Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

glucagon receptors and to -adrenergic receptors. In such

cases, the amount of cAMP produced is the sum of that in-

duced by the individual hormones. G proteins may also act

in other ways than by activating AC: They are known, for

example, to stimulate the opening of K



channels in heart

cells and to participate in the phosphoinositide signaling

system (Section 19-4A).

Some ligand–GPCR complexes inhibit rather than acti-

vate AC (Fig. 19-17, right). These include the 

-adrenergic

receptor and receptors for somatostatin and opioids. The

inhibitory effect is mediated by “inhibitory” G protein, G

which may have the same  and  subunits as does “stimula-

tory”G protein,G

, but has a different  subunit, G

i

(41 kD).

acts analogously to G

in that on binding to its corre-

sponding ligand–GPCR complex, its G

i

subunit exchanges

bound GDP for GTP and dissociates from G



. However,

i

inhibits rather than activates AC, through direct inter-

actions and possibly because the liberated G



binds to and

sequesters G

s

. The latter mechanism is supported by the

observation that liver cell membranes contain far more G

than G

. The activation of G

in such cells would therefore

release enough G



to bind the available G

s

and G

i

are members of a family of related proteins,

many of which have downstream effectors other than AC.

This family also includes:

1. G

q

, which forms a link in the phosphoinositide sig-

naling system (Section 19-4Ba).

2. Transducin (G

t

), a variant of G

i

, which transduces

visual stimuli by coupling the light-induced conformational

Section 19-2. Heterotrimeric G Proteins 691

Figure 19-17 Mechanism of receptor-mediated activation/

inhibition of AC. The binding of hormone to a stimulatory

receptor, R

(left), induces it to bind G

protein, which, in turn,

stimulates the G

s

subunit of this G

s



heterotrimer to

exchange its bound GDP for GTP. The G

s

 GTP complex then

dissociates from G



and, until it catalyzes the hydrolysis of its

bound GTP to GDP, stimulates adenylate cyclase (AC) to

convert ATP to cAMP. The binding of hormone to the inhibitory

Stimulatory

external

signal

Inhibitory

external

signal

Cholera

toxin

4ATP

Pertussis

toxin

phosphodiesterase

4cAMP

+ R

⋅ cAMP

+ 2C

Protein-

Cellular

response

phosphoprotein

phosphatase

4PP

ADP

ATP

GDP

GTP GTP

ββ

GDP

Plasma

membrane

Cytosol

4AMP

sα

⋅ GDP

+ P

iα

⋅ GDP

+ P

sα sα iα iα

Protein

inactive( )

active ( )

receptor, R

(right), triggers an almost identical chain of events

except that the presence of G

i

 GTP complex inhibits AC from

synthesizing cAMP. R

represents protein kinase A (PKA),

whose catalytic subunit, C, when activated by the dissociation of

the regulatory dimer as R

 cAMP

(Section 18-3Cb), activates

its target cellular proteins by catalyzing their phosphorylation.

The sites of action of cholera and pertussis toxins are indicated.

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 691

change of rhodopsin to the activation of a specific phos-

phodiesterase, which then hydrolyzes cGMP to GMP. This

cGMP-phosphodiesterase (cGMP-PDE) is an ␣␤␥

het-

erotetramer that is activated by the displacement of its

inhibitory ␥ subunits (PDE␥) by their tighter binding to

t␣

ⴢ GTP. A cation-specific transmembrane channel (Sec-

tion 20-3A) that is held open by the binding of cGMP

closes on the resulting reduction in [cGMP], thereby trig-

gering a nerve impulse (Section 20-5B) indicating that light

has been detected.

3. G

olf

, a variant of G

s␣

, which is expressed only in olfac-

tory sensory neurons and participates in odorant signal

transduction.

4. G

12␣

and G

13␣

, which participate in the regulation of

the cytoskeleton.

This heterogeneity in G proteins occurs in the ␤ and ␥ sub-

units as well as in the ␣ subunits. In fact, 21 different ␣ sub-

units, 6 different ␤ subunits, and 12 different ␥ subunits

have been identified in humans, some of which appear to

be ubiquitously expressed whereas others are expressed

only in specific cells.Thus, a cell may contain several closely

related G proteins of a given type that interact with varying

specificities with receptors and effectors. This complex sig-

naling system presumably permits cells to respond in a

graded manner to a variety of stimuli.

a. G Proteins Often Require Accessory Proteins

to Function

The proper physiological functioning of a G protein

often requires the participation of several other types of

proteins:

1. A GTPase-activating protein (GAP), which as its

name implies, stimulates its corresponding G protein to hy-

drolyze its bound GTP. This rate enhancement can be

⬎2000-fold. The downstream effectors of G

t␣

and G

q␣

cGMP-PDE (Section 19-3E) and PLC-␤ (Section 19-4Ba),

respectively, exhibit GAP activities toward G

t␣

and G

q␣

(which otherwise would hydrolyze GTP at physiologically

insignificant rates), but AC does not exhibit GAP activity

toward either G

s␣

or G

i␣

. However, in humans, a diverse

family of 37 RGS proteins (for regulators of G protein sig-

naling) function as GAPs for G

␣

subunits by binding most

avidly to them when they are in the transition state confor-

mation for hydrolyzing GTP.

