Transcript Document

1. Fibroblast Chemotaxis: more about positive
feedback loops.
2. Autoregulatory Mechanisms of Eukaryotic
Chemotaxis System Components: Receptors, Gproteins, GEFs, PI3K, Kinases, phosphatases. How
evolution has selected for components with
autoregulation and integral feedback control.
Fibroblasts chemotax toward growth factors
0
3 hrs
8 hrs
QuickTime™ and a
Photo - JPEG decompressor
are needed to see this picture.
12 hrs
21 hrs
PDGF-stimulated wound healing in mouse embryo fibroblasts
PI3K p110 Family Members
Class Ia
110 a p85
Ras
binding binding
C2
PIK
110 b
110 d
Tissue Regulation
Kinase
All Tyr Kinase
Tyr Kinase
All + b/g
Blood Tyr Kinase
Cells
Class Ib
110 g
Blood
Cells
b/g
Deletion of Class Ia PI3K genes in
mouse embryo fibroblasts impairs
PDGF-dependent cell migration.
Brachmann et al., 2005 Mol. Cell.
Biol. 25, 2593.
0
3 hrs
8 hrs
QuickTime™ and a
Photo - JPEG decompressor
are needed to see this picture.
12 hrs
21 hrs
WT
PI3K-Ia
Deletion
Woundhealing 10ng/ml PDGF
80
3h
Migrated Cells
70
8h
15h
21h
60
50
Ly294003
PI3K inhibitor
40
30
20
10
0
Wild type
PI3K Ia deletion
P85a-/-;p85b-/-
Defect in PDGF-induced lamellipodia formation
in MEFs defective in class Ia PI3K
Brachmann et al., 2005 Mol Cell Biol 25, 2593
unstimulated
Wild Type
PI3K Ia
deletion
PDGF
PDGF + WM
Class Ia PI3K has multiple domains
for signal input, allowing it to act as an
‘AND GATE’ or possibly an ‘OR GATE’
GTP
CDC42
SH3
Cdc42
binding
GPCR
Membrane
Tyr Kinase
GTP
bg
P-Tyr
P-Tyr
Ras
SH2
SH2
Ras
C2 PIK Catalytic
Binding
p85 regulatory
Class Ia PI 3-Kinase
p110 catalytic
p110b can also be activated by
bg subunits of G proteins, but
only when bound to a
phosphoTyr protein (AND
GATE).
Growth
Factor
Class Ia PI3K mediates growth factordependent cortical actin formation
Receptor
Tyr
Kinase
P-Tyr
Ras
GTP
p85 p110
SH2 PI3K
PIP3
PH
AKT
PIP2
PIP3
Rac GTP
GEF? Rac
PTE
N
Cortical Actin
Cell Migration
Deletion of class Ia PI3K genes appears to impair (but not
eliminate) Ras activation (as judged by impaired activation of
the downstream protein kinase, Erk)
Brachmann et al., 2005 Mol Cell Biol. 25, 2593
5 min PDGF [ng/ml]:
0
1
3
10
0
1
3
10
Erk-P
Erk
Wild Type MEFs PI3K Ia deleted
Thus, as in Dictyostelium, there appears to be a positive
feedback loop between PI3K and Ras in fibroblasts.
Reduced PDGF-induced Rac activation
in MEFs lacking class Ia PI3K
Brachmann et al., 2005 Mol Cell Biol. 25, 2593
Control
PDGF:
-
+
Double KO
-
+
Rac GTP
GST-CRIB pulldown
Rac
p85a
Erk-P
Erk
Overexpression of a Rac GEF (Vav2) induces lamellipodia formation
in MEFs lacking Class Ia PI3K
Brachmann et al., 2005 Mol Cell Biol. 25, 2593
Wild Type
PI3K Ia deleted
RhodaminePhalloidin
(Actin)
Vav2
PI3K is involved in both local Ras and local Rac positive
feedback loops
PDGF
Receptor
p85
Ras
GTP
?
p110
PIP3
+
?
