Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

Section 19-3. Tyrosine Kinase–Based Signaling 701

Figure 19-27 Schematic diagrams of

RTKs. (a) The PDGF receptor and (b) the

EGF receptor.Their autophosphorylation

sites and the proteins that are activated by

binding to these sites via their SH2 domains

(all of which are discussed in this chapter)

are indicated. Note that almost all of the

autophosphoryated Tyr residues that bind

to other proteins lie outside the tyrosine

kinase domains. [After Pawson, T. and

Schlessinger, J., Curr. Biol. 3, 435 (1993).]

Plasma membrane

Extracellular

Cytosol

SH2 domain

Tyrosine

kinase

domain

Tyrosine

kinase

domain

Tyrosine

kinase

domain

Kinase

insert

sequence

Phospho-Tyr

residue

PI3

kinase

RasGAP

P Src

PDGF

receptor

EGF

receptor

SHP-2

PLC-γ 1

Grb2

Shc

(a) (b)

Figure 19-26 X-ray structure of the 2:2:2 complex of FGF2, the D2–D3 portion of FGFR1,

and a heparin decasaccharide. The view is normal to the dimeric complex’s 2-fold axis of

symmetry, with the plasma membrane below. The polypeptides are represented in

ribbon form with the FGF molecules cyan and blue, the D2 and D3

domains of one FGFR monomer lavender and gold, and those of the other

magenta and orange. The two heparin decasaccharides are shown in

space-filling form with C green, N blue, O red, and S yellow. The D2

and D3 domains of FGFR each have an immunoglobulin fold: a ␤

sandwich comprised of a 3-stranded and a 4-stranded antiparallel

␤ sheet (Section 8-3Bg). [Based on an X-ray structure by Moosa

Mohammadi, New York University School of Medicine. PDBid 1FQ9.]

JWCL281_c19_671-743.qxd 6/4/10 10:53 AM Page 701

19-3D, this causes the activation of the proteins that partic-

ipate in executing the instructions implied by the extracel-

lular presence of the protein growth factor.

c. The Insulin Receptor PTK Undergoes Major

Conformational Changes on Autophosphorylation

How does autophosphorylation activate a PTK? The

comparison of the X-ray structures of the PTK domain of

the insulin receptor in its inactive unphosphorylated and

active triphosphorylated states has done much to answer

this question. The insulin receptor is expressed as a single

1382-residue precursor peptide that is proteolytically

processed to yield the disulfide-linked ␣ and ␤ subunits

(731 and 619 residues) of the mature receptor (Fig. 19-25).

The X-ray structure of a 306-residue PTK-containing seg-

ment of the ␤ subunit that was phosphorylated at its three

autophosphorylation sites, Tyr residues 1158, 1162, and

1163 (using the precursor numbering system), and in com-

plex with the nonhydrolyzable ATP analog AMPPNP and

an 18-residue peptide substrate, was determined by Stevan

Hubbard. It reveals (Fig. 19-28a) that this PTK structurally

resembles other PTKs of known structure as well as pro-

tein Ser/Thr kinases such as PKA (Fig. 18-15) and the phos-

phorylase kinase ␥ subunit (Fig. 18-21).

Comparison of this structure with that of the unphospho-

rylated and uncomplexed protein, determined by Hubbard

and Wayne Hendrickson, reveals that, on phosphorylation

and substrate binding,the PTK’s N-terminal lobe undergoes a

nearly rigid 21° rotation relative to the C-terminal lobe

about the long axis of the protein (Fig. 19-28b). This dra-

matic conformational change closes the active site cleft

about the AMPPNP and, presumably, correctly positions

702 Chapter 19. Signal Transduction

(a)

(b)

Figure 19-28 X-ray structure of the PTK domain of the

insulin receptor. (a) The PTK phosphorylated at Tyr residues

1158, 1162, and 1163 and in complex with AMPPNP and an

18-residue polypeptide substrate. The PTK is shown in the

“standard” protein kinase orientation with its N-terminal domain

pink, its C-terminal domain cyan, and its activation loop light

blue. Its three phosphorylated Tyr side chains are shown in

ball-and-stick form with C green, N blue, O red, and P yellow.

The AMPPNP, which is identically colored, is shown in

space-filling form.The substrate polypeptide is orange (only 6 of

its residues are visible) and its phosphorylatable Tyr residue is

shown in ball-and-stick form with C magenta and O red. (b) The

polypeptide backbones of the phosphorylated and

unphosphorylated forms of the insulin receptor PTK domain as

superimposed on their C-terminal lobes.The phosphorylated

protein is green with its activation loop blue, and the

unphosphorylated protein is yellow with its activation loop red.

The axis (black line) and arrow (blue) indicate the rotation

required to align the N-terminal domain of the unphosphorylated

protein with that of the phosphorylated protein. [Part a based on

an X-ray structure by and Part b courtesy of Stevan Hubbard,

New York University Medical School. PDBids 1IR3 for the

phosphorylated protein and 1IRK for the unphosphorylated

protein.]

