“Understandably, the search for these cross connections ... · RA Weinberg, The Biology of Cancer, 2007, p202. Genetic Variation at the Low Density Lipoprotein Receptor‐Related

“Understandably, the search for these cross connections has been largely postponed until the operations of each primary pathway are elucidated.” RA Weinberg, The Biology of Cancer, 2007, p202.

Genetic Variation at the Low Density Lipoprotein Receptor‐Related Protein 5 (LRP5) Locus Modulates Wnt Signaling and the Relationship of Physical Activity with Bone Mineral Density in Men

Kiel et al, Bone, 2007, 40(3): 587–596.

WT

alle

le

Hap

loty

pe1

Hap

loty

pe2

G-T

-G-T

A-T

-G-T

WT

alle

le

TOP

flash

activ

ity in

HE

K29

3 ce

llstra

nsfe

cted

with

LR

P5

hapl

otyp

es

Wnt3a - + - + - +

… “the TT genotype was associated with lower BMD in men with higher physical activity scores, conversely with higher BMD in men with lower physical activity scores”.

essential & known components

non-essential known components

“novel” components

MB Major et al, Science Signaling, 2008, 1(45): ra12

BW Miller et al, Mol. Sys. Biol., 2009, (5):315

Cytoscape visualization of the Wnt pathway from cancer.cellmap.org (netpath.org data)

GSK3β

APC

Axin1

Dvl2

Wnt Planar Cell Polarity Pathway

KEGG Canonical Wnt signaling

KEGG colorectal cancer pathway

netpath.org

WikiPathways

APC

APC

PORCWLS

free Wnt(short range)

complexed Wnt(long range)

Flottillin-2

retromer

GPC3

Morphogendiffusion profile

ligand processing & delivery

signal transduction

regulation of transcription

degradation complex

Wnt ligand processing and secretion

heparan sulfate proteoglycans: GPC3

PORCWLS

Flottillin-2

retromer

complexed Wnt(long range)

free Wnt(short range)

GPC

3

Morphogendiffusion profile

Xin

g et

al,

Gen

es &

Dev

. 200

3 17

: 275

3-27

64

β-catenin degradation in the absence of Wnt signaling

Axin:Apc:Gsk3β:Diversin:CKIα/ε:PP2A:WTX+

β-catenin

Axin:Apc:Gsk3β:Diversin:CKIα/ε:PP2A:WTX:β-catenin|p

Axin:Apc:Gsk3β:Diversin:CKIα/ε:PP2A:WTX

+

β-cateninβ-Trcp, Skp1, Cul1, Rbx

recy

cle

|p

|p

Wnt signaling – (1) Receptor Complex Formation & signal amplification

+ Lrp+ Fz

LFzW

+ LFzWDFrGαβγ

Wnt ------------------PP1 , PP2A,B

CK1ε

(CK1ε)p

+ Frat-1

Dvl

(Dvl)pp (Dvl)pp:Fr

LFzWDFr + Ck1γ

+ Axin:GSK3LFzWD(AGC)Fr LFWD(AGC)

PP

destruction complex

freeβ-catAPC

to nucleus

MACF1

+

Axin:GSK3

LFWD(AGC)

PP

Model

destruction complex

Wnt Axin,Gsk3β

β-catAPC

Receptorcomplex

n

Wnt signaling – (2) Receptor Aggregation & Clustering

Lrp aggregation

PIP5K1A, PI4KIIα

LFzWDFr(AGC)PIP2

L FWD(AGC)

PP m

L FWD (AGC)

PP m

L FWD (AGC)

PP mq

n.mn

n.m

L FWD (AGC)

PP mq

n.m.k

n

Overall Model

destruction complex

Wnt A,G,C

β-catAPC

Signalamplification

Receptoraggregation

Receptorclustering

PP

Dvlmediatedreceptorclustering

In the nucleus

+ phosphorylation of Tcf and Tle by Ck1 and Ck2

Wntnuclear(APC)P

Cytoplasmic β-cat

Nuclear β-cat

transcription

repression

Histone modifications

SET1 H3K4me

β-cat:Pygo bind H3K4me, dissociate from TCF

CtBP

KRAS/RAF1/RAC1

BCL-9, Pygode

laye

d / s

low CBP, P300, TRRAP,

SWI/SNF, BRG1,Mediator, COMPASS

PAF1, HRPT2, MLL1/2

export /

degradation RanBP3, RanGTPβ-Trcp, Skp1

Cul1, Rbx

HDAC-1,SHARP,TLE, CtIP,CtBP1/2X

Data: BW Miller et al, Molecular Systems Biology, 2009, (5):315

My comments:

Less DVL sequestration

Canonical W

nt pathway genes e

xtracte

d from Moon lab gen

ta notome-wide screen (M

ichael M

ajor, Jason Berndt et al)

Unpublished da

shown.