2. A guanine nucleotide exchange factor [GEF; alter-

natively guanine nucleotide releasing factor (GRF)], which

induces its corresponding G protein to release its bound

GDP. The G protein subsequently binds another guanine

nucleotide (GTP or GDP, which most G proteins bind with

approximately equal affinities), but since cells maintain a

GTP concentration that is 10-fold higher than that of GDP,

this, in effect, exchanges the bound GDP for GTP. For het-

erotrimeric G proteins, the agonist–GPCR complexes

function as GEFs.

3. A guanine nucleotide-dissociation inhibitor (GDI).

A G

␤␥

may be regarded as its associated G

␣

’s GDI because

GDP dissociates slowly from isolated G

␣

subunits but is es-

sentially irreversibly bound by heterotrimers.

b. The X-Ray Structures of G

␣

Proteins Rationalize

Their Functions

The X-ray structures of the C-terminal 325 residues of

the 350-residue bovine transducin-␣ (G

t␣

) in its complexes

with GDP (Fig. 19-18a,b) and with the poorly hydrolyzable

GTP analog GTP␥S

(Fig. 19-18c,d) were determined by Heidi Hamm and Paul

Sigler. G

t␣

consists of two clearly delineated domains con-

nected by two polypeptide linkers: (1) a highly conserved

GTPase domain that is structurally similar to those in other

G proteins of known structure (and hence is often de-

scribed as a Ras-like domain), and (2) a helical domain

that is unique to heterotrimeric G proteins. Guanine

nucleotides bind to G

t␣

in a deep cleft that is flanked by

these domains. The X-ray structures of G

i␣

ⴢ GTP␥S and

s␣

ⴢ GTP␥S, both determined by Gilman and Stephen

Sprang, closely resemble that of the G

t␣

ⴢ GTP␥S.

Comparison of the structures of the G

t␣

ⴢ GDP and

t␣

ⴢ GTP␥S complexes reveals that the presence of GTP’s

␥ phosphate group promotes significant conformational

shifts in three loops known as switch regions, all of which

are located on the facing side of G

t␣

in Fig. 19-18. The ␥

phosphate hydrogen bonds to side chains on Switches I and

II, thereby pulling these polypeptide segments in toward it

and causing Switch II to contact Switch III in a way that

pulls it to the right in Fig. 19-18.These concerted conforma-

tional shifts cause an extensive cavity over the GDP-

binding site to largely fill in the GTP␥S complex.

Switches I and II have counterparts in other G proteins

of known structure. Portions of these polypeptide segments

have been implicated in the interactions of G

t␣

with the

cGMP-PDE it activates and in the interactions between

the closely related G

s␣

with its target AC (Section 19-2D).

c. The X-Ray Structures of Heterotrimeric G Proteins

The X-ray structures of heterotrimeric G proteins were

determined by Gilman and Sprang (G

ⴢ GDP; Fig. 19-19)

and by Hamm and Sigler (G

ⴢ GDP). These structures

reveal that the G

␤

subunit (Fig. 19-19b) consists of an

N-terminal helical domain followed by a C-terminal domain

comprising seven 4-stranded antiparallel ␤ sheets arranged

like the blades of a propeller¬a ␤ propeller whose blades

are each formed by a WD40 sequence motif (Section

12-4Cb)¬that surround a water-filled central channel. The

WD40 motif occurs in a functionally diverse group of 4- to

8-bladed ␤-propeller proteins, including the 7-bladed

N-terminal domain of the clathrin heavy chain (Section

12-4Cb). The G

␥

subunit consists mainly of two helical

GTP␥S

⫺

OH OH

OP PO

⫺

692 Chapter 19. Signal Transduction

JWCL281_c19_671-743.qxd 6/4/10 10:52 AM Page 692

segments joined by a polypeptide link (Fig. 19-19b). It is

closely associated with G



along its entire extended length

through mainly hydrophobic interactions and hence has no

tertiary structure. The X-ray structure of isolated G



essentially identical to that in the G



 GDP–G



complex,

thereby indicating that the structure of G



is unchanged

by its association with G



 GDP.



and G



associate mainly via highly conserved

contacts between the Switch I and II regions of G



and

the loops and turns at the bottom of G



’s  propeller

(Fig. 19-19). In addition, there is a less extensive interaction

between the N-terminal helix of G



(which is disordered

in G



alone) and the first blade of the G



propeller

(back side of Fig. 19-19a). Comparison of the structure

Section 19-2. Heterotrimeric G Proteins 693

Figure 19-18 Structural differences between the inactive and

active forms of G

t

(transducin). The change in structure is

indicated by the comparison of the X-ray structure of G

t

 GDP

in its (a) ribbon and (b) space-filling forms with that of G

t



GTPS in its (c) ribbon and (d) space-filling forms, all viewed

from the same direction. In the ribbon drawings, helices and

sheets are green; the segments linking them are gold; the guanine

nucleotides are magenta, except for the  phosphate of GTPS,

which is yellow; and the bound Mg

2

ion is represented by a

blue ball.The protein’s three switch regions (I, II, and III)

are highlighted in cyan. In Part c, the two loop regions of

the protein that are implicated in its interaction with the

(a) (b)

cGMP-phosphodiesterase subunit to which it binds (PDE) are

pointed out with yellow labels, whereas the three loop regions

that are implicated in the interaction of the homologous G

s

with

adenylate cyclase (AC) are indicated with pink labels. The space-

filling models are colored similarly to the ribbon diagrams except

for the yellow residues, which here represent those that appear

to propagate or stabilize the structural transitions induced by the

binding of the  phosphate group. The box in the space-filling

models outlines the cavity in G

t

 GDP that closes when the GDP

is replaced by GTPS and that has been implicated in modulating

the affinity of G

t

for G



and for the receptor. [Courtesy of Paul

Sigler,Yale University. PDBids 1TAG and 1TND.]

