P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
Annu. Rev. Biophys. Biomol. Struct. 2001. 30:191–209
c 2001 by Annual Reviews. All rights reserved
Copyright °
STRUCTURAL RELATIONSHIPS AMONG REGULATED
AND UNREGULATED PHOSPHORYLASES
Jenny L. Buchbinder1, Virginia L. Rath2, and
Robert J. Fletterick1
1Department of Biochemistry and Biophysics, University of California, San Francisco,
San Francisco, California 94143; e-mail: [email protected]; [email protected]
2Exploratory Medicinal Sciences, Central Research, Pfizer Inc., Eastern Point Road,
Groton, Connecticut 06340; e-mail: [email protected]
Key Words allosteric, enzymes, molecular structure, site-directed mutagenesis,
X-ray diffraction
■ Abstract Species and tissue-specific isozymes of phosphorylase display differences in regulatory properties consistent with their distinct roles in particular organisms and tissues. In this review, we compare crystallographic structures of regulated
and unregulated phosphorylases, including maltodextrin phosphorylase (MalP) from
Escherichia coli, glycogen phosphorylase from yeast, and mammalian isozymes from
muscle and liver tissues. Mutagenesis and functional studies supplement the structural
work and provide insights into the structural basis for allosteric control mechanisms.
MalP, a simple, unregulated enzyme, is contrasted with the more complicated yeast and
mammalian phosphorylases that have evolved regulatory sites onto the basic catalytic
architecture. The human liver and muscle isozymes show differences structurally in
their means of invoking allosteric activation. Phosphorylation, though common to both
the yeast and mammalian enzymes, occurs at different sites and activates the enzymes
by surprisingly different mechanisms.
CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
E. coli MALTODEXTRIN PHOSPHORYLASE (MalP) . . . . . . . . . . . . . . . . . . . . .
An Unregulated Catalytic Enzyme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Domain Structure—Preserved in Bacterial and Human Enzymes . . . . . . . . . . . . .
Oligosaccharide Binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
YEAST PHOSPHORYLASE, AN EVOLUTIONARY ADVANCE IN
REGULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Regulation at Its Simplest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effectors at the Dimer Interface—A Phosphopeptide Competes with Glc-6-P . . . .
MUSCLE PHOSPHORYLASE—THE MOST ELABORATELY REGULATED . . .
A Competition by Activators and Inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1056-8700/01/0610-0191$14.00
192
193
193
193
194
194
194
197
197
197
198
191
P1: GDL
April 14, 2001
192
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
Mutagenesis Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
HUMAN LIVER PHOSPHORYLASE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Regulation in the Unselfish Organ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Crystallographic Structures Show Changing Interfaces . . . . . . . . . . . . . . . . . . . .
Inhibitors of Liver Phosphorylase for Treatment of
Non-Insulin-Dependent Diabetes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONCLUSIONS: Evolution Provides Regulatory Diversions . . . . . . . . . . . . . . . . .
200
201
201
202
204
205
INTRODUCTION
In the early biotic world, the ability to maintain carbohydrate reserves must have
granted selective advantage to those organisms with even primitive management of
glycogen. In human beings, glucose is stored as glycogen in high concentrations
in skeletal muscle and liver tissues, where glycogen degradation is exquisitely
regulated by the control of the activity of glycogen phosphorylase. Phosphorylase
responds to metabolites within the cell, including AMP, ATP, glucose 6-phosphate
(Glc-6-P), and glucose, and also to hormonal signals from epinephrine at the cell
surface. Regulation is achieved primarily by covalent phosphorylation and by
binding of metabolite effectors, which mediate conformational changes between
inactive and active forms of the enzyme. Isozymes with specialized regulatory
properties are expressed in different cell types and provide tissue-specific control
of glycogen degradation. Three isozymes of phosphorylase are found in human
beings and are named after the tissues where they are predominantly expressed:
brain, liver, and muscle. They share ∼80% sequence identity. All are activated
by phosphorylation at Ser-14 but differ primarily in their responses to allosteric
effectors. In this review, we examine phosphorylases from the primitive versions
in Escherichia coli and yeast to the elaborate sensory molecules found in human
tissues. We compare the most simple unregulated enzyme with the more complicated species and tissue-specific isozymes that have evolved distinct mechanisms
of allosteric control.
Biochemical and physiological studies of phosphorylase began in the late 1930s
and continue to this day (23, 44, 45). Phosphorylases from all species are active as
homodimers with molecular weights of ∼200,000. All have highly conserved
active sites and catalyze phosphorolysis of glycogen by the same catalytic mechanism. The product, Glc-1-P, is converted to Glc- 6-P for entry into the glycolytic
pathway to provide energy to the cell. Nearly three decades of crystallographic
studies of phosphorylases from different species provide structures of phosphorylated and unphosphorylated forms of the enzyme in assorted complexes with
activators, substrates, and inhibitors (6, 20, 30, 32). Regulated and unregulated
phosphorylases adopt a range of conformational states, which differ substantially in
structure and subunit association. Mutagenesis and functional studies have helped
define the mechanisms linking changes in structure to changes in function. We
have organized this review from an evolutionary perspective, primarily to achieve
focus and simplicity, though at the expense of certain details. We start with a brief
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
193
description of the structure of the E. coli homologue, the simplest, unregulated
phosphorylase, and then proceed to discuss the regulated enzymes from yeast and
mammals. We focus on the mammalian isozymes from muscle and liver tissues.
The brain isozyme is omitted from the review because no crystallographic structures are available for it. We address the questions: How are the regulated enzymes
inhibited and activated by effectors binding at sites distant from the active site?
What is the structural basis for allosteric control? How do structural elements link
the allosteric sites with one another and with the active site? How are effector sites
related, from an evolutionary perspective, among the phosphorylases?
E. coli MALTODEXTRIN PHOSPHORYLASE (MalP)
An Unregulated Catalytic Enzyme
MalP is a homologue of glycogen phosphorylase in E. coli that catalyzes the
phosphorolysis of α (23, 30, 44, 45) glucosyl linkages from linear rather than from
branched oligosaccharides. It has less than 1% activity with branched glycogen
(60). Unlike regulated phosphorylases, the enzyme is not inhibited by Glc-6-P nor
activated by AMP or phosphorylation. Enzyme levels are controlled instead by
transcriptional regulation (12). MalP represents a basic catalytic phosphorylase
with no allosteric controls.
Domain Structure—Preserved in Bacterial
and Human Enzymes
Crystallographic studies show that the enzyme has an α/β fold consisting of two
domains (Figure 1; 68). A distinct surface structure within the N-terminal domain contains a site for binding maltodextrose. Muscle phosphorylase (mGP)
attaches to glycogen particles at this position. The C-terminal domain has the
familiar nucleotide binding fold similar in structure to lactate dehydrogenase,
but with a coating of additional α-helices. The PLP cofactor is covalently linked
to the side chain of Lys-680 within the nucleotide binding fold adjacent to the
substrate. The active site of the enzyme is located in a crevice between the
N-terminal and C-terminal domains.
The N-terminal’s hundred or so amino acids are especially important for allosteric control in regulated phosphorylases. Two simple structural elements, the
“cap” loop and the α2 helix, which directly follows the cap, form approximately
half the intersubunit contacts in the phosphorylase dimer. The phosphoserine and
AMP binding sites in mGP nestle close to each other between the twofold related
cap and the α2 helix. These regulatory sites lie about 40 Å away from the active
site. MalP lacks the phosphoserine site of mGP because the first 17 residues of
mGP, which form the first α helix of the structure, are missing from its sequence.
