Transcript HLA-C, rs9264942

Determinants of host response to HIV-1:
the role of rare and common variants
Host Genetics portfolio
Genetics of vaccine trials
Genetics of
viral control
Genetics of
resistance
Exposure
Infection
Phenotype
Telenti A & Goldstein DB, Nat Rev Microbiol 2006
the EuroCHAVI consortium
Danish Cohort
Denmark
N. Obel
Guy Kings St.Thomas Hospital
United Kingdom
P. Easterbrook
Clinics Hospital
Barcelona, Spain
J.M. Gatell
Swiss HIV Cohort
University Hospital, Lausanne
Switzerland (coordinating center)
A. Telenti
P. Francioli
IrsiCaixa
Barcelona, Spain
B. Clotet
San Raffaele Hospital
Milan, Italy
A. Castagna
Royal Perth Hospital
Perth, Australia
S. Mallal
Modena Cohort
Modena, Italy
A. Cossarizza
I.CO.NA Cohort
Rome, Italy
A. De Luca
WGAViewer: gene context annotation (HLA-C, HLA-B,
HCP5)
HLA-C,
rs9264942
HLA-B*5701/HCP5,
rs2395029
Showing all SNPs
genotyped in this region
sorted by p-value or
functionality
http://www.genome.duke.edu/centers/pg2/downloads/wgaviewer.php
WGAViewer: SNP annotation (HLA-C, rs9264942)
Showing all HapMap
SNPs not genotyped in
this region sorted by r2 or
functionality
http://www.genome.duke.edu/centers/pg2/downloads/wgaviewer.php
CHAVI set point study: global results
Gene & SNP
P-value for association P-value for association
with HIV-1 viral load
with protection against
at setpoint
progression (CD4 <350)
N=2362
N=1071
HCP5 / HLA-B*5701
rs2395029
4.5E-35
1.2E-11
HLA-C
rs9264942
5.9E-32
7.4E-12
ZNRD1 / RNF39
rs9261174
1.1E-04
3.8E-08
CCR5 Δ32 het
rs333
1.7E-10
2.6E-06
Bonferroni threshold for genome-wide significance: 5E-08
Independence of the HCP5 and HLA-C
association signals
The HCP5 and HLA-C variants are in partial LD
(r2=0.06, D’=0.86)
the combined strength of their associations is less
than the sum of the signals measured separately
nonetheless, a nested regression model clearly
demonstrates that each of these variants is
independently genome-wide significant:
• rs2395029: p=1.8E-23
• rs9264942: p=2.4E-20
Independence of the ZNRD1
association signal
 The variants in the ZNRD1 region, 1Mb away from
HLA-B and HLA-C, are not in LD with the top 2 SNPs
 The strength of their association signal is the same in
models including the HCP5/HLA-C SNPs
 The identified association signal is likely to be
synthetic (high LD in a 150kb region that includes 12
genes or pseudogenes, notably HLA-A)
Independent replications of associations
2008;3(12):e3907. Epub 2008 Dec 24.
2008 Dec 30. [Epub ahead of print]
√ HCP5/B*5701
√ HLA-C: rs9264942 was
not genotyped, but is in LD
with the top hit,
rs10484554, which also
associates with HLA-C
expression
√ HCP5/B*5701
√ ZNRD1
2008;3(11):e3636. Epub 2008 Nov 4.
√ HCP5/B*5701
√ HLA-C
√ ZNRD1: in haplotypes that
contain HLA-A10
2009 Jan 2;23(1):19-28
√ HCP5/B*5701
√ HLA-C
Independent replications of associations
Not yet published:
Mary Carrington’s lab
Rasmi Thomas et al., in revision
√ HLA-C, including protein
expression
International HIV Controllers Study
Paul de Bakker, manuscript in preparation
√ HCP5/B*5701
√ HLA-C
√ ZNRD1
Nef counteracts HLA-C mediated immune
control of HIV-1
The HLA-C –35 “C” allele associates with better control of HIV
To help understand how, Frank Kirchhoff elegantly tested whether
the HIV-1 accessory protein Nef can neutralize the C-related
protective effect, by comparing –35 CC subjects with low vs. high
viral loads
Results :
 high VLs in subjects with the CC genotype do not associate with
an increase in Nef-mediated downmodulation of HLA-C
 But they associate with enhanced potency in other Nef functions
that impair antigen-dependent T cell activation
HIV-1 Nef functions
possibly contributing
to high viral loads in
individuals that have
a ‘protective’ HLA-C
-35 CC genotype
Anke Specht, Frank Kirchhoff et al., in preparation
Importance of host genetics to a measure of
disease progression
ZNRD1/RNF39 (Genome-wide significant
determinant of progression)
HCP5 (Genome wide significant determinant of
progression and viremia)
HLA-C (Genome wide significant determinant
2
of viremia)
1.5
1
Progression was defined on the
basis of observed or predicted drop
in CD4 counts to below 350 for
individuals with and without
protective alleles:
- in blue the average time to CD4
drop is 2 years for individuals without
any protective alleles
- in red the average time is 8 years
for subjects with 1 or 2 protective
allele(s) in at least 4 of those variants
.5
variants, not currently genome wide
significant)
0
distribution and kernel density
CCR5 delta32
CCR2 V64I (Widely accepted functional
100
1'000
10'000
time to progression (number of days, logarithmic scale)
Data from Fellay et al. Science 2007 & the Euro-CHAVI
Consortium, part of the Center for HIV/AIDS Vaccine
Immunology (CHAVI)
The impact of common variants
After study of 500 subjects, three common
variants explain 14% of the variation in
viral load at set point
And…
After study of 2600 subjects, three
common variants explain 14% of the
variation in viral load at setpoint
Height
Marc Gasol
Pau Gasol
Height
Heritability is > .8
The most important common variant, in
HMGA2, explains one third of one percent
of variation in height general population
– Weedon et al 2007 .
