated variants in
CREBRF (encoding CREB3 regulatory factor), rs150207780 & rs373863828
•
Because of high LD in the region, conditional analyses were not able to
distinguish between these two top variants on statistical grounds.
•
Annotation - neither rs12513649, between ATP6V0E1 and CREBRF, nor
rs150207780, in intron 1 of CREBRF, had any predicted regulatory function
•
•
•
Bayesian Fine Mapping
rs373863828
PP =
rs150207780
•
Strong support to follow-up rs373863828
•
Performed genotyping of rs373863828 in all study participants
0.80
0.22
ENCODE
0.92
0.34
rs373863828 in CREBRF
BMI Association (all genotyped)
• Discovery: 7.0 × 10−13
• Replication: 3.5 × 10−9
• Joint: 1.4 × 10−20
Effect Size
• Discovery: 1.36 kg/m2 per copy
• Replication: 1.45 kg/m2 per copy
Percent of BMI Variation Explained
• Discovery: 1.93% of variance
• Replication: 1.08% of variance
rs373863828
in CREBRF
(CAMP [Cyclic adenosine monophosphate] Responsive Element
Binding Protein 3 [CREB3] Regulatory Factor =>CREBRF)
• CREBRF:c.1370G>A => A is missense mutation
• CREBRF:p.Arg457Gln => glutamine produced not arginine
• CAMP Responsive Element Binding Protein 3 is a
Protein Coding gene
• Highly Conserved: GERP RS (Genomic Evolutionary Rate
Profiling) Score: 5.49
• Predicted Deleterious – based on amino acid change on
protein function, using two of several algorithms
SIFT (Sorting Intolerant From Tolerant): 0.03
PolyPhen2 (Polymorphism Phenotyping) : 0.996
rs373863828 Allele Frequencies
Samoans (all samples)
0.259
Exome Consortium
0.0000412
• (5 observed alleles of 121,362 chromosomes)
1000 Genomes
• East Asians
• Admixed Americans
• South Asians
• Europeans
• Africans
0.000
0.000
0.000
0.000
0.000
Phenotype Wide - Discovery
•
•
•
•
•
BMI
% Body Fat
Waist Circ.
Hip Circ.
WHR
• Obesity
⬆ 1.36 kg/m2
⬆ 2.20 %
⬆ 2.84 cm
⬆ 2.36 cm
—
1 × 10−13
2 × 10−10
2 × 10−12
1 × 10−12
2 × 10−3
⬆ 1.305 OR
1 × 10−5
Phenotype Wide – Replication
•
•
•
•
•
BMI
% Body Fat
Waist Circ.
Hip Circ.
WHR
• Obesity
⬆ 1.45 kg/m2
⬆ 1.34 %
⬆ 3.22 cm
⬆ 2.72 cm
—
8 × 10−10
7 × 10−4
5 × 10−10
4 × 10−9
0.017
⬆ 1.441 OR
8 × 10−6
Phenotype Wide - Discovery –
metabolic traits
•
•
•
•
•
Glucose
Insulin
HOMA-IR
Adiponectin
Leptin
• Diabetes
⬇ 2.25 mg/dL
—
—
—
—
7 × 10−8
0.592
0.754
0.412
0.159 / 0.213
⬇ 0.586 OR
7 × 10−9
Phenotype Wide Replication –
metabolic traits
•
•
•
•
•
Glucose
Insulin
HOMA-IR
Adiponectin
Leptin
• Diabetes
⬇ 2.09 mg/dL
—
—
—
n/a
8 × 10−6
0.004
0.015
0.068
n/a
⬇ 0.742 OR
0.029 n.s.
Weight, Obesity & Diabetes by
genotype of rs373863828
Genotype Weight
GG
-
Obesity
52.1%
Diabetes
19.7%
GA
~4kg⬆
58.0%
14.2%
AA
~8kg⬆
65.2%
9.5%
Comparison of Effect Sizes CREBRF p.Arg457Gln vs FTO - Loos 2016. Nature Genetics 48, 976-78.