See Interactive Exercise 13

JWCL281_c19_671-743.qxd 10/19/10 7:36 AM Page 702

critical residues for substrate binding and catalysis. All

three phosphorylated Tyr residues occur on the PTK’s acti-

vation loop (residues 1149–1170; recall that many Ser/Thr

kinases are also phosphorylated on their activation loops;

Section 18-3Cb). The unphosphorylated activation loop

threads through the PTK’s active site so as to prevent the

binding of both ATP and protein substrates. However, on

phosphorylation, the activation loop assumes a conforma-

tion that does not occlude the active site (Fig. 19-28b) but

instead forms part of the substrate recognition site. The

phosphate group on Tyr 1163 bridges the activation loop by

forming a hydrogen bond with the side chain of the con-

served Arg 1155 on one side of the loop and with the main

chain N of Gly 1166 on its other side.The phosphate group

of Tyr 1162 makes two hydrogen bonds to the side chain of

the conserved Arg 1164. However, the phosphate group of

Tyr 1168 makes no protein contacts, which suggests that it

forms a docking site for downstream signaling proteins

(see below). These observations are in agreement with ex-

periments indicating that the tyrosine kinase activity of the

insulin receptor increases with the degree of phosphoryla-

tion at its three autophosphorylatable Tyr side chains and

that full activity is not achieved until Tyr 1163 is phospho-

rylated. Indeed, nearly all known RTKs have between one

and three autophosphorylatable Tyr residues in their acti-

vation loops (a major exception being the members of

EGFR subfamily; Fig. 19-27b) and, in all phosphorylated

protein kinases of known structure, assume similar confor-

mations.

Only the 6 centrally located residues, GDYMNM, of

the 18-residue substrate peptide are seen in the foregoing

X-ray structure. These include the YMXM sequence found

in all efficient substrates of the insulin receptor. This seg-

ment associates with the kinase as a strand in a  sheet. Its

Met side chains fit into adjacent hydrophobic pockets of

the protein, whereas its phosphorylatable Tyr side chain ex-

tends toward the  phosphate group of the AMPPNP (Fig.

19-28a). The specificity of the protein for phosphorylating

Tyr rather than Ser or Thr is explained by the observation

that the side chain of Tyr, but not those of Ser or Thr, are

long enough to reach the active site.

B. Cancer: The Loss of Control of Growth

Before we continue our discussion of signaling pathways,

let us consider cancer, a group of diseases that is character-

ized by defects in signal transduction causing uncontrolled

growth. Indeed, studies of cancer have greatly increased

our understanding of signal transduction and vice versa.

The cells of the body normally remain under strict devel-

opmental control. For instance, during embryogenesis, cells

must differentiate, proliferate, migrate, and even die in the

correct spatial arrangement and temporal sequence to

yield a normally functioning organism. In the adult, the

cells of certain tissues, such as the intestinal epithelium, the

blood-forming tissues of the bone marrow, and those of

hair follicles, continue to proliferate. Most adult body cells,

however, have permanently ceased doing so.

Cells occasionally lose their developmental controls and

commence excessive proliferation. The resulting tumors

can be of two types:

1. Benign tumors, such as warts and moles, grow by sim-

ple expansion and often remain encapsulated by a layer of

connective tissue. Benign tumors are rarely life threaten-

ing, although if they occur in an enclosed space such as in

the brain or secrete large quantities of certain hormones,

they can be lethal.

2. Malignant tumors or cancers grow in an invasive

manner and shed cells that, in a process known as metasta-

sis, colonize new sites in the body. Malignant tumors are al-

most invariably life threatening; they are responsible for

20% of the mortalities in the United States.

The most obvious and the medically most significant

property of cancer cells is that they proliferate uncontrol-

lably. For instance, when grown in a tissue culture dish, nor-

mal cells form a monocellular layer on the bottom of the

dish and then, through a process termed contact inhibition,

cease dividing (Fig. 19-29a). In contrast, the growth of ma-

lignant cells is unhampered by intercellular contacts; in cul-

ture they form multicellular layers (Fig. 19-29b). Moreover,

even in the absence of contact inhibition, normal cells are

far more limited in their capacity to reproduce than are

cancer cells. Normal cells, depending on the species and age

of the animal from which they were taken, will only divide

in culture 20 to 60 times before they reach senescence (a

stage at which they cease dividing) and die (a phenomenon

that, no doubt, is at the heart of the aging process; Section

30-4Db). Cancer cells, on the other hand, are immortal; there

is no limit to the number of times they can divide. In fact,

some cancer cell lines have been maintained in culture

through thousands of divisions spanning 6 decades. Immor-

tal cells, however, are not necessarily malignant: The hall-

mark of cancer is immortality combined with uncontrolled

growth.

Section 19-3. Tyrosine Kinase–Based Signaling 703

Figure 19-29 Growth pattern of vertebrate cells in culture.

(a) Normal cells stop growing through contact inhibition once they

have formed a confluent monolayer. (b) In contrast, transformed

cells lack contact inhibition; they pile up to form a multilayer.