+
Rac
GTP
PH
GEF
Conclusions
1. Growth Factor Receptors stimulate Class Ia PI3K through
PhosphoTyr residues of receptors binding to SH2 domains, while
GPCRs stimulate Class Ib PI3K through bg subunits binding to
the catalytic subunit.
2. In both cases, PI-3,4,5-P3 is in a local positive feedback
amplification loop involving Rac (and Ras?) that allows nonisotrophic localization of cortical actin, providing directionality
to chemotaxis.
How is perfect adaptation achieved in eukaryotic chemotaxis?
Shutoff mechanisms must exist to adapt the system to a given
level of stimulation, allowing a temporal increase in receptor
stimulation to be sensed. The adaptation should be slow
compared to the stimulation to insure significant directional
migration prior to adaptation.
What is known about shutoff mechanisms of GPCRs and
Receptor Tyr Kinases?
GPCR ACTIVATION
Hormone
Receptor
a
GDP
bg
Receptor
a bg
GDP
GPCR ACTIVATION
attractant
Receptor
a
GDP
Ligand-induced
Conformational
Change <100 msec
msec to sec
Receptor
bg
a
GDP
bg
Receptor
a bg
GTP
Receptor
bg
Effector 2
GTP
PI 3-kinase etc.
a
GTP
Effector 1
(Phospholipase C, etc.)
Signal Termination, Downregulation and Reset to Basal State
Receptor
bg
Effector 2
RGS
Seconds
a
GTP
Effector 1
Minutes
a
GDP
Signal Termination, Downregulation and Reset to Basal State
Receptor
bg
G-Receptor
Kinase (GRK)
RGS
Seconds
a
GTP
Effector 1
Minutes
a
GDP
Signal Termination and Reset to Basal State
Inactive
Receptor
PPP
Receptor
PPP
bg
Arrestin Phosphatase
G-Receptor
Kinase (GRK)
Seconds
Effector 1
Receptor
a bg
GDP
Dephosphorylation
And rebinding of
Ga and bg. (minutes)
RGS
a
GTP
Basal State
Minutes
a
GDP
Only activated receptors are
phosphorylated and downregulated.
This effect is slow (minutes)
compared to activation (seconds).
During this perturbation from steady
state, PI3K activation occurs, driving
directional motility.
Integral Feedback Control
Analogous to model in Yi, Huang, Simon&Doyle 2000 PNAS 97, 4649
If we assume that only activated receptors are
phosphorylated (and thus inactivated) and that the
phosphatase that dephosphorylates the GPCR
operates at saturation and is less active than the
G-protein Receptor Kinase (GRK), then the
model is analogous to integral control of bacterial
chemotaxis receptors. Inhibition of active
chemotaxis receptors by demethylation is
analogous to inactivation of active GPCRs by
phosphorylation. This is a consequence of the
fact that GRKs only phosphorylate receptors
associated with active bg proteins.
The rate of receptor phosphorylation is: dRP/dt = VPmax - VKmax(A)/(KK+A) (where A is
the concentration of activated receptors, KK is the KM of the GRK for activated
receptors, VPmax is the maximal activity of the phosphatase and VKmax is the maximal
activity of the kinase, GRK ).
Thus, the activity at steady state will be: Ast= KKVPmax/(VKmax-VPmax)
This is the set point (y0 in the model above). y is defined as the difference between the
activity at time t (y1) and the activity at steady state (y0). Thus, at steady state, y = 0.
Increased ligand binding acutely increases u and elevates y1 to a value above y0, giving
a transient positive value for y (resulting in PI3K activation). At steady state, (y = 0)
the rate of phosphorylation and dephosphorylation are equal. If one assumes that GRK
only acts on active receptors (whether or not ligand is bound) then the net rate of
phosphorylation at any instantaneous time will be directly proportional to y (the
transient excess in active receptors over the steady state value). When y = 0
phosphorylation and dephosphorylaiton cancel out.
