between both types of protein kinase, but the protein-tyrosine kinases and protein-serine/threonine kinases are distinguished by specific signature motifs (3).
SERIALREVIEWSI PROTEIN KINASES 8
How
do protein
kinases
serine/threonine from the insulin SUSAN S. TAYLOR,1
discriminate
between
and tyrosine? Structural receptor protein-tyrosine
ELZBIETA
RADZIO-ANDZELM,
insights kinase
AND TONY HUNTER
Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, California 92093, USA; and Molecular Biology and Virology Laboratory, The Salk Institute, La Jolla, California 92037, USA
The
rapid
advances
in the solution
of the 3-dimensional
protein kinases have provided the useful possibility of making comparisons that are important to our understanding of the mechanism, regulation, and specificity of these enzymes. In this review by Taylor et al., a detailed comparison is made between the active conformation of cAMP-dependent protein kinase, structures
of several
the first time how activation might be achieved.-Taylor, S. S., Radzio-Andzelm, E., Hunter, T. How do protein kinases discriminate between serine/threonine and tyrosine? Structural insights from the insulin receptor protein-tyrosine kinase. FASEBJ. 9, 1255-1266 (1995)
the paradigmatic Ser/Thr protein kinase, and an inactive conformation of the catalytic region of the insulin receptor that contains tyrosine phosphorylation activity.
Key word.,: P-site catalytic subunit Lion insulin receptor kinase domain
In spite of the overall similarity, useful structural differences are noted that explain the lack of activity of the tyrosine kinase and its specificity toward tyrosine residues. The role of autophosphorylation of tyrosine residues occupying the active site is also made evident by this analysis. This review also points to several unsolved
PROTEIN-TYROSINE PHOSPHORYLATION, first encountered 1979 as an activity of viral transforming gene products
questions
regarding
the
regulation
of the receptor
tyro-
was
quickly
recognized
transducing growth plasma membrane. tyrosine
to
other
protein-kinase P1(4
play
an
important
also
cellular
regulates
processes
into
an active
tary
to the
protein-serine/threonine kinases immediately sue of how amino acid substrate specificity
(FASEB (FASEB
conformation.
This
review
is complemen-
series by Walsh and by Hunter
ABSTRACT The eukaryotic protein kinases that directly phosphorylate proteins are divided into two major classes: those that phosphorylate tyrosine and those that phosphorylate serine and threonine. Until recently, the similarities between these two classes of enzymes, which now total more than 400, were based primarily on sequence aligmnents. A recent report of the structure of the kinase domain (IRK) of the insulin receptor protein-tyrosine kinase now allows the features of these two families to be compared at the structural level. We review here this first tyrosine-specific protein kinase structure, and compare and contrast it to the structure of the serine/threonme-specific cAMP-dependent protein kinase. Although the general fold of the polypeptide backbone is conserved as predicted, unique features at the IRK active site provide a basis for understanding the differences in specificity for the phosphate acceptor amino acid. The structure of this inactive, dephosphorylated protein-tyrosine kinase also defines for
0892-6638/95/0009-1
255/$O1
.50. © FASEB
Purified
role
protein-tyrosine
kinases
in the that
function
the cell
transcription, and synaptic transmission. a second class of protein kinase to add
in (1),
across clear
protein
including
sine kinases, the main one being the mechanism through which agonist binding shifts the catalytic region topics discussed in this J. 8, 1227-1236; 1994) J. 9, 576-596; 1995).
phosphoryla-
factor receptor signals With time it has become
phosphorylation
many
.
in
cycle,
The discovery of to the well-known
show
raised the isis achieved.
an exquisite
selec-
tivity for tyrosine and do not phosphorylate serine or threonine even when they are substituted for tyrosine in a known
acceptor
sequence
serine/threonine though
there
(2).
kinases are
Conversely,
a few
dual-specificity
as the
MAP kinase kinases,
phosphorylate
both serine/threonine
such By that
1982 the
it had
superfamily,
become
protein-tyrosine
ine/threonine
kinases which
clear
protein-
tyrosine,
protein
from
sequence
and
the
to the same
a catalytic
domain
can
analysis protein-ser-
protein
kinase
of about
260
amino acids. A number of residues within this domain highly conserved between both types of protein kinase, the protein-tyrosine kinases and protein-serine/threonine kinases are distinguished by specific signature motifs A priori,
however,
it was
not
clear
whether
were important in determining amino whether they might reflect the nature
1f whom correspondence at: Department of Chemistry La Jolla, CA 92093, USA.
and reprint
these
are but (3).
motifs
acid specificity or of the first protein
requests should
and Biochemistry,
al-
kinases,
that physiologically and tyrosine.
kinases
belong has
most
do not phosphorylate
be addressed,
UCSD, 9500 Gilman
Dr.,
1255
SERIAL REVIEW kinase that underwent the evolutionary transition that allowed it to phosphorylate tyrosine. An early attempt to alter the specificity
by simply
switching
a protein-serine/threonine kinase resulted in a dead
one of the motifs
from
kinase into a protein-tyrosine enzyme (4), which did not resolve
the issue!
