Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

GTP (or the 10-fold less abundant GDP) can subsequently

displace Sos from Ras.

The different families of small G proteins interact with

different classes of GEFs, whose catalytic domains share

no sequence similarity and are structurally unrelated [e.g.,

rhodopsin (Fig. 19-16) is a GEF for G

t␣

]. Nevertheless,

many of these GEFs share the same general mechanism for

promoting GDP–GTP exchange, which suggests that this

mechanism arose on several occasions through convergent

evolution.

i. GAPs Function to Turn Off Ras-Mediated Signals

Ras hydrolyzes its bound GTP with a rate constant of

0.02 min

⫺1

(vs 2–3 min

⫺1

for G

␣

subunits), too slowly for ef-

fective signal transduction. This led to the discovery of a

120-kD GTPase activating protein, RasGAP, that, on bind-

ing Ras ⴢ GTP, accelerates the rate of GTP hydrolysis by a

factor of 10

. RasGAP’s physiological importance as a

regulator of Ras-mediated signal transduction is demon-

strated by the observation that the relative biological

activities of Ras mutants are better correlated with their

resistance to regulation by RasGAP than by their intrinsic

GTPase activity.

The mechanism whereby RasGAP activates the GTPase

activity of Ras was revealed by the X-ray structure, deter-

mined by Alfred Wittinghofer, of the 334-residue GTPase-

activating domain of RasGAP (GAP334) bound to Ras in

its complex with GDP and AlF

(Fig. 19-39). GAP334,

which consists of two all-helical domains, interacts with

Ras over an extensive surface that includes its Switch I and

II regions. The AlF

, which has trigonal planar symmetry,

binds to Ras at the expected position of GTP’s ␥ phosphate

group, with the Al atom opposite a bound water molecule

that presumably would be the attacking nucleophile in the

GTPase reaction. Since Al¬F and P¬O bonds have simi-

lar lengths and phosphoryl-transfer reactions occur via a

trigonal bipyramidal transition state (Fig. 16-6b), the

GDP–AlF

–H

O assembly presumably resembles the

GTPase reaction’s transition state, with the AlF

mimicking

the planar PO

group. Note that Ras ⴢ GDP does not by

itself bind AlF

GAP334 binds to Ras with GAP334’s exposed so-called

finger loop inserted into the Ras active site such that the

finger loop’s Arg 789 side chain interacts with both the

Ras-bound GDP’s ␤ phosphate and the AlF

(Fig. 19-39).

In Ras ⴢ GTP, this Arg side chain would be in an excellent

position to stabilize the developing negative charge in the

GTPase reaction’s transition state. Indeed, the catalytically

more efficient G

␣

subunits contain an Arg residue (Arg 178

in G

i␣

), whose guanidinium group occupies a nearly identi-

cal position (in G

s␣

, this is the Arg side chain that is ADP-

ribosylated by cholera toxin; Section 19-2Cd). The main

chain carbonyl O of Arg 789 hydrogen bonds to the side

chain N of Ras’ catalytically important Gln 61. The O of

this side chain is thereby positioned to hydrogen bond with

the nucleophilic water molecule while its NH

group inter-

acts with an F atom of AlF

(Fig. 19-37), an arrangement

that presumably stabilizes the GTPase reaction’s transition

state.

j. Oncogenic Mutants of Ras Are GAP-Insensitive

Mutations in Ras of Gly 12 and Gln 61 are its most com-

mon oncogenic mutations (an oncogenic form of which is

found in ⬃30% of human cancers). These mutations pre-

vent RasGAP from activating Ras to hydrolyze its bound

GTP and hence lock Ras into its active conformation. The

foregoing X-ray structure reveals why these mutants are

GAP-insensitive. Gly 12 is in such close proximity to the

finger loop that even the smallest possible residue change

(to Ala) would sterically interfere with the geometry of the

transition state through steric clashes with the main chain

of Arg 789 (of RasGAP) and the side chain NH

of Gln 61.

The observation that Gly 12 mutants of Ras bind GTP with

nearly wild-type affinity therefore suggests that larger

side chains can be tolerated at Ras residue 12 in the

Ras–RasGAP Michaelis complex but not in the transition

state. The apparent participation of Gln 61 in transition

state stabilization confirms that this residue has an essen-

tial role in catalysis.

Section 19-3. Tyrosine Kinase–Based Signaling 711

Figure 19-39 X-ray structure of the GAP334 ⴢ Ras ⴢ GDP ⴢ

AlF

complex. The active site regions of the proteins are

shown as ribbons with Ras cyan, its Gly 12 magenta, GAP334

yellow, and its finger loop red.The GDP, AlF

, and the side

chains of Ras Asn 61 and GAP334 Arg 789 are drawn in stick

form with C green, N blue, O red, F yellow-green, P orange, and

Al pink; the nucleophilic water molecule is represented by a red

sphere; and hydrogen bonds are shown as dashed lines. [Based

on an X-ray structure by Alfred Wittinghofer, Max-Planck-Institut

für Molekulare Physiologie, Dortmund, Germany. PDBid

1WQ1.]