Kofahl & Wolf, Biochem. Soc. Trans., 2010, (38):1281–1285;

Kruger & Heinrich, Genome Informatics, 2004, 15(1): 138-148.Full model : 15 ODEs, 31 parameters Reduced : 7 ODEs, 8 algebraic equations, 19 parameters(Same model as Lee et al, 2003)

Bold face = measured in Xenopus egg

extracts

Italics = estimated

3 decimal points!

Lee et al, PLoS Biology, 2003, 1(1):116-123.

guess-timatedfrom above

Oksana Schaeffer-Tymchyshyn, PhD Thesis, Dept CS, U. Birmingham, 2008.

46 rate equations, modeled with ODE’s and also using stochastic p-calculus

(simian) COS cells + in-vitro

Xenopus egg extractsSW480 + COS + in-vitroin-vitro

in-vitro

(mouse) C57MG cells

in-vitro

(drosophila) S2 cells(human) 3T3 cells

(mouse) C57MG cells

in-vitro(human) MB231 cells

(human) MB231 cellsXenopus egg extracts

(simian) COS cells + in-vitro

Cho, Baek & Sung, FEBS Letters, 2006, (580):3665–70.

18 “elementary” reactions

Based on: Lee, Salic, Kruger, Heinrich, Kirschner, PLoS Biology, 2003, (1):E10.

For a mutant with n missing sites:

KD(n) = cn * KD(WT)

Cho, Baek & Sung, FEBS Letters, 2006, (580):3665–70.

Immunoblots in engineered

mouse F9 cells

Yokoyama et al, Journal of Molecular Signaling 2007, (2):11

Kinetics of signalosome formation in HeLa cells

Bilic

eta

l, Sc

ienc

e, 2

007,

(316

):161

9–16

22.

Axin=green, Lrp6=red, arrows=membrane-associated co-aggregates

Mirams, Byrne & King, J Math Biol. 2010,60(1):131-60.

1-equation formal reduction of Lee et al (2003) model (for Xenopus):

ββ β

Where:

Mirams’ extension to include Axin2 feedback

ββ

ββ

β

Qualitative predictions using Miram’s 1-equation model Wnt dose response

β−catenin

Wnt

Bie

chel

eet

al,

Che

mis

try &

Bio

logy

, 200

7, (1

7):1

177–

1182

.

log (Wnt)

β−catenin

Predictions using Miram’s 1-equation model

GSK3 RNAi

DVL RNAi

contradicts Moon lab RNAi data & Miller et al, 2009, MSB, (5):315

X

X

√√

β−catenin time course

~ 8

hour

s

√GSK3 OE

APC RNAiAPC OE

√DVL OE

Wnt=1

Predictions using Miram’s 1-equation model

GSK3 RNAicontradicts Miller et al, 2009, MSB, (5):315

XGSK3 OE

APC RNAiAPC OE

√DVL OE

Wnt=0

β−catenin time course

Goentoro and Kirschner’s 1-equation model, Molecular Cell, 2009, (36):872–884

Wnt+Wnt

β-cat β-cat− β-cat

β-cat

APC-mediated degradation of Axin

steady state β-catenin concentration is a saturating function of APC & GSK3β levels

stea

dy s

tate

β-c

aten

in

GSK3β

GSKβ OE

X

stea

dy s

tate

β-c

aten

in

APC

APC RNAi

X

Power Law approximate kinetics – Saturable and Cooperative formalism

condition1 condition2 condition3

30 p

aram

eter

s

Input X1 X2v1 v2v3

v5

X3 v4

X4 v6

Sorribas et al, Biotechnology and Bioengineering, 2007, 97(5):1259-1277.

( )GinputfionmaxExpresstimeScaledtdG

Ginputfkkk

dtdGGkinputfk

dtdG

d

tddt

−=

−=−=

)(..

)(. .)(.