The binding site for AMP is also absent in MalP because of sequence differences
in residues from the cap and α2 helix. Arg-309 and Arg-310, which bind the
P1: GDL
April 14, 2001
194
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
phosphate of AMP in mGP, have their side chains pointing in opposite directions
in the E. coli enzyme. The intersubunit contacts in MalP are unrelated to those
in structures of phosphorylases from other species. Regulatory sites apparently
developed at the dimer interface by the incorporation of side chains that were
appropriately positioned to bind effectors.
Oligosaccharide Binding
The structure of MalP, cocrystallized with maltohexaose, provides the only phosphorylase structure featuring an oligosaccharide bound at the catalytic site (51).
Comparison to the structure of the enzyme alone indicates that an induced-fit
mechanism closes the N- and C-terminal domains over the oligosaccharide substrate. An active site element, the 380’s loop moves by about 7 Å in response to
oligosaccharide binding to close the domain interface. Domain closure is accompanied by formation of ionic and van der Waals interactions that stabilize the loop
in this position. Although the 380’s loop residues form no direct interactions with
the oligosaccharide, site-directed mutagenesis studies show that the substitution
E382A reduces the affinity for oligosaccharide and suggest the importance of these
interactions for proper positioning of active site residues (16).
Perhaps correlated with the structural reorganization, oligosaccharides are
known to bind weakly to the active sites of phosphorylases. Although MalP was
crystallized with maltohexaose, the structure shows only two glucose units bound
at the active site (Figure 2). All hydroxyls of the sugar form hydrogen bonds to
residues at the active site. The side chain of the universally conserved Tyr-280
points into the active site, where it forms a stacking interaction with the hydrophobic face of the sugar. Site-directed mutagenesis studies confirm the importance of
many of these residues for oligosaccharide recognition and catalytic activity (16).
A Y280A substitution increases the Km for maltoheptaose 200-fold and reduces
kcat/Km by 104 (51).
The constitutive activity of MalP contrasts with regulated phosphorylases and
may be accounted for in part by differences in the position of a movable gate, the
280’s loop, which controls access to the active site (51, 68). The 280’s loop in MalP
is located away from the active site. The 280’s loop, in structures of inactive mGP,
enters the active site, blocks substrate binding, and furthermore binds inhibitors
(33, 34, 64). In structures of inhibited mGP, Arg-569, which binds the phosphate
and oligosaccharide substrates in MalP, is not positioned correctly to bind either
substrate (4, 22, 64).
YEAST PHOSPHORYLASE, AN EVOLUTIONARY
ADVANCE IN REGULATION
Regulation at Its Simplest
Regulation of yeast phosphorylase (yGP) is relatively simple compared to the more
complicated mammalian isozymes. It is activated solely by phosphorylation and
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
195
inhibited only by binding of Glc-6-P (5, 21). A yeast cAMP-dependent protein
kinase phosphorylates Thr-−10 to activate yeast yGP. Thr-−10 is located in an
N-terminal extension (residues −1 to −39, numbered backwards from residue 1
of the mGP sequence) not found in either MalP or mGP. Structures of unphosphorylated and phosphorylated forms of yGP provide a glimpse of the allosteric
transition between inhibited and activated states (39, 40, 57, 58).
Phosphorylation control mechanisms are known for only a handful of enzymes,
and a similarity between yeast and higher eukaryotes is often assumed. It comes
as a surprise that the mechanism by which phosphorylation activates yGP is utterly unrelated to that of mGP. Sequence differences at the N-terminus of the
yeast enzyme eliminate the phosphorylation site found in mGP (28). Yeast GP
has an Asn in place of Ser-14, the site of regulatory phosphorylation in mGP,
and in the region that corresponds to the N-terminus of mGP, the two enzymes
have only one identical residue. It is surprising that the structure of yeast GP b
(unphosphorylated phosphorylase) shows that this region adopts a conformation and position more similar to that of the phosphorylated than to that of
the unphosphorylated muscle enzyme (57, 63). This observation implies that an
ancient precursor enzyme was active with its N-terminus in a helical conformation bound between the subunits. The mammalian enzyme became inactive
when its N-terminus became extended and no longer bound at the dimer interface. Phosphorylation is required in the muscle enzyme to initiate the refolding of its N-terminus into a helix that binds between the subunits. Electrostatic
repulsion between the phosphate and surrounding acidic residues may be required in the muscle enzyme to drive the movement of the N-terminus to the
subunit interface (11, 31). In the yeast enzyme, differences in the sequence allow this region to adopt a helical conformation and dock at the subunit interface in the absence of phosphorylation. Moreover, phosphorylation of the yeast
enzyme has little effect on the structure of this region (40, 57). yGP is inactive though because the subunit contacts have evolved to stabilize the inactive
form.
Ionic interactions that form in mGP between the negatively charged serinephosphate and Arg-43 and Arg-69 are not conserved in the yeast enzyme. Another
difference between the yeast and muscle enzymes is that the C-terminus of yGP
is shorter than in the muscle enzyme. C-terminal residues in mGP compete with
the phosphorylated N-terminus for binding at the subunit interface (63). In yGP,
the corresponding C-terminal residues that impede binding of the N-terminus in
mGP are absent.
The location of the N-terminal extension of yGP depends on the phosphorylation
of Thr-−10 (Figure 3). In unphosphorylated yGP, residues −27 to −38 of the Nterminal extension bind between the N- and C-terminal domains to block the
entrance to the active site (57). The position of the N-terminus indicates that
it regulates activity by controlling access of substrates to the active site. Both
hydrophobic and ionic interactions fuel the conformational change. In response
to phosphorylation of the enzyme, the N-terminus refolds and moves away from
the active site to the subunit interface (40). Residues from 5 to −7 assume an
P1: GDL
April 14, 2001
196
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
α-helical conformation, and residues from −8 to −10 form a hairpin structure that
binds between the subunits to build a hydrophobic cluster. The phosphorylated
Thr-−10 forms ionic interactions with the side chains of Arg-309 and Arg-310 at
a site that corresponds to the AMP binding site in mGP (40). In striking mimicry
of mammalian AMP activation, the phosphopeptide competes with Glc-6-P for
binding the pair of arginines. Ionic interactions with the phosphopeptide serve to
properly position the N-terminal peptide at the dimer interface. Phosphorylation
is accompanied by a small increase in the separation of the two subunits. The
movement of the subunits apart, as a result of phosphorylation, contrasts with
mammalian activation in which phosphorylation or AMP binding draw the subunits
closer together.
The structural basis for allosteric control extends to the active site through
specific linkages. In the unphosphorylated, inhibited state, the 380’s loop forms
contacts with the N-terminus near the entrance to the active site (57). The
N-terminus and the 380’s loop together block binding of substrates to the enzyme in the inactive state. These contacts are disrupted during phosphorylation
by the refolding and movement of the N-terminus, and as a result, the 380’s loop
moves away from the active site (40). The two active sites are linked by the two
tower helices and active site gates. The change in position of the tower helices
as a result of phosphorylation brings together a cluster of conserved hydrophobic residues Phe-252, Leu-254, Phe-257, Tyr-262, Tyr-1630 , Val-2780 , Tyr-2800 ,
Pro-2810 , Phe-2850 , and Leu-2910 near the active site, which stabilize the activated
enzyme. The conservation of these residues in all phosphorylases suggests that
the condensation of hydrophobic side chains is a common feature in all active
phosphorylases (39). Figure 3 shows the hydrophobic residues, which form a cluster in the phosphorylated form of the enzyme. The aromatic ring of Tyr-280 flips
180◦ during activation to point into the active site. Although no structure has ever
been determined of yGP with oligosaccharide bound at the active site, the structure of MalP with maltohexaose shows that Tyr-280 is a key residue involved in
oligosaccharide recognition (51).