JWCL281_c19_671-743.qxd 7/1/10 1:12 PM Page 711

D. MAP Kinase Signaling Cascades

The signaling pathway downstream of Ras consists of a

linear series of Ser/Thr kinases that form a so-called MAP

kinase cascade (Fig. 19-40). Many of the proteins that par-

ticipate in MAP kinase cascades are the products of proto-

oncogenes:

1. Raf, a Ser/Thr protein kinase, is activated by direct

interaction with Ras ⴢ GTP (although other signaling

712 Chapter 19. Signal Transduction

Figure 19-40 The Ras-activated MAP kinase cascade. This

signaling cascade begins when an RTK binds its cognate growth

factor, thereby inducing the autophosphorylation of this RTK’s

cytosolic domain. Grb2/Sem-5 binds to the resulting

phosphoTyr-containing peptide segment via its SH2 domain

and simultaneously binds to Pro-rich segments on Sos via its two

SH3 domains.This activates Sos as a guanine nucleotide

exchange factor (GEF) to exchange Ras’ bound GDP for GTP,

which activates Ras to bind to Raf. Then, Raf, a Ser/Thr kinase,

Myc

Transcription

factors

Gene expression

Nucleus

DNA

Jun

P PYY

Sos

Other effectors

Other

effectors

Other effectors

Sos

Other

effectors

(GAPs)

GAPs Cytosol

Extracellular

medium

Kinase

cascade

Grb2/

Sem–5

SH3

domain

SH3

domain

SH2

domain

Protein

growth

factor

RTK

Raf

MEK

MAPK

Other

kinases

Fos

Jun

Plasma membrane

Ras

(inactive)

GTP

Ras

(active)

GDP

phosphorylates MEK, which in turn phosphorylates MAPK,

which then migrates to the nucleus, where it phosphorylates

transcription factors such as Fos, Jun, and Myc, thereby

modulating gene expression.The MAP kinase cascade eventually

returns to its resting state through the actions of protein

phosphatases (Section 19-3F) after a GTPase activating protein

(GAP) deactivates Ras by inducing it to hydrolyze its bound

GTP to GDP. [After Egan, S.E. and Weinberg, R.A., Nature 365,

782 (1993).]

See the Animated Figures

JWCL281_c19_671-743.qxd 6/4/10 10:54 AM Page 712

pathways may activate Raf by phosphorylating it at multi-

ple Ser and Thr residues;see below).The X-ray structure of

the Ras homolog Rap1A in complex with GMPPNP and

the Ras binding domain of Raf (RafRBD), determined by

Wittinghofer, reveals that the two proteins associate

largely by mutually extending their antiparallel  sheets

across a mainly polar interface (Fig. 19-41). Although

Ras  GTP has 1300-fold greater affinity for binding to Raf

than does Ras  GDP, it is unclear from this structure how

GTP hydrolysis by Ras affects the Ras–Raf interface. Quite

possibly the conformational change in Ras Switch I per-

turbs the Ras–Raf interface to the point that it dissociates.

2. Activated Raf phosphorylates a protein alternatively

known as MEK and MAP kinase kinase (MKK) at specific

Ser and Thr residues, thereby activating it as a Ser/Thr kinase.

[Raf is therefore a MAP kinase kinase kinase (MKKK)].

3. Activated MEK phosphorylates a family of proteins

named mitogen-activated protein kinases (MAP kinases or

MAPKs) or extracellular-signal-regulated kinases (ERKs).

For more than marginal activation, a MAPK must be phos-

phorylated at both its Thr and Tyr residues in the sequence

Thr-X-Tyr. MEK (which stands for MAP kinase/ERK-

activating kinase) catalyzes both phosphorylations and

thus has dual specificity for Ser/Thr and Tyr. The X-ray

structure of the unphosphorylated MAP kinase ERK2, de-

termined by Elizabeth Goldsmith, reveals that this protein

structurally resembles other protein kinases of known

structure and that its Tyr residue that becomes phosphoryl-

ated blocks the peptide-binding site in its unphosphoryl-

ated form.

4. The activated MAPKs phosphorylate a variety of cy-

toplasmic and membrane-associated proteins, including

Sos and EGFR, at Ser/Thr-Pro motifs. In addition, the

MAPKs migrate from the cytosol to the nucleus, where

they phosphorylate a large variety of transcription factors

including Jun/AP-1, Fos, and Myc. These activated tran-

scription factors, in turn, induce the transcription of their

target genes (Section 34-4Bd). The effects commissioned

by the extracellular presence of the protein growth factor

that initiated the signaling cascade are thereby produced.

MAP kinase cascades can be activated in other ways be-

sides by liganded RTKs. For example, Raf may also be acti-

vated via its Ser/Thr phosphorylation by protein kinase C,

which is activated via the phosphoinositide signaling sys-

tem described in Section 19-4. Alternatively, Ras may be

activated by subunits of certain heterotrimeric G proteins.