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 693

of G

␣

ⴢ GDP–G

␤␥

with that of G

␣

ⴢ GTP␥S reveals why

␣

cannot simultaneously bind GTP and G

␤␥

: In

ⴢ GDP–G

, the Switch II segment of G

contacts G

in a

way that prevents Switch II from assuming the conforma-

tion it requires to bind GTP’s g phosphate. Moreover, the

conformational changes in Switch II are coordinated with

those in Switch I so that, together, they close over the GDP

bound to G

␣

–G

␤␥

, thereby accounting for its tight binding

relative to that in G

␣

ⴢ GDP.

The question remains, how does a liganded GCPR in-

duce its target G

␣

subunit to exchange its bound GDP for

GTP? The X-ray structure of opsin in an activated confor-

mation in complex with the 11-residue C-terminal segment

of G

t␣

, determined by Oliver Ernst, suggests a partial an-

swer to this question.This helical segment (the right end of

helix ␣5 in Fig. 19-18a and the left end of the nearly hori-

zontal gray helix at the top of Fig. 19-19b), which has been

shown to be a major site of interaction with activated

opsin, binds at the cytosolic face of opsin. This prompted

the construction of the model in which the remainder

of G

t␣

␤␥

is appended onto the G

t␣

C-terminal segment

(Fig. 19-20).To avoid a steric clash between G

t␣

␤␥

and the

membrane, G

t␣

␤␥

must be tilted upward by 40°. A simple

mechanism for this process is a reorientation of the loop

connecting helix ␣5 and strand ␤6 (Fig. 19-18a). Since this

loop participates in binding the guanine ring of GDP, a rea-

sonable assumption is that this conformational adjustment

expels GDP from its binding pocket, which when opsin

subsequently releases G

t␣

␤␥

, allows GTP to bind in its

place.

d. Cholera Toxin Stimulates Adenylate Cyclase by

Permanently Activating G

s␣

The major symptom of cholera, an intestinal disorder

caused by the bacterium Vibrio cholerae, is massive diar-

rhea that, if untreated, frequently results in death from de-

hydration. This dreaded disease is not an infection in the

usual sense since the vibrio neither invades nor damages

tissues but merely colonizes the intestine, much like E. coli.

The catastrophic fluid loss that cholera induces (often over

6 liters per hour!) occurs in response to a bacterial toxin.

Indeed, merely replacing cholera victims’ lost water and

salts enables them to survive the few days necessary to im-

munologically eliminate the bacterial infestation.

Cholera toxin (CT; also known as choleragen) is an

87-kD protein of subunit composition AB

in which the B

subunits (103 residues each) form a pentagonal ring to which

the A subunit (240 residues) is bound. Previous to CT’s se-

cretion, its A subunit is cleaved at a single site by a bacter-

ial protease to yield two fragments, A1 (the N-terminal

694 Chapter 19. Signal Transduction

(a)

(b)

Figure 19-19 X-ray structure of the heterotrimeric G protein

. (a) The G

␣

subunit is violet with its Switch I, II, and III

segments green, blue, and red, respectively, and with its bound

GDP shown in space-filling form with C green, N blue, O red,

and P yellow. The G

␤

subunit’s N-terminal segment is blue and

each blade of its ␤ propeller has a different color.The G

␥

subunit

is gold.The view is perpendicular to the axis of the G

␤

subunit’s

␤ propeller.The plasma membrane would be at the top of the

drawing as inferred from the positions of the N terminus of G

␣

and the neighboring C terminus of G

␥

, which, in vivo, are

lipid-linked to the plasma membrane. However, the orientation

of the protein relative to the plasma membrane is unknown (but

see Fig. 19-20). (b) View related to that in Part a by a 90° rotation

about its horizontal axis and thus looking from the general

direction of the plasma membrane. The protein is colored as in

Part a except that the G

␣

subunit is mainly gray. [Based on an

X-ray structure by Alfred Gilman and Stephan Sprang,

University of Texas Southwestern Medical Center. PDBid

1GP2.]