What are the biological mechanisms?
Approach – focus on altered cellular metabolism
Studies in mice & fruit flies (CREBRF ortholog
REPTOR) suggest role in
•
•
•
•
Adipogenic differentiation
Fat accumulation
Bioenergetic profile
Starvation resistance
• Use 3T3-L1 preadipocyte model & eGFP
(Green Fluorescent Protein model) - explore
overexpression of wild type and variant
CREBRF
• Explore nutrient starvation and its inhibition
CREBRF overexpression induces adipogenic
differentiation in the absence of hormonal
treatment
One-way analysis of variance (ANOVA), two-sided Games–Howell post-hoc
test ***P < 1 x 10−4 compared to control, #P < 0.05 compared to eGFP.
CREBRF increases fat accumulation and
the variant provides a further increase
One-way analysis of variance (ANOVA), two-sided Games–Howell post-hoc test
*P < 1 x 10−3, *P < 1 x 0.03 compared to control, #P < 0.05 compared to eGFP.
CREBRF increases fat accumulation and
the variant provides a further increase
WT CREBRF increases both mitochondrial
function and glycolysis, but the variant
decreases both
One-way analysis of
variance (ANOVA), twosided Games–Howell posthoc test *P < 1 x 0.03
compared to control, #P <
0.05 compared to eGFP.
Crebrf is rapidly induced by starvation
and by TORC1 inhibition
One-way ANOVA and two-sided Bonferroni post-hoc tests. **P = 0.002, ***P < 1 x 10−11
compared to cells at 0 h; #P = 0.02, ###P = 8.8 x 10−13 compared to cells at 24 h (refed)
WT and variant CREBRF equally protect
against starvation
One-way ANOVA and two-sided Games–Howell post-hoc tests. ***P < 5 x 10−5
compared to cells transfected with control eGFP construct
Summary
Condition
WT CREBRF
Variant CREBRF
Adipogenic
differentiation
+++
++
Fat accumulation
++
+++
Mitochondrial
function
+++
--
Glycolysis
+++
--
Starvation
resistance
+++
+++
SCENARIOS FOR NATURAL SELECTION?
Nutritional Stressors during Voyages?
Notes
Island Ecosystems Nutritional Stressors?
Population Cycles
Human Carrying Capacity
Tropical Storms
Political Conflict
Infectious Diseases
Selective Sweep Approaches using 2010 Samoan GWAS SNP data
•
Under natural selection, a new beneficial mutation will rise in
frequency in the population.
•
Above schematic shows polymorphisms along a chromosome,
including the selected allele, before and after selection.
Ancestral alleles in grey and derived (non-ancestral) alleles are
shown in blue.
•
As a new positively-selected allele (red) rises to high
frequency, nearby linked alleles on the chromosome ‘hitchhike’
along with it to high frequency, creating a ‘selective sweep.’
•
Haplotype blocks around selected allele- set of closely linked
alleles/markers on a chromosome that, over evolutionary time,
tend to be inherited together.
Signatures of
Natural
Selection
• (Left) The core haplotype carrying the derived BMI-increasing allele has
elevated extended haplotype homozygosity (EHH) compared to haplotypes
carrying ancestral allele.
• (Right) Haplotypes carrying the derived allele extend longer than haplotypes
carrying the ancestral allele.
Genomic Evidence for Natural Selection
Left graph of wild type or ancestral allele shows much branching away from center due
to recombination over time and meiotic events
Right graph shows derived allele/missense mutation has far less branching indicating
less variation around it, likely due to selective sweep
Conclusions
•
Association observed and replicated of rs373863828 with BMI and
other adiposity and metabolic phenotypes in Samoans.
