here - University of Sussex
Download
Report
Transcript here - University of Sussex
Associative Learning by
Single Cells
Dr. Chrisantha Fernando
Systems Biology Centre
University of Birmingham
Questions
• How can associative learning be
implemented in single cells?
• How can we go about trying to find if
these implementations exist?
• How can we make associative learning
devices and what are they good for?
Coincidence detectors
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Pre-Synaptic (Eccles)
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Post-Synaptic (Hebb)
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Coincidence Detection
and Memory
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
• Pre-synaptic AC coincidence detection
– 5-HT (G-protein) + Ca2+/Calmodulin (Eccles)
• Post-synaptic NMDA coincidence detection
– Ca2+ + Glutamate (Hebbian)
• Short and Long Term State Storage
– AC --> cAMP [15mins] --> PKA --> Decreased K+ conductance
– MAPK, Prion like CREB --> CRE gene expression
– Increased NMDA localization to membrane,PKC --> AMPA
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
A Model of Pre-Synaptic AC
based Learning
• Gingrich and Byrne (J. Neurophys.
1987)
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Paramecia Exhibit Classical
Conditioning
• Todd Hennessey et al
• Shock (UCS) + Vibration (CS) classical
conditioning of ‘avoidance response’ in
paramecia.
UCS = Shock
CS = Vibration
R = Avoiding Response
Sensory Mechanisms in
Paramecia
• Mechano: Eckert, Naitoh and Friedman.
J. Exp. Biol. (1972)
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Ca2+ current
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
K+ current
MACHEMER & ECKERT
1973
Applying
depolarization
produces
reversal
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
2+
Ca channels
are on the
membrane surrounding the
cilia Voltage gated Ca
2+
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Shaving cilia
abolishes Ca2+
current, until they
grow back.
channels are essential
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Behaviour of voltage gated
Ca2+ channels can be
modulated
Vibration??
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
2+
Ca -Calmodulin
Ciliary
activates ciliary AC.
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Is AC acting as a coincidence detector in classical conditioning
in paramecia? How is AC activity influenced by vibration? Is
spatial distribution of membrane de/hyper-polorization relevant?
Possible Associative Learning
Mechanisms in Paramecia
• Is learning occurring by a mechanism
analogous to pre-synaptic facilitation in
Aplysia, i.e. using an AC coincidence
detector, and cAMP dependent state
changes mediating memory?
Gene mediated
memory
Reduced CDI
PKA
Ca2+ channel
Ca2+
ATP
Vibration
cAMP
AC
Depolorization
Ciliary beat
reversal
An Intra-cellular “Hebbian”
Learning Mechanism
• I propose an abstract organization for an
intra-cellular “Hebbian” mechanism, i.e. that
depends on the extent of ciliary activity (“postsynaptic” effect) and not just on the
coincidence between shock and vibration.
• This can be implemented for example using a
PK, PKK cascade with positive feedback.
mPK*2
mPK
mPK*1
mPK
PKK + u1
V
S
u1
U1*
u2
U2*
PKK + u2
10
10
PKKu1
PKKu2
oPKK 0.005
Cilia feedback signal PKK
Existing Components
• oPKK activated along with the effecter by at
least two iPKs
• Two iPKs themselves activated by another
mPK only when they are bound to signal
molecules or signal molecules themselves
are phosp. directly.
• The oPKK should bind to signal molecules
and specifically activate the appropriate mPK
• The mPK should have a very slow equilibrium
compared to the other PKs.
Kinase Cascades with
Positive Feedback
NS Phosphatase
A more general mechanism
PK*2
PK
PK*1
PK
PKK + u1
u1
U1*
u2
U2*
Promotor
PKK + u2
Gene
10
PKKu1
10
PKKu2
PKK
0.005
Constructing an Associative
Learning Circuit
• Are such components known?
• How to go about finding networks in
existence?
• How to go about making them and
seeing if the idea works?
Acknowledgements
•
•
•
•
T. Hennessey
D. Stekel
E. Szathmary
J. Rowe