.2
.1
0
Effect of the Nth snp
.3
Height effect sizes and fitted exponential
0
5
observed points
10
Rank of each height snp
15
least square fitted exponential
20
How many SNPs to explain 80
percent of the variation in height?
1. Effect size of SNP N
=0.0008242+0.3502509*0.8912553^N
2. 80 =
N*0.0008242+0.3502509*.8912553^N/LN
(.8912553) 0.0008242+0.3502509*.8912553/LN(.891
2553)
3. N=93,000
Where to next?
Other racial/ethnic groups
New cohorts (to assess acquisition)
Screens for rare variants (structural and
single site)
Malawi EU Study
500 positives/1000 negatives (exposure)
– Will add another 250 positives
Exposure criteria
– Visited STD clinic
– Older than 23
No genome-wide significant p-values for SNP association
– Still evaluating results
CNV analysis currently being run
Structural Variants
WGA screen for structural variants
– EuroCHAVI
– MACS
Deletions and duplications were inferred by using
publically available intensity software (PennCNV)
CNV region on chromosome 19 showed association with
setpoint and progression
Rare: 2.8% deletion
3.3% duplication
5
Setpoint by CNV state
4.66755
4.26229
4.07537
4
3.83744
1
2
3
3.10918
n=72
n=1977
n=86
n=2
0
1
2
3
4
0
n=2
CNV
Viral load setpoint decreases with chr19 CNV state
KIR:
Killer Cell Immunoglobulin-like Receptor
Methods in Molecular Biology, Martin & Carrington, 2008
-Multiple known haplotypes with different combinations
of KIR genes
-Most common duplication
-Most common deletion
Bw6 only
at least one HLA Bw4
5
p=0.5
4.47
p=6E-05
4.67
4.32 4.30
4.30
4.04
4
3.59
3.11
3
2
0
16
401
21
1
2
3
4
2
38
734
46
2
0
1
2
3
4
KIR copy number variants
Complete resequencing of
individuals with ‘extreme
phenotypes’
Extreme traits resequencing:
proposed framework
WG resequencing of a few individuals with extreme
phenotypes - likely to be enriched for rare causal
variants
2. Selection of a subset of the identified variants
(bioinformatics: genetic function, candidate genes…)
3. Genotyping of the best candidates in large
populations
1.
Hemophilia project
Study design: case/control study
 up to 1000 patients intravenously exposed to HIV between 1979-1984
 HIV infected individuals already analyzed in other Host Genetics projects
Exposure: The high prevalence of CCR5 d32 homozygosity in “exposed,
yet uninfected” haemophilia patients (known to be 15-25%) proves a very
high rate of effective exposure to HIV in this population:
CCR5d32 homozygosity
0.3
0.25
0.2
0.15
0.1
0.05
0
0%
20%
40%
60%
80%
Effective exposure to HIV
100%
SequenceVariantAnalyzer, a dedicated software
infrastructure to manage, annotate, and
analyze the large number of very unique
variants detected from a resequencing project.
Processed variant data including genomic coordinates (single site,
small and large copy number changes)
SVA GUI application
In-house statistical
module
External SIFT
RefSeq
HapMap & Illumina
program
Ensembl core database
Variation sets
KEGG pathway
Ensembl variation database
database
Exon-level prediction of
variant function
Functional impact of
NS SNPs on proteins
Binary output
Pathway filter
Presence in
existing databases
Fisher’s exact test
“Load ” test
for association with phenotype
32
The big question …
• Is whether the causal variants are
‘recognizable’
With thanks to
• NIH (CHAVI)
– NIAID, DAIDS, OAR
• Bill & Melinda Gates Foundation
Dr. Jacques Fellay
Dr. Kevin Shianna
Dr. Dongliang Ge
Dr. Woohyun Yoon
Dr. TJ Urban
Dr Anna Need
Liz Cirulli
Nicole Walley
Curtis Gumbs
Kiim Pelak
Dr. Amalio Telenti
Dr. Sara Colombo
Dr. Bart Haynes
Dr. Norm Letvin
Dr. Andrew McMichael
Dr. Lucy Dorrell
Dr. Seph Borrow
Dr. Mary Carrington
Dr. Nelson Michael
Dr. Amy Weintrob