Computational models of the transcriptional machinery and spatial patterning of the early mammalian embryo
Systems Biology of Stem Cells, Irvine, CA May 24-25, 2010
Carsten Peterson
Computational Biology & Biological Physics Lund University, Sweden
http://cbbp.thep.lu.se/
Lund Stem Cell Center, Lund University, Sweden
Cdx2 Oct4 Sox2
Nanog Gata-6
Overview
Transcriptional machinery
The basic ESC switch architectureSwitch properties
An embryonic ground state Reprogramming
Trophectoderm/endoderm extensions
Overview
Spatial patterning
Geometrical constraints in the embryo
Mechanics and biochemistry Trophectoderm formation
Endoderm formation
Transcriptional machinery
The basic ESC switch architectureSwitch properties
An embryonic ground state Reprogramming
Trophectoderm/endoderm extensions
Early “single gene” experiments identified OCT4, SOX2 and NANOG as key players Niwa (2000); Chambers (2003)
Followed by ChIP-chip experiments for their binding sites and microarray profiling for further insights
Boyer (2005); Loh (2006); Ivanova (2006)
A core structure emerges:
Core transcriptional network of embryonic stem cells
Early “single gene” experiments identified OCT4, SOX2 and NANOG as key players
Followed by ChIP-chip experiments for their binding sites and microarray profiling for further insights
A core structure emerges:
Core transcriptional network of embryonic stem cells
Nature of interactions
explored by the dynamics
ON OFF OFF
ON Switch ON =>
OFF =>
Early “single gene” experiments identified OCT4, SOX2 and NANOG as key players
Followed by ChIP-chip experiments for their binding sites and microarray profiling for further insights
A core structure emerges:
Core transcriptional network of embryonic stem cells
ON OFF OFF
ON Switch ON =>
OFF =>
Simple model considerations:
Everyone is an activator
Deterministic rate equations
Solve iteratively the Shea-Ackers rate equations:
Concentrations:
[A] = External signal [O] = OCT4
[S] = SOX2 [N] = NANOG
[OS] = OCT4/SOX2 complex
OS Binds first
Then recruits NANOG
……..
…….. Similar structure
……..
Turning the switch “ON”
ON OFF
NANOG
Signal A+
OCT4-SOX2
ON OFF
Wnt + ..
How does the Oct4-Sox2-Nanog system respond to external activating signals, e.g. Wnt?
The switch is very robust against parameter variations
Turning the switch “OFF”
OCT4-SOX2
NANOG
Signal B-
ON
ON
OFF
OFF
p53 How does the Oct4-Sox2-Nanog system respond to
external repressive signals, e.g. p53?
The switch is very robust against parameter variations
An ESC “ground state”
Signal A+
By over-expressing NANOG one obtains an irreversible switch.
Once ON, the stem cell can continue to self-renew in the absence of external
factors – a “ground state” Ying (2008)
Reprogramming
Expressing Oct4/Sox2 when OFF turns the switch ON
- Mechanism: Oct4 recruits Nanog to turn on the switch - Only expressing Nanog is not sufficient
The other known reprogramming factors Klf4 and c-Myc
are not part of our simplified network, but …
Takahashi, 2006
Lineage specification
Rossant (2006)
Lineage specification – the extensions
OCT4
OCT4
TARGETTARGET
Global expression profiling of Oct4 manipulated ES cells combined with (ChIP) assays
=>
Genes show both activation and repression depending on Oct4 expression levels
Matoba (2006)
OCT4 Trophectoderm Endoderm
OCT4
ES
T E
Lineage specification - trophectoderm
GCNF
Trophectoderm extension
OCT4–SOX2
NANOG CDX2
OCT4 SOX2
Lineage specification -endoderm
GCNF
Endoderm extension
OCT4–SOX2
NANOG
OCT4 SOX2
GATA6
Lineage specification – closing a loop
GCNF
Endoderm extension
OCT4–SOX2
NANOG
OCT4 SOX2
GATA6
Trophectoderm
CDX2
extension
GCNF
Endoderm extension
OCT4–SOX2
NANOG
OCT4 SOX2
GATA6
Trophectoderm
CDX2
extension
Suppress OCT4 -Trophectoderm lineage
GATA6, CDX2, GCNF
OCT4, SOX2, NANOG
SN
Suppress OCT4 -Trophectoderm lineage
GCNF
Endoderm extension
OCT4–SOX2
NANOG
OCT4 SOX2
GATA6
Trophectoderm
CDX2
extension
Overexpress OCT4 - endoderm lineage
Connecting two worlds
Gene expression meets cell division and mobility
1. Trophectoderm formation Oct4/Cdx2
2. Endoderm formation
Nanog/Gata-6
t<E0.5 n=1 t=E1.0 n=2 t=E1.5 n=4 t=E2.0 n=8
t=E2.5 n=8 t=E3.5 n=32
Laboratoire de Biologie de la Reproduction; Lausanne
The early embryonic development
Live imaging data emerging with geometric and some genetic info
Rossant (2009); Zernicka-Goetz (2005)
The early embryonic development
Develop a computational modeling framework for simulating the patterning of the embryo
Mechanics meets biochemistry
Mechanistic model of embryogenesis