Thus, the MAP kinase cascade serves to integrate a variety

of extracellular signals.

a. Scaffold and Anchoring Proteins Organize

and Position Protein Kinases

Eukaryotic cells contain numerous different MAPK

signaling cascades, each with a characteristic set of com-

ponent kinases, which in mammals comprise at least 14

MKKKs, 7 MKKs, and 12 MAPKs (Fig. 19-42). Although

each MAPK is activated by a specific MKK, a given MKK

can be activated by more than one MKKK. Moreover, sev-

eral pathways may be activated by a single type of recep-

tor. How then does a cell prevent inappropriate cross talk

between closely related signaling pathways? One way that

this occurs is through the use of scaffold proteins, proteins

that bind some or all of the component protein kinases of a

particular signaling cascade so as to ensure that the protein

kinases of a given pathway interact only with one another.

In addition, a scaffold protein can control the subcellular

location of its associated kinases.

The first known scaffold protein was discovered through

the genetic analysis of a MAP kinase cascade in yeast,

which demonstrated that this protein, Ste5p, binds the

MKKK, MKK, and MAPK components of the pathway

and that, in vivo, the scaffold’s absence inactivates the

pathway. Evidently, the interactions between successive

kinase components of this MAP kinase cascade are, by

themselves, insufficient for signal transmission.

Section 19-3. Tyrosine Kinase–Based Signaling 713

Figure 19-41 X-ray structure of the Ras binding domain of

Raf (RafRBD; orange) in complex with Rap1A  GMPPNP

(blue). The Switch I and II regions of Rap1A are magenta and

green, and its bound GMPPNP is shown in space-filling form

with C green, N blue, O red, and P yellow. Rap1A  GMPPNP

and Ras  GMPPNP have nearly identical structures. [Based on

an X-ray structure by Alfred Wittinghofer, Max-Planck-Institut

für Molekulare Physiologie, Dortmund, Germany. PDBid

1GUA.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 713

JIP-1 (for JNK interacting protein-1; Fig. 19-43a) is a

scaffold protein that simultaneously binds HPK1 (for

hematopoietic progenitor kinase-1), a Ras analog and

hence an MKKK kinase (MKKKK); the MKKKs MLK3

and DL3; MKK7; and the MAPK JNK (for Jun N-terminal

kinase). MEKK1 is a somewhat different type of scaffold

protein (Fig. 19-43b); this functional MKKK binds its sub-

strate, MKK4, as well as the latter’s substrate, JNK.

Protein Ser/Thr kinases may similarly be individually

tethered to their sites of action by anchoring proteins. For

example, protein kinase A (PKA), which participates in nu-

merous parallel signaling pathways including that regulat-

ing glycogen metabolism (Section 18-3), associates with

several unrelated so-called A-kinase anchoring proteins

(AKAPs). The different AKAPs, which all bind the re-

gulatory (R) subunits of PKA, target PKA to different

subcellular locations (e.g., to vesicle or plasma membranes

or to particular receptors) and may also bind other signaling

proteins (e.g., PP1, the protein phosphatase that removes

phosphate groups installed by PKA; Section 18-3Cg), thus

functioning to integrate intracellular signals.

b. Anthrax Lethal Factor Specifically

Cleaves MAPKKs

Anthrax, an infectious disease caused by the bacterium

Bacillus anthracis, affects mainly herbivorous animals such

as cattle, sheep, and goats. On rare occasions, however, it

may be transmitted to humans (but not between humans),

in whom it is often fatal, if untreated, through massive

714 Chapter 19. Signal Transduction

Figure 19-43 Some examples of scaffold proteins that

modulate mammalian MAP kinase cascades. (a) JIP-1 binds all

protein components of the MAP kinase cascade in which HPK1

phosphorylates MKL3 or DLK (MKKKs), which then

phosphorylates MKK7, which then phosphorylates JNK (an

MAPK). (b) MEKK1 (an MKKK) is the kinase for MKK4 and

also binds JNK, MKK4’s target MAPK. [After Garrington,T.P.

and Johnson, G.L., Curr. Opin. Cell Biol. 11, 213 (1999).]

Figure 19-42 MAP kinase cascades in mammalian cells. Each

MAP kinase cascade consists of an MKKK, an MKK, and an

MAPK.Various external stimuli may each activate one or more

MKKKs, which in turn may activate one or more MKKs.

However, the MKKs are relatively specific for their target

MAPKs.The activated MAPKs phosphorylate specific

Growth factor

Extracellular

stimulus:

Stress, growth factor,

differentiation factor

Transcription

factors and

other kinases:

Cellular

response:

MEKK4,

DLK

TAK1, ASK1,

MLK3

PAKMKKK

MEKK1-3,

Tpl-2

Raf-1, A-Raf,

B-Raf, Mos

MKK3, MKK6MKK4, MKK7MKK5

MKK

MEK1,

MEK2

JNK1, JNK2,

JNK3

p38α, p38β,

p38γ, p38δ

ERK5

MEF2C

c-Jun

MAPK

ERK3,

ERK4

ERK1,

ERK2

c-Jun, ATF-2, Elk-1, p53,

DPC4, NFAT4

MAPKAP kinase, ATF-2,

Elk-1, Chop, Max, MEF2C

p90

rsk

, S6 kinase, Sos,

phospholipase A

EGF receptor, Elk-1, Ets1,

Sap1a, c-Myc, Tal, STATS

Growth, differentiation,

survival, apoptosis

Cytokine production,

apoptosis

Growth,

differentiation

Stress

MLK3 or DLK

Cytokines, stress, etc. Stress, etc.