In both the unphosphorylated and phosphorylated forms of yGP, the separation
between the N-terminal and C-terminal domains is greater than in either the complex of MalP with oligosaccharide or the complex of mGP inhibited with glucose
(40, 51, 57, 64). That is, the domains are rotated apart such that the active site cleft
would be open for substrate binding in the unphosphorylated enzyme if not for
the blockage by the N-terminus and the 380’s loop. The active site gate of yGP
a is 10 Å away from its position in structures of inhibited mGP, and it is in the
position identical to that in MalP. The movement of the N-terminus and the 380’s
loop thus appears to promote substrate binding and link the two active sites. In
structures of both unphosphorylated and phosphorylated forms of yGP, the gate is
held in a position where it does not block the active site (40, 58). The side chain
of Arg-569 in both unphosphorylated and phosphorylated forms of yGP occupies
the same position observed for activated mGP and is in the proper position to bind
the substrate phosphate.
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
197
Mutagenesis studies further support a role for the N-terminal extension in regulation of activity (38). An enzyme variant that has the first 42 residues of yGP
deleted displays significant activity, though it is incapable of being phosphorylated.
Removal of the N-terminus results in partial activation of the enzyme, presumably
because substrates can bind at the unblocked active site. Without the N-terminal
phosphopeptide at the dimer interface to induce further conformational changes,
however, the mutant does not achieve full activation.
Effectors at the Dimer Interface—A Phosphopeptide
Competes with Glc-6-P
Glc-6-P inhibits yGP by displacing the phosphorylated N-terminus from the effector site (40). Residues that bind Glc-6-P in yGP are mostly the same as in mGP,
though the mode of binding of Glc-6-P differs in the two enzymes (58). Muscle
GP binds the α anomer of Glc-6-P, whereas yGP binds the α anomer. The precise
positioning of the inhibitor and the specific contacts it makes with the enzyme differ accordingly in the two structures. Notably, the hexose rings of Glc-6-P are
oriented differently, by 90◦ , in the two enzymes. In yGP, the position of Trp-67
matches that of the activated muscle enzyme, which necessitates the change in position of the hexose ring of Glc-6-P. Like the muscle enzyme, Arg-309 and Arg-310
form ionic interactions with the phosphate group of the inhibitor. The phosphate
group of Glc-6-P forms ionic interactions with the same two arginine residues
as the Thr-−10-phosphate of yGP. Since interactions of the phosphate groups of
activators and inhibitors at this site are similar, other interactions with the effectors
must differentiate the allosteric response. At the subunit interface, interactions with
hydrophobic residues of the N-terminus differ from those with Glc-6-P and provide the discrimination. Hydrophobic interactions of the phosphopeptide inserted
between the subunits lead to the separation of the subunits that triggers activation.
Unlike mGP, AMP does not activate yGP because it lacks the residues in mGP that
interact with the adenine ring of AMP (28).
MUSCLE PHOSPHORYLASE—THE MOST
ELABORATELY REGULATED
A Competition by Activators and Inhibitors
In skeletal muscle, phosphorylase activity is activated during muscle contraction
when glycogenolysis is required to produce ATP for energy. The enzyme is regulated by metabolites of glycolysis, including ATP and glucose-6-P, which act as
feedback inhibitors, and AMP, which serves as an allosteric activator (14, 15, 35, 50,
54). In addition, the enzyme is activated during stress by covalent phosphorylation
at residue Ser-14. Phosphorylase is interconverted between unphosphorylated and
phosphorylated states that have different allosteric properties by phosphorylase
P1: GDL
April 14, 2001
198
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
kinase and protein phosphatase PP-1, whose activities in turn are controlled by a
complex multienzyme cascade responsive to hormonal regulation, nerve stimulation, and intracellular Ca2+ concentration (13, 17–19, 29, 37).
Structure
The overall structure of the enzyme is similar to other phosphorylases; however,
the subunits associate in different orientations than in E. coli, MalP, or yGP, and
mGP has regulatory sites not present in the other enzymes (20, 32). The site of
covalent phosphorylation and the binding site for AMP in mGP reside in an activation subdomain consisting of the first six N-terminal helices. Inhibitors ATP
and Glc-6-P bind competitively with AMP at overlapping regulatory sites in this
region (46, 50, 54, 67). Activation of the enzyme by phosphorylation at Ser-14 is
associated with a structural rearrangement of the N-terminus and movement of
the N-terminus over a distance of 35 Å to the subunit interface (Figure 4; 63). In
phosphorylase b, N-terminal residues 5–16 are disordered, and residues 10–22,
which have an extended conformation, form intrasubunit contacts (47, 69). Phosphorylation causes the N-terminus to refold into a 310 helix, and upon relocation
to the subunit interface, the phosphoserine forms ionic interactions with Arg-430
of the cap (residues 40–45) and Arg-69 of the α2 helix within the activation
subdomain.
AMP binds to the subunit interface between the cap and the adjacent α2 helix at
a site approximately 10 Å away from the phosphoserine. The phosphate group of
AMP forms ionic interactions with Arg-309 and Arg-310, which were also shown
to have a role in binding Glc-6-P in yGP. The purine ring of AMP forms an aromatic
stacking interaction with the aromatic ring of Tyr-75 in the third helix of the
activation subdomain, and the ribose group forms a hydrogen bonding interaction
with the side chain of Asp-420 from the cap. The formation of intersubunit contacts
mediated by AMP or the phosphoserine draws the cap0 and α2 helix closer together
(3, 62, 63, 66). This movement of secondary structural elements triggers a transition
in quaternary structure. The subunits rotate in opposite directions by 5◦ about
two axes approximately perpendicular to the molecular twofold (3, 4, 22, 66, 65).
Further tertiary structural changes propagate allosteric effects to the active site.
At the opposite end of the subunit interface, the tower helices (helix α7, residues
264–277) form intersubunit contacts that link the two active sites. These tower
helices are named for their obvious extension away from the body of the monomer.
Intersubunit contacts between the helices are altered by the rotation of the subunits
(4, 22). In the inactive state, the tower helices assume an approximately antiparallel
configuration. Rotation of the subunits changes the geometry of the helices such
that the angle formed by the helices crossing each other is approximately 80◦ . The
reorientation of the tower helices is accompanied by the movement of a loop of
amino acids (residues 280–288 are the active site gate), which directly follows each
tower helix. The role of the active site gate is similar to the N-terminal peptide of
yGP that controls access of substrates to the active site. The gate blocks the active
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
199
site when the enzyme is in an inactive conformation and moves away from the
active site when the enzyme is in an activated conformation (4, 22, 70). Mimicry
as seen at the Glc-6-P site in yGP is also apparent here. Residue Asp-283 mimics
the substrate phosphate at the active site (4, 22). In the inactive conformation,
the carboxylate side chain occupies the phosphate subsite of the active site and
electrostatically and sterically interferes with substrate binding. The carboxylate
group of Asp-283 forms charged hydrogen bonds with the side chains of His-571
and Asn-284. These interdomain contacts contribute to the stabilization of the
inactive conformation of the enzyme in which the N- and C-terminal domains are
rotated toward each other to close the active site.
The two residues following Asp-283, Asn-284 and Phe-285, of the active site
gate loop participate in binding allosteric inhibitors. Asn-284 forms a hydrogen
bond with the C2 hydroxyl group of glucose, which binds at the active site at the
position of the substrate, glucose 1-phosphate (Glc-1-P) (64). The aromatic ring
of Phe-285 forms aromatic stacking interactions with purine inhibitors that bind
just outside the active site (61). The identity of the physiological inhibitor that may
bind at this conserved purine site in vertebrate phosphorylases remains a mystery.