Plastic

tissue culture

dish

(a) Normal cells

(b) Transformed cells

Growth medium

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 703

a. Cancer Is Caused by Carcinogens,

Radiation, and Viruses

Most cancers are caused by agents that damage DNA or

interfere with its replication or repair. These include a

great variety of man-made and naturally occurring sub-

stances known as chemical carcinogens (Section 30-5F),

as well as radiation, both electromagnetic and particulate,

with sufficient energy to break chemical bonds. In addi-

tion, certain viruses induce the formation of malignant

tumors in their hosts (see below).

Almost all malignant tumors result from the transfor-

mation of a single cell [conversion to the cancerous state;

this term should not be confused with the acquisition of

genetic information from exogenously supplied DNA

(Section 5-2A)], which, then being free of its normal de-

velopmental constraints, proliferates. Yet, considering, for

example, that the human body consists of around 10

cells,

transformation must be a very rare event. One of the major

reasons for this, as the age distribution of the cancer death

rate indicates (Fig. 19-30), is that transformation requires a

cell or its ancestors to have undergone several independent

and presumably improbable carcinogenic changes. Conse-

quently, exposure to a carcinogen may prime many cells for

transformation, but a malignant tumor may not form until

decades later when one of these cells suffers a final trans-

forming event.

The viral induction of cancer was first observed in 1911

by Peyton Rous, who demonstrated that cell-free filtrates

from certain chicken sarcomas (malignant tumors arising

from connective tissues) promote new sarcomas in chick-

ens (Fig. 19-31). Although decades were to pass before the

significance of this work was appreciated (Rous was

awarded the Nobel prize in 1966 at the age of 85), many

other such tumor viruses have since been characterized.

The Rous sarcoma virus (RSV), as are all known RNA tu-

mor viruses, is a retrovirus (an RNA virus that replicates its

chromosome by copying it to DNA in a process mediated

by a virally encoded reverse transcriptase, inserting the

DNA into the host’s genome, and then transcribing this

DNA). It contains a gene, v-src (“v” for viral, “src” for sar-

coma), which encodes a protein named v-Src that mediates

704 Chapter 19. Signal Transduction

Figure 19-30 Variation of the cancer death rate in humans

with age. The linearity of this log–log plot can be explained by

the hypothesis that several randomly occurring mutations are

required to generate a malignancy. The slope of the line suggests

that, on the average, five such mutations are required for a

malignant transformation.

Figure 19-31 Transformation of cultured chicken fibroblasts

by Rous sarcoma virus. (a) Normal cells adhere to the surface of

the culture dish, where they assume a flat extended

20 8030 50 706040

Annual death rate/milliion

Age

100

1000

(a)

(b)

conformation. (b) On infection with RVS, these cells become

rounded and cluster together in piles. [Courtesy of G. Steven

Martin, University of California at Berkeley.]

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 704

host cell transformation. v-src has therefore been termed

an oncogene (Greek: onkos, mass or tumor).

What is the origin of v-src and what is its viral function?

Hybridization studies (Section 5-3Cb) by Michael Bishop

and Harold Varmus in 1976 led to the remarkable discov-

ery that uninfected chicken cells contain a gene, c-src (“c”

for cellular), that is homologous to v-src. Moreover, c-src is

highly conserved in a wide variety of eukaryotes that span

the evolutionary scale from Drosophila to humans.This ob-

servation strongly suggests that c-src, which antibodies di-

rected against v-Src indicated is expressed in normal cells,

is an essential cellular gene. In fact, both v-Src and its

normal cellular analog, c-Src, function to stimulate cell

proliferation (Section 19-3Ea). Apparently, v-src was origi-

nally acquired from a cellular source by an initially non-

transforming ancestor of RSV. By maintaining the host

cell in a proliferative state (cells are usually not killed by

RSV infection), v-Src presumably enhances the viral repli-

cation rate.

b. Viral Oncogene Products Mimic the Effects of

Protein Growth Factors and Hormones

The proteins encoded by many viral oncogenes are

analogs of various growth factor and hormone system com-

ponents. For instance:

1. The v-sis oncogene of simian sarcoma virus encodes

a protein secreted by infected cells that is nearly identical

to PDGF. Hence, the uncontrolled growth of simian sar-

coma virus–infected cells apparently results from the

continuous and inappropriate presence of this PDGF

homolog.

2. Nearly half of the more than 20 known retroviral

oncogenes, including v-src, encode PTKs. For example, the

v-erbB oncogene specifies a truncated version of the EGF

receptor (Fig. 19-27b) that lacks the EGF-binding domain

but retains its transmembrane segment and its protein ki-

nase domain. Evidently, oncogene-encoded PTKs inappro-

priately phosphorylate the target proteins normally recog-

nized by RTKs, thereby driving the afflicted cells to a state of

unrestrained proliferation.