HPK1

MKK7

JNK

Response

JNK

MKK4

MEKK1JIP1

(a) (b)

transcription factors (e.g., Elk-1, Ets1, p53, NFAT4, Max) as well

as specific kinases (e.g., p90

rsk

, S6 kinase, MAPKAP kinase). The

resulting activated transcription factors and kinases then induce

cellular responses such as growth, differentiation, and apoptosis

(programmed cell death; Section 34-4E). [After Garrington, T.P.

and Johnson, G.L., Curr. Opin. Cell Biol. 11, 212 (1999).]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 714

septic shock (Section 19-1Lb). Anthrax spores are signifi-

cant agents of biological warfare because their inhalation

results in inhalational anthrax, a form of the disease that is

nearly always fatal. This is because by the time the symp-

toms of inhalational anthrax become apparent,the bacterial

infection has already released so much toxin that eliminat-

ing the infection through antibiotic treatment does not

reverse the progress of the disease.

Anthrax toxin consists of three proteins that act in con-

cert: protective antigen (PA), lethal factor (LF), and

edema factor (EF; oedema factor in British English). PA,

which is named for its use in vaccines, is a 735-residue, 4-

domain protein that binds to its host cell-surface receptor

(a single pass transmembrane protein) via its C-terminal

domain. Most of PA’s N-terminal domain is then cleaved

away by a cell-surface protease, whereon the remaining

membrane-bound portions of PA form cyclic heptamers

reminiscent of the cyclic pentamers formed by cholera

toxin (Fig. 19-22).The heptameric PA then binds LF and/or

EF by their homologous N-terminal domains and mediates

their endocytotic uptake into the cell. Indeed, the intra-

venous administration of only PA and LF rapidly kills ani-

mals. EF is a calmodulin-activated adenylate cyclase whose

action upsets water homeostasis and hence is probably re-

sponsible for the massive edema (abnormal buildup of in-

tercellular fluids) seen in cutaneous anthrax infections.

LF is a 776-residue monomeric protease that has only

one known cellular target: It cleaves members of the

MAAPKK family of proteins near their N-termini so as to

excise the docking sequences for their cognate downstream

MAPKs. It thereby disrupts the signal transduction path-

ways in which these proteins participate. However, anthrax

infection targets mainly macrophages, a type of white

blood cell (mice whose blood has been depleted of

macrophages are resistant to anthrax). Low levels of LF,

which occur in early stages of anthrax infection, cleave

MAPKK-3, which inhibits macrophages from releasing but

not producing the inflammatory mediators NO (Section

19-1Lb) and tumor necrosis factor- (TNF; a cytokine

that has opposite effects to most protein growth factors

and is largely responsible for the wasting seen in chronic

infections). This has the effect of reducing and/or delaying

the immune response. In contrast, high levels of LF, which

occur in late stages of infection, trigger macrophage lysis,

causing the sudden release of NO and TNF-, which pre-

sumably results in the massive septic shock that causes

death.

E. Tyrosine Kinase–Associated Receptors

Many cell-surface receptors are not members of the recep-

tor families that we have discussed and do not respond to

ligand binding by autophosphorylation. These include the

receptors for the cytokines (Section 19-3Eb) and T cell re-

ceptors [which control the proliferation of immune system

cells known as T lymphocytes (T cells); Section 35-2D].

Ligand binding induces these tyrosine kinase–associated

receptors to dimerize (and, in some cases, to trimerize or

form even higher oligomers), often with different types of

subunits, in a way that activates associated nonreceptor ty-

rosine kinases (NRTKs). The domain organization for the

major subfamilies of NRTKs is diagrammed in Fig. 19-44.

a. The Structure of Src Reveals Its

Autoinhibitory Mechanism

Many of the NRTKs that are activated by tyrosine

kinase–associated receptors belong to the Src family, which

contains at least nine members including Src, Fyn, and

Lck. Most of these ⬃530-residue membrane-anchored (by

myristoylation) proteins have both an SH2 and an SH3

domain and all have a PTK domain. Hence, an Src-related

kinase may also be activated by association with an au-

tophosphorylated RTK. Although Src-related kinases are

each associated with different receptors, they phosphorylate

Section 19-3. Tyrosine Kinase–Based Signaling 715

Figure 19-44 Domain organization of the major NRTK

subfamilies. The N-termini for these polypeptides, which are

drawn approximately to scale, are on the left and the domain

Abl

tyrosine kinase

pseudo-kinase

SH3

SH2

integrin binding

focal adhesion binding

DNA binding

F-actin binding

pleckstrin homology

Jak homology

Zap70

Jak

Btk

Fak

Src

Csk

Fes

identification is provided on the right. [Courtesy of Stevan

Hubbard, New York University School of Medicine.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 715

overlapping sets of target proteins. This complex web of

interactions explains why different ligands often activate

some of the same signaling pathways.

