Transcript - Institut de Génétique Humaine

Giacomo Cavalli
In vivo protein-DNA
interactions
UE Méthodologie, 11 April, 2014
Institut de Génétique Humaine, CNRS
Montpellier, France
10,000 nm
DNA compaction
in the nucleus
11 nm
30nm
1bp (0.3nm)
compact size
DNA
length
compaction
nucleus (human)
2 x 23 = 46 chromosomes
92 DNA molecules
10 mm ball
12,000 Mbp
4 m DNA
400,000 x
mitotic chromosome
2 chromatids, 1 mm thick
2 DNA molecules
10 mm long X
2x 130 Mbp
2x 43 mm DNA
10,000 x
DNA domain
anchored DNA loop
1 replicon ?
60 nm x 0.5 mm
60 kbp
20 mm DNA
35 x
chromatin fiber
approx. 6 nucleosomes per ‘turn’ of 11 nm
30 nm diameter
1200 bp
400 nm DNA
35 x
nucleosome
disk 1 ? turn of DNA (146 bp) + linker DNA
6 x 11 nm
200 bp
66 nm DNA
6 - 11 x
1 bp
0.33 nm DNA
1x
base pair
0.33 x 1.1 nm
Compaction of DNA by histones
Compaction by higher order determinants
Implication of PcG proteins in dynamic gene regulation
Cellular
Memory
System
Differentiation
Cell fate
determination
Proliferation
Cell cycle
Control
HOX
genes
Signaling
genes
PcG proteins
Cell cycle
genes
TFs of
developmental
networks
Cancer
development
Stem cell
plasticity
Developmental
pathways
Schematic mechanism of Polycomb mediated silencing
Histone Methyl Transferase
Me3
K27
Core PRC2
Me3
K27
E(z)
Esc
Su(z)12
Nurf-55
PRE
H3
Target gene
H3
H3
H3
Schematic mechanism of Polycomb mediated silencing
Histone Methyl Transferase
Me3
K27
Core PRC2
Me3
K27
E(z)
Esc
Su(z)12
Nurf-55
PRE
H3
H3
H3
H3
Target gene
Psc
Ph
Core PRC1
dRing
Pc
Me3
K27
H3
Me3
K27
H3
H3
H3
Schematic mechanism of Polycomb mediated silencing
Histone Methyl Transferase
Me3
K27
Core PRC2
Me3
K27
E(z)
Esc
Su(z)12
Nurf-55
PRE
H3
H3
H3
H3
Target gene
Psc
Ph
Core PRC1
Ub E3 ligase
dRing
Pc
Me3
K27
H3
Ub
K119
H3
Me3
K27
H3
Ub
K119
H3
Schematic mechanism of Polycomb mediated silencing
Histone Methyl Transferase
Me3
K27
Core PRC2
Me3
K27
E(z)
Esc
Su(z)12
Nurf-55
PRE
H3
H3
H3
H3
Target gene
Psc
Ph
Core PRC1
Ub E3 ligase
dRing
Pc
Me3
K27
H3
H2A
Ub
K119
H3
H2A
Me3
K27
H3
H2A
ATP-dependent chromatin remodeling
Ub
K119
H3
H2A
PcG and trxG proteins associate to multiple genomic loci
Polytene chromosome staining shows around 100 bands for each PcG protein
Genome-wide identification of downstream PcG target genes
« ChIP-on-chip » approach
The ChIP on chip approach
ChIP
Cross-link
chromatin
Produce
soluble
chromatin
DNA chip
1st generation microarrays
Produce 2 KB PCR fragments of overlapping
genomic DNA fragments
2nd generation microarrays
IP step
whole genome coverage with 1,000,000 long
oligonucleotides, i.e. 1 Oligo per 120 bp of
euchromatin
Produce fluorescent labeled probes
Protein IP
Control IP
Hybridize to the DNA chip
Obtain the profile
Dynamic function of Polycomb proteins and cell proliferation
Embryos -Schuettengruber et al 2009
H3K27me3
200Kb
PC
Ph
S2 cells data-Schwartz et al 2006
H3K27me3
PC
Psc
http://www.purl.org/NET/polycomb
ChIP on chip validation: Comparing ChIP on chip data with a chromatin profiling using an
independent technology called DamID
In DamID, the chromatin protein of interest is fused to the bacterial Dam-methylase
and the construct is transfected into the cells of interest. The protein of interest
drives the Dam partner to its targets, and the methylase puts a methyl mark at the
“A” of GATC sequences. Methylated DNA is then isolated and hybridized onto
microarrays of interest
Correspondence between ChIP on chip and DamID data
Signaling pathways interacting with RDGN genes:
26.4%
Maternal genes
73.6%
N=53
23.1%
Gap genes
76.9%
N=13
PcG target genes
regulate genes at
multiple layers of
transcriptional
cascades
toy / PAX6
ey / PAX6
eyg / PAX6(5A)
eya / EYA1-4
so / SIX1/2
27.3%
Optix / SIX3/6
Pair-rule genes
shf / WIF1
dac / DACH1-2
72.7%
N=11
Eye specification
Segment polarity
genes
40.7%
59.3%
N=54
Homeotic genes
100 %
N=8
No target
PcG target
Direct Hox
gene targets
N=21
52.4%
47.6%
FLY
toy2
ey 1
eyg (toe)1
Optix1
shf
eya2
so1
dac1
hh1
dpp1
MOUSE
Pax6
Pax6
Six6
Wif1
Eya1-4
Six1
Dach1
Shh
Bmp2
HUMAN
PAX6
PAX6
SIX6
WIF1
EYA1-4
SIX1
DACH1
SHH
BMP2
Additional factors involved in eye development:
oc1
Otx1
OTX1
ato3
Atoh1-8
ATOH1-8
tsh2
bi1
Tbx2
TBX2
The evolution of ChIP: massive sequencing of the immunoprecipitated chromatin DNA
H3K36
ChIP
100bp
1Kb+
ChIP-Seq Library construction
~5-10ng
Polish ends
5’
3’