3. The v-ras oncogene encodes a protein, v-Ras, that

functionally resembles the monomeric G-protein c-Ras

(Section 19-3Cf) in that it is localized on the cytoplasmic

side of the mammalian plasma membrane where, when

binding GTP, it activates a variety of cellular processes by

stimulating the phosphorylation of numerous proteins at

specific Ser and Thr residues. Although v-Ras hydrolyzes

GTP to GDP, it does so much more slowly than c-Ras. The

restraint to protein phosphorylation that GTP hydrolysis

would normally impose on c-Ras is thus greatly reduced in

v-Ras, thereby transforming the cell.

4. Several viral oncogenes, including v-jun and v-fos,

encode nuclear proteins whose corresponding normal cellu-

lar analogs are synthesized in response to growth factors

such as EGF and PDGF that induce mitosis (cell division).

Many such proteins, including the v-jun and v-fos gene prod-

ucts, bind to DNA, strongly suggesting that they influence

its transcription and/or replication. Indeed, v-jun is 80%

identical in sequence to the proto-oncogene (normal cellu-

lar analog of an oncogene) c-jun, which encodes a tran-

scription factor named Jun (also called AP-1; Section 19-

3D). Moreover, Jun/AP-1 forms a tight complex with the

protein encoded by the proto-oncogene c-fos, which

greatly increases the ability of Jun/AP-1 to stimulate tran-

scription from Jun-responsive genes.

Oncogene products therefore appear to be functionally

modified or inappropriately expressed components of elab-

orate control networks that regulate cell growth and differ-

entiation. The complexity of these networks (as we shall

see, cells generally respond to a variety of growth factors,

hormones, and transcription factors in partially overlap-

ping ways) is probably why malignant transformation re-

quires several independent carcinogenic events. Note,how-

ever, that few human cancers are virally induced; nearly all

of them arise from genetic alterations involving proto-

oncogenes. We discuss the nature of these alterations in

Section 34-4C.

C. Relaying the Signal: Binding Modules, Adaptors,

GEFs, and GAPs

Many autophosphorylated RTKs can directly phosphory-

late their target proteins. Surprisingly, however, not all

RTKs do so. How, then, do they activate their target pro-

teins? The answer, as we shall see, is through a highly di-

verse and complicated set of interconnected signaling

pathways involving cascades of associating proteins.

a. Two-Hybrid Systems Identify Proteins That

Interact in vivo

Before we consider the interacting proteins that partici-

pate in RTK-mediated signal transduction, let us discuss

one of the most often used methods to detect their associa-

tions in vivo, the two-hybrid system. This ingenious experi-

mental technique, which was formulated by Stanley Fields,

is based on the peculiar bipartite nature of many transcrip-

tion factors (proteins that bind to the promoters and other

upstream control regions of eukaryotic genes and, in doing

so, influence the rate at which RNA polymerase initiates

the transcription of these genes; Section 5-4Ab). Such tran-

scription factors, as we further discuss in Section 34-3Bi,

contain a DNA-binding domain (DBD) that targets the

transcription factor to a specific DNA sequence and an ac-

tivation domain (AD) that recruits RNA polymerase to

initiate transcription at a nearby transcriptional initiation

site. These two domains function independently, so that a

genetically engineered hybrid protein with the DBD of

one transcription factor and the AD of another will acti-

vate the transcription of the gene for which the DBD is tar-

geted. Moreover, it makes little difference as to whether

the DBD is on the N-terminal or the C-terminal side of the

AD, regardless of how they are arranged in their parent

proteins. Evidently, as long as a DBD and an AD are held

in proximity, they function as a transcription factor for the

gene(s) to which the DBD is targeted.

Section 19-3. Tyrosine Kinase–Based Signaling 705

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 705

The two-hybrid system employs two different plasmids

in yeast (Fig. 19-32): One encodes a hybrid protein consist-

ing of a DBD fused to a so-called bait or probe protein; the

other encodes a hybrid protein consisting of an AD fused

to a so-called fish or target protein.The DBD is targeted to

a reporter gene that has been engineered into the yeast

chromosome such as the E. coli lacZ gene, which encodes

the enzyme -galactosidase. On culture plates containing

X-gal (a colorless compound that turns blue on being hy-

drolyzed by -galactosidase; Section 5-5Ca), yeast colonies

expressing -galactosidase turn blue, thereby indicating

that the bait and fish proteins they encode associate with

one another. Using this technique,cells can be screened for

fish proteins that specifically interact with a particular bait

protein by properly inserting the various cDNAs (Section

5-5F) derived from the cells into the AD-containing plas-

mid. A fish protein selected in this way can then be identi-

fied by sequencing its cDNA.

b. SH2 Domains Mediate Signal Transduction

Now let us return to our discussion of RTK-mediated

signaling. Many (100) of the diverse cytoplasmic proteins

that bind to autophosphorylated receptors, for example,

the cytoplasmic PTK c-Src (henceforth referred as just

Src), certain GTPase activating proteins (GAPs), and

phospholipase C- (Section 19-4Bc), contain one or two

conserved ⬃100-residue modules known as Src homology

domain 2 (SH2; so named because they were first noticed

in tyrosine kinases related to Src; SH1 refers to their cat-

alytic domains). SH2 domains specifically bind phosphoTyr

residues in their target peptides with high affinity; they bind

their unphosphorylated target peptides weakly if at all.