:lyequivalent

:lyequivalent

maxExpression1

maxExpression2

timeScale1

timeScale1 G (A

.U.)

input

Time (A.U.)timeScale2

inputper threshold1 inputs offunction logic and sevaluation node theoforder in the encoded is

set is :models logicIn

⇒=

=

f(input) timeScale

1 ionmaxExpress

RNAi RNAi OE OE

– Wnt + Wnt – Wnt + Wnt

Ub Core UBE2M, CUL1, DUSP18, βTRCP

SKP1A ↑lower β-cat degradation

↑ ↑

APC ↑AXIN 1/2 ↑ ↑ M ↑ M ↓ M

SIAH1 ↑ (degrades β-cat)

SMURF1/2 ↓ M (degrades Axin) ↓ (activates PCP)

TCF/LEF1 ↓ ↑ ↑LRPs/FZs ↓

Wnt4/5B/11 ↓ M

sFRP2 ↓ M

WIF1 ↑FRAT1/GNAi1 ↓ ↓ ↑ ↑

PP1γ ↓ (as Axin OE)

DVL1/2/3 ↑ X (see idea) Miram’s ↓ ↑ Miram’s ↑ ↑ Miram’s ↑NKD1 ↓ M ↓ M

GSK3α/β ↓ X GSK3β titration? α↓ ↓β M Miram’s ↑ ↓ M Miram’s ↓CSNK2A1/2 ↑ (β-cat degradation) ↑ (β-cat degradation)

CSNK1ε ↓ (signalosome) ↑ (signalosome) ↑ (signalosome)

CSNK1δ ↑ (in DC) ↓ (in signalosome) ↑ (in signalosome) ↑ (in signalosome)

CTNNB1 ↓ ↓ ↑ ↑

proteasomal degradation of Axin and APC

CSK2 is constitutively active

SMURF2 binds Axin and destabilizes it

HEK293T

HE

K29

3T

Kim & Jho, Journal of Biological Chemistry, 2010, 285(47):36420–36426.

Narimatsu et al, Cell, 2009, (137):295–307.

(in mice)

HEK293 cells over‐expressing Dvl‐2 (green)blue=nuclear stain

These cytoplasmic Dvl aggregates are not required for Wnt/β‐catenin signaling

Sm

alle

y et

al J

ourn

al o

f

red=F-actin

Cel

l Sci

ence

, 200

5, (1

18):5

279-

5289

HEK293s with Dvl‐2 (green) at ~ physiological level

Hypothesis: Over-expressed Dvl recruits Axin out of the destruction complex β-catenin signaling(strong inter-DIX-domain interaction between Dvl and Axin)

A parsimonious model of canonical Wnt 24-48hr response in HEK293

Nkd1

Axin1+Axin2Dvl(with GSK3 etc)

Destruction Complex

Wnt signalosome

β-catenin

NKD1AXIN2

X

Fzd’s sFRP’sCanonical Wnt’s

X

non‐canonical Wnt’sβ-catenin

(Adherent) A375 cells expressing Wnt3a

Wnt3a gradient due to flow

Cimetta et al, Lab on a Chip, 2010, (10):3277–3283.

Reported graded response to ligand concentration in canonical Wnt signaling

minutesunst

imua

ted

(60ng/ml rWnt3a)

(β-catenin firefly luciferase reporter in HEK293T cells)

Naik and Piwnica-Worms, PNAS, 2007, 104(44):17465–17470.

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

Out

put (

A.U

.)

Normally distributed thresholds

mean=0.5, std=0.05Ave

rage

of 2

000

step

func

tions

log (x)

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

Out

put (

A.U

.)A

n ex

ampl

e st

ep fu

nctio

n

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

Out

put (

A.U

.)A

n ex

ampl

e M

M fu

nctio

n

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

Out

put (

A.U

.)

Sum of 2000 MM functions Normally distributed KMs, mean=0.1, std=0.01

Ave

rage

of 2

000

MM

func

tions

log (x)

β-catCk2 TLE1

in C2C12 mouse myoblast cells

S W

ang

& K

A J

ones

Minimal model

Cytoplasmic β-cat nuclear(APC)P

Wnt

Nuclear β-cat

transcription

repression

Histone modifications

SET1 H3K4me

β-cat:Pygo bind H3K4me, dissociate from TCF

export /

CtBP

dela

yed

/ slo

wdegradation

X

YX

in

X

Y

in=low in=high

4321

4

21

,,

...1

.3

..1

.

kkkk

ykxkink

inkdtdy

xkink

inkdtdx

xj

i

<

−++

=

−+

=

(x slower than Y)

Nkd1 RNAi ↓ β-catenin

Nkd1 OE ↓ β-catenin

Wnt

β-ca

teni

n

Dvl

β-ca

teni

n

Axin1

β-ca

teni

n

GSK3β

β-ca

teni

n

Nkd1l

β-ca

teni

n

Axin2

β-ca

teni

n

Predicted steady state β-catenin levels

Dashed red lines indicate model’s un-optimized operating point

Gujral & MacBeath, PLoS ONE, 2010, 5(4):e10024.

