Biophysics at ECM: from Hydra to Neurons - ECM-UB

Biophysics at ECM: from Hydra to Neurons
Jordi Soriano-Fradera
Ramón y Cajal fellow
Barcelona, February 6th, 2009
MYSELF
1998 -2003 ---------------- PhD thesis at ECM
(rough interfaces in disordered media)
2003 – 2008: a postdoc’s life (biophysics)
Bayreuth
(Ott’s lab)
Weizmann
(Moses’ lab)
Soriano et al, PRL (2006)
Axis formation in Hydra Kücken et al., Biophys. J. (2008)
Soriano et al, Biophys. J (2009)
- Connectivity in living neural networks:
Breskin et al, PRL (2006)
Eckman et al., Phys. Rep. (2007)
Soriano et al, PNAS (2008)
Jacobi et al, J. Neurophysiol. (2009)
-  Mechanisms of synchronization in neural networks.
-  Oscillations in gene-regulatory networks.
-  Cognitive neuroscience and schizophrenia.
SELF-ORGANIZATION IN HYDRA
HYDRA. THE “RAMBO“ OF NATURE
-  One of the earliest multicellular
organism (600 m. years old).
-  State of constant growth and tissue
replacement.
Tentacles
Permanent, immortal “embryo”
Head
Hypostome
Astonishing regeneration
capabilities
Bud
-  Complex patterning involved in...
Asexual reproduction (budding).
Body structure maintenance
and regeneration.
Great model system!
Foot
REGENERATION AND AXIS FORMATION
Regeneration mechanisms:
a) 1 % of Hydra tissue from the body column makes a normal hydra.
b) Dissociated cells reorganize and form a new Hydra.
Isotropic configuration
1h
broken symmetry
Regeneration process in Hydra
H:
H: Head
B: Body column
F: Foot
Example of the closing process.
First 40 min.
75X real time.
1 mm
Regeneration process in Hydra
H:
H: Head
B: Body column
F: Foot
1 mm
3000X real time.
(2 days total)
Past results: Mechanical oscillations influence regeneration
Axis formation
model
Faster oscillations
faster regeneration
physical modeling
cell tension induces patterning
oscillations
pattern
axis
Future experiments at ECM:
-  Mechanisms of self-organization in Hydra aggregates.
- Patterning in aggregates.
time
Small sphere: one head
Large aggregates: multiple heads
…and multiple mess.
Regeneration of cells aggregates:
NEUROPHYSICS
move 600 million years ahead in evolution…
· Computing: From I/O to CPU (thought, emotion, intent).
· Language: Transmission of complex ideas through
“noises from our mouths”.
· Memory and Learning: We learn in milliseconds and
remember for decades.
· Sleep: An open mystery (synaptic resetting?)
· Networks: Construction of an emergent “self”
Neural networks. The quest for connectivity
a) Dragonfly. 90% of the brain is flight control.
The same neurons cultured in a dish just form a stupid network.
b) Human brain: 1011 neurons, 1014 connections.
Frenetic development during the first 15 years defines our personality.
-  Which is the relation between connectivity and function?
-  Can we unravel neural connectivity?
EXPERIMENTS AT WEIZMANN
Development of a new experimental tool to describe connectivity:
- Experiment: individual monitoring of ~1000 neurons.
- Theory: Graph-percolation approach.
Neuron. Integrate (over inputs) and fire:
Axon (1mm – 1m)
Cell body (soma)
Dendrites
Terminal branches
of the Axon
•  Firing threshold ~ 30 mV
•  Individual synaptic contribution ~ 2 mV
Synaptic connection
~ 15 inputs to fire
NEURAL CULTURES
rat Hippocampus
(19 days embryo)
neurons dissociation
plating
(day 0)
mature culture
(day 14)
MODEL: neurons fire either if they are selfexcited or get at least m inputs.
EXPERIMENT
The model reproduce many key elements of experimental data
For a Gaussian distribution of connections:
FUTURE DIRECTIONS AT ECM
1) From random connectivity to defined one:
- Neurons confined in traps.
- Constrained connectivity.
2) Relation between connectivity and function.
PDMS mold (transparent)
FUTURE DIRECTIONS AT ECM
1) From random connectivity to defined one:
- Neurons confined in traps.
- Constrained connectivity.
2) Relation between connectivity and function.
3) Three-dimensional cultures
PDMS mold (transparent)
PROBING THE BRAIN
MODEL
Disintegration of the network:
Control parameter m : “effective” number of inputs.
Voltage threshold (30 mV)
synaptic voltage (2 mV)
gsyn decreases with the amount of blocking (and m grows)
1
gsyn 
1 + [CNQX] / Kd
m0 = 15
The model contains the essential ingredients to reproduce the exp. results.
It provides a scenario to understand the experimental data.
HYPERCOLATION MODEL
= degree probability distribution (links per node)
f = fraction of neurons that fire in response to voltage V
F = total fraction of neurons that fire
Solved for a given
by considering the generating function
(or z-transform) of the
degree distribution
and its derivatives.
with z = 1-F.
DEVELOPMENT OF THE NETWORK
We have studied the emergence and growth of the giant component as
a function of time, and compared embryonic and postnatal hippocampal
cultures.
4h
plating
….
4h
giant component
DEVELOPMENT OF THE NETWORK
We observe a very similar behavior as
the disintegration of the network with
CNQX.
The giant component emerges
abruptly and grows very fast. It
can be characterized as a
percolation transition:

g ~ | 1 – t / t0 |
CONCLUSIONS AND FUTURE PERSPECTIVES
•  Novel experimental technique based on collective excitation of neurons.
•  Hypercolation model on random graphs contains the key integrate-andfire elements to reproduce the experimental results.
•  The model permits to extract statistical information of the network, such
as
- Gaussian connectivity.
- Average number of inputs per neuron ~ 150.
(in the brain ~ 7000)
- Ratio between excitation and inhibition.
•  Current directions:
- Effect of nerve growth factors on connectivity.
- Connectivity on brain slices.
- Global excitation applied to synchronization in neurons
Why neural cultures
-  Good “model systems”.
-  From a few to 105 neurons.
-  Accessible in the lab (easy to culture, study and manipulate).
-  Interests: propagation of information, activity, learning, electric and
magnetic stimulation, ...
rat Hippocampus
(19 days embryo)
neurons dissociation
plating
(day 0)
mature culture
(day 14)
ks1 expression pattern for adult Hydra and buds
2002, C. Colombo and T. Bosch (U. Kiel)