Most of the phosphoTyr residues to which SH2 binds are

located in the juxtamembrane (just after the transmem-

brane helix), kinase insert, and C-terminal regions of

RTKs; those in the activation loop function mainly to stim-

ulate PTK activity. Indeed, RTK autophosphorylation oc-

curs in two phases: First the activation loops of the RTK

are phosphorylated and then the resulting activated PTK

phosphorylates the other sites on the opposing subunit of

the RTK.

The X-ray and NMR structures of SH2 domains from

several proteins, both alone and in complex with phospho-

Tyr (pY)-containing polypeptides, have been determined.

SH2 is a hemispherically shaped domain, which contains a

central 5-stranded antiparallel  sheet that is sandwiched

between two nearly parallel  helices.The N- and C-terminal

residues of SH2 are in close proximity on the surface oppo-

site the peptide-binding site, which suggests that this do-

main can be inserted between any two surface residues on

a protein without greatly disturbing its fold or function. In-

deed, the sequences of a variety of SH2-containing pro-

teins reveal no apparent preference for the location of this

domain in a protein.

John Kuriyan determined the X-ray structure of the

SH2 domain of Src in complex with an 11-residue polypep-

tide containing the sequence pYEEI, a tetrapeptide seg-

ment that binds to this SH2 domain with high affinity. The

11-residue peptide binds to the SH2 domain in an extended

conformation, with contact being made primarily by the

pYEEI tetrapeptide (Fig. 19-33a). The phosphoTyr side

chain inserts into a small cleft formed, in part, by three

highly conserved positively charged residues, including the

side chain of an invariant Arg, which contacts the phos-

phate group. The Ile side chain is similarly inserted into a

nearby hydrophobic pocket and the entire tetrapeptide

segment interacts very tightly with SH2, although the side

chains of the peptide’s two central Glu residues do not

706 Chapter 19. Signal Transduction

Figure 19-32 The two-hybrid system.

Two plasmids are expressed in yeast: One

encodes a hybrid protein consisting of the

DNA-binding domain (DBD) of a

transcription factor fused to a “bait”

protein, and the other encodes the

activation domain (AD) of a transcription

factor fused to a “fish” protein. The DBD

specifically binds to the promoter of a

reporter gene, here lacZ. If the “fish”

protein associates with the “bait” protein,

its attached AD recruits RNA polymerase

to initiate the transcription of the reporter

gene. The -galactosidase encoded by the

lacZ gene is readily detected through the

use of X-gal, which turns blue when

hydrolyzed by -galactosidase. Yeast

colonies expressing “bait” and “fish”

proteins that do not interact remain

colorless, as do colonies that do not express

both plasmids.

Yeast

cell

Plasmid 1 Plasmid 2

Expression

Promoter

DBD Bait Fish

lacZ reporter gene

Transcriptional

initiation

RNA

polymerase

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 706

project toward SH2. Thus, the peptide resembles a two-

pronged plug that is inserted into a two-holed socket on

SH2 (Fig.19-33b). Comparison of this structure with that of

uncomplexed Src SH2 indicates that, on binding peptide,

SH2 undergoes only small conformational changes that are

localized at its peptide-binding site. These structures pro-

vide a simple explanation for why SH2 does not bind the

far more abundant phosphoSer- and phosphoThr-containing

peptides: The side chains of these residues are too short

for their phosphate groups to contact the invariant Arg

side chain at the bottom of the phosphoTyr-binding pocket.

c. PTB Domains Also Bind PhosphoTyr-

Containing Peptides

A second type of motif that specifically binds to

phosphoTyr-containing target peptides is known as the

phosphotyrosine-binding (PTB) domain. PTB domains

specifically bind the consensus sequence NPXpY (where X

is any residue) and hence recognize the sequence on the

N-terminal side of pY rather than that on its C-terminal

side, as do SH2 domains.

The NMR structure of the 195-residue PTB domain of

Shc, an adaptor protein (see below), in complex with a

12-residue target peptide containing the centrally located

sequence NPQpY was determined by Stephen Fesik. The

structure consists of a ␤ sandwich comprising two nearly

perpendicular antiparallel ␤ sheets flanked by three ␣ he-

lices (Fig. 19-34). The N-terminal segment of the target

phosphopeptide assumes an extended conformation that,

Section 19-3. Tyrosine Kinase–Based Signaling 707

(a)

(b)

Figure 19-33 X-ray structure of the 104-residue Src SH2

domain in complex with an 11-residue polypeptide

(EPQpYEEIPIYL) containing the protein’s pYEEI target

tetrapeptide. (a) A cutaway view of the complex, in which its

solvent-accessible surface is represented by red dots, the protein

(pink) is shown in ribbon form with its side chains in stick form

(light blue), and the N-terminal 8-residue segment of the bound

polypeptide is shown in space-filling form with its backbone

yellow, its side chains green, and its phosphate group white [the

Figure 19-34 The NMR structure of the PTB domain of Shc

in complex with a 12-residue polypeptide (HIIENPQpYFSDA)

from the Shc binding site of a nerve growth factor (NGF)

receptor. The PTB domain is colored in rainbow order from its

N-terminus (blue) to its C-terminus (red) and the target peptide

is magenta.The target peptide’s phosphoTyr (pY) side chain and

the side chain of Arg 67, which form a salt bridge (dashed line),

are drawn in stick form with C green, N blue, O red, and P

orange. Note that the pY side chain extends from a ␤ turn on the

peptide ligand. [Based on an NMR structure by Stephen Fesik,

Abbott Laboratories,Abbott Park, Illinois. PDBid 1SHC.]