The 3-dimensional structure of cAMP-dependent protein kinase (PKA)2 catalytic (C) subunit bound to ATP and an inhibitor peptide (PKI:5-24) provided a template for the general conserved folding of the protein kinase polypeptide chain and also gave clues as to how serine/threonine-contaming substrates are recognized (5-8). The small NH2terminal lobe binds ATP, the larger lobe is important for catalysis
and
peptide
binding,
and
the active
site
lies
in a
cleft between the two lobes. Basic residues in PKI NH2terminal to the phosphotransfer site, or P-site, interact with surface acidic residues in the COOH-terminal lobe, thus positioning the target hydroxyamino acid residue in the catalytic cleft. The hydroxyl group in a substrate, replaced with an Ala in PKI, would be located less than 3 A from the hydroxyl group of the ‘y-phosphate of ATP. In a substrate:ADP complex of PKA, the hydroxyl of the P-site serine is 2.7 A from the carboxyl oxygen of Asp, the proposed catalytic base (9), and is in an ideal position to undergo a concerted phosphotransfer reaction. Because the serine carbon only
and
threonine
atom,
they
difference
hydroxyls are equivalent
is that
accommodated
are
in the
the active
both
in terms
extra site
methyl
linked
of catalysis; group
without
to the has
interfering
-
the to be with
phosphate transfer if threonine is to be phosphorylated. Several other protein-serine/threonine kinase structures have subsequently been reported, and their overall structures are very similar (10-13). However, the other structures have either been of catalytically inactive forms (Cdk2 and the ERK2 MAP kinase) or have not been solved with a bound substrate (twitchin and casein kinase I). Nevertheless, it appears that the general mechanism of substrate recognition is likely to be similar, with the target bydroxyamino acid poised for phosphotransfer at the active site and residues on either the NH2-terminal or COOH-terminal side of it in a position to interact with complementary sites on the surface of the catalytic domain. From the structure of the PKA:PKI complex it is clear that if the substrate peptide backbone were maintained in the same position, then the tyrosine side chain would not fit correctly into the active site so that the 04-hydroxyl group
could
be
positioned
deed,
when
the
structure
to receive of the
the
catalytic
phosphate. domain
Inof the
EGF receptor
protein-tyrosine kinase was modeled based on the PKA structure (14), the predicted general folding of the EGF’ receptor catalytic domain was similar to that of
2Abbreviations: PKA, cAMP-dependent P-site, phosphotransfer site; IRK, insulin multiple endocrine neoplasia.
1256
Vol.9
October
1995
protein kinase; C, catalytic; receptor kinase domain; MEN,
PKA, but the backbone of the substrate peptide had to be positioned differently at the active site in order to accommodate the tyrosine and prevent its hydroxyl from protruding into the ATP-binding site. An important corollary is that if the catalytic cleft of the protein-tyrosine kinases is designed to position the hydroxyl group of tyrosine to accept phosphate, this will preclude the phosphorylation of serine and threonine, because their hydroxyl groups would not penetrate deep enough to act as acceptors. An analogy can be drawn with SH2 domains and the protein-tyrosine phosphatases, which typically interact with phosphotyrosme, but not phosphoserine or phosphothreonine, and have deep clefts that accommodate phosphotyrosine. Enzymes such as the dual-specificity protein kinases (e.g., MAP kinase kinase) and phosphatases (e.g., Cdc25 and MKP-1) can recognize the hydroxyls of serine/threonine and tyrosine;
the
structures
of such
enzymes
with
bound
peptide
substrates are needed in order to learn how they achieve a similar positioning of both aliphatic and aromatic hydroxyls. For these and many other reasons, a structure of a protein-tyrosine kinase has report of the 3-dimensional
been
eagerly structure
tor protein-tyrosine largely superseded
kinase all these
(15). This structure
has a number
and reveals in part why these tyrosine. Crystals were obtained
awaited. The of the insulin
recent recep-
catalytic domain has now theoretical considerations
of interesting
features,
enzymes are specific for of a fragment of the human
insulin receptor kinase domain (IRK), residues Va1978Lys’283 containing Cys981Ser and Tyr’Phe mutations. The atomic model includes all but three NH2-terminal residues. Although this fragment, expressed in insect cells, can be activated by autophosphorylation, the protein that was
crystallized
was
unphosphorylated
and
thus
inactive.
The fact that the catalytic domain is unphosphorylated is important, because it is known that the first step in the activation of the insulin receptor upon insulin binding is an autophosphorylation event that occurs in trans between two catalytic domains. As many as seven tyrosines can be autophosphorylated, but kinetically the first to be phosphorylated
is either
Tyr1162
or Tyr’,
followed
by Tyr’
(16-19). When a synthetic peptide containing all three tyrosines is phosphorylated by IRK, Tyr’t62 is phosphory. lated first and then Tyr’, followed by Tyrflm (20). All three tyrosines lie in subdomain VIII, the activation loop, which contains activating
a region called phosphoryla-
tion sites in many protein-serine/threonine and protein-tyrosine kinases. Indeed, in PKA the phosphate esterified to Thr’97 in the activation loop is very stable and appears to play an important role in holding the catalytic domain in an active conformation. For example, replacement of this threonine with alanine leads to a substantial loss in catalytic activity due to a decrease in Keai and an increase in K,, (ATP) (21). One can also deduce from the structures of Cdk2 and the ERK2/MAP kinase that phosphorylation of residues in their activation loops must alter the conformation of these proteins to convert them from an inactive to an active state that more closely resembles PKA.
The FASEB Journal
TAYLOR
ET AL.
SERIAL REVIEW GENERAL STRUCTURAL INSULIN RECEPTOR COMPARISON WITH
FEATURES
KINASE PKA
OF THE
AND
IRK defines
for the first time the struckinase. It emphasizes those elements that are conserved throughout the entire family of enzymes that transfer phosphate to serine, threonme, and tyrosine, and identifies those features that are unique to the protein-tyrosine kinase family. To highlight these similarities and differences, we shall describe the structure of IRK and compare it with that of the C-subunit of PKA (Fig. 1). The C-subunit represents an active enzyme complexed with MgATP and an inhibitor peptide, and is in a closed conformation (7). The dephosphorylated IRK represents an inactive state and is also in a more open conformation. The most striking finding is the remarkable similarity of these two structures (Fig. 1). As predicted from the sequence similarities (3), all members of this diverse family have conserved the same overall folding of the polypeptide
This
structure
tural
hallmarks
of
of a protein-tyrosine
chain. Inserts can readily be accommodated at the loops on the surface, and these inserts provide a major basis for structural diversity. The insulin receptor is the first solved protein kinase structure that has a large insert between aD and ctE. In the PDGF receptor, this insert is nearly 100 residues and contains several tyrosine phosphorylation sites. Spatially this insert protrudes from the same general region as the aG-aH insert common to members of the cyclin-dependent kinase and MAP kinase families. The COOH-terminal tail of IRK, which in IRK and frequently in other receptor protein-tyrosine kinases contains tyrosines
that
become
conserved.