Src, as indicated in Fig. 19-44, consists of, from N- to

C-terminus, a myristoylated N-terminal “unique” domain

that differs among Src family members, an SH3 domain, an

SH2 domain, a PTK domain, and a short C-terminal tail.

Phosphorylation of Tyr 416 in the PTK’s activation loop

activates Src, whereas phosphorylation of Tyr 527 in its

C-terminal tail deactivates it. In vivo, Src is phosphorylated

at either Tyr 416 or Tyr 527, but not at both. The dephos-

phorylation of Tyr 527 or the binding of external ligands to

the SH2 or the SH3 domain activates Src, a state that is

then maintained by the autophosphorylation of Tyr 416.

When Tyr 527 is phosphorylated and no activating phos-

phopeptides are available, Src’s SH2 and SH3 domains

function to deactivate its PTK domain, that is, Src is then

autoinhibited.

The X-ray structure of Src  AMPPNP lacking its

N-terminal domain and with Tyr 527 phosphorylated, deter-

mined by Stephen Harrison and Michael Eck, reveals the

structural basis of Src autoinhibition (Fig. 19-45). As bio-

chemical studies had previously shown, the SH2 domain

binds phosphoTyr 527, which occurs in the sequence

pYNPG rather than the pYEEI sequence characteristic of

high affinity Src SH2 target peptides. Although the pYNP

segment binds to SH2 as does the pYEE segment in Fig.

19-33b, the succeeding residues are poorly ordered in the

X-ray structure and, moreover, the SH2 pocket in which

the Ile side chain of pYEEI binds is unoccupied. Appar-

ently, the phosphoTyr 527-containing peptide segment

binds to the Src SH2 domain with reduced affinity relative

to its target peptides.

The SH3 domain binds to the linker connecting the SH2

domain to the N-terminal lobe of the PTK domain.

Residues 249 to 253 of this linker form a polyproline II he-

lix that binds to the SH3 domain in much the same way as

do SH3’s Pro-rich target peptides (Fig. 19-33). However,

the only Pro in this segment is residue 250. The polar side

chain of Gln 253, which occupies the position of the second

Pro in SH3’s normal Pro-X-X-Pro target sequence, does

not enter the hydrophobic binding pocket that this second

Pro would occupy (Fig. 19-35), and hence the path of the

peptide deviates from that of Pro-rich target peptides at

this point. Apparently, this interaction is also weaker than

those with Src’s SH3 target peptides.

Src’s SH2 and SH3 domains bind on the opposite side of

the PTK domain from its active site. How, then, does the

conformation shown in Fig.19-45 inhibit the PTK’s activity?

The two lobes of Src’s PTK domain are, for the most part,

closely superimposable on their counterparts in the PTK

domain of phosphorylated and hence activated Lck (a Src

family member) as well as the C subunit of activated PKA

(Fig. 18-15). However, Src helix C (the only helix in the

PTK’s N-terminal lobe) is displaced from the interface be-

tween the N- and C-terminal lobes relative to its counter-

parts in Lck and PKA. Helix C contains the conserved

residue Glu 310 (using Src numbering), which in activated

Lck and PKA projects into the catalytic cleft,where it forms

a salt bridge with Lys 295, an important ligand of the sub-

strate ATP’s  and  phosphates. However, in inactive Src,

Glu 310 forms an alternative salt bridge with Arg 409,

whereas Lys 295 instead interacts with Asp 404. In activated

Lck,Arg 409 forms a salt bridge with phosphoTyr 416.

The foregoing structural observations suggest the fol-

lowing scenario for Src activation (Fig. 19-46):

1. The dephosphorylation of Tyr 527 and/or the binding

of the SH2 and/or SH3 domains to their target peptides

(for which SH2 and SH3 have greater affinity than their in-

ternal Src binding sites) releases these domains from their

PTK-bound positions shown in Fig. 19-45, thus relaxing

conformational constraints on the PTK domain. This al-

lows the PTK’s active site cleft to open, thereby disrupting

the structure of its partially helical activation loop (which

occupies a blocking position in the active site cleft; Fig.

19-45) so as to expose Tyr 416 to autophosphorylation.

2. The resulting phosphoTyr 416 forms a salt bridge

with Arg 409, which sterically requires the structural reor-

ganization of the activation loop to its active, nonblocking

conformation.The consequent rupture of the Glu 310–Arg

409 salt bridge frees helix C to assume its active orientation

which, in turn, allows Glu 310 to form its catalytically im-

portant salt bridge to Lys 295, thereby activating the Src

PTK activity.