•
Evidence of inverse association with glucose and type 2 diabetes
•
Found only in Polynesians – so far
•
Evidence of positive selection for the BMI-increasing allele
•
p.Arg457Gln CREBRF decreases cellular energy utilization while
promoting lipid accumulation
•
CREBRF is a starvation response gene, conferring protection
against cell death
Future Directions - funded
Competing Renewal NIH Grant - 2016-2020
Aim 1 - Identify, characterize and validate the gene networks mediating
the effects of CREBRF and the CREBRFR457Q variant on energy
homeostasis and energy substrate metabolism in
• cultured cell models
• metabolically-relevant tissues from mice
• subcutaneous adipose tissue from 120 Samoans
Aim 2 - Identify expression quantitative trait loci (eQTLs) associated with
the CREBRF network.
- Impact of CREBRFR457Q variant on whole body and tissue-specific energy
homeostasis and energy substrate metabolism using CREBRFR457Q
knock-in mice.
Future Directions - funded
Aim 3 - Impact of CREBRFR457Q variant on metabolic and behavioral traits
affecting energy homeostasis in 500 Samoan GWAS participants selected
based on genotype with new fieldwork.
•
Oral Glucose Tolerance Tests (OGTT)
•
DXA scans for body composition - adipose tissue in risky or less risky places?
•
Hunger, satiety and food orientation interviews and experiments
Aim 4 - Characterize the interrelations of the CREBRFR457Q variant using
comprehensive statistical approaches including
• multivariate analyses for pleiotropy (one gene ->several phenotypes)
• pathway analyses (try to construct biological mechanisms)
• testing for G x E interactions with focus on diet and physical activity
• testing for G x G interactions focusing on genes identified in Aims 1 and 2
• testing for association with newly gathered phenotypes from Aim 3.
• testing for selective signatures
Future Directions - planned
Collaborate with other Groups to identify the CREBRFR457Q variant in other Pacific Island groups and look
for associations with cardiometabolic phenotypes
Submitted NIH Grants
Energy Metabolism and Balance
• Study longitudinally whole body energy metabolism of Samoans by genotype with emphasis on energy
expenditure , incl. total daily EE by DLW, resting metabolic rate, thermic effect of food, substrate
utilization by indirect calorimetry & changes in body composition and obesity.
Whole Genome Sequencing
• Impute from 1,300 Samoans with WGS to the other 3,700 individuals and identify new associations
with cardiometabolic phenotypes and G x E and G x G influences.
Planned NIH Grant submissions
Metabolomics
• Determine biological pathways interacting with the CREBRF variant on metabolic and behavioral traits
affecting energy homeostasis.
Epigenome wide association study
• Explore DNA methylation effects and gene expression pathways focused on the CREBRF variant and
other variants to be identified with focus on cardiometabolic traits.
Ethical, legal and social implications of the genetic research in Samoa
• Explore understanding and meaning of the genetic findings; return of results (RoR); incidental findings
(IF); and wider data use - using mixed methods
Acknowledgements
Pittsburgh
• Daniel Weeks
• Ryan Minster
• Zsolt Urban
• Erin Kershaw
• Chi-Ting Su
• Olive Buhule
• Jerome Lin
Sāmoa
• Muagututi‘a Sefuiva Reupena
• Satupa‘itea Viali
• Take Naseri
Cincinnati
• Ranjan Deka
• Guangyun Sun
• Hong Chen
•
People of the two Sāmoas
•
Samoan Public Health, Medical and Political Leaders
•
NIH Grants: R01-HL093093 (PI: S.T.M.), R01-AG09375 (PI:
S.T.M.), R01-DK59642 (PI: S.T.M.), R01-HL52611 (PI: I.
Kamboh), P30 ES006096 (PI: S.M. Ho), R01-DK55406. (PI:
R.D.), R01-HL090648 (PI: Z.U.), and R01-DK090166 (PI:
E.E.K),
Brown University student research funds.
American Sāmoa
• John Tuitele
Yale
• Nicola Hawley
•