INSULIN
RECEPTOR PROTEIN-TYROSINE
KtNASE
in IRK) are also indicated.
and
act
as
substrate
Invariant
residues
positioned
throughout
the
catalytic core so far, with one exception, are located in the same position in all of the protein kinase structures. Al-
Figure 1. The structure of the insulin receptor (IRK) (right) is compared with the structure of the (PKA) (left). Ribbon structures of residues 981-1283 of the insulin receptor and residues 22-312 marked. The a-carbons of conserved residues are indicated as white dots and in some cases side shown are indicated in Table 1. Inserts are highlighted in yellow; MgATP is yellow and PKI(5-24) is shaded darker. The side chains of the phosphorylated residues in the activation loops (T197 phenylalanine in the DFG loop (F185 in PKA; F1051 binding site in PKA is shown as a CPK structure.
phosphorylated
docking sites, extends from the same general surface, a region that is well removed from the active site cleft, even though such inserts as well as the COOH-terminal tail can reach to the active site and influence peptide recognition (6, 22). With the exception of the inserts, which all lie in loops on the surface, the general secondary structure is
A detergent
catalytic of PKA chains is red. in PKA;
molecule
subunit of cAMP-dependent protein kinase are shown; the NH2 and COOH termini are are shown. The specific conserved residues The linker segment that joins the two lobes Y1158, Y1162 and Y1163 in IRK) and of
(octanoyl-N-methylglucamnide)
filling
the myristyl
1257
SERIAL REVIEW though there are some differences in the small lobe and in the orientation of the two lobes relative to one another, the major differences in the two enzymes, emphasized in the superimposed structures in Fig. 2, are in the activation loop in the large lobe near the cleft interface. Clearly, the conformation of this ioop in most protein kinases depends critically on the phosphorylation state of the protein. To emphasize the specific features that differ in the two enzymes, the two lobes will be discussed separately. The residues of particular importance in PKA and IRK, along with their nomenclature, are indicated in Table 1.
THE
NH2-TERMINAL
ATP-BINDING
LOBE
The NH2-terminal ATP-binding lobe constitutes an ATPbinding motif that is unique to the protein kinase family. The ATP-binding lobes of IRK and PKA are shown with conserved features of the sequence highlighted in Fig. 3. In IRK, this lobe is actually in a more open conformation relative to the large lobe, in part due to the position of the activation loop, but in Fig. 3 they are superimposed. The
5-stranded 13-sheet is highly conserved, in particular, the glycine-rich loop between 131 and f32 and the invariant lysine in 133, whereas the helical regions differ in several regards. The small B-helix is so far unique to PKA. The C-helix, in contrast, which contains a conserved glutamate (Glu9’ in PKA and Glut0’17 in IRK) that is part of the ATP-binding site, is present but is displaced in IRK relative to the 13-sheet (Fig. 3), and is also somewhat shorter. Consequently, this structural motif is located relatively far from the active site cleft in the overall structure. In the active conformation, this helix must move into place so that G1u1047 can come into close proximity to Lys’#{176}3#{176} and the catalytic loop between 136 and 137in the large lobe. In the ternary complex of PKA with MgATP and PKI(5-24), Glu91 positions Lys72 for binding to the a- and 13-phosphates of ATP, but also is itself less than 4 A from the catalytic base, Asp’, at the site of phosphotransfer. Of the 11 conserved residues highlighted in Fig. 1, Glu’#{176}47 is the only one that is displaced significantly relative to PKA in any of the protein kinase structures solved so far. In Cdk2, this helix is displaced even further and twisted so that the conserved glutamate is actually fully exposed to solvent. Cyclin is
Figure 2. Superimposition of the a-carbon backbones of IRK (cyan) and PKA (blue). The side chains of the phosphorylated residues are shown in white; M5ATP is yellow. A) Standard view of the two enzymes (in stereo).
B
The superimposed structures are rotated 1800 to emphasize the activation loops that are oriented very differently in the two enzymes.
1258
Vol.9
October1995
The FASEB Journal
TAYLOR ETAL.
SERIAL REVIEW TABLE
1. Functional
residues in P1(4 and IRK
PKA
Location
IRK
Location and Function
C 50#{176}
I31-2 Loop
G 1003
Glycine-rich
loop
G 52’
31-J32 Loop
C 1005
Glycine-rich
loop
FM
1-f2
F 1007
At tip of loop. Folds
Loop
Close
G55a
G1008
Glycine
to Phe
on top of P-site
residue.
185
rich-
loop
K 72#{176}
33
K 1030
Binds
E91#{176}
aC
E1047
Ion pairs
with Lys72
R 165
136-7
R 1131
Ion pairs
with P.Thr
D 166#{176}
Cat loop
D
K168
Cat loop
A 1134
Binds
to Thr201,
E 170
Cat loop
R 1136
Binds
to P-2 site Arg and ATP in PKA;
loop
1132
to a- and p-phosphates
of ATP
197
Catalytic base
Tyr1162,
Asp’,
Asp32,
y-phosphate
Asp1161,
and
of ATP and P-2 binds
Arg
to
Trp 1175 in IRK
2+
N 171
Cat loop
N 1137
Binds
to inhibitory
Mg
D 184#{176}
138-139 links
D 1150
Binds
to activating
Mg2
F 1151
Binds
on Outer surface ofP siteresiduein
K 1155
Ion pairs
Y
Phosphorylation
that bridges
3-and
y-phosphate
(PICA)
F 185
8-9
links
PKA;
fills in adenine
pocket
in IRK
(PKA) K 189
J39 (PICA)
K 192
9
(PICA)
W 196
1158
197
with P.Thr
site in IRK
D 1161 (Y1162)
Y 1162 (Y1163)
Essential
L 198
P+lloop
G1169
P+ 1 recognition
C
P+1
L1170
W 197
199
loop
phosphorylation
G200
P+lloop
L1171
1201
P+lloop
P1172
Binds to Lys
P 202
P+1 loop
V 1173
P+1 recognition
168
E 203
P+1
loop
R 1174
P-6 recognition
Y 204
P+1 loop
W 1175
Hydrogen
L 205
P+lboop
E 208#{176}
aF
D220
R 280#{176}
site
site
166
and Asp site site
bonds
to
M1176
P+ 1 recognition
E 1179
Ion pairs with Arg
D 1191
Hydrogen
R 1253
Ion pairs
bonds
G1u230(P-2 recognition site)
site 280.
in PICA and Arg
to backbone
with Glu
1253.
of catalytic
in IRK
base in PKA and IRK
208
‘Residues indicated in Fig. 1.
thought to bind to and alter the conformation of this region of Cdk2 generating an active conformation (10, 23) (see “Note added in proof”); in the insulin receptor, the plasma membrane or the other molecule in the dimer may influence the conformation of this region upon insulin binding.