HEK293

Nkd1:Dvl3 aggregation

Nkd1 Dvl3

HEK293 cells+ rWnt3a

Nkd1

HEK293 cellsNkd1:Dvl3 aggregation (Guo et al)

Guo et al, PLoS ONE, 2009, 4(11):e7982.

Nkd1 localizes to the cell membrane

Nkd1 mutations blocking Dvl interaction increase β-catenin in HEK293 cellsG. J. Lau, MS Thesis, Dept Biochem, U Toronto, 2008

Nkd1 interacts with Axin1B

W M

iller e

t al,

Mol

ecul

ar S

yste

ms

Bio

logy

, 200

9, (

5):3

15

(Over)expressed Nkd1: Dvl sequestration β-catenin/TCF ↓Nkd1 RNAi: Axin1 sequestration ↓ β-catenin/TCF ↓

Gar

net J

ean

Lau,

MS

The

sis,

Dep

t Bio

che

HEK293

m, U

Tor

onto

, 200

8

HEK293

HEK293

(Over)expressed Nkd1: Dvl sequestration β-catenin/TCF ↓Nkd1 RNAi: Axin1 sequestration ↓ β-catenin/TCF ↓

WT NKD1 Nkd1

Axin1+Axin2(+ etc)

Dvl

signalosomeWnt Destruction Complex

β-catenin

NKD1AXIN2

Nkd1

Axin1+Axin2

signalosome

Dvl

Destruction Complex

β‐catenin

NKD1AXIN2

Wnt

Nkd1

Axin1+Axin2

signalosome

Dvl

Destruction Complex

β‐catenin

NKD1AXIN2

Wnt

OE RNAi

Wnt4 inhibits canonical Wnt signaling by redirecting β-catenin to the cell membrane

HEK293T cells transfected with Wnt3a or Wnt4

Bernard et al, Biology of the Cell, 2008, (100):167–177

sFRP2 enhances Wnt signaling & β-catenin nuclearization

HEK293 cells

Von Marschal & Fisher, Biochemical and Biophysical Research Communications, 2010, (400):299–304.

sFRP2 up-regulates Wnt target genes

mR

NA

/ co

ntro

l

HEK293 cells

HEK293 cells

Von Marschal & Fisher, Biochemical and Biophysical Research Communications, 2010, (400):299–304.

But are expressed in response to canonical Wnt

Gujral & MacBeath, PLoS ONE, 2010, 5(4):e10024.

HEK293 cells + 24 hours 200ng/ml purified rWnt3aNon-canonical Wnts do not activate TOPflash

HEK293s transfectedwith indicated ligands

TOP

flash

activ

ity

Lu et al, PNAS, 2004, 101(9):3118–3123.

HEK293 cells + 200ng/ml purified rWnt3a

HEK293 cells + 200ng/ml purified rWnt3a

Gujral & MacBeath, PLoS ONE, 2010, 5(4):e10024.

WT X

Fzd’s sFRP’sCanonical Wnt’s

X

non‐canonical Wnt’sβ-catenin

sFRP’s

non-canonical Wnt’s

Fzd’s

β-catenin

X

Canonical Wnt’s

XsFRP RNAi

Wnt4/5B/11 RNAisFRP’s

X

Fzd’s Canonical Wnt’s

X

non‐canonical Wnt’sβ-catenin

Model prediction: Large under- and over-expression of bothnon-canonical WNT’s and sFRP’s can reduce TOPflash activity

0.2 0.4 0.6 0.8 1.0 1.2 0.2 0.4 0.6 0.8 1.0 1.2Ste

ady-

stat

e le

vel o

f ava

ilabl

e Fz

dre

cept

ors

(A.U

.)RNAi

RNAi

“WT total non-canonical WNT level”“WT SFRP level”

OE

OE

“WT total non-canonical WNT level”“WT SFRP level”

O

O

O

O

O

O

Non-canonical WNT’s, amount (A.U.) sFRP’s , amount (A.U.)