N-terminal Pro side chain (left) is largely obscured in this view

and the C-terminal three residues are disordered]. (b) The

molecular surface of the protein only, as viewed toward the

peptide binding site and colored according to its local

electrostatic potential with the most positive regions deep blue

and the most negative regions deep red.The binding pockets for

the phosphoTyr (right) and Ile (left) side chains are circled in

yellow and important residues are identified by red arrows.

[Courtesy of John Kuriyan, The Rockefeller University.]

JWCL281_c19_671-743.qxd 6/11/10 6:57 AM Page 707

in effect, forms an additional antiparallel  strand of one

of the  sheets. The NPQpY segment forms a  turn in

which the phosphate group contacts an Arg side chain that

mutational studies indicate is essential for the binding of

target peptide.

d. SH3 Domains Bind to Proline-Rich Peptides

Many of the RTKs that contain SH2 domains also have

one or more 50- to 75-residue SH3 domains. Moreover,

SH3 is contained in several membrane-associated proteins

that lack an SH2. The SH3 domain, which is unrelated to

SH2, binds Pro-rich sequences of 9 or 10 residues contain-

ing the motif Pro-X-X-Pro, with the residues surrounding

this motif targeting these sequences to specific SH3 do-

mains. The physiological function of SH3 is less apparent

than that of SH2 because SH3 occurs in a greater variety of

proteins, including receptor and nonreceptor tyrosine ki-

nases, adaptor proteins such as Grb2 (see below), and

structural proteins such as spectrin and myosin. However,

the observation that the deletion of the SH3 domain–

encoding segments from the proto-oncogenes Src and Abl

(which both encode PTKs) converts them to oncogenes

suggests that SH3, much like SH2 and PTB, functions to

mediate the interactions between kinases and regulatory

proteins. SH2, PTB, and SH3 have therefore been called

“molecular velcro.”

The X-ray and NMR structures of SH3 domains from

several proteins indicate that the SH3 core consists of two

3-stranded, antiparallel  sheets that pack against each

other with their strands nearly perpendicular.As with SH2,

the close proximity of SH3’s N- and C-termini suggests that

this domain could be modularly inserted between two

residues on the surface of another protein without greatly

perturbing either structure. The X-ray structures of the

SH3 domains from the tyrosine kinases Abl and Fyn in

complex with two different 10-residue Pro-rich polypep-

tides to which they tightly bind were determined by

Andrea Musacchio and Matti Saraste. Both decapeptides

assume nearly identical conformations with their C-terminal

7 residues in the polyproline II helix conformation (Sec-

tion 8-1Bb). The peptides bind to SH3 over their entire

length in three geometrically complementary cavities

(Fig. 19-35), which are mostly occupied by Pro side chains.

e. Other Binding Modules

Several other binding modules have been implicated in

mediating signal transduction.These include:

1. The WW domain (named after its two highly con-

served Trp residues), an ⬃40-residue module that binds to

Pro-rich sequences on its target proteins.

2. The pleckstrin homology (PH) domain (so named

because was first recognized in pleckstrin [for platelet and

leukocyte C kinase substrate protein)], an ⬃120-residue

module that is present in 100 proteins. It structurally re-

sembles the PTB domain (Fig. 19-34) but binds to the inos-

itol head groups of phosphoinositides (Section 19-4). It

therefore targets its attached proteins to the inner surface

of the plasma membrane. In addition to its role in intracel-

lular signaling, the PH domain participates in cytoskeletal

organization, regulation of intracellular membrane trans-

port, and modification of membrane lipids. Its structure is

discussed in Section 19-4Bb.

3. The PDZ domain (named after the three proteins in

which it was first described: PSD-95, Dlg, and ZO-1), an

⬃100-residue module that mainly binds to the C-terminal

tripeptide, Ser/Thr-X-Val, of its target proteins.

Many of the proteins that participate in signal transduction

consist of several modular units that also occur in several, if

not many, other such proteins. These modules may have en-

zymatic activities (e.g., PTK activity) or bind to specific

molecular motifs, such as a phosphoTyr residue within a

specific sequence (e.g., SH2 domains) or another protein

module (e.g., SH3 domains). Apparently, signaling proteins

have arisen through the evolutionary shuffling of these

modules to generate different combinations of interactions

and activities. Indeed, we shall see that the complex behav-

ior of these signaling proteins is a consequence of the inter-

actions among their various modules.

f. Ras Is Activated by Phosphorylated RTKs

via a Grb2–Sos Complex

c-Ras (or just Ras), a proto-oncogene product, is a

monomeric membrane-anchored (by prenylation) G protein

708 Chapter 19. Signal Transduction

Figure 19-35 X-ray structure of the SH3 domain from

Abl protein in complex with its 10-residue target Pro-rich

polypeptide (APTMPPPLPP). The protein is represented by

its surface diagram (cyan) and the peptide is drawn in stick form

with Pro C magenta, other C green, N blue, O red, and S yellow.