716 Chapter 19. Signal Transduction

Figure 19-45 X-ray structure of Src  AMPPNP lacking its

N-terminal domain and with Tyr 527 phosphorylated. The SH3

domain is orange, the SH2 domain is magenta, the linker joining

the SH2 domain to the PTK domain is green with its 5-residue

polyproline II helix gold, the N-terminal lobe of the PTK domain

is pink, the C-terminal lobe is cyan with its activation loop blue,

and the C-terminal tail is orange. The AMPPNP is shown in

space-filling form and Y416 and pY527 are shown in ball-and-stick

form, all with C green, N blue, O red, and P yellow. [Based on an

X-ray structure by Stephen Harrison and Michael Eck, Harvard

Medical School. PDBid 2SRC.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 716

The above mechanism is, perhaps unexpectedly, criti-

cally dependent on the rigidity of the 8-residue linker join-

ing the SH2 and SH3 domains. Thus, replacing three of

these linker residues with Gly (whose lack of a C



atom

makes it the least conformationally restricted residue) re-

sults in a protein that is no longer deactivated by the phos-

phorylation of Tyr 527. This is corroborated by molecular

dynamics simulations (Section 9-4) indicating that the ther-

mal motions of the SH2 and SH3 domains are highly corre-

lated (move as a unit) when Tyr 527 is phosphorylated but

that this correlation is significantly reduced when Tyr 527 is

dephosphorylated or when Gly replaces the three linker

residues.

b. The JAK-STAT Pathway Relays

Cytokine-Based Signals

Cytokines form a diverse group of small soluble pro-

teins that, when secreted by cells, act in an autocrine,

paracrine, or endocrine fashion to induce a great variety of

responses, including the immune response (Section 35-2Aa),

cell proliferation, growth, differentiation, apoptosis (pro-

grammed cell death; Section 34-4E), and chemotaxis (the

movement of a motile cell along a concentration gradient

of a specific substance).They include the 35 different inter-

leukins (IL-1 to IL-35), the colony stimulating factors [which

include macrophage colony-stimulating factor (M-CSF),

granulocyte colony-stimulating factor (G-CSF), and granu-

locyte macrophage colony-stimulating factor (GM-CSF);

macrophages and granulocytes are types of white blood

cells], growth hormone (Section 19-1J), erythropoietin

(EPO; which stimulates the production of erythrocytes),

the interferons (which protect against viral infection; Sec-

tion 32-4Ab), the tumor necrosis factors (TNFs), nerve

growth factor (NGF), and the chemokines (which induce

chemotaxis in responsive cells). Abnormalities in specific

cytokines or their receptors have been implicated in a large

variety of diseases and, conversely, several cytokines are

therapeutically useful in alleviating certain pathological

states [e.g., EPO is used to treat anemia resulting from

chronic kidney disease (EPO is produced by the kidneys)

and GM-CSF is used to stimulate white cell production fol-

lowing chemotherapy (which kills fast-growing cells such

as white cells)].

Most cytokine receptors, as we saw for human growth

hormone (Section 19-1J), are activated through ligand-

induced receptor aggregation of two or more receptor

components. For example, the GM-CSF receptor consists

of two different types of subunits: GMR, which is specific

for GM-CSF, and 

, which is a common subunit for the

GM-CSF, IL-3, and IL-5 receptors. Both of these subunits

consist of an N-terminal cytosolic domain, a single trans-

membrane helix, and a C-terminal ectodomain, which in

GMR consists of two fibronectin type III domains (Sec-

tion 19-1J) and in 

consists of four such domains.

The X-ray structure of GM-CSF in complex with its re-

ceptor’s ectodomain, determined by Angel Lopez and

Section 19-3. Tyrosine Kinase–Based Signaling 717

Figure 19-46 Schematic model of Src activation. See the text

for an explanation.The coloring scheme largely matches that in

Fig. 19-45, as does the viewpoint. [After Young, M.A.,

Autoinhibited Form Active Form

SH3–SH2

linker

ATP

SH3

SH2

PTK domain

(N-lobe)

Lys 295

αC

Glu 310

Tyr 416

PTK domain

(C-lobe)

Tyr 527

Src

activato

Substrate

Arg 409

Lys 295

Gonfloni, F., Superti-Furga, G., Roux, B., and Kuriyan, J., Cell

105, 115 (2001).]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 717

Michael Parker, reveals that it consists of a 2:2:2 complex

of GM-CSF, GMR, and 

(Fig. 19-47). GM-CSF, as had

previously been determined, consists of an up–up–down–

down four-helix bundle, a topology that occurs only in

helical cytokines. Each GM-CSF molecule interacts with

both domains of a GMR subunit and with the N-terminal

domain of one 

subunit (D1) and with the C-terminal do-

main of the other 

subunit (D4).The latter phenomenon,

in which the domain from one subunit is exchanged with

the same domain from an identical subunit to form an in-

tertwined dimer, is known as domain swapping. It is a com-

mon mechanism for oligomer assembly.

Although the X-ray structures of the ectodomains of

several cytokine receptors have been determined, the

structures of their cytosolic domains are, as yet, unknown.