RNAi RNAi OE OE

– Wnt + Wnt – Wnt + Wnt

Ub Core UBE2M, CUL1, DUSP18, βTRCP

SKP1A ↑lower β-cat degradation

↑ ↑

APC ↑AXIN 1/2 ↑ ↑ M ↑ M ↓ M

SIAH1 ↑ M (degrades β-cat)

SMURF1/2 ↓ M (degrades Axin) ↓ (activates PCP)

TCF/LEF1 ↓ ↑ ↑LRPs/FZs ↓

Wnt4/5B/11 ↓ M

sFRP2 ↓ M

WIF1 ↑FRAT1/GNAi1 ↓ ↓ ↑ ↑

PP1γ ↓ (as Axin OE)

DVL1/2/3 ↑ X (see idea) Miram’s ↓ ↑ Miram’s ↑ ↑ Miram’s ↑NKD1 ↓ M ↓ M

GSK3α/β ↓ X GSK3β titration? α↓ ↓β M Miram’s ↑ ↓ M Miram’s ↓CSNK2A1/2 ↑ (β-cat degradation) ↑ (β-cat degradation)

CSNK1ε ↓ (signalosome) ↑ (signalosome) ↑ (signalosome)

CSNK1δ ↑ (in DC) ↓ (in signalosome) ↑ (in signalosome) ↑ (in signalosome)

CTNNB1 ↓ ↓ ↑ ↑

proteasomal degradation of Axin and APC

CSK2 is constitutively active

WT DVL Nkd1

Axin1+Axin2(+ etc)

Dvl

signalosomeWnt Destruction Complex

β-catenin

NKD1AXIN2

Nkd1

Axin1+Axin2

signalosome

Dvl

Destruction Complex

β-catenin

NKD1AXIN2

Wnt

Nkd1

Axin1+Axin2

signalosome

Dvl

Destruction Complex

β-catenin

NKD1AXIN2

Wnt

OE RNAi ?

Interesting hypotheses to test experimentally:

1. Is there a Wnt response threshold in single cells?

2. Do OE and RNAi of sFRPs and non-canonical Wnts both reduce TOPflash activity?

3. Do DVL RNAi and OE both reduce TOPflash activity?

4. Does Nkd1 bind DVL more strongly than Axin?

5. Does over-expressed DVL recruit Axin out of the destruction complex?

Only mechanistically correct models can extrapolate reliability

“The man who has fed the chicken every day throughout its life at last wrings its neck instead, showing that more refined views as to the uniformity of nature would have been useful to the chicken.”

– Bertrand Russell in The Problems of Philosophy, 1912, Chapter 6. .

The Yin-Yang of genetics:

The “a few bad apples” scenario:

A small number of distinct causes underlie the vast majority of cases of any disease.

Once we know the perpetrator(s), the same intervention(s) can be applied to all cases.

The Anna Karenina scenario:(Every unhappy family is unhappy in its own way.)

Dysregulation is the result of different combinations of causes in each individual.

Interventions must be individualized.

( )

( )( )

( )

( )

( )

( )

( )

( )( )

( )

( )

( )

( ) ( )

( )

( ) ( )

Wnt pathway gene expression in 108

normal and tumor prostate samples from 54 patients.

Wissman et al, J Pathology , 2003, (201): 204–212.

DNA sequence variations affecting cellular signaling genes in two individuals

overlapWatson total Venter total (non-synonymous + frameshift)

VEGF Wnt

18 32 15 29 26 27

Watson Venter Watson Venter(883) (1273) total entries

in dbSNP

Notch Phosphatidylinositol TCR

17 3 16 21 34 27 27 22 20

Watson Venter Watson Venter Watson Venter(1250) (1010) (1141)

Ca

49 53 46

Watson Venter(2740)

Hh

7 17 15

Watson Venter(501)

BCR

26 29 23

Watson Venter(805)

ErbB

17 15 12

Watson Venter(1122)

JakStat

55 22 25

Watson Venter(1468)

MAPK

63 66 61

Watson Venter(3792)

mTOR

11 9 9

Watson Venter(1068)

TLR

23 20 19

Watson Venter(1033)

TGFβ

7 20 8

Watson Venter(720)

(Stephen Quake’s genome)

Thanks to my collaborators:

Laboratories of Randy Moon (HHMI) and Andy Chien at UW

Jason Berndt, Travis Biechele, Richard James, et al.

Thanks to my collaborators:

Laboratories of Randy Moon (HHMI) and Andy Chien at UW

Jason Berndt, Travis Biechele, Richard James, et al.