[Based on an X-ray structure by Andrea Musacchio, European

Molecular Biology Laboratory, Heidelberg, Germany. PDBid

1ABO.]

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 708

that lies at the center of an intracellular signaling system: It

regulates such essential cellular functions as growth and dif-

ferentiation through the phosphorylation and hence activa-

tion of a variety of proteins. Ras is the prototypic member

of the superfamily of small G proteins, whose 154 members

form five principal families, those of (1) Ras; (2) Arf and

(3) Rab, which both participate in vesicle trafficking (Sec-

tions 12-4Cd and 12-4Db); (4) Rho, which is mainly in-

volved in regulating the cytoskeleton (Section 35-3Ed);

and (5) Ran, which regulates nuclear import and export.

In the signaling pathway described in Section 19-3D, the

binding of ligand to RTKs ultimately activates a guanine

nucleotide exchange factor (GEF; Section 19-2Ca) to

exchange a Ras-bound GDP for GTP. Only Ras  GTP

is capable of further relaying the signal. However, as do

the homologous and structurally similar  subunits of the

heterotrimeric G proteins, Ras eventually hydrolyzes its

bound GTP to GDP, thereby halting further signal trans-

duction and limiting the magnitude of the signal generated

by the binding of ligand to the receptor. Indeed, the struc-

ture of Ras closely resembles those of the GTPase domains

of the heterotrimeric G protein G



subunits, including their

Switch I and Switch II regions (Fig. 19-18). Mammalian

cells express four Ras homologs: H-Ras, N-Ras, K-Ras 4A,

and K-Ras 4B.

Molecular genetic analyses of signaling in a variety of

distantly related organisms (notably humans, mice, Xeno-

pus, Drosophila, and the nematode worm Caenorhabditis

elegans) have revealed a remarkably conserved pathway in

which RTKs funnel the signal that they have bound ligand

to Ras, which, in turn, relays the signal, via a so-called MAP

kinase cascade, to the transcriptional apparatus in the nu-

cleus (Section 19-3D). Nevertheless, the way in which mes-

sages are passed between the RTKs and Ras remained

enigmatic for several years until investigations in numer-

ous laboratories revealed their major details. In particular,

these studies demonstrated that two previously character-

ized proteins, Grb2 and Sos, form a complex that bridges

activated RTKs and Ras in a way that induces Ras to ex-

change its bound GDP for GTP, thereby activating it (i.e.,

they act as a GEF).

The mammalian protein Grb2, a 217-residue homolog

of drk in Drosophila and Sem-5 in C. elegans, consists al-

most entirely of an SH2 domain flanked by two SH3 do-

mains. Sos protein (the 1596-residue product of the Son of

Sevenless gene, so named because Sos interacts with the

Sevenless gene product, an RTK that regulates the devel-

opment of the R7 photoreceptor cell in the Drosophila

compound eye), which is required for Ras-mediated signal-

ing, contains a central domain homologous to known Ras-

GEFs and a Pro-rich sequence in its C-terminal segment

similar to known SH3-binding motifs. Moreover, mam-

malian homologs of Sos (mSos) have been shown to

specifically stimulate guanine nucleotide exchange in

mammalian Ras proteins. Western blotting techniques

(Section 6-4Bc) using anti-Grb2 and anti-mSos antibodies

indicated that Grb2 binds to the C-terminal segment of

mSos but not when one of the Pro residues in Sos’ SH3-

binding motif has been replaced by Leu or in the presence

of synthetic polypeptides that have these Pro-rich se-

quences. Similar studies indicated that in the presence of

epidermal growth factor (EGF), the EGF receptor (an

RTK; Figs. 19-25 and 19-27b) specifically binds the

Grb2–mSos complex. However, this interaction is blocked

by the presence of a phosphopeptide with the sequence of

the peptide segment containing one of the activated EGF

receptor’s phosphoTyr residues. Evidently, Grb2’s SH2

domain binds a phosphoTyr-containing peptide segment in

an activated RTK while its two SH3 domains bind the Pro-

rich sequences of Sos. The GEF function of Sos is thereby

stimulated to activate Ras.

g. Grb2, Shc, and IRS Are Adaptors That

Recruit Sos to the Vicinity of Ras

The X-ray structure of Grb2, determined by Arnaud

Ducruix, reveals that neither of its SH3 domains contacts

its SH2 domain (Fig. 19-36). Moreover, although the two

SH3 domains are in contact, their interface area is rela-

tively small and hence is more likely to be an artifact of

crystallization than a structural feature of Grb2 in solution.