Binding of glucose and purine inhibitors stabilizes the inactive conformation of the
active site gate loop and hinders substrate binding. yGP is not inhibited by glucose
even though all of the residues that are involved in binding glucose in mGP are
conserved, perhaps because the active site gate does not assume the correct position
to bind the inhibitor.
During activation of mGP by either AMP or phosphorylation, the N- and
C-terminal domains rotate apart by 5◦ , the active site gate moves away from the
active site, and Arg-569 moves into the correct position to bind the substrate, phosphate (3, 4, 22, 63, 66). The opening of the domains and the movement of the active
site gate make the active site accessible to oligosaccharide and phosphate. Further
structural changes may be required for activation; however, the crystallographic
structures of mGP, either phosphorylated or with AMP bound, do not depict full
activation since the enzyme crystallized as a tetramer and not as a functional dimer.
Glc-6-P inhibits phosphorylase by competing with AMP (67). The binding site
for the inhibitor overlaps that of the activator. The phosphate group of Glc-6-P,
similarly to AMP, forms ionic interactions with the side chains of Arg-309 and
Arg-310, and in addition, Arg-242 contributes an ionic interaction. The position of
the glucopyranose ring differs from that of the ribose ring of AMP. Phosphorylase
senses the state of phosphorylation of the nucleotide in the activator site so that
only AMP activates. ATP and ADP, like Glc-6-P, inhibit the enzyme by competing
with AMP for binding at the subunit interface. Discrimination among nucleotides is
based on the charged interactions of the phosphate groups and different placements
of their adenine rings. For ATP, the α- and β-phosphate groups form ionic interactions with Arg-309 and Arg-310 (62). Additional ionic interactions are formed
between the γ -phosphate of ATP and Lys-410 and the β-phosphate and Arg-242.
The ribose ring of ATP forms one hydrogen bonding interaction with the enzyme.
The adenine ring of ATP is shifted from the position observed for AMP because of
P1: GDL
April 14, 2001
200
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
the longer triphosphate cluster. The adenine ring extends out into the solvent and
forms no productive interactions with the enzyme. ATP and Glc-6-P form a different set of interactions than the activator AMP and similarly inhibit the enzyme by
mediating intersubunit contacts that stabilize the inactive quaternary association of
the subunits. Unlike AMP, these inhibitors do not cause the appropriate movement
of the cap and α2 helix closer together than is necessary for triggering activation.
It is only the correct condensation of the three elements of structure, the cap, the
α2 helix, and the arginine cluster, that activates and only AMP can perform the
condensation.
Mutagenesis Studies
Allosteric activation of mGP can be mimicked by the binding of divalent transition state metals between the cap and α2 helix (7). Site-directed mutagenesis was
used to replace Val-45 of the cap and Tyr-75 of the helix with histidine residues to
create a metal binding site at the subunit interface. As metal ions bind to residues
from both monomers, the cap and α2 helix move closer together, simulating the
effects of AMP binding or phosphorylation. Activation levels of up to 10% of that
for AMP have been achieved by the binding of metal ions at engineered divalent
cation sites. mGP can also be partially activated by acidic substitutions at Ser-14,
the site of regulatory phosphorylation (11). An S14D mutant exhibits a 1.6-fold
increase in Vmax, a tenfold decrease in the apparent dissociation constant for AMP,
and a threefold decrease in the S0.5 for Glc-1-P. An S14E mutant shows a 2.2fold increase in Vmax, a sixfold decrease in the apparent dissociation constant for
AMP, and a twofold decrease in the S0.5 for Glc-1-P. Both mutants still require
AMP for activity and show cooperativity for Glc-1-P in contrast to the phosphorylated enzyme. Native GP a (phosphorylated phosphorylase) neither displays
substrate cooperativity nor requires AMP for activation because phosphorylation
converts the enzyme almost entirely to the active state. An acidic substitution at the
N-terminus apparently does not induce the quaternary transition associated with
allosteric activation on its own. In the presence of AMP, however, which triggers
the subunit rotation, the mutants are activated to a greater extent than native mGP b.
The mutations may, therefore, enable the N-terminus to refold and dock at the subunit interface when the monomers are oriented in their activated configuration.
Molecular modeling indicates that a carboxylate side chain at the position of the
serine phosphate may participate in salt-bridge interactions with the arginines that
normally interact with the phosphoserine. Differences in the charge and in the geometry of carboxylate versus phosphate groups and the weaker ionic interactions
with charged residues presumably prevent full activation.
The crystallographic studies suggest the importance of the tower helix and active
site gate in coupling structural changes induced by allosteric effectors binding at
the subunit interface to changes at the active site. The tower helices undergo a large
change in tertiary structure during the allosteric transition, though the movement of
the tower helices during activation could simply be the result of the subunit rotation
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
201
triggered by AMP or phosphorylation. Mutagenesis studies designed to test the
functional significance of particular tower helix, gate interactions, and structural
changes in these regions show that mutations in the tower helix and active site gate
do affect allosteric regulation (9, 10). The entire tower helix of mGP was replaced
with the tower helix from yGP to test the importance of specific intersubunit
interactions on allosteric activation and catalytic activity. Only 33% of the amino
acids in the helix of yGP are identical to those of mGP, and the helix replacement
changes helix-helix intersubunit interactions. Kinetic studies of this mutant show
that the replacement with the yGP tower helix results in an eightfold reduction
in Vmax, a loss in cooperativity for both AMP and Glc-1-P, but little change in
the Ka for AMP, and the [S]0.5 for the Glc-1-P is reduced twofold. The ability of
this mutant to be activated by AMP shows that the helix replacement does not
destroy the coupling between the AMP and active sites. The relatively modest
loss in catalytic activity shows that the specific interactions of the helices are not
essential for allosteric activation or catalytic activity, though they do determine
the precise response to allosteric effectors. The helices have been tuned during
evolution to provide a calibrated response at the active site to regulatory signals
from the opposite face of the dimer.
Some amino acids in the active site have multiple functions. The importance of
active site gate residue Asp-283 for catalytic activity and allosteric regulation was
tested by site-directed mutagenesis (9). Replacement with either Asn or Ala results
in a tenfold stronger association of Glc-1-P with retention of full substrate cooperativity. The increased apparent affinity for Glc-1-P suggests that the substrate
binds to the inactive conformer in these mutants because of the loss of competition between the substrate phosphate and the negatively charged carboxylate side
chain of Asp-283. The 15-fold reduced levels of catalytic activity in the D283A
and D283N mutants suggest that Asp-283 plays more than an inhibitory role at the
active site. Although no structure is available for mGP with oligosaccharide bound
at the active site, the structure of MalP with maltohexaose suggests that residues
of the active site gate loop participate in oligosaccharide recognition. In MalP, a
D283A substitution results in a sevenfold increase in the Km for maltoheptaose and
a threefold decrease in kcat. The structure of MalP shows that Asp forms no direct
interactions with the oligosaccharide but may contribute secondary interactions
that affect the positioning of residues at the active site.
HUMAN LIVER PHOSPHORYLASE
Regulation in the Unselfish Organ
The liver and muscle isozymes fulfill different physiological requirements and
accordingly respond differently to allosteric effectors. In muscle, where the role
of phosphorylase is to mobilize carbohydrate for intracellular use, the enzyme
displays a greater sensitivity to ligands, such as AMP and ATP, which signal the
energy state of the cell. The unphosphorylated form of mGP is potently activated
P1: GDL
April 14, 2001
202
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
by AMP and is sensitive to ATP or Glc-6-P, which competitively inhibit AMP
binding (8, 25, 26). In contrast, the primary role of the liver enzyme is to break
down glycogen for release as free glucose into the bloodstream to ensure a constant
supply of this substrate to other tissues. The liver isozyme is consequently less
sensitive to intracellular ligands. The unphosphorylated form of liver glycogen
phosphorylase (lGP) is inactive, shows only weak activity in the presence of AMP
(10%–20%), and is insensitive to inhibition by ATP or Glc-6-P (36, 42, 43). The
enzyme, instead, depends strongly on phosphorylation to achieve activity. The
phosphorylation state of the enzyme is subject to hormonal control; thus, lGP
activity is regulated primarily extracellularly in keeping with the function of this
isozyme in providing glucose outside of the cell. It is uncertain from an analysis
of sequence differences between the isozymes what accounts for their differential
responses to ligands. The liver and muscle enzymes are about 80% identical and,
with one exception, all of the residues that bind AMP are conserved.