See Interactive Exercise 12

JWCL281_c19_671-743.qxd 10/19/10 7:36 AM Page 694

⬃195 residues) and A2 (the C-terminal ⬃45 residues), that

remain joined by a disulfide bond. On the binding of CT to

its cell surface receptor, ganglioside G

(Sections 12-1D

and 25-8Cd), the nicked A subunit (but not the B subunits)

is taken into the cell via receptor-mediated endocytosis and

travels backward through the secretory pathway (Section

12-4B) to the Golgi apparatus. From there, it is conducted

into the endoplasmic reticulum (ER) via the binding of

A2’s C-terminal KDEL sequence to a KDEL receptor

(which normally functions to retrieve ER-resident proteins

that have escaped the ER; Section 12-4Ch). The A1 frag-

ment is then released from A2 and enters the cytoplasm

through the translocon (which normally conducts grow-

ing and still unfolded polypeptides into the ER; Section

12-4Bd) via a process in which A1 is unfolded through the

chaperonelike action of protein disulfide isomerase (PDI;

Section 9-2A).

In the cytoplasm, A1 catalyzes the irreversible transfer

of the ADP–ribose unit from NAD

⫹

to a specific Arg side

chain of G

s␣

(Fig. 19-21). This reaction is greatly acceler-

ated by the interaction of A1 with the small Ras-like G

protein ADP-ribosylation factor (ARF) in complex with

GTP, which normally functions to prime the formation of

clathrin-coated vesicles (Section 12-4Cd).

Section 19-2. Heterotrimeric G Proteins 695

Figure 19-21 Mechanism of action of cholera toxin. The

cholera toxin’s A1 fragment in complex with ARF  GTP

catalyzes the ADP-ribosylation of a specific Arg residue on G

s␣

Figure 19-20 Model for signal transmission from an activated

GPCR to its target heterotrimeric G protein. (a) The X-ray

structure of activated bovine opsin (orange, oriented relative to

that in Fig. 19-16 by an ⬃90° rotation about the vertical axis) in a

complex with the 11-residue C-terminal helical segment from G

t␣

(blue).The X-ray structure of G

t␣

␤␥

(gray) is positioned over

the opsin structure such that this portion of the G

t␣

’s C-terminal

helix (the remainder of which is red) is superimposed on the

helical segment bound to opsin.The dashed lines delineate the

membrane in which opsin is embedded. Note that this model

results in a steric clash between the G protein and the membrane.

(b) To alleviate this steric clash, the G protein has been rotated

by 40° (lower yellow arrow) through a conformational change in

the loop preceding the C-terminal helix of G

t␣

.This loop

participates in binding guanine nucleotides (GDP and GTP) and

hence it is postulated that its reorganization expels GDP (drawn

in space-filling form with C magenta, N blue, and O and P

orange) from its binding pocket (upper yellow arrow). On the

subsequent dissociation of opsin, the G

t␣

binds GTP (which the

cell maintains at an ⬃10-fold higher concentration than GDP),

and consequently dissociates from G

␤␥

. [Courtesy of Oliver

Ernst, Charité–Universitätsmedizin Berlin, Germany. PDBids

3DQB and 1GOT.]

(a)

(b)

–

(CH

)

OHOH

denosine

NAD

Nicotinamide

Arg

s␣

–

(CH

)

OHOH

Adenosine

ADP-ribosylated G

s

s␣

A1 subunit of

cholera toxin

ARF  GTP

by NAD

⫹

, thereby rendering this subunit incapable of hydrolyz-

ing GTP.

JWCL281_c19_671-743.qxd 6/30/10 2:22 PM Page 695

ADP-ribosylated G

s

 GTP can activate AC but is in-

capable of hydrolyzing its bound GTP. As a consequence,

the AC remains “locked” in its active state. The epithelial

cells of the small intestine normally secrete digestive fluid

(an -rich salt solution) in response to small increases

in [cAMP] that activate intestinal Na



pumps through their

phosphorylation by PKA (ion pumps are discussed in Sec-

tions 20-3 and 20-4). The ⬃100-fold rise in intracellular

[cAMP] induced by CT causes these epithelial cells to pour

out enormous quantities of digestive fluid, thereby produc-

ing the symptoms of cholera. CT also affects other tissues

in vitro but does not do so in vivo because CT is not ab-

sorbed from the gut into the bloodstream.

The remarkable X-ray structure of CT (Fig. 19-22a), de-

termined by Graham Shipley and Edwin Westbrook, re-

veals that its A2 segment forms an unusual extended helix

whose C-terminal end inserts into the B

pentamer’s solvent-

filled central pore, where it is noncovalently anchored.

The N-terminal segment of A2 extends beyond the B

pen-

tamer so as to tether the wedge-shaped A1 segment to B

much like a balloon on a string.The X-ray structure of only

in complex with the pentasaccharide from its G

re-

ceptor (Fig. 19-22b), determined by Wim Hol, indicates that

this pentasaccharide binds through an extensive hydrogen

bonded network to each B subunit on the face of B

opposite

HCO



that which binds A.The binding of the A subunit or the re-

ceptor pentasaccharide to B

causes only modest structural

changes at their respective binding sites without altering

’s subunit interfaces. A1 contains an elongated crevice in

the vicinity of a catalytically implicated residue, Glu 112,

that presumably forms its active site.

Certain strains of E. coli cause a diarrheal disease (trav-

elers’ diarrhea) similar to, although considerably less se-

vere than, cholera through their production of heat-labile

enterotoxin (LT), a protein that closely resembles CT

(their A and B subunits are 80% identical and form AB

toxins that have closely similar X-ray structures) and has

the same mechanism of action. The reasons for the differ-

ence in severity of these infections are unclear (cholera can

be fatal within hours, whereas enterotoxic strains of E. coli

usually only temporarily incapacitate adults, although they

are responsible for the deaths of hundreds of thousands of

children annually). It might be due to the modest structural

differences between the toxins, differences in the amounts

of toxin secreted, and/or variations in microbial ecology.