Blastomers are incompressible ellipsoids
Elastic response is lumped into principal axes Measure deformation in cell overlap
Elastic, adhesion and drag forces Total force = Felastic + F
adhesion + F
drag
Overdamped mechanics (no acceleration)
• F
elastic• F
adhesionAttracts nearby cells in proportion to overlap area
Tangential drag force proportional to relative velocities and overlap area
• F
dragMechanistic model of embryogenesis
Blastomers are incompressible ellipsoids
Elastic response is lumped into principal axes Measure deformation in cell overlap
Elastic, adhesion and drag forces Total force = Felastic + F
adhesion + F
drag
Overdamped mechanics (no acceleration)
• F
elastic• F
adhesionAttracts nearby cells in proportion to overlap area
Tangential drag force proportional to relative velocities and overlap area
• F
dragCell division
• Cell cycle times sampled from experimental distribution Daughters share parental cell volume
• Selection of division plane is either
- random or directional (with respect to pellucid zone)
• Partition rules for cell content - Equal partition
- Asymmetric partition
Ready to go
Each cell has a set of internal data
- concentration of proteins, cell cycle length, etc.
Different “cell species” can have different division/growth rules, interaction parameters
Neighborhood determined by Voronoi diagram relatio
n
Track cell lineages, protein concentration, elastic energy, etc.
Analyze statistics of different cell lineages - explore hypothesis
Trophectoderm formation
Current conceptional models
Position-based model (inside-outside):
Inner or outer position of a cell dictates its Cdx2 level Polarity-based model:
Outer cells, which are known to be polarized, polarize Cdx2 as well Asymmetric divisions for cells with high/low Cdx2 content
Model the alternatives with mechanics and a simplified biochemical network with Cdx2 and Oct4 only
Trophectoderm formation
Cdx2 levels Inner/outer Tracking
Polarity-based model
Trophectoderm formation
Position-based model Polarity-based model
Trophectoderm formation
Computational model outcome:
Position-based model (inside-outside) or polarity-based model?
Both models give rise to the desired pattern
However, the inside-outside model is more robust
Blastocoel expansion
The fluid-filled blastocoel is formed after the 32-cell stage A slowly expanding spherically shaped region
Endoderm formation
After the blastocoel is formed:
Nanog and Gata-6 cells randomly distributed
Problem: How do these separate (cluster) in a directional manner?
Proposed mechanisms:
Differential adhesion and directional signaling
Model the system and evaluate the impact from such mechanisms
Differential adhesion
For moving cells; randomly, through homing signals or cell divisions, adhesion properties could be important
We explore such effects in endoderm formation by assigning different adhesion and cross-adhesion strengths
With Nanog/Nanog > Gata-6/Gata-6 > Nanog/Gata-6, the two cell populations segregate
However, a homing signal from the blastocoil surface is needed for robust endoderm formation
Towards the endoderm
Differential adhesion only
Directional signal only Nanog
Gata-6
Towards the endoderm
Nanog Gata-6
Differential adhesion + directional signal
Efficiency versus adhesion strength
Summary
The ESC switch including lineage sub-switches
Core Oct4/Sox2/Nanog model (2006) still captures essentials Bistability
Reprogramming with Oct4/Sox2
“ground state” with no external signals
The Cdx2 and Gata-6 “plug-ins” handles trophectoderm and endoderm formation
Ongoing and future work
(in progress; Chickarmane, Olariu)Epigenetics and more components
Noise – transcriptional versus epigenetic
Benefit from novel data; e.g on Nanog knockdowns (Lu, 2010)
Summary
Patterning mammalian embryonic development
Mechanics is important
Trophectoderm formation Endoderm formation
A simulation modeling framework essential for testing hypothesis
Future work
Implementing signalling and extend mechanics to describe further development
Key collaborators & publications
Vijay Chickarmane Biology Division, Caltech
V. Chickarmane et al, Transcriptional dynamics of the embryonic stem cell switch, PloS Comp Bio e123 (2006)
V. Chickarmane et al, A computational model for understanding stem cell, trophectoderm and endoderm lineage determination, PloS One e3470 (2008) P. Krupinski et al, Simulating the mammalian blastocyst – how biochemical and mechanical interactions pattern the embryo, submitted (2010)
Pawel Krupinski
Computational Biology, Lund