Crystallographic Structures Show Changing Interfaces
The first structures of human lGP in both active and inhibited conformations
have recently been determined (56). The phosphorylated enzyme was crystallized in an active conformation with AMP bound and in an inactive conformation complexed with a potent but nonphysiological glucose analog, N-acetyl-β-Dglucopyranosylamine (71). Overall, the structures of both the active and inactive
conformations are similar to their counterparts in mGP. A comparison of the inactive and active conformations of lGP and mGP shows that in both isozymes,
the solvent accessible surface area of the subunit is reduced when the enzyme is
activated. Hydrogen bonds specific for the inactive conformation are replaced by
a greater number of hydrogen bonds tailored to stabilize the active conformation.
Activation of both isozymes involves a rotation of the subunits about a conserved
axis accompanied by domain closure, and these changes are similar in magnitude
in both isozymes.
Despite these overall similarities, there are significant differences between mGP
and lGP in both the active and inactive conformations. The crystal structures revealed that mGP and lGP do not share the same subunit interface in either conformation. This is unexpected because all but three of the residues involved in
the subunit interface are conserved between the two enzymes (59). Repacking of
the subunit interface is the result of changes in the relative orientation of the hydrophobic cores of the subunits and in hydrogen bonds and van der Waals contacts
between the subunits. Half of the hydrogen bonds exclusive to either the active or
inactive conformation are isozyme specific; the remainder are found in both mGP
and lGP. Repacking of the two subunits creates a lGP dimer, which is distinct from
the mGP dimer and could not have been predicted by sequence comparison.
The crystal structure of the dimeric, activated complex of lGP/AMP also reveals
how more than 40 residues of the catalytic site work together to activate the enzyme
(Figure 5). On activation of lGP a, residues 250–290 undergo order or disorder
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
203
transitions, changes in secondary structure, and/or repacking, resulting in a larger
subunit interface with new hydrogen bonds and van der Waals contacts. In the
activated lGP structure, the tower helix is shorter by two turns and does not form
the hydrogen bonds to its mate in the other subunit, which are seen in inactive lGP
a. The gate is well ordered and is moved up and away from the entrance to the
active site such that substrate has direct access to the catalytic residues. In its open
position, the gate residues make van der Waals contacts with the 250’s loop of the
other subunit, and the length of the gate is shortened by extending the helix that
follows it (helix 8). This shorter gate can no longer contact the 380’s loop, which
is less well defined in the electron density. Thus, when open, the gate contacts the
250’s loop of the second subunit, and as a consequence, the 380’s loop of its own
subunit becomes more mobile, probably to permit the interaction with substrate
observed in the MalP complex with maltoheptaose.
These 40 residues include most of the cluster of conserved hydrophobic residues,
which stabilize the active enzyme in yGP. The structure of lGP/AMP shows that
the hydrophobic cluster is preserved structurally in the liver enzyme (although yGP
and mGP share only 49% sequence identity) and suggests that the more closely
related mammalian phosphorylases (80% sequence identity between human liver
and human mGP) are also stabilized in the same way. This result cannot be confirmed in mGP because we lack a crystal structure of the fully activated, dimeric
mGP enzyme. In all the mGP structures solved to date, the 250’s loop, a key part
of the hydrophobic condensation, is disordered, and the enzyme is either inhibited
in some way or forms a tetramer in which the catalytic site is blocked.
The lGP structures also suggest why AMP is not a strong activator of lGP. The
loss of AMP cooperative binding and weak enzyme activation in the liver isozyme
results from changes at the subunit interface, where the AMP site is located,
notably in the conformation of the adenine loop (residues 315–325). Despite the
conservation of binding site residues, many of the interactions with AMP seen in
mGP are absent in the complex of the liver enzyme with AMP. The loss of contacts
to binding site residues is primarily the result of a rigid body rotation between the
two subunits and secondarily of alterations in side chain positions. The rotation of
the subunits with respect to each other increases the distance between Cα carbons
of the cap and the α2 helix by 0.5 Å (on average), with the loss of five hydrogen
bonds seen in the mGP complex. The fact that AMP forms fewer hydrogen bonds
to residues of the α2 helix partly explains the loss of cooperative binding in lGP
because the signal that AMP is bound to one site is propagated through the α2
helix to the other subunit.
In mGP, the adenine N6 amino group of AMP is recognized by the main chain
carbonyls of residues within the adenine loop (66). These contacts are absent in
the lGP complex because the loop adopts a different conformation that moves
it away from the AMP site. Hydrogen bonds, which stabilize the conformation
of the adenine loop in mGP, are absent in lGP as a result of sequence changes.
The adenine loop connects the AMP site to the active site gate through helix α8.
Without the contacts between AMP and the adenine loop, local structural changes
P1: GDL
April 14, 2001
204
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
induced by AMP binding may be poorly coupled to the active site and may account
for weaker AMP activation (36).
Inhibitors of Liver Phosphorylase for Treatment
of Non-Insulin-Dependent Diabetes
Recently, there has been a growing interest in using inhibitors of lGP in treating
noninsulin-dependent or Type 2 diabetes. Type 2 diabetes mellitus is a disease characterized by high levels of glucose in the plasma and leads to complications such as
nerve and kidney damage, blindness, premature atherosclerosis, and heart disease.
The molecular basis of the disease is not well understood but is characterized by
peripheral insulin resistance and pancreatic defects in insulin secretion. Intensive
control of blood glucose levels prevents or delays the onset of diabetic complications (1) but is rarely achieved with oral antidiabetic agents. In diabetic subjects,
breakdown of glycogen by lGP remains an important contributor to hepatic glucose output even when blood glucose levels are high, establishing a potential role
for phosphorylase inhibitors in diabetes therapy.
Inhibitors of phosphorylase include glucose analogs (active site) (24, 48), an
AMP site inhibitor (71), and hydroxylated piperidines and pyrrolidines (2). Recently, two series of indole-2-carboxamide inhibitors were reported (49, 27). The
indole-2-carboxamides inhibit synergistically with glucose, a desirable property
that could minimize the risk of hypoglycemia, a potentially severe side effect
of many antidiabetic agents. Preliminary results have shown that the indole-2carboxamide inhibitors show both cellular and oral activity in reducing blood
glucose levels in ob/ob mice (49, 27).
To identify the binding site for these compounds on the enzyme, the crystal
structure of lGP a complexed with one of these inhibitors was determined at 2.3 Å
resolution (Figure 6; 55). The electron density of the bound inhibitor was located
in a difference map of a crystal of lGP a grown in the presence of excess compound
and a glucose analog, GlcNAc (63). The compound binds to a new, highly specific,
allosteric site on the enzyme. This site is distant from the catalytic site and has not
been reported previously. The compound is bound within the solvent cavity, which
forms part of the dimer interface, very close to the axis of the molecular twofold
symmetry operator. The binding site for the inhibitor is made up of residues from
both subunits and consists of a primarily hydrophobic half and a mixed hydrophobic and polar half. The secondary structural elements that form this novel binding
site are conserved in the known crystal structures of yGP, mGP, and lGP, and they
are expected to be present in the brain isozyme as well.