Taq extend
600bp
500bp
400bp
300bp
200bp
100bp
A
Ligate Solexa Linkers
A
Illumina sequencing
Laser
C
A
Linker ligated DNA
Amplify to form clusters
T
G
Sequence
one base at a time
Flow cell imaging by microscopy
60 X objective: thousands tiff images / hundred thousands of images per run.
Chromatin Immunoprecipitation Tag Sequencing
After obtaining the sequences, they are positioned on the
genome by automated algorythms (like Blast but quicker) and
each tag is thus assigned its position on the genome.
These profiles can then be quantified and analyzed just like
normal ChIP on chip profiles
Identification of new PcG target genes
•PcGtargets (PC/PH/H3K27me3)
305
maintained
PH
181
embryos
PC
145
New domain
0
0
H3K27me3
PH
eye
discs
350
275
0
0
PC
353
0
H3K27me3
0
(+)
(-)
Anna Delest
fd96Ca
fd96Cb
danr
dan
PRE position is highly conserved in Drosophila species
D.Melanogster vs D.Yakuba
PH Mel
PH Yak
PC Mel
PC Yak
K27 Mel
K27 Yak
PHO Mel
PHO Yak
DSP1 Mel
DSP1 Yak
K4 Mel
K4 Yak
Wnt4
wg
→ species-specific differences can be used to study PRE sequence features
Bernd Schüttengruber
• Exploiting In vivo protein-DNA interactions to
learn about the three dimensional
conformation of chromatin
PREs are sometimes located at positions overlapping the proximal gene promoter,
but in other instances they can be at tens of kilobases away from it.
How can PcG proteins repress transcription in all these cases?
> 30 kb
Mecanisms
?
PcG proteins
28 kb
"SPREADING versus LOOPING"
Two models have been proposed in order to explain how PcG proteins repress their target genes:
1. They might spread from the PRE into flanking chromatin, covering the whole domain including
the target promoter
2. Alternatively, they might reach the promoter via direct looping of the PRE and establishment of
protein-protein contacts.
"LOOPING"
PRE
"SPREADING"
PcG proteins
PcG proteins
Interestingly, at some endogenous target genes PREs are located at very large distance
from the promoter and they are flanked by elements called: "chromatin insulators"
Insulators
• Insulators are divided into three classes depending on their abilities
Enhancer
blockers
En.
Ins.
En.
Gene
Chromatin boundaries
Ins.
Insulators
that can be
"bypassed"
En.
En.
Ins.
• One insulator can have many of these properties
Gene
Ins.
Ins.
Gene
The gypsy insulator
● DNA element isolated from the drosophila gypsy retrotransposon
● This sequence contains 12 binding sites for the Su(Hw) protein, that is required for insulator function
En.
Insulator bypass model
Ins.
Ins.
Gene
Model of nuclear chromosomal architecture
based on insulators interaction
Domain A
Domain B
Insulating proteins
Gerasimova et al, Mol. Cell, 2000
Bypass of the gypsy insulator by a PRE



yellow
PRE
Insulator

white
yellow
PRE
Insulator
Expression of white
Expression of white
red
red
brown
brown
orange
orange
yellow
yellow
white
white
PRE
Enhancer
Insulator
Insulator
white
Insulator
Gene
Gene
Insulator
Yes! the PRE can
bypass 2 insulators
ChIP analyisis of the molecular landmarks of insulator bypass
35
1kb
Fold enrichment
30
25
PC
20
pupal stage
● PcG proteins bound to the PRE
can reach a downstream promoter
without coating an insulated
chromatin domain
15
10
5
0
yellow
PRE
Insulator
● PcG proteins are able to spread
from a PRE into a neighboring
region of several kb. This spreading
is blocked by one insulator
white
Insulator
0
Fold enrichment
5
● Two insulators build a chromatin
domain fully shielded from invasion
by PcG proteins
10
15
20
PH
25
pupal stage
30
35
1kb
The data shown before provide good evidence for a spreading
process
Can we get direct evidence for looping?