The fraction of phosphorylated receptors (x) at any time t is then determined by the
number of receptors in the phosphorylated state at time zero, x0 (e.g. prior to the
perturbation due to increased ligand binding) plus the number of receptors that get
phosphorylated during the interval in which the system was perturbed. This latter term
is the integral from the time at which the perturbation (e.g. ligand unbinding) occurred
t=0 to time t of ydt.
So x(t) = x0 +
t
ydt
0
Notice that y can be + or - depending on whether ligand decreases or increases.
Thus dx/dt = y = k(u-x) - y0
At steady state, dx/dt=y=0 and y1=y0
Notice that since k and y0 are constants, an increase in u (rapid binding of ligand) is
ultimately offset by a slow decrease in x so that at steady state k(u-x) = y0.
P-Tyr
Tyr-P
Kinase
Tyr-P
SH2
Kinase
Kinase
Regulation of protein-Tyr kinases
P-Tyr
PI3K
Autopho-transphorylation of low activity monomeric protein kinases in the
ligand-induced dimer stabilizes the active state of each monomer, allowing
further transphosphorylation at sites that recruit signaling proteins.
P-Tyr
Kinase
Kinase
Tyr-P
SHP2
SH2
Tyr-P
P-Tyr
SH2 containing phosphoTyr phosphatases (e.g. SHP2) are preferentially
recruited to activated receptors and play a dual role of transmitting
additional signals (Ras activation) and turning off receptors.
INSULIN RECEPTOR CATALTIC DOMAIN (INACTIVE)
ATP
Pocket
Tyr 1158
Tyr 1162
Tyr 1163
Prior to stimulation, protein-Tyr kinases have floppy activation loops (region containing Tyr
1157, 1162 and 1163 of the insulin receptor). As a consequence the enzyme has a low
probability of being in the active conformation (~1%). Despite this low activity, when brought
in proximity with a another low activity Tyr kinase (due to growth factor binding), crossphosphorylation of respective activation loops can occur. Phosphorylation of the residues on
this loop stabilizes the active conformation of the protein giving a ~100 fold increase in activity.
Activated
Insulin Receptor
Peptide substrate
Integral Control of Receptor Protein-Tyr Kinases
The preferential dephosphorylation of activated Protein-Tyr kinases by SH2containing phosphatases provides a potential mechanism for integral control. In
response to an acute elevation in the level of ligand, the receptor will be rapidly
activated, but in the continuous presence of the ligand, the phosphatase will
ultimately return the kinase to a steady state activity that is determined by the
affinity of the phosphatase for the activated kinase, the Vmax of the phosphatase
and the Vmax of the kinase for transphosphorylation. Analogous to the set point for
bacterial chemotaxis receptors one can show that:
Ast = KM-SHP2VKinmax/(VSHP2max - VKinmax)
This simplified system does not reset to the same steady state as prior to receptor
stimulation since VKinmax is dependent on receptor ligation. Modeling predicts an
overshoot followed by return to a steady state that depends on ligand occupation.
This is in agreement with observations at intermediate times (0 to 30 min.)
following PDGF stimulation
Exclusive ubiquitinylation, of activated protein-Tyr kinases (due to SH2-containing
E3 ligases (e.g cbl)), leads to receptor internalization, providing a second
mechanism of longer term shut-off that also models as integral feedback control.
Integral Control of PI3K
PI3K, when activated, phosphorylates lipids at a high rate but
also autophosphorylates (on regulatory and catalytic subunits) at
a slow rate, leading to inactivation.
Assuming that the phosphatase that dephosphorylates PI3K is
saturated by substrate, this could also lead to integral control of
this enzyme.