The discovery of this series suggests that screening for allosteric inhibitors
(or, potentially, activators) may be generally useful in identifying new classes of
drugs. The indole-2-carboxamide series was identified by screening the Pfizer
sample bank for compounds that would inhibit enzyme activity in the presence
of physiological concentrations of glucose. Because glucose itself is an active
site inhibitor, the screen amounted to a search for compounds that would inhibit
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
205
phosphorylase activity by allosteric mechanisms. Phosphorylase has multiple effector sites, each of which could be the target of a separate screen. The discovery
of a new, highly specific, allosteric inhibitor site on such a well-studied enzyme
reveals the remarkable complexity and plasticity of phosphorylase.
CONCLUSIONS: Evolution Provides Regulatory Diversions
In conclusion, allosteric control mechanisms appear to have evolved independently
in the yeast and mammalian enzymes. Mammalian and yeast enzymes differ in
their subunit interactions and in their mode of binding phosphopeptides and Glc6-P. Their phosphorylation sites differ and though N-terminal phosphorylation
activates both types of enzymes, the mechanisms of their activation differ. The
structural changes triggered at the interface by effectors are distinct. During activation of the mammalian enzyme, the subunits draw closer together; in yGP, the
subunits separate. Both mammalian and yeast enzymes are inhibited by Glc-6-P;
however, they display different anomer preferences and bind the inhibitor in different orientations. The enzymes have in common a binding pocket at the subunit
interface that recognizes the phosphate groups of either activators or inhibitors.
Regulation is achieved by the competition of effectors for binding at this site. In
yGP, this site binds either the phosphate of the phosphorylated N-terminus or Glc6-P. In mammalian phosphorylases, this site binds either the phosphate of AMP or
Glc-6-P or the triphosphate of ATP. The ancestral phosphorylase was presumably
a simple constitutively active catalytic enzyme similar to MalP that over time, in
response to environmental conditions, evolved distinct sets of binding amino acids
for the pertinent effectors. Because all phosphorylases must come to the same final
configuration of active site amino acids around the substrates, it is the family of
inactive states that varies among species. The binding of a competitive activator
that displaces inhibitors to drive the stabilization of an active conformation represents a common feature of the regulatory mechanisms of phosphorylases. Recent
evidence suggests that in the yeast and mammalian phosphorylases, stabilization
of the active state involves the condensation of 12 conserved hydrophobic side
chains into a hydrophobic cluster near the active site. In the inactive states, these
side chains are found in different positions in the yeast and mammalian enzymes.
Crystallographic studies indicate that regulated phosphorylases adopt at least
three conformational states. The unphosphorylated enzyme assumes an inactive
conformation in which the active site is inaccessible to substrates. In yGP, the
N-terminus and 380’s loop block the entrance to the active site. In mammalian
phosphorylases, the N- and C-terminal domains are rotated together to close the
active site cleft, and the active site gate occupies the phosphate subsite of the
active site. Allosteric inhibitors, including Glc-6-P, glucose, and purines, stabilize
the enzyme in an inactive conformation. Phosphorylation and/or AMP binding
promote a transition to an activated conformation: The N-terminus in yGP and the
active site gate in mGP move away from the active site to relieve inhibition. The
P1: GDL
April 14, 2001
206
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
N- and C-terminal domains rotate apart to open the active site to substrate binding,
and key residues move into positions to increase affinities for phosphate and sugar
substrates. Finally, substrates bind at the active site, and the enzyme undergoes
an oligosaccharide-induced domain closure involving movements of the 280s and
380s loops to create the catalytically competent ternary complex.
ACKNOWLEDGMENTS
The authors would like to thank Jay Pandit (Pfizer Inc, Groton, CT) for access to
unpublished data. This work was supported by NIH Grant DK32822 to RJF.
Visit the Annual Reviews home page at www.AnnualReviews.org
LITERATURE CITED
1. Abraira C, Colwell JA, Nuttall FQ, Sawin
CT, Nagel NJ, et al. 1995. Veterans Affairs Cooperative Study on glycemic control
and complications in type II diabetes (VA
CSDM). Results of the feasibility trial. Veterans Affairs Cooperative Study in Type II
Diabetes. Diabetes Care 18:1113–23
2. Anderson B, Rassov A, Westergaard N,
Lundgren K. 1999. Inhibition of glycogenolysis in primary rat hepatocytes by 1,4dideoxy-1,4-amino-D-arabinol. Biochem. J.
342:545–50
3. Barford D, Hu SH, Johnson LN. 1991. Structural mechanism for glycogen phosphorylase control by phosphorylation and AMP.
J. Mol. Biol. 218:233–60
4. Barford D, Johnson LN. 1989. The allosteric
transition of glycogen phosphorylase. Nature 340:609–16
5. Becker J-U, Wingender-Drissen R, Schiltz
E. 1983. Purification and properties of
phosphorylase from baker’s yeast. Arch.
Biochem. Biophys. 225:667–78
6. Browner MF, Fletterick RJ. 1992. Phosphorylase: a biological transducer. Trends
Biochem. Sci. 17:66–71
7. Browner MF, Hackos D, Fletterick R. 1994.
Identification of the molecular trigger for allosteric activation in glycogen phosphorylase. Nat. Struct. Biol. 1:327–33
8. Buc H. 1967. On the allosteric interaction between 50 AMP and orthophosphate on
phosphorylase b. Quantitative kinetic predictions. Biochem. Biophys. Res. Commun.
28:59–64
9. Buchbinder JL, Fletterick RJ. 1996. Role of
the active site gate of glycogen phosphorylase in allosteric inhibition and substrate
binding. J. Biol. Chem. 271:22305–9
10. Buchbinder JL, Guinovart JJ, Fletterick RJ.
1995. Mutations in paired alpha-helices at
the subunit interface of glycogen phosphorylase alter homotropic and heterotropic
cooperativity. Biochemistry 34:6423–32
11. Buchbinder JL, Luong C, Browner MF,
Fletterick RJ. 1997. Partial activation of
muscle phosphorylase by replacement of
serine 14 with acidic residues at the site of
regulatory phosphorylation. Biochemistry
36:8039–44
12. Chapon C. 1982. Role of the catabolite activator protein in the maltose regulon of Escherichia coli. J. Bacteriol. 150:722–29
13. Cohen P. 1989. The structure and regulation of protein phosphatases. Annu. Rev.
Biochem. 58:453–508
14. Cori CF, Cori GT. 1936. Mechanism of formation of hexose monophosphate in muscle and isolation of a new phosphate ester.
Proc. Soc. Exp. Biol. Med. 34:702–5
15. Cori GT, Cori CF. 1940. The kinetics
of the enzymatic synthesis of glycogen
from glucose-1-phosphate. J. Biol. Chem.
135:733–56
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
16. Drueckes P, Boeck B, Palm D, Schnizel R.
1996. Mutational analysis of the oligosaccharide recognition site at the active site
of Escherichia coli maltodextrin phoshorylase. Biochemistry 35:6727–34
17. Fischer EH, Graves DJ, Krebs EG. 1957.
Phosphopeptides from P32-labeled phosphorylase a. Fed. Proc., Fed. Am. Soc. Exp.