Chromosome Conformation Capture (3C) technology:
3C technology allows to
convert chromosomal
interaction events into DNA
ligation events that can be
analyzed by PCR
Biological material
1
5
DNA purification and
quantitative PCR analysis
4
Main steps of
3C technology
Formaldehyde-fixed nuclei
preparation
2
3
Chromatin digestion
Extensive dilution
Ligation
Two gypsy insulators build a chromatin loop
Interaction level in percentage of input
(P)(S)YSW-22E lines - Adult
H3C - distal gypsy insulator anchor
0.5%
0.4%
0.3%
0.2%
0.1%
0.0%
-15kb
-10kb
PRE
CG4238-RF
-5kb
Prox.yellow
Ins.
0
+5kb
+10kb
Nplp4-RA
Dist. mini-white
Ins.
CG33543-RC
CG15353-RA
(P)(S)YSW transposon
tRNA:CR31940-RA
tRNA:CR31669-RA
tRNA:CR31939-RA
Anchor
tRNA:CR31943-RA
tRNA:CR31944-RA
+15kb
In summary, both spreading and looping models could be correct, each one
accounting for a particular context
yellow
PRE
b
w
PcG proteins
mini-white
yellow
PRE
br
b
Dist.
Ins.
w
mini-white
br
Prox.
Ins.
PRE close to
its target
promoter
Dist.
Ins.
PRE distant from
its target promoter
"SPREADING"
+
Comet et al, Dev. Cell 2006
PRE
"LOOPING"
How PcG proteins and insulators might work in the cell nucleus
Nucleus
PcG bodies
Insulator bodies
Insulator-binding protein
complexes
 High-resolution 3C is appropriate to study chromatin conformation
Analysis of Hox gene contacts by 4C
• We developed a new 4C method based on “biotinylated primer extension”
streptavidin bead
Biotinylated Primer
GGGGG
CCCCC
• The amplified material is then hybridized to a Microarray (Roche Nimblegen)
• We used the Fab-7 PRE sequence as a bait, which negatively regulates the
Abd-B gene in the BX-C
Itys COMET
Modification and control of the 4C procedure
3C preparation
20
18
16
14
12
10
8
6
4
2
0
Copies number ratio
1.
Biotinylated-primer
extension
Copies number ratio
Anchor
fragment
2 Ins. / 1 Ins.
2 Ins. / 1 Ins.
Unknown
partner
3C
2.
Affinity purification on
streptavidin beads
4C
INPUT
BEFORE
amplification
20
18
16
14
12
10
8
6
4
2
0
4C
BEFORE
amplification
4C
AFTER
amplification
3.
“In situ” linker synthesis
1kb
4.
Quantitative amplification
by real-time PCR
500bp
400bp
300bp
Unknown partners
ligated to the anchor
fragment
200bp
5.
Genomic DNA-Chip
Hybridization
Anchor fragment
100bp
Primer dimers
Denaturing 4% agarose gel
4C data analysis by generation of domainograms
Normalized profile intensities for each probe i are transformed into rank based scores Qi,
which are combined into Siw multiscale scores and transformed as Piw probabilities using
Fisher's Chi square law.
Piw at scale =
N probes
probabilities at
scale = 3 probes
probabilities at
scale = 1 probe
Legend:
N is the total number of probes
ri is the rank of probe i
Benjamin LEBLANC
The Piw values represent
probabilities of 4C events as
a function of chromatin
domain size
Piw in false color
Piw at
Log scale =
N probes
Major Fab-7 4C hits are Polycomb bound
regions
4C Domainogram
1Mb
1Mb
3R
3R
ANT-C
grn
prospero
hth
E5-ems
Fab-7
BX-C
srp-pnr ss
Polycomb ChIP Domainogram
NK-C
pnt
Drop
10-5000
10-500
10-50
10-10
10-1
Simplified Hi-C procedure
•Fix nuclei of 16-18 hr embryos
•Digestion with 4-cutter DpnII
•Ligation and DNA purification as 3C
•Sonication and selection of ~800 bp
•Deep paired-end sequencing
Hi-C efficiently reproduces known 3C contacts
Chromatin contact features
2.
Matrix diagonal is not homogeneous
Polycomb-mediated interactions
Bantignies et al., 2011
http://www.igh.cnrs.fr/equip/cavalli/link.PolycombTeaching.html
References: Schüttengruber et al. (2009) PLoS Biol 7(1): e1000013; Comet et al. Dev Cell 11, 117-124
and PNAS , 108(6):2294-9; Bantignies et al. Cell 144, 214-26, Sexton et al. Cell 148, 458-472
Bernd SCHÜTTENGRUBER
Nicolas NEGRE
Benjamin LEBLANC
Anna DELEST
Itys COMET
Tom SEXTON
ERC
EU - 7FP
CNRS, ARC
French ministry of research