It therefore appears that Grb2’s SH3 domains are flexibly

linked to its SH2 domain. How does the binding of such a

pliable adaptor (a linker that lacks enzymatic activity) to a

phosphorylated RTK stimulate Sos to act as Ras’s GEF?

Grb2 and Sos bind one another so tightly that they are es-

sentially permanently associated in the cell. Hence, when

Section 19-3. Tyrosine Kinase–Based Signaling 709

Figure 19-36 X-ray structure of Grb2. Its SH2 domain (green)

is linked to its flanking SH3 domains (cyan and orange) via

apparently unstructured and hence flexible 4-residue linkers.

[Based on an X-ray structure by Arnaud Ducruix, Université de

Paris-Sud, Gif sur Yvette Cedex, France. PDBid 1GRI.]

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 709

Grb2 binds its target phosphorylated RTK, it recruits Sos

to the inner surface of the plasma membrane, where Sos’s

then increased local concentration causes it to more read-

ily bind to the membrane-anchored Ras.

Shc proteins are also adaptors that link activated RTKs

to Ras. Shc proteins consist of an N-terminal PTB domain

(Fig.19-34),a central effector region (CH1),and a C-terminal

SH2 domain that binds to certain activated RTKs. More-

over, Shc proteins are also major targets of various RTKs,

which phosphorylate them within their CH1 domains at

sequences that then form binding sites for the Grb2 SH2

domain. Activated RTKs may therefore bind Grb2 indi-

rectly via Shc as well as directly. Moreover, in some cases,

an Shc–Grb2–Sos complex that lacks a bound RTK may

activate Ras.

The activated (autophosphorylated) insulin receptor

does not directly interact with SH2 domain–containing

proteins. Rather, it mainly phosphorylates an ⬃1300-

residue protein named the insulin receptor substrate (IRS;

actually a family of four homologous proteins named

IRS1–4, each of which is expressed in a tissue-specific man-

ner). The IRS proteins all have an N-terminal “targeting”

region consisting of a PH domain that localizes the IRS to

the inner surface of the plasma membrane, followed by a

PTB domain that binds the IRS to a phosphoTyr residue of

an activated insulin receptor (Fig. 19-37).The insulin recep-

tor thereupon phosphorylates the IRS at one or more of its

6 to 8 Tyr residues, converting them to SH2-binding sites

that then couple this system to SH2-containing proteins

(Section 19-4F). Adaptors with multiple SH2-binding sites

such as the IRS proteins and Shc are also known as dock-

ing proteins because they function as platforms for the re-

cruitment of a variety of downstream signaling molecules

in response to the activation of their corresponding RTK.

Thus, a docking protein increases the complexity and regu-

latory flexibility of its RTK-initiated signaling pathway as

well as amplifying its signal.

h. Sos Functions to Pry Open Ras’s

Nucleotide Binding Site

The X-ray structure, determined by Kuriyan, of Ras in

complex with a 506-residue GEF-containing segment of

Sos reveals how Sos induces Ras to exchange its normally

tightly bound GDP for GTP. This Sos segment consists

of two  helical domains, of which only the C-terminal

so-called catalytic domain contacts Ras. Ras is packed

against the center of the elongated bowl-shaped catalytic

domain (Fig. 19-38). Those portions of Ras that interact

with the catalytic domain include both its Switch I and

Switch II regions as well as the loop that binds the  and 

phosphates of GDP and GTP, the so-called P-loop (al-

though GMP binds to Ras with 10

-fold less affinity than

does GDP, so that the  phosphate of GDP is largely re-

sponsible for its tight binding to Ras). This interaction dis-

places Switch I relative to its position in the X-ray structure

of Ras in complex with the nonhydrolyzable GTP analog

GMPPNP:

Ras’s nucleotide binding site is thereby partially opened up

and a Leu and a Glu side chain from Sos are, respectively,

introduced into Ras’ Mg

2

-binding site and the site that

binds the  phosphate group of GDP/GTP. However, this

interaction does not significantly occlude Ras’s guanine

and ribose binding sites. This rationalizes how the Sos–Ras

interaction can be strong enough to displace the normally

tightly bound GDP from Ras but yet weak enough so that

–

Guanosine-5-(,-imido)triphosphate

(GMPPNP)

OH OH

–

P NH

710 Chapter 19. Signal Transduction

Figure 19-37 Structure of an insulin receptor substrate (IRS).

An IRS contains a PH and a PTB domain at its N-terminus

followed by multiple phosphoTyr-containing binding sites for the

SH2 domains of downstream signaling proteins.

PH PTB

PPPP PP P

Figure 19-38 X-ray structure of the complex between Ras and

the GEF-containing region of Sos. The N-terminal domain of the

Sos region is blue, its catalytic domain is green, and Ras is mainly

gray, with its Switch I and II regions orange and its P loop red.

Conserved regions (the SCRs) among the Ras family of GEFs

are cyan. [Courtesy of John Kuriyan, The Rockefeller University.

PDBid 1BDK.]

JWCL281_c19_671-743.qxd 3/16/10 7:17 PM Page 710