THE ACTIVATION
LOOP
The most significant
differences between IRK and PKA are found in the activation loop in the large lobe, a segment that displays considerable conformational diversity depending on the phosphorylation state of the enzyme. Probably the most unexpected feature of the IRK structure is INSULIN
RECEPTOR PROTEIN-TYROSINE
KINASE
the position of this unphosphorylated loop and the apparent stability of this inactive conformation. This segment, shown in Fig. ‘IA, is bridged by two conserved residues whose positions remain more or less fixed. In PKA, Asp at the end of 138 binds to the activating Mg2 ion in the C-subunit that bridges the and ‘y-phosphates of ATP (Fig. 4C). At the other end of this segment is G1u208, which is fixed by its interaction with Arg#{176}. This buried ion pair, conserved so far in nearly all of the protein kinase structures, is also present in IRK. The equivalent residues in IRK are Asp’#{176},G1u1179, and Arg’253. Contained within this region in PKA flanked by Asp184 and G1u208 are three structural motifs: 139, an activating phosphorylation site, and the P+1 loop, which is the docking site for the P+1 residue in substrates of PKA. This region, with the excep-
13-
1259
SERIAL REVIEW to be stable and involves numerous hydrogen bonding interactions. Where the equilibrium lies in solution in the cell where ATP concentrations are millimolar is unknown. What is clear, however, is that the conformation of this loop will change substantially when the enzyme is phosphorylated.
The
two residues
that
interact
with
P.Thr197
in
PKA, Arg165 and Lys’, are both arginines in IRK (Arg”31 and Arg”55) and are positioned to interact with a phosphate, although the side chain of Arg1155 is disordered. 13-strand 6, which interacts with 139in PKA, is also in place in IRK waiting for its partner strand, which is missing in this structure. Hubbard et al. (15) predict that Tyr”62 corresponds to Thr197; however, based on alignment with other receptor protein-tyrosine
kinases
that
have
only
a single
tyrosine
in
this region (24), it is also possible that Tyr”63 corresponds to Thr’97. Whichever residue is equivalent to Thr’97, once phosphorylated itmay bind to Arg”3’ and Arg”55, leaving the
other
two
phosphates
on
the
surface.
This
region
is
unoccupied PKA Trp1
Figure 3. The ATP binding domains of IRK (cyan) and PKA (blue). The small NH2-terminal lobes (residues 40-127 in PICA; 993-1083 in IRK) are superimposed. The following conserved residues in this domain are indicated: the three glycines in the glycine-rich loop (represented by a-carbon CPK structures), the phenylalanine in the glycine-rich loop (F1007 in IRK; FM in PKA), the lysine in 3 (K72 in PICA; K1030 in IRK), and the glutamate in aC (E 1047 in IRK; E91 in PKA). The position of the C-helix and the linker segment that joins the two lobes is indicated. MgATP from the ternary complex of PKA indicates the position of the nucleotide relative to the ATP-binding domain. An alignment of the sequences in these regions of IRK and PICA (subdomains I-IV) is shown.
in the unactivated, truncated IRK structure. In is a major determinant for recognition of the R-subunit (25). Tyrt on the other hand, would now lie at the edge of the cleft interface in the position that corresponds to Lys’92 in PKA. The role that phosphorylation of Tyr”8, Tyr”62, and Tyr1 plays in insulin receptor activation is discussed below.
THE SITE OF CATALYSIS A closer comparison
lion of most of the P+ 1 loop, is ordered very differently in IRK (Fig. 4B). Furthermore, unlike the structure of unphosphorylated Cdk2 where this region is also ordered very differently from PKA, thissegment in IRK is well-defined, indicating that the inactive conformation of this region may be quite stable. Two featuresof the IRK structureare particularly remarkable. First is the conserved DFG motif where Phe”5’ folds over into the ATP-binding site. By filling the adenine pocket, this side chain precludes ATP binding (Fig. ‘IA). After this is a very stable region that contains the three tyrosinephosphorylationsites.The polypeptide chain is positionedso thatone of the tyrosines,Tyr1162,isoriented with its hydroxyl group lying in approximately the same position as the P-site serine hydroxyl in a substrate complex of PKA. The following residue, Tyr’163, lies in the somewhat distorted P+ 1 binding site. As discussed later, Pro72 is responsible for positioning the P-site tyrosine in the phosphotransfer site, whereas in PKA the P-site serine is positioned by interactions with the peptide backbone. Whether
this
position
is stable
when
ATP
is bound
or even
whether AlP can bind to the inactive enzyme is not known. In the absence of ATP, under the crystallization conditions used here this conformation of the activation ioop appears
1260
Vol.9
October
1995
of the regions surrounding the actual site of phosphotransfer reveals some of the most fundamental differences between the protein-serine/threonine and protein-tyrosine kinases (Fig. 5), as would be expected because this region determines whether the enzyme will accept serine/threonine or tyrosine. The two regions that are most characteristically distinct for the two enzyme families are in the catalytic loop (YRDLKPEN in PKA and HRDLAARN in IRK) and the P+ 1 loop (TPEYLAPE in PKA and PVRWMAPE in IRK). As predicted in the model of the EGF receptor kinase domain (14), these regions, shown in Fig. 4 and Fig. 5 (subdomains VIB and VIII) (24),converge at the siteof catalysisand interactwith one another as well as with the substrate. Protein-serine/threonine kinases have a lysine in the conserved catalytic loop whereas protein-tyrosine kinases have RAA (cytoplasmic
protein-tyrosine
kinases)
or AAR
(receptor
protein-tyrosine kinases).In PKA (Fig.5B), Lys’, interacts with Thr 201 in the P+ 1 loop and also binds to the P-2 backbone carbonyl in the substrate, thus helping to orient the peptide for phosphotransfer. Lys’68 also binds to the y-phosphate of ATP (7). In the IRK, (Fig. 5A), Arg”36 in AAR interacts with Asp”61 and Tyr”62 in the activation loop, Asp’132, the catalytic base, and Trp”75 in the P+1 loop. The Arg”36 guanidinium group is bent away from the y-phosphate binding site. It also comes close to the aromatic ring of the P-site tyrosine. Because there is still no PKA structure available that does not contain peptide, it
The FASEB Journal
TAYLOR
ET AL.