However, as James Darnell elucidated, the signal that heli-

cal cytokines have been bound by their cognate receptors

is transmitted within the cell by the JAK-STAT pathway.

These cytokine receptors form complexes with proteins of

the Janus kinase (JAK) family of NRTKs, so named be-

cause each of its four ⬃1150-residue members (JAK1,

JAK2, JAK3, and Tyk2) has two PTK domains (Janus is

the two-faced Roman god of gates and doorways),

although only the C-terminal domain is functional (Fig.

19-44). STATs (for signal transducers and activators of

transcription) comprise a family of seven ⬃800-residue

proteins that are the only known transcription factors

whose activities are regulated by Tyr phosphorylation and

that have SH2 domains.

The JAK-STAT pathway functions as is diagrammed in

Fig. 19-48:

1. Cytokine binding induces the cytokine receptor to

oligomerize.

2. The cytokine receptor’s two associated JAKs are

thereby brought into apposition (in the case of the GM-CSF

receptor,JAK2 binds to the cytosolic domain of 

), where-

upon they reciprocally phosphorylate each other and then

their associated receptors, a process resembling the au-

tophosphorylation of dimerized RTKs (Section 19-3Ab).

Note that unlike most NRTKs, JAKs lack both SH2 and

SH3 domains.

3. STATs bind to the phosphoTyr group on their cog-

nate activated receptor via their SH2 domain and are then

phosphorylated on a conserved Tyr residue by the associ-

ated JAK.

4. Following their dissociation from the receptor, the

phosphorylated STATs homo- or heterodimerize via the

association of their phosphoTyr residue with the SH2 do-

main on the opposing subunit.

5. The STAT dimers are translocated to the nucleus,

where these now functional transcription factors induce

the expression of their target genes in much the same way

as do the transcription factors that are phosphorylated by

the MAPKs (Fig. 19-40).

c. PTKs Are Targets of Anticancer Drugs

The hallmark of chronic myelogenous leukemia (CML)

is a specific chromosomal translocation (Section 34-4C)

forming the so-called Philadelphia chromosome in which

the Abl gene (which encodes the NRTK Abl) is fused with

the Bcr gene (which encodes the protein Ser/Thr kinase

Bcr). The Abl portion of this Bcr–Abl fusion protein is con-

stitutively (continuously, without regulation) activated,

probably because its Bcr portion oligomerizes. Hematopoi-

etic stem cells (from which all blood cells are descended)

bearing the Philadelphia chromosome are therefore primed

to develop CML (malignancy requires several independent

genetic alterations; Section 19-3Ba). Without a bone mar-

row transplant (a high-risk procedure that is unavailable to

718 Chapter 19. Signal Transduction

(a)

(b)

Figure 19-47 X-ray structure of the extracellular portions of

the GM-CSF receptor in complex with GM-CSF. The structure of

this 2-fold symmetric 2:2:2 complex of GM-CSF, GMR, and 

is viewed (a) parallel to the plane of the plasma membrane and

(b) from the extracellular side of the membrane. The GM-CSF is

purple, the GMR is yellow, one 

monomer (chain a) is

magenta and the other (chain b) is green.The labels indicate the

domain names.The observed N-linked carbohydrates are drawn

in stick form with Asn C green, carbohydrate C gray, N blue, and

O red.The disordered peptides that link the C-termini of the

receptor chains to their transmembrane segments are

represented by dashed lines. [Modified from a drawing by Angel

Lopez, Hanson Institute,Adelaide,Australia, and Michael

Parker, St. Vincent’s Institute of Medical Research, Fitzroy,

Victoria, Australia. PDBid 3CXE.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 718

which was developed by Brian Druker and Nicholas

Lydon, has caused the remission of symptoms in ⬃90% of

CML patients with almost no serious side effects. This un-

precedented performance occurs, in part, because Gleevec

is essentially inactive against other PTKs (an exception be-

ing the PDGF receptor) as well as protein Ser/Thr kinases.

Abl resembles Src (Fig. 19-44) but lacks Src’s C-terminal

regulatory phosphorylation site (Figs. 19-45 and 19-46).

The X-ray structure of Abl’s PTK domain in complex with

a truncated form of Gleevec,

Section 19-3. Tyrosine Kinase–Based Signaling 719

most individuals due to the lack of a suitable donor), CML

is invariably fatal, with an average survival time of ⬃6 years.

An inhibitor of Abl would be expected to prevent

the proliferation of, and even kill, CML cells. However,

to be an effective anti-CML agent, such a substance must

not inhibit other protein kinases because this would al-

most certainly cause serious side effects. Derivatives of

2-phenylaminopyrimidine bind to Abl with exceptionally

high affinity and specificity. One such derivative, imatinib

(trade name Gleevec),

Imatinib (Gleevec

)

Figure 19-48 The JAK-STAT pathway for the intracellular relaying of cytokine signals. See the

text for details. [After Carpenter, L.R., Yancopoulos, G.D., and Stahl, N., Adv. Protein Chem. 52,

109 (1999).]