The foregoing results provide a structural basis for the

design of ligands that interfere with the binding of CT and

LT to their receptors. Since these receptors occur on the

surface of the intestinal epithelium, ligands that compete

with them need not pass through any membrane. This

696 Chapter 19. Signal Transduction

Figure 19-22 X-ray structure of cholera toxin. (a) The entire

complex as viewed parallel to the presumed direction of the

plane of the plasma membrane to which it binds, extracellular

side up. The A1 segment is cyan, the A2 segment is gray, and each

B subunit has a different color.Although the A1 and A2

segments in this structure form a continuous polypeptide chain,

residues 193–195, which immediately precede the peptide bond

that is cleaved on toxin activation, are disordered and hence not

visible here (upper left). The C-terminal end of the A2 helix

binds in the pentamer’s central pore. [Based on an X-ray

(a)

(b)

structure by Graham Shipley, Boston University School of

Medicine, and Edwin Westbrook, Northwestern University.

PDBid 1XTC.] (b) The structure of only the B

pentamer in

which each subunit is binding CT’s G

receptor pentasaccharide.

The structure is viewed as from the bottom of Part a. The

subunits of the B

pentamer are colored as in Part a and the

pentasaccharides are shown in ball-and-stick form with C green,

N blue, and O red. Note the pentamer’s large central pore.

[Based on an X-ray structure by Wim Hol, University of

Washington. PDBid 2CHB.]

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 696

greatly increases the usual ⬃500-D size limit for an effec-

tive drug candidate (Section 15-4Ba). Moreover,a large lig-

and is unlikely to enter the bloodstream and hence would

have minimal side effects. Consequently, the synthesis of

multivalent ligands that simultaneously bind with high

affinity to all five receptor binding sites on an AB

mole-

cule has yielded promising lead compounds against CT

and LT.

e. Pertussis Toxin ADP-Ribosylates G

i

Bordetella pertussis, the bacterium that causes pertussis

(whooping cough; a disease that is still responsible for

⬃300,000 infant deaths per year worldwide), produces an

protein, pertussis toxin (PT), that ADP-ribosylates a

specific Cys residue in G

i

. In doing so, it prevents G

i

from

exchanging its bound GDP for GTP and therefore from in-

hibiting AC. PT’s X-ray structure, determined by Randy

Read, reveals that its A and B subunits are structurally ho-

mologous to those of CT and LT, although the A subunit of

PT extends from the opposite face of its B pentamer rela-

tive to that in CT. Moreover, PT’s B pentamer consists of

four different subunits (one in two copies), each of which is

only ⬃15% identical to the B subunits of CT and LT.

f. Receptors Are Subject to Desensitization

One of the hallmarks of biological signaling systems is

that they adapt to long-term stimuli by reducing their re-

sponse to them, a process named desensitization. These sig-

naling systems therefore respond to changes in stimulation

levels rather than to their absolute values. What is the mech-

anism of desensitization? In the case of -adrenergic re-

ceptors, an epinephrine–receptor complex, but not recep-

tor alone, is phosphorylated at one or more of the Ser or

Thr residues on its C-terminal tail by -adrenergic recep-

tor kinase (ARK). This cytosolic protein is recruited to

the plasma membrane through its interaction with G



when it is not bound to G

s

, which is also a consequence of

receptor activation. The phosphorylation of the receptor

decreases the ability of epinephrine to influence it, at least

in part by reducing the receptor’s epinephrine-binding

affinity. The phosphorylated receptor, in turn, is bound by

either of two 78% identical proteins known as -arrestins

in a way that blocks the activated receptor from activating

its target G protein. Moreover, the -arrestin binds to the

adaptor protein AP2 in clathrin-coated pits (Section 12-4Cd),

whereupon the -adrenergic receptor–-arrestin complex

is endocytotically sequestered (Section 12-5Bc) in spe-

cialized vesicles. The receptor’s otherwise extracellular,

epinephrine-binding surface then faces the interior of the

vesicle and the -arrestin is located on its outside, from

which it is subsequently released. The vesicles are devoid

of both heterotrimeric G protein and AC,thus further attenu-

ating the cell’s response to epinephrine. If, however, the ep-

inephrine level is reduced, the receptor is slowly dephos-

phorylated by a phosphatase and exocytotically returned

to the cell surface, thus restoring the cell’s epinephrine sen-

sitivity.Alternatively, the vesicles may fuse with endosomes

for the delivery of their contents to lysosomes, where the

receptor is proteolytically degraded (Fig. 12-91). In this latter

case, the restoration of epinephrine sensitivity requires the

synthesis of additional receptor.

ARK is a member of a family of seven proteins known

as GCPR kinases (GRKs) that,together with several other

types of protein kinases including PKA, phosphorylate the

C-terminal tails and/or the cytosolic loops of most agonist-

occupied GPCRs. As above, the phosphorylated GPCRs

are bound by a -arrestin and sequestered into endocytotic

vesicles, thereby isolating them from their corresponding

heterotrimeric G proteins. Moreover, the sites of a given

GCPR that are phosphorylated vary in a tissue-specific

way that provides further regulatory flexibility.