Biol. 16:180
18. Fischer EH, Krebs EG. 1955. Conversion
of phosphorylase b to phosphorylase a in
muscle extracts. J. Biol. Chem. 216:121–32
19. Fischer EH, Graves DJ, Crittenden ERS,
Krebs EG. 1959. Structure of the site phosphorylated in the phosphorylase b to a reaction. J. Biol. Chem. 234:1698–1704
20. Fletterick RJ, Madsen NB. 1980. The structures and related functions of phosphorylase a. Annu. Rev. Biochem. 49:31–61
21. Fosset M, Muir LW, Nielsen LD, Fischer
EH. 1971. Purification and properties of
yeast glycogen phosphorylase a and b. Biochemistry 10:4105–13
22. Goldsmith EJ, Sprang SR, Hamlin R,
Xuong NH, Fletterick RJ. 1989. Domain
separation in the activation of glycogen
phosphorylase a. Science 245:528–32
23. Graves DJ, Wang JH. 1972. α-Glucan
phosphorylases—chemical and physical
basis of catalysis and regulation. Enzymes
7:435–82
24. Gregoriou M, Noble MEM, Watson KA,
Garman EF, Krulle TM, et al. 1998. The
structure of a glycogen phosphorylase
glucopyranose spirohydantoin complex at
1.8 Å resolution and 100 K: the role of the
water structure and its contribution to binding. Protein Sci. 7:915–27
25. Guenard D, Morange M, Buc H. 1977.
Comparative study of the effect of 50 AMP
and its analogs on rabbit glycogen phosphorylase b isoenzymes. Eur. J. Biochem.
76:447–52
26. Helmreich E, Cori CF. 1964. The role of
adenylic acid in the activation of phosphorylase. Proc. Natl. Acad. Sci. USA 51:131–
36
207
27. Hoover DJ, Lefkowitz-Snow S, BurgessHenry JL, Martin WH, Armento SJ, et al.
1998. Indole-2-carboxamide inhibitors of
human liver glycogen phosphorylase. J.
Med. Chem. 41:2934–38
28. Hwang PK, Fletterick RJ. 1986. Convergent and divergent evolution of regulatory
sites in eukaryotic phosphorylases. Nature
324:80–84
29. Ingebritsen TS, Cohen P. 1983. Protein
phosphatases: properties and role in cellular regulation. Science 221:331–38
30. Johnson LN, Barford D. 1990. Glycogen
phosphorylase. The structural basis of the
allosteric response and comparison with
other allosteric proteins. J. Biol. Chem.
265:2409–12
31. Johnson LN, Barford D. 1994. Electrostatic effects in the control of glycogen
phosphorylase by phosphorylation. Protein Sci. 3:1726–30
32. Johnson LN, Hajdu J, Acharya KR, Stuart
DI, McLaughlin PJ, et al. 1989. Glycogen
phosphorylase b. In Allosteric Enzymes, ed.
G Herve, pp. 81–127. Boca Raton, FL:
CRC Press. 288 pp.
33. Kasvinsky PJ, Madsen NB, Sygusch J,
Fletterick RJ. 1978. The regulation of
glycogen phosphorylase alpha by nucleotide derivatives. Kinetic and x-ray
crystallographic studies. J. Biol. Chem.
253:3343–51
34. Kasvinsky PJ, Shechosky S, Fletterick RJ.
1978. Synergistic regulation of phosphorylase a by glucose and caffeine. J. Biol.
Chem. 253:9102–6
35. Kihlman B, Overgaard-Hansen K. 1955.
Inhibition of muscle phosphorylase by
methylated oxypurines. Exp. Cell. Res.
8:252–55
36. Kobayashi M, Soman G, Graves DJ. 1982.
A comparison of the activator sites of liver
and muscle glycogen phosphorylase b. J.
Biol. Chem. 257:14041–47
37. Krebs EG. 1972. Protein kinases. Curr.
Top. Cell Regul. 5:99–133
38. Lin K, Hwang PK, Fletterick RJ. 1995.
P1: GDL
April 14, 2001
208
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
17:31
Annual Reviews
BUCHBINDER
¥
RATH
¥
AR128-09
FLETTERICK
Mechanism of regulation in yeast glycogen
phosphorylase. J. Biol. Chem. 270:26833–
39
Lin K, Hwang PK, Fletterick RJ. 1997. Distinct phosphorylation signals converge at
the catalytic center in glycogen phosphorylase. Structure 5:1511–23
Lin K, Rath VL, Dai SC, Fletterick RJ,
Hwang PK. 1996. A protein phosphorylation switch at the conserved allosteric site
in GP. Science 273:1539–42
Lorek A, Wilson KS, Sansom MS, Stuart DI, Stura EA, et al. 1984. Allosteric
interactions of glycogen phosphorylase b.
A crystallographic study of glucose 6phosphate and inorganic phosphate binding to di-imidate-cross-linked phosphorylase b. Biochem. J. 218:45–60
Maddaiah VT, Madsen NB. 1966. Kinetics of purified liver phosphorylase. J. Biol.
Chem. 241:3873–81
Madsen NB. 1964. Allosteric properties of
phosphorylase b. Biochem. Biophys. Res.
Commun. 15:390–95
Madsen NB. 1986. Glycogen phosphorylase. In The Enzymes, ed. PD Boyer,
EG Krebs, 17:365–94. London: Academic
Press
Madsen NB, Avramovic-Zikic O, Lue PF,
Honikel KO. 1976. Studies on allosteric
phenomena in glycogen phosphorylase b.
Mol. Cell Biochem. 11:35–50
Madsen NB, Shechosky S. 1967. Allosteric
properties of phosphorylase b. II. Comparison with a kinetic model. J. Biol. Chem.
242:3301–7
Martin JL, Johnson LN, Withers SG. 1990.
Comparison of the binding of glucose and
glucose 1-phosphate derivatives to T-state
glycogen phosphorylase b. Biochemistry
29:10745–57
Martin JL, Veluraja K, Ross K, Johnson
LN, Fleet GWJ, et al. 1991. Glucose analogue inhibitors of glycogen phosphorylase: the design of potential drugs for diabetes. Biochemistry 30:10101–16
Martin WH, Hoover DJ, Armento SJ, Stock
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
IA, McPherson RK, et al. 1998. Discovery
of a human liver glycogen phosphorylase
inhibitor that lowers blood glucose in vivo.
Proc. Natl. Acad. Sci. USA 95:1776–81
Morgan HE, Parmeggiani A. 1964. Regulation of glycolysis in muscle. III. Control
of muscle glycogen phosphorylase activity.
J. Biol. Chem. 239:2440–45
O’Reilly M, Watson KA, Schinzel R, Palm
D, Johnson LN. 1997. Oligosaccharide
substrate binding in Escherichia coli maltodextrin phosphorylase. Nat. Struct. Biol.
4:405–12
Palm D, Goerl R, Weidinger G, Zeier R,
Fischer B, Schnizel R. 1987. E. coli maltodextrin phosphorylase: primary structure
and deletion mapping of the C-terminal
site. Z. Naturforsch. Teil C 42:394–400
Palm D, Thanner F, Schwenk L. 1975. Purification and properties of E. coli maltodextrin phosphorylase. In Metabolic Interconversion of Enzymes, ed. S Shaltiel,
pp. 37–43. New York : Springer-Verlag
Parmeggiani A, Morgan HE. 1962. Effect
of adenine nucleotides and inorganic phosphate on muscle phosphorylase activity.
Biochem. Biophys. Res. Commun. 9:252–
56
Rath VL, Ammirati ML, Hoover DJ. 2000.
Eur. Patent Off. App. EP 0 978 279 A1
Rath VL, Ammirati ML, Lemotte PK,
Fennell KF, Mansour MN, et al. 2000.
Activation of human liver glycogen phosphorylase by alteration of the secondary
structure and packing of the catalytic core.
Mol. Cell 6:139–48
Rath VL, Fletterick RJ. 1994. Parallel evolution in two homologues of phosphorylase. Nat. Struct. Biol. 1:681–90
Rath VL, Lin K, Hwang PK, Fletterick RJ.