Parallels between low molecular weight G protein (Ras, Rac Rho)
regulation and heterotrimeric G protein regulation
GDP/GTP Exchange Factor (GEF) activate: analogous to GPCR
GEF (SOS)
basal
Ras-GDP
GTPase Activating Protein
Analogous to RGS
slow
GAP
Ras-GTP
Effector
Effectors such as Raf
(Ser/Thr kinase) or PI3K
bind to activated Ras
Heterotrimeric and low molecular weight GTP binding proteins have been retained and
expanded during evolution because they have unstable activated states and can
spontaneously return to inactive states. Inactivation can also be accelerated by GAPs.
Signal Transduction in Eukaryotic cells is usually initiated by recruitment of
signaling proteins to the plasma membrane.
We have discussed three major mechanisms for acute and reversible protein
relocation in response to cell stimulation. These mechanisms have the potential to
amplify small signals.
More importantly, recruiting signaling proteins from a 3-dimensional space
(cytosol) to a 2-dimensional space (membrane) provides a mechanism for
facilitating unfavorable multimeric interactions.
Protein phosphorylation
to create docking site
Membrane
Tyr
P-Tyr
Activation of
GTP-binding
proteins
Membrane
GDP-Ras GTP-Ras
Generation of lipid
second messengers
Membrane
PI-4,5-P2
PI-3,4,5-P3
PH
SH2
Ras
GEF
Stochiometric
PI3K
Small Amplification
AKT
Large Amplification
Sarraste
Plasma Membrane
1
1
5
4
3
R28 3
The PH domain of BTK binds the head group of PI-3,4,5-P3 and
crystallizes as a dimer with the two binding pockets on the same surface.
However, in solution, it behaves as a monomer.
Moving signaling proteins from the three dimensional environment
of the cytosol to the two dimensional environment of the plasma
membrane decreases the entropy difference between a monomeric
and dimeric state.
Many signaling proteins may have evolved very weak free energies
of homo or hetero-dimerization to insure that dimerization only
occurs when confined on a two dimensional surface.
Signal
Dimers
Monomers
Membrane
Pólya's Random Walk Constants http://mathworld.wolfram.com/PolyasRandomWalkConstants.html
Let p(d) be the probability that a random walk on a d-D lattice returns to the origin. Pólya (1921) proved that
(1) p(1) = 1; p(2) = 1
but
(2) p(d) < 1
for d > 2. Watson (1939), McCrea and Whipple (1940), Domb (1954), and Glasser and Zucker (1977) showed
that
(3) p(3) = 1 - 1/u(3) = 0.340537….
where
u(3) = 3/(2p)3
p p p
_____dxdydz______
-p -p -p 3-cosx-cosy-cosz
Finch, S. R. "Pólya's Random Walk Constant." §5.9 in Mathematical Constants. Cambridge, England: Cambridge University Press, pp. 322331, 2003.
Domb, C. "On Multiple Returns in the Random-Walk Problem." Proc. Cambridge Philos. Soc. 50, 586-591, 1954.
Glasser, M. L. and Zucker, I. J. "Extended Watson Integrals for the Cubic Lattices." Proc. Nat. Acad. Sci. U.S.A. 74, 1800-1801, 1977.
McCrea, W. H. and Whipple, F. J. W. "Random Paths in Two and Three Dimensions." Proc. Roy. Soc. Edinburgh 60, 281-298, 1940.
Montroll, E. W. "Random Walks in Multidimensional Spaces, Especially on Periodic Lattices." J. SIAM 4, 241-260, 1956.
Sloane, N. J. A. Sequences A086230, A086231, A086232, A086233, A086234, A086235, and A086236 in "The On-Line Encyclopedia of
Integer Sequences." http://www.research.att.com/~njas/sequences/.
Watson, G. N. "Three Triple Integrals." Quart. J. Math., Oxford Ser. 2 10, 266-276, 1939.
Eric W. Weisstein. "Pólya's Random Walk Constants." From MathWorld--A Wolfram Web Resource.
http://mathworld.wolfram.com/PolyasRandomWalkConstants.html
Further websites for random walks
http://www.rwc.uc.edu/koehler/biophys.2ed/java/walker.html
http://www.krellinst.org/UCES/archive/modules/monte/node4.html