D. Adenylate Cyclases

The heterotrimeric G proteins G

and G

function to con-

trol the activities of adenylate cyclase (AC). In fact, mam-

mals have 9 known membrane-bound isoforms of AC,AC1

through AC9 (alternatively, AC-I through AC-IX), which

are each expressed in a tissue-specific manner and differ

in their regulatory properties. These ⬃120-kD transmem-

brane glycoproteins each consist of a small N-terminal do-

main (N), followed by two repeats of a unit consisting of a

transmembrane domain (M) followed by two consecutive

cytoplasmic domains (C), thus forming the sequence

(Fig. 19-23). The ⬃30% identical C

and C

domains associate to form the AC’s catalytic core,

whereas C

, as well as C

and C

, bind regulatory mole-

cules. Thus, G

i

inhibits AC1, 5, and 6 by binding to C

s

activates all AC isoforms but AC9 by binding to C



in-

hibits AC1, 3, and 8 but activates AC2, 4, and 7 by binding

Section 19-2. Heterotrimeric G Proteins 697

Figure 19-23 Schematic diagram of a typical mammalian AC.

The M

and M

domains are each predicted to contain six

transmembrane helices. C

and C

form the enzyme’s

pseudosymmetric catalytic core. The domains with which various

regulatory proteins are known to interact are indicated. [After

Tesmer, J.J.G. and Sprang, S.R., Curr. Opin. Struct. Biol. 8, 713

(1998).]

iα

sα,

PKC, G

βγ

ATP

2Mg

cAMP + PP

CaM,

PKA

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 697

to C

; and Ca

2⫹

–calmodulin (Ca

2⫹

–CaM; Section 18-3Ce)

activates AC1, 3, and 8 by binding to C

. Moreover, the C

of AC2, 5, and 7 are activated by the phosphorylation of

specific Ser/Thr control sites, for example, by protein

kinase C (PKC; Section 19-4C), whereas the C

of AC5

and 6 are similarly inhibited by PKA, for example. Clearly,

cells can respond to a great variety of stimuli in determin-

ing their cAMP levels.

No X-ray structure of an intact AC isoform has yet been

reported. However, Sprang has determined the X-ray

structure of a hybrid catalytic core consisting of the C

domain of AC5 (VC

) and the C

domain of AC2 (IIC

) in

complex with G

s␣

ⴢ GTP␥S, ATP␣S (an isomer of ATP␥S

with the S atom on the ␣ phosphate), and forskolin

(a product of the plant Coleus forskohlii that activates all

ACs but AC9 and functions to lower blood pressure). The

ⴢ IIC

catalytic core is enzymatically active and is sen-

sitive to both G

s␣

ⴢ GTP and forskolin. Its X-ray structure

(Fig. 19-24) reveals that VC

and IIC

form a pseudosym-

metric heterodimer that binds ATP␥S and forskolin at

pseudosymmetrically related sites of their interface.

s␣

ⴢ GTP␥S interacts with IIC

mainly via its Switch II

helix, which binds in a cleft on IIC

The X-ray structure of the catalytically inactive C

homodimer in complex with two symmetrically arranged

forskolin molecules, by James Hurley, provides a perhaps

crude model for the inactivated heterodimer. Comparison

of these structures (Fig. 19-24) suggests that the binding of

s␣

ⴢ GTP to the C

ⴢ C

catalytic core pries open the

Switch II binding cleft on C

in a way that mechanically

forces C

to rotate ⬃10° with respect to C

. This is postu-

lated to reorient the complex’s active site residues such

that they can efficiently catalyze the conversion of ATP to

cAMP. The conformational change that G

s␣

undergoes on

hydrolyzing its bound GTP to GDP (Fig. 19-18) apparently

reorients its Switch II region such that it no longer can bind

to C

, thereby causing AC to revert to its inactive confor-

mation.

has a cleft that corresponds to the G

s␣

-binding cleft

on IIC

. This suggests that this cleft on VC

provides the

binding site for G

i␣

. Indeed, mutagenesis studies on VC

are consistent with this hypothesis. However, the cleft on

is too narrow to accommodate the binding of a Switch

II helix. This further suggests that the binding of

i␣

ⴢ GTP to C

pries open this cleft in a way that reorients

Forskolin

the complex’s catalytic residues so as to reduce its cat-

alytic activity.

E. Phosphodiesterases

In any chemically based signaling system, the signal molecule

must eventually be eliminated in order to control the ampli-

tude and duration of the signal and to prevent interference

with the reception of subsequent signals. In the case of cAMP,

this second messenger is hydrolyzed to AMP by enzymes

known as cAMP-phosphodiesterases (cAMP-PDEs).