1996. The evolution of an allosteric site in
phosphorylase. Structure 4:463–73
Rath VL, Newgard CB, Sprang GEJ, Fletterick RJ. 1987. Modeling the biochemical differences between rabbit liver phosphorylase. Proteins: Struct. Funct. Genet.
2:225–35
P1: GDL
April 14, 2001
17:31
Annual Reviews
AR128-09
PHOSPHORYLASES
60. Schwartz M, Hofnung M. 1967. La maltodextrine phosphorylase d’Escherichia
coli. Eur. J. Biochem. 2:132–45
61. Sprang S, Fletterick R, Stern M, Yang D,
Madsen N, Sturtevant J. 1982. Analysis of
an allosteric binding site: the nucleoside
inhibitor site of phosphorylase alpha. Biochemistry 21:2036–48
62. Sprang S, Goldsmith E, Fletterick R.
1987. Structure of the nucleotide activation switch in glycogen phosphorylase a.
Science 237:1012–19
63. Sprang SR, Acharya KR, Goldsmith EJ,
Stuart DI, Varvill K, et al. 1988. Structural changes in glycogen phosphorylase induced by phosphorylation. Nature
336:215–21
64. Sprang SR, Goldsmith EJ, Fletterick RJ,
Withers SG, Madsen NB. 1982. Catalytic
site of glycogen phosphorylase: structure
of the T state and specificity for alpha-Dglucose. Biochemistry 21:5364–71
65. Sprang SR, Madsen NB, Withers SG.
1992. Multiple phosphate positions in
the catalytic site of glycogen phosphorylase: structure of the pyridoxal-50 pyrophosphate coenzyme-substrate analog. Protein Sci. 1:1100–11
66. Sprang SR, Withers SG, Goldsmith EJ,
Fletterick RJ, Madsen NB. 1991. Structural
67.
68.
69.
70.
71.
209
basis for the activation of glycogen phosphorylase b by adenosine monophosphate.
Science 254:1367–71
Wang JH, Tu JI, Lo FM. 1970. Effect of
glucose 6-phosphate on the nucleotide site
of glycogen phosphorylase b. A general
approach for negative heterotropic interactions. J. Biol. Chem. 245:3115–21
Watson KA, Schinzel R, Palm D, Johnson LN. 1997. The crystal structure of Escherichia coli maltodextrin phosphorylase
provides an explanation for the activity
without control in this basic archetype of a
phosphorylase. EMBO J. 16:1–14
Weber IT, Johnson LN, Wilson KS, Yeates
DG, Wild DL, Jenkins JA. 1978. Crystallographic studies on the activity of glycogen phosphorylase b. Nature 274:433–
37
Withers SG, Madsen NB, Sprang SR, Fletterick RJ. 1982. Catalytic site of glycogen
phosphorylase: structural changes during
activation and mechanistic implications.
Biochemistry 21:5372–82
Zographos SE, Oikonomakos NG, Tsitsanou KE, Leonidas DD, Chrysina ED,
et al. 1997. The structure of glycogen phosphorylase b with an alkyl-dihydropyridinedicarboxylic acid compound, a novel and
potent inhibitor. Structure 5:1413–25
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 1 Ribbon diagram of MalP dimer. The binding site for maltodextrose is shown in
magenta. The PLP cofactor (yellow) is covalently linked to the side chain of Lys-680 within
the nucleotide binding fold motif of the C-terminal domain. The active site of the enzyme
is located in a crevice between the N-terminal and C-terminal domains and is shown with
bound maltose (purple), glycerol (purple), and sulfate (pink).
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 2 View of the active site of MalP shows PLP cofactor (yellow) and bound maltose
(dark blue), glycerol (dark blue), and sulfate (pink). Hydrogen bonding interactions are
indicated with dashed lines (RCSB Protein Data Bank entry 1AHP).
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 3 Comparison of (A) phosphorylated (RCSB Protein Data Bank entry 1YGP) and
(B) unphosphorylated yeast phosphorylase (22). The phosphorylase dimer is depicted as a
Connolly surface with one monomer colored blue and the other purple. N-terminal residues
1-22, corresponding to the N-terminus of muscle phosphorylase, are shown as a ribbon in
white. The unique N-terminal extension of yeast phosphorylase (residues -1 through -39)
is drawn as a ribbon in pink. The structure of yGP a is of a truncated form of the enzyme,
which contained a deletion of the first 22 N-terminal residues; therefore, only residues -1
through -14 of the N-terminal extension are shown. The structure of yGP b is of the fulllength enzyme; however, residues -12 through -22 of yGP b were disordered and are not
shown. In the unphosphorylated enzyme, the N-terminal extension blocks the entrance to
the active site. Phosphorylation results in the movement of the N-terminal extension to an
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 3 (Continued )
allosteric site at the subunit interface where it displaces the inhibitor, glucose 6-phosphate
(orange). Thr-10, the site of regulatory phosphorylation, is colored pink. The position of
the active site is indicated by the PLP cofactor (yellow) and in yGP a, by the additional
presence of a bound phosphate (pink). Hydrophobic residues (Phe-252, Leu-254, Phe-257,
Tyr-262, Tyr-1630 , Val-2780 , Tyr-2800 , Pro-2810 , Phe-2850 and Leu-2910 ), shown in coral,
condense to form a hydrophobic cluster near the active site in yGP a.
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 4 Comparison of (A) phosphorylated (27) and (B) unphosphorylated muscle phosphorylase (RCSB Protein Data Bank entry 2GPB). The phosphorylase dimer is depicted
as a Connolly surface with one monomer colored blue and the other purple. N-terminal
residues (5-22 in mGP a and 12-22 in mGP b) are drawn as a ribbon in white. Residue
Ser14 is shown in dark pink and the PLP cofactor is shown in yellow. AMP (light pink) is
bound at the allosteric site of mGP a. The inhibitor glucose (orange) is bound at the active
site of mGP b. In response to phosphorylation, the N-terminus of mGP refolds, assuming a
helical conformation, and moves to the dimer interface. This structural change triggers the
allosteric transition and results in the rotation of the subunits and further tertiary structural
changes that lead to activation.
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 5 Refolding and repacking of the protein core. Cartoon showing the structural
changes that occur on the catalytic face of the enzyme in the inactive and active conformations of lGP a. The subunits of the functional dimer are shown as circles, with the N-terminal
and C-terminal domains indicated. In the inactive structure, hydrogen bonds between the
tower helices are shown as orange-dotted lines; regions of the 250’s loop that are disordered
are shown as dashed lines; the tower helix is two turns longer (shown as a solid orange
cylinder). In the active structure, the subunit rotation axis is shown in green; helix 8 is
extended by 1 turn (shown as a solid purple cylinder).
P1: FQP
April 24, 2001
14:44
Annual Reviews
AR128-09-COLOR
Figure 6 The new allosteric inhibitor site. Human liver glycogen phosphorylase a is
depicted as a ribbon diagram, one subunit in purple (helices) and pink (sheets) and the other
in green (helices) and blue (sheets). AMP (gray); Ser-14-P (pink, red phosphates); PLP
(pyridoxal phosphate, the essential cofactor in red); GlcNAc, the glucose analog (purple,
marks the glucose binding site); and CP-403,700 (carbon, pink; nitrogen, blue; oxygen, red)
are shown in CPK. The twofold symmetry operator relating the subunits is located between
the two molecules of CP-403,700, orthogonal to the plane of the page. The binding sites for
Ser-14-P and AMP are located close to each other but do not overlap. To show all the binding
sites in one image, a composite of two crystal structures was made; AMP and residues 5-22
(including Ser-14-P) from the crystal structure of HLGP a complexed with AMP (62) and
the rest from the complex of HLGP a complexed with GlcNAc and CP-403,700 (71).