SERIAL REVIEW
4. Comparison of the activation loops of IRK and cAPK. The two bps. which are highlighted in Fig. 2B, are shown (residues 1150-1179 in IRK (cyan); residues 184-208 in PKA (blue)), a-carbons
Figure
activation
of the two conserved carboxylates at each end of the loop are shown as CPK structures. The sides chains of the phosphorylated residues are shown in yellow. A) The hydrogen bonding in the IRK activation loop. Also shown (in red) is n-strand 6 and Arg#{176}3’ where the activation loop will presumably dock following phosphorylation of Tyri and/or Tyr’ ‘‘ B) The activation loops in A and C are superimposed. G The hydrogen bonding in the PKA activation loop. Also shown in p-strand ATP (red) and a portion of the inhibitor peptide (R”3RNAlt”). An alignment of the sequences in these regions of IRK and PKA (subdomains VII and VIII) is shown. In this case, Tyr#{176}is aligned with Thr’97.
6,
cannot be ascertained whether Lys’ also folds away from the y-phosphate binding site when peptide is missing. This lysine in PKA is known to be protected from modification with acetic anhydride both in the free enzyme and in the ternary complex (26). A fundamental difference between the protein-tyrosine and protein-serine/threonine kinases is that the P-site phosphate acceptor is anchored differently. In both structures the hydroxyl itself is positioned by hydrogen bonding
for phosphotransfer. backbone carbonyl Ser
(27).
In
acceptor
the
group
Ser53 hydrogen-bonds and the backbone amide protein-tyrosine
kinases
is too large
to
both
the
of the P-site the
to be efficiently
phosphate
stabilized
seen in Fig. 6B and Fig. 6C, the Tyr”62 ring stacks onto a conserved proline, Pro”72, in the P+1 loop, as was predicted through
peptide
in the model
backbone
of the EGF
contacts.
receptor
Instead,
kinase
domain
as
(14).
This
catalytic base (Asp’66 in PKA and Asp”32 in IRK) (9). In PKA, the substrate is anchored in addition by interactions of the peptide backbone, with Ser53 in the glycine-
proline is the equivalent of Thr 201 in PKA. The tyrosine ring isalso positionedby Arg”36, the spatial equivalent of Lys’ in PKA, and by an axial polar interaction with the indole ring of Trp75. Whether residues from the small
rich
lobe,
to the
loop
and
Lys’68
in
the
catalytic
loop.
Lys’68
is
such
as Phe’#{176}#{176}7 in the glycine-rich
loop
and
Met1t53
hydrogen-bonded to the P.2 carbonyl and also to Thr20’ in the P+1 loop. These three residues, Ser53, Lys’68, and Thr201, are all highly conserved in most protein-ser-
in the largelobe,alsocontributeto docking of the tyrosine ring cannot be determined from this inactive conformation.
ine/threonine
the substrate For example,
bonding
INSULIN
network
kinases that
and serves
provide
a concerted
to position
RECEPTOR PROTEIN-TYROSINE
KINASE
the P-site
hydrogen hydroxyl
Recognition
of amino
acid
side
chains
of the residues
in
that flank the P-site must also be conserved. in PKA one of the residues in the P+ 1 loop,
1261
SERIAL REVIEW crosses
over
and
indirectly
stabilizes
the
P-2
recognition
site by hydrogen-bonding to G1u230; in IRK, Trp”75 fills the same space, and this tryptophan is further stabilized by axial polar interactions with the guanidinium group of Arg”36.
ACTIVATION CATALYTIC
OF INSULIN ACTIVITY
RECEPTOR
What do we learn about insulin receptor activation from the structure? Autophosphorylation accompanies insulinmediated activation, and the fact that individual or combined mutation of Tyr”58, Tyr”62, or Tyr” to phenylalanine reduces kinase activity (29-35) indicates that autophosphorylation of the activation loop plays an important role in activation. In this regard, the most striking feature of the insulin receptor catalytic domain structure is that Tyr’’62, one of the first tyrosines to be autophosphorylated,
Figure 5. The site of catalysis in IRK (A) and PICA(B). The two regions that most characteristically discriminate between protein-serine/threonme kinases and protein-tyrosine kinases are shown. This includes the catalytic loop (residues 1132-1137 in IRK; residues 166-171 in PKA) and the P+1 loop (residues 1169-1176 in IRK; residues 198-205 in P1(A). In the PKA structure a fragment of the PKI peptide (P-2 Arg-P+ 1 lie) is shown in yellow. The white arrow indicates the P-site. In IRK the segment of the protein that occupies the peptide binding site, Aspi _Tyr 2Tyr11. is also shown in yellow. Hydrogen bonding is shown in white. An alignment of the sequences and PICA (subdomain VIB) is shown.
in these
regions
Glu203, also helps to recognize the P-6 Arg in the inhibitor peptide (Fig. 6B). In IRK this residue is replaced with Arg1174 (Fig. 6A), which is consistent with IRK’s preference for acidic side chains (28). In PKA, Tyr204 also
Vol. 9
October
1995
in the
catalytic
cleft,
with
of IRK
SPECIFICITY
1262
is located
the 04-hydroxyl positioned to accept the ‘y-phosphate of ATP. However, Phe115’ in the DFG loop is folded into the adenine binding site, thereby blocking access of ATP to the active site. The apparent inability of the insulin receptor to bind ATP in this configuration would preclude autophosphorylation of Tyr”62 in cis,although the authors have not yet obtained a structure of an ATP-bound complex to establish that this is the case. In addition, the N112-terminal lobe is rotated 26#{176} and translated 0.8 A relative to the active,closed PKA structure, suggesting that insertion of Tyr”62 into the cleft may actually preclude the enzyme from adopting an active conformation. The position of Phe”51 also contributes to the more open conformation. The authors propose that Tyr”62 is, in fact, phosphorylated first in trans by the neighboring catalytic domain in the insulin receptor dimer, and that this could be triggered by an insulin-induced conformational change that allows ATP binding to occur. This model is consistent with the general requirement for transphosphorylation in the activation of receptor protein-tyrosine kinases, but is hard to imagine topologically, especially without knowing how the two protomers in the dimer, both of which are anchored to the membrane, interact with one another. Once phosphorylated, Tyr”62 would no longer be bound in the active site and presumably would bind somewhere else, thus stabilizing an active conformation. The other two tyrosines in the activation loop would then, in principle, be accessible for phosphorylation. At exactly what stage an exogenous substrate or the autophosphorylation sites in the juxtamembrane domain and the COOH-terminal tail would have access to the active site is unclear.