The PDE superfamily, which includes both cAMP-

PDEs and cGMP-PDEs, is encoded in mammals by at least

20 different genes grouped into 12 families (PDE1 through

PDE12). Moreover, many of the mRNAs transcribed from

these genes have alternative initiation sites and alternative

splice sites (Section 34-3C), so that mammals express ⬃50

PDE isoforms. These are functionally distinguished by

their substrate specificities (for cAMP, cGMP, or both) and

kinetic properties, their responses (or lack of them) to var-

ious activators and inhibitors (see below), and their tissue,

698 Chapter 19. Signal Transduction

Figure 19-24 X-ray structure of an AC catalytic core. This core

consists of dog VC

and rat IIC

in complex with bovine G

s␣

ⴢ

GTP␥S and forskolin, and is shown with a model of ATP. VC

tan, IIC

is violet, and G

s␣

, which is only shown in part, is dark

gray, with its IIC

-contacting segments, Switch II and the ␣3–␤5

loop, highlighted in red and blue. The forskolin and the ATP are

shown in stick form with C gray, N cyan, O red, and P green.The

brown ribbon shows the nonoverlapping portions of the

catalytically inactive rat IIC

homodimer in which one of its

subunits is superimposed on IIC

in the VC

–IIC

complex.

[Courtesy of Heidi Hamm, Northwestern University Medical

School.The X-ray structures of the VC

–IIC

–Gs

s␣

complex and

the IIC

homodimer were determined by John Tesmer and

Stephen Sprang, University of Texas Southwestern Medical

Center, and by James Hurley, NIH. PDBids 1AZS and 1AB8.]

␣1⬘–␣2⬘

␣2

Forskolin

Gs␣

␣3–␤4

␣1–␣2

ATP

IIC

␣3–␤5

␣3⬘

Adenylyl Cyclase

JWCL281_c19_671-743.qxd 6/30/10 1:14 PM Page 698

cellular, and subcellular distributions. The PDEs have

characteristic modular architectures with a conserved

⬃270-residue catalytic domain near their C-termini and

widely divergent regulatory domains or motifs, usually in

their N-terminal portions. Some PDEs are membrane-

anchored, whereas others are cytosolic.

PDE activity, as might be expected, is elaborately con-

trolled. Depending on its isoform, a PDE may be activated

by one or more of a variety of agents including Ca

2

–CaM;

phosphorylation by PKA, insulin-stimulated protein ki-

nase (Section 18-3Cg), and calmodulin-dependent protein

kinase II; and the binding of cGMP to a noncatalytic site.

However, for some PDEs, cGMP is inhibitory. Phosphory-

lated PDEs are dephosphorylated by a variety of protein

phosphatases including Ca

2

–CaM-dependent phosphatase

and protein phosphatase-2A. Thus, the PDEs provide a

means for cross talk between cAMP-based signaling sys-

tems and those using other types of signals.

PDEs are inhibited by a variety of drug agents that influ-

ence such widely divergent disorders as asthma, congestive

heart failure, depression, erectile dysfunction, inflammation,

and retinal degeneration. Sildenafil (trade name Viagra),

a compound used to treat erectile dysfunction, specifically

inhibits PDE5, which hydrolyzes only cGMP. Sexual stimu-

lation in males causes penile nerves to release NO, which

activates guanylate cyclase to produce cGMP. This induces

vascular smooth muscle relaxation in the penis, thereby in-

creasing the inflow of blood, which results in an erection.

This cGMP is eventually hydrolyzed by PDE5. Sildenafil is

therefore an effective treatment in men who produce in-

sufficient NO and hence cGMP to otherwise generate a

satisfactory erection.

3 TYROSINE KINASE–BASED

SIGNALING

See Guided Exploration 17: Mechanisms of hormone signaling

involving the receptor tyrosine kinase system

We have seen that

glycogen synthesis and breakdown are regulated by the

phosphorylation/dephosphorylation of the enzymes that

Sildenafil (Viagra™

™

)

SOO

catalyze these metabolic processes as well as of many of the

enzymes that catalyze these modification/demodification

processes (Section 18-3). Numerous other processes in

multicellular organisms are similarly regulated. In fact, over

one-third of the proteins in vertebrates are subject to re-

versible phosphorylation, and the human genome contains

518 protein kinase genes (the so-called kinome; the Protein

Kinase Resource at http://pkr.genomics.purdue.edu/ and

kinase.com at http://kinase.com/ are databases for the

kinome).The vast majority of the phosphorylated amino acid

residues are Ser or Thr; only about 1 in 2000 is Tyr. Never-

theless, as we discuss in this section, Tyr phosphorylation is

of central importance in regulating a variety of essential

cellular processes.

A. Receptor Tyrosine Kinases

Many protein growth factors variously control the differ-

entiation, proliferation, migration, metabolic state, and sur-

vival of their target cells by binding to their cognate recep-

tor tyrosine kinases (RTKs). The RTKs form a diverse

family of 58 transmembrane glycoproteins (Fig. 19-25) that

each have a C-terminal cytoplasmic protein tyrosine kinase

(PTK) domain and a single-pass transmembrane segment

that is presumably an  helix. As their name indicates,

PTKs catalyze the ATP-dependent phosphorylation of

their target proteins at specific Tyr residues:

The PTK domains of RTKs are homologous to and, as we

shall see, structurally resemble the far more abundant

Ser/Thr-specific protein kinases such as PKA (Fig. 18-15).

RTKs are activated by the binding of a cognate protein

growth factor to their ectodomains. It seems unlikely that

the single transmembrane helix of monomeric RTKs such

as platelet-derived growth factor receptor (PDGFR) and

epidermal growth factor receptor (EGFR; Fig. 19-25) has

the structural complexity to transmit the ligand-binding

state of its ectodomain to its cytoplasmic tyrosine kinase do-

main. Rather, as we have seen for human growth hormone

receptor (which is not an RTK), ligand binding induces

receptor dimerization (Fig.19-10).This, in turn,activates the

RTK’s PTK activity, as we discuss below. For RTKs that are

permanent dimers, such as the insulin receptor (InsR;

Fig. 19-25), the PTK is thought to be activated by a ligand-

induced structural change (probably a counter-rotation of

the two protomers that preserves the dimer’s 2-fold axis of

symmetry) that is transmitted across the membrane.

OHCH

Protein

O O



PCH

Protein

ATP

ADP

Section 19-3. Tyrosine Kinase–Based Signaling 699

JWCL281_c19_671-743.qxd 3/16/10 7:16 PM Page 699

a. FGF and Heparin Sulfate Are Required to

Activate FGF Receptor

The mammalian fibroblast growth factors (FGFs) form

a family of at least 21 structurally related proteins

(FGF1–21) that regulate a variety of critical biological

processes including cell growth, differentiation, and migra-

tion and that are expressed in specific spatial and temporal

patterns in embryos and adults. FGF-stimulated processes

are mediated by four FGF receptors (FGFR1–4), which

each bind a unique subset of the FGFs, thereby accounting

for the diversity and tight regulation of the foregoing

processes. FGFR dimerization in solution requires the

presence of heparan sulfate proteoglycans (Section 11-3A)

in addition to that of FGF.

FGF receptors each consist of, from N- to C-terminus

(Fig. 19-25), three extracellular immunoglobulin-like do-

mains (D1–D3 for domains 1–3), a single transmembrane

helix, and a cytoplasmic domain with PTK activity. Of

these, only the D2 and D3 domains are involved in FGF

binding (in general, only a few of the domains in the extra-

cellular portions of RTKs participate in ligand binding).

Moosa Mohammadi determined the X-ray structure of the

2:2:2 complex of FGF2, the D2–D3 segment of FGFR1,

and a heparin decasaccharide (Fig. 11-21). It reveals

(Fig. 19-26) that each FGF monomer binds to the D2 and

D3 domain on one FGFR subunit and, more tenuously, to

the D2 domain on the other subunit, whereas the heparin

cross-links each FGF monomer to both D2 domains

(whose contacts in the absence of FGF and heparin are in-

sufficient to support appreciable FGFR dimerization).

b. RTK Dimers Are Activated by

Autophosphorylation

The dimerization of an RTK (or its conformational

change in the case of the insulin receptor subfamily) brings

its cytoplasmically located PTK domains into apposition,

such that they cross-phosphorylate each other on specific

Tyr residues on their activation loops (Fig. 19-27a). This

autophosphorylation activates the PTK in much the same

way as we saw that activation loop phosphorylation in-

duces PKA to phosphorylate its target proteins (Section

18-3Cb). In many cases, the activated PTK further phos-

phorylates the opposing RTK subunit at specific Tyr

residues outside of the PTK domain (Fig. 19-27).This, as we

shall see in Sections 19-3Cb and 19-3Cc, provides binding

sites for certain cytoplasmic proteins. The activated PTK

may also phosphorylate specific Tyr residues on a variety of

cytoplasmic proteins. In both cases, as we discuss in Section

700 Chapter 19. Signal Transduction

EGFR

ErbB2

ErbB3

ErbB4

InsR

IGF1R

IRR

PDGFR

CSF1R

Kit

Flk2

Flt1

KDR

Flt4

FGFR1

FGFR2

FGFR3

FGFR4

TrkA

TrkB

TrkC

Ror1

Ror2

MuSK Met

Ron

Sea

Axl

Eyk

Tyro3

Nyk

Tie

Tek

EphA1

EphB1

Ret Ryk DDR1

DDR2

Ros

SAM

tyrosine

kinase

kringle

discoidin

cadherin

leucine-rich

EGF

(immunoglobulinlike)

fibronectin

type III

cysteine-rich

Figure 19-25 Domain organization in a variety of receptor

tyrosine kinase (RTK) subfamilies. One to five members of each

subfamily are depicted.The narrow rectangle that extends

horizontally across the diagram represents the plasma

membrane, with the extracellular region above it and the cytosol

below. The polypeptides are shown only approximately to scale,

with their N-termini above. EGFR, InsR, PDGFR, and FGFR

refer to epidermal growth factor receptor, insulin receptor,

platelet-derived growth factor receptor, and fibroblast growth

factor receptor, respectively. The RTKs’ extracellular portions

are modularly constructed from a variety of often repeating

domains that are identified at the right of the diagram. Note that

the tyrosine kinase domains of PDGFR and Flt1 subfamilies are

interrupted by ⬃100-residue kinase inserts and that the members

of the InsR subfamily are ␣

␤

heterotetramers, whose subunits

are disulfide-linked (short horizontal lines). [Courtesy of Stevan

Hubbard, New York University School of Medicine.]

JWCL281_c19_671-743.qxd 6/30/10 1:15 PM Page 700