P P

STAT

JAK JAK

STAT

ADPATP

ADP

ATP

JAK JAKJAK JAK

βββ β

122 21

JAK JAK

DNA

Gene expression

Nucleus

Extracellular space

Plasma

membrane

Cytoplasm

Cytokine receptor

subunit

Cytokine receptor

subunit

Cytokine

JAK

Receptors

STAT

JWCL281_c19_671-743.qxd 6/11/10 6:57 AM Page 719

determined by Kuriyan (Fig. 19-49), reveals, as expected,

that the truncated Gleevec binds in Abl’s ATP-binding site

[the piperazinyl group that this inhibitor lacks relative to

Gleevec does not significantly alter its target discrimina-

tion but does increase Gleevec’s solubility and hence its

bioavailability (Section 15-4Ba); it probably binds in a sol-

vent accessible groove at the back of Abl]. Abl thereby

adopts an inactive conformation in which its activation

loop, which is not phosphorylated, appears to mimic the

way in which substrate peptides bind to PTKs (such as the

insulin receptor, Fig. 19-28a); that is, the activation loop as-

sumes an autoinhibitory conformation. As a consequence,

the N-terminal end of the activation loop, which has the

highly conserved sequence Asp-Phe-Gly (whose Asp side

chain, in the active PTK, ligates an Mg

2

ion that is essen-

tial for catalysis), assumes a conformation that is quite

different from that observed in the X-ray structures of

inactive Abl and Src (Fig. 19-45) because this latter confor-

mation would block the binding of Gleevec.

Gleevec was the first of several compounds that inhibit

specific protein kinases to be approved by the FDA for

clinical use against certain cancers. In addition, several

monoclonal antibodies (Section 6-1Da) that bind to spe-

cific PTKs or their ligands are in clinical use as anticancer

agents. For example, cetuximab (trade name Erbitux), a

chimeric mouse/human monoclonal antibody, is effective

against colorectal and certain head and neck cancers. It

specifically binds to the ectodomain of EGFR so as to

block its ligand binding and hence prevent its activation,

thus resulting in impaired cell growth and proliferation.

Such receptor-targeted therapies hold enormous promise

for controlling, if not curing, cancers by specifically target-

ing the aberrant proteins that cause the cancers. In con-

trast, most chemotherapeutic agents that are presently in

use indiscriminately kill fast-growing cells and hence al-

most always have debilitating side effects.

d. Increased Chaperone Activity Facilitates Cancer

Epidemiological studies have shown that individuals with

age-related neurodegenerative diseases such as Alzheimer’s

and Parkinson’s diseases have a far lower incidence of can-

cer than the general population. What is the biochemical

basis of this intriguing observation? These neurodegena-

tive diseases are all characterized by the deposition of

plaques containing amyloid fibrils (Section 9-5) in the af-

fected brain cells. In Alzheimer’s disease, these fibrils con-

sist mainly of A protein (Section 9-5B) and in Parkinson’s

disease they contain mainly -synuclein (a 140-residue sol-

uble protein of unknown function that normally occurs in

presynaptic terminals). These amyloid fibrils apparently

form because of age-associated reductions in the levels of

activity of the chaperone proteins that normally prevent

their aggregation.

Many oncogene products, as we have seen, are mutant

forms of proteins that participate in signal transduction and

hence have altered functionality. Such mutant proteins are

usually less stable than their unmutated forms and hence re-

quire greater than normal attention from chaperone pro-

teins to maintain their active conformations. In particular,

the Hsp90 chaperone proteins (Section 9-2C) facilitate the

late stage folding of numerous signaling proteins and, pre-

sumably, their oncogenic mutants. In fact, many unmutated

signaling proteins are unstable unless they are binding their

corresponding ligand or protein and hence are present in

cells in complex with Hsp90 (which accounts for Hsp90’s

high abundance, normally 1–2% of a cell’s soluble proteins).

Hsp90 is overexpressed in many types of cancers, a situ-

ation that is correlated with resistance to therapy and

hence a poor prognosis (and which accounts for the obser-

vation that individuals with neurodegenerative diseases

rarely get cancer). Consequently, Hsp90 inhibitors are anti-

cancer drugs. Several Hsp90 inhibitors, all of which inter-

fere with its ATPase function, are presently in clinical trials.

These substances, unlike most targeted anticancer drugs

(e.g., Gleevec), are likely to be effective against a broad

range of cancers.

720 Chapter 19. Signal Transduction

Figure 19-49 X-ray structure of the Abl PTK domain in

complex with a truncated derivative of Gleevec. The protein is

viewed from the right of the “standard” view of protein kinases

(e.g., Figs. 19-28a and 19-45), with its N-terminal lobe pink, its

C-terminal lobe cyan, and its activation loop blue. The truncated

Gleevec, which occupies the PTK’s ATP-binding site, is shown in

space-filling form with C green, N blue, and O red. [Based on an

X-ray structure by John Kuriyan, The Rockefeller University.

PDBid 1FPU.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 720