OF INSULIN
RECEPTOR-SUBSTRATE
INTERACTIONS
Although the mechanism of autoactivationof the insulin receptor is still not resolved, the structure does allow us to
The FASEB Journal
TAYLOR E AL.
SERIAL REVIEW
Figure 6. CPK models of the P+ 1 loops in IRK (A) and PICA (B). Metit7S in IRK is pink with a yellow sulfur atom, and the corresponding residue in PKA, L205. is turquoise. The P-site (indicated by an arrow) and P+ 1 residues (Tyrl iC2 and Tyr” in IRK, and Ala377 and lie378 in PKI) are also shown. The side chains of these residues are white and the backbones are grey. All nitrogen and oxygen atoms are blue and red respectively. (C). The Y1162-Y1163 peptide has been separated from the enzyme to expose the P+1 recognition site in IRK.
make some predictions about the primary sequence specificity of the insulin receptor. Four glutamates (Glu’70, G1u127, G1u203, and Glu230) in PKA interactwith the P-2 and P-3 basic amino acids in PKA substrates. The equivalent residues in the insulin receptor catalytic domain are Arg”36, Asp’#{176}83, Arg”74, and G1u1201. Additional residues that are potential candidates for interaction with the P-2 and P-3 residues in the substrate are Arg’089, Lys’085, and Lysl#{174}2. This configuration of basic residues provides an explanation for IRK having a preference for phosphorylating tyrosines with acidic residues upstream. Indeed, several of the tyrosines known to be phosphorylated in the preferred insulin receptor protein-tyrosine kinase substrate IRS-1 are in this sequence context (36), and an analysis of the preferred sequences phosphorylated in a degenerate peptide library confirms the preference for glutamates upstream of the tyrosine (28). The IRK also displays a preference for methionine or phenylalanine at the P+1 position in peptide substrates.
INSULIN
RECEPTOR PROTEIN-TYROSINE
KINASE
In one sense the activation loop in the inactive conformation acts as a surrogate peptide substrate with Tyr11 replacing the P+ 1 hydrophobic residue. This can best be seen in a space filling model of the P+ 1 loop where the P-site serine (designated as Ala377 in PKI) and P+1 isoleucine fill the region in PKA (Fig. 6A), and Tyr”62 and Tyr”’ fill the corresponding region in IRK (Fig. 6B). One apparent difference in the P+ 1 loop is the position of the glycine. In PKA, the residues that contribute most to this hydrophobic pocket are Leu’98, Pro202, and Leu205. G1y20#{176} contributes little. In contrast, in IRK the equivalent of Leu’98 is a glycine and the hydrophobic contribution to the loop comes from Leu’171, the equivalent of G1y2#{176}#{176} in PKA. The two leucines there fill the same space in PKA and IRK, although they have a LeuXGly and GlyXLeu motif, respectively. Also noteworthy in thisloop isthe positionof Met”76 (Fig. 6B). This methionine is conserved in all receptor
protein-tyrosine
nonreceptor
protein-tyrosine
kinases,
but
kinases.
is a threonine Replacement
in all of this
1263
SERIAL REVIEW methionine by threonine in the RET sine kinase is the cause of multiple (MEN) type 2B (37), and it has been may be a result of an altered RET through a change in P+1 selectivity
receptor protein-tyroendocrine neoplasia speculated
substrate
that
this
specificity
(28).
lated after insulin consistent
QUESTIONS
The structure of the unactivated insulin receptor catalytic domain provides many insights, but leaves a number of questions. Most of these concern how insulin binding to the intact receptor dimer increases catalytic activity. Because activation of the intact full-length receptor is triggered by insulin binding to the extracellular domain, it may not be possible to establish the precise mechanism that triggers the activation of the actual receptor from this truncated form of IRK. The plasma membrane, the juxtamembrane domain, and/or the COOH-terminal tail, as well as the presence of ATP, could all change the conformation of catalytic center, particularly the orientation of the two lobes relative to one another. The two noncatalytic domains as well as the membrane will also almost certainly affect how the nonconserved segments at both ends of the catalytic domain interact with the catalytic core. To understand the mechanism of activation at the molecular level we need to know the orientation of the two catalytic domains in the dimer and how they interact, what insulin-induced conformational change triggers transphosphorylation, how phosphorylation of the activation loop is achieved in trans, and what conformation it adopts after phosphorylation. One unanswered question is whether ATP can even bind to this inactive form of the IRK and whether this form will exist in the cell where ATP concentrations are millimolar. Because ATP binding to the soluble IRK will automatically lead to activation, a structure of IRK bound to a nonhydrolyzable analog of ATP may be needed to answer this question. Another question unanswered by the IRK structure is the position of the COOH-terminal tail, which contains two sites of autophosphorylation, because the tail is deleted in the crystallized
protein.
It will
be important
to determine
whether the COOH-terniinal tail binds to the active site and serves as an autoinhibitor as the COOH-terminal tail of twitchin appears to do. The fact that the insulin receptor COOH-terminal tail can be removed by proteolysis without activating the kinase might suggest that it does not occupy the active site (38). To explain insulin-dependent activation of the inactive IRK, Hubbard et al. (15) have proposed a cis-inhibition trans-activation model. In this model the IRK can adopt two conformations: one without ATP bound and with Tyr’162 in the active site, and the other with ATP bound and Tyr”62 flipped out and accessible. Insulin is proposed to induce a conformational change in the receptor dimer, thus allowing Tyr”62 in one subunit to be phosphorylated in trans by the other subunit,which perforcemust alsobe in the ATP-bound state. This model fits with the observation that Tyr”62 is one of the first sites to be phosphory-
1264
Vol. 9
October
1995
being
However,
an alternative
one of the first tyrosines
model, to be
phosphorylated and itbeing more accessiblethan Tyr1t62 in the inactive IRK, is that Tyr11 is in fact phosphorylated first in trans. Once phosphorylated, P.Tyr” could bind into a pocket created by Arg31 and Arg”53, which would result
UNRESOLVED
binding.
with Tyr”58
in a further
conformational
change
of the activation
loop exposing Tyr”62 and allowing it to be phosphorylated. Once phosphorylated, P.Tyr62 could bind to another site or displace P.Tyr””, stabilizing the active conformation. In principle, one ought to be able to distinguish between these models based on the properties of insulin receptor mutants in which single activation loop tyrosines have been mutated to phenylalanine. Unfortunately, the autophosphorylation site mutagenesis studies are not completely in agreement as to the roles of the individual tyrosines (29-35). None of the tyrosines is absolutely essential for insulin receptor activation. In at least some studies, the Tyr””Phe mutation by itself has a significant inhibitory effect on activation (31, 35), suggesting that it could play an important role. However, it is clear that the Tyr’ ‘62Phe mutation alone has a dramatic effect, and ultiinatelyphosphorylationof Tyr”62 must be requiredforthe insulin receptor to achieve an activated state. The Tyr’ UPhe mutation has the least effect, but the phosphorylation of Tyr’ ‘ appears to be required to attain maximal activation. Ultimately, it may turn out that phosphorylation of any of the tyrosines in the IRK activation loop is capable of triggering the conformation change that allows additional tyrosines to be phosphorylated. If phosphorylation of accessible activation loop tyrosines can trigger this conformational change, then it is possible that phosphorylation of activation loop tyrosines in trans by another protein-tyrosine kinase could trigger insulin-independent activation. It is known that Src can phosphoi-ylate activation loop tyrosines in the insulin receptor in vitro, leading to its activation (39). What does the IRK structure tell us about other receptor protein-tyrosine kinases and protein-tyrosine kinases in general? The members of the insulin receptor protein-tyrosine kinase subfamily are unique in that they exist as disulfide-bonded dimers, and therefore ligand binding can in principle simply elicit an activating conformational change rather than causing dimerization. For this reason, the inactive conformation of the insulin receptor may be designed to preclude accidental transphosphorylation. The insertion of Tyr”62 into the active site and the apparent inability of this conformation to bind ATP would be an efficient way to achieve this. Indeed, if the stability of the dephosphorylated
activation
loop
is maintained
in the cell,
it may mean that the unphosphorylated enzyme is more stablethan the activephosphorylated enzyme. This could be a mechanism to ensure that the insulin receptor rapidly returns to the inactive state once it is dephosphorylated. However, for most other receptor protein-tyrosine kinases, which are activated by ligand-induced dimerization, it is not clear that this fail-safe mechanism is necessary. Indeed, the activation loops of about half the known receptor
The FASEB Journal
TAYLOR
Er AL.
SERIAL REVIEW protein-tyrosine
kinase
subfamilies
have
only
a single
ty-
rosine, which appears to be equivalent to Tyr”. For this type of receptor protein-tyrosine kinase, phosphorylation of the activation loop tyrosine in trans, followed by its binding to a surface pocket, might be sufficient to activate the catalytic domain. The fact that insulin receptor kinase activity becomes ligand-independent after autophosphorylation (40), whereas this is not the case for other studied receptor protein-tyrosine kinases, also suggests that the activation mechanism of the insulin receptor may be specialized. Nevertheless, it will be interesting to learn whether other protein-tyrosine kinases are cis-inhibited by insertion of an activation
loop
autophosphorylation
site
tyrosine
into
REFERENCES 1.
2.
3.
4.
5.
6.
the
activesite,and whether activationoccurs by phosphorylation in trans within a dimer. There are now several examples of protein-serine kinases that are activated by transphosphorylation of one or more residues in the activation loop by a separate protein kinase. One recent example is protein kinase C, where phosphorylation of the activation loop threonine equivalent to Thr’97 in PKA is first phosphorylated in trans by another protein kinase, thus activating subsequent autophosphorylation (41). Transphosphorylation, followed by true autophosphorylation, may be a common activation mechanism for protein-serine kinases and could also be a mechanism for activation of protein-tyrosine kinases as well. Another question is how tyrosines lying outside the catalytic domain are autophosphorylated? The COOH-terminal tail of the insulin receptor, which contains two other autophosphorylation sites, is located on the face opposite the catalytic cleft. Although it could be phosphorylated in cis if it were to fold around the catalytic domain, it is plausible that itcould be phosphorylated in trans. Likewise the sites in the juxtamembrane domain would be most easily phosphorylated in trans. The kinase insert in the PDGF receptor family, which contains several autophosphorylation sites, is also on the opposite face to the catalytic cleft, and in this case it can be phosphorylated only in trans. Structures of the complete cytoplasmic domain of the insulin receptor dimer with peptide substrate and ATP bound will be eagerly awaited so that a full understanding of the mechanisms of activation and autophosphorylation of the insulin receptor protein-tyrosine kinase and its interaction with substrates can be attained.
7.
8.
9.
10.
11.
12. 13.
14.
15.
16.
17.
18. 19.
20. Note
added
complex
has now been
K., Gibbs, Cdk2
reported
E., Hurwitz,J.,
376,
Nature
undergoes
313-320},
[Jeffrey,
and
ioop,
the position
therefore
N. P. (1995)
the activation
RECEPTOR PROTEIN-TYROSINE
ioop in
change
upon
cleft.
However,
site
a phosphate
in the fully
21.
22.
cyclin the
on Thr16#{176} in the
conformation
A/Cdk2 complex are not known.
INSULIN
AICdk2
A. A., Polyak,
active
the precise
a cyclin
P. D., Russo.
conformational
of the Thr’6#{176} phosphate
of
and Pavletich,
and as predicted
a significant
and
J.,
Massague,
binding, which removes it from the Cdk2 in this structure did not contain activation
structure
The crystal
in proof:
23.
of this loop active
cyclin
24.
Eckhart. W., Hutchinson, M. A., and Hunier, 1. (1979) An activity phosphoiylating tyrosine in polyoma T antigen immunoprecipitates. Cell 18. 925-933 Weinmaster,
C., and Pawson. T. (1986) Protein kinase activity of FSV (Fujinami sarcoma virus) Pl3Ogag-fps shows a strict specificity for tyrosine residues. J. Biol. C/tern. 261,328-333 Hanks, S. K., Quinn, A. M., and Hunter. T. (1988) The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science 241.42-52 Moran, M. F., Koch, C. A., Sadowski. I., and Pawson,T. (1988) Muiational analysis of a phosphotransfer motif essential for v-ips tyrosine kinase activity.
Oncogene Knighton, Sowadski, subunit of ence 253, Knighton, Taylor. S. subunit of ence 253, Zheng, J.,
3,665-672 D. R., Zheng, J.. Ten Eyck, L F.. Xuong, N-H., Taylor, S. S., and J. M. (1991) Structure of a peptide inhibitor bound to the catalytic cyclic adenosine monophosphate-dependent protein kinase. Sci414-420 D. R., Zheng, J., Ten Eyck, L. F., Ashford. V. A., Xuong, N-H., S., and Sowadski, J. M. (1991) Ciystal siructure of the catalytic cyclic adenosine monophosphale-dependeni protein kinase. Sci407-414
Knighton. D. R., ten Eyck. L. F.. Karlsson, R.. Xuong. N., Taylor. S. S., and Sowadskj. J. M. (1993) Crystal structure of ihe catalytic suhunii of (AMP-dependent protein kinase coniplexed wiih MgATP and pepiide inhibitor. Biochemistry 32.2154-2161 Bossemeyer. D., Erigh, R. A.. Kinzel, V., Ponstingi, El., and Huher, K. (1993) Phosphotransferase and substrate binding mechanism of the cAM P-dependent pJotein kinase catalytic subunit from porcine heart as deduced from the 2.0 A structure of the complex with Mn2 a(lenyl imidodiphosphate and inhibitors peptide PKI(5-24). EMBO). 12.849-859 Madhusudan, A., Trafny, E. A., Xuong, N. H., Adams, J. A., Ten Eyck, 1. F., Taylor, S. S., and Sowadski. J. M. (1994) cAMP-dependent Protein kinase: crystallographic insighis into substrate recognition and phosphotransfer. Protein Sci. 3, 176-187 Dc Bondt, H. L., Rosenblati, J., Jancarik, J., Jones, ii. D., Morgan, D. 0., and Kim, S. (1993) Crystal structure of cyclin-dependent kinase 2. Nature (London) 363, 595-602 Zhang. F.. Strand. A., Robbins, D.. Cobb, M. H., and Coldsnmiih, E. J. (1993) Atomic structure of the MAP kinase ERK2 at 2.3 A resolution. Nature (London) 367,704-711 flu, S.-H., Parker, M. W., Lei, J. Y., Wilce, M. C. J., Benian, C. M., and Kemp, B. E. (1994) Insights into autoregulation from the crystal stnicture of twitchin kinase. Nature (London) 369, 581-584 Xu, R.-M., Carmel, C., Sweet, R. M., Kuret, J., and Cheng, X. (1995) Crystal structure of casein kinase-1, a phosphate-directed J)roiein kinase. EMBO J. 14, 1015-1023
Knighton, D. K., Cadena. D. L., Zheng, J., Ten Eyck, L. F.. Taylor. S. S., Sowadski, J. M., and Gill, C. N. (1993) Structural features thai specify tyrosine kinase activity deduced froni homology modeling of the epidertnal growth factor receptor. Proc. Nail. Acad. Sci. USA 90. 5001-5005 Hubbard, S. R., Wem, L, Ellis, L., and Ilendricksoim. W. A. (1994) Crystal structure of tlte tyrosine kinase domain of tie’ human insulin receptor. Nature (London) 372, 746-754 Tornqvist. H. E.,Cunsalus.J. R., Nemenoff, R. A., Frackelton, A. R., Pierce, M. W., an(l Avruch. J. (1988) Identification of the insulin recepior tyrosine residues undergoing insulin-stimulated phosphorylation in intact rat hepatotna cells. J. Biot. C/tern. 263, 350-359 White, M. F., Shoelson, S. E., Keutmammn, II., and Kahn, C. K. (1988) A cascade of tyrosmne autophosphorylation in the -suhunit activates the phosphotransferase of the insulin receptor.). Biot. Chern. 263. 2969-2980 Dickens, M.. and Tavare. J. M. (1992) Analysis of the order of autophosphorylation of human insulin receptor tyrosines 1158, 1162. and I 163. Bioehern. Biophys. Res. Corn,nun. 186, 244-250 Tavare, J. M.. and Dickens. M. (1991) Changes in insulin-receptor tyrusine. serine and threonine phosphorylaiion as a result of substitution of tyrosine1162 with phenylalanine. Bioc/tern.J. 274, 173-179 Levine, B. A., Clack. B., and Ellis, L. (1991) A soluble insulin receptor kinase catalyzes ordered phosphouylatmon at multiple tymsines of dodecapeptide substrates.). Biol. C/mm. 266, 3565-3570 Adams. J. A., McGlone. M. L., Cibson. K., and Taylor, 5.5. (1995) Phosphory’lation modulates catalytic function and regulation ot the cAMP-dependent protein kinase. Biochemistry 34, 2447-2454 Kallunki. T.. Su, B., Tsigelny. I., Sluss. H. K.. De.rijard, B., Moore, C.. Davis, R.. and Karin, M. (1994) JNK2 contains a specificity-determining region responsible for efficient c-Jun binding and phosphorylation. Gene.c Dec. 8, 2996-3007 Marcote, M. J., Knightoo, D. R., Basi, C., Sowadski, J. M., Bramhilla. P., Draetta, C., and Taylor, S. S. (1993) A three-dimensional model of the Cdc2 protein kinase: localization of cyclin- and Sucl-binding regions and phosphorylation sites. Mol. Cell. Biol. 13, 5122-5131 Hanks, S. K. (1991) Eukaryotic protein kinases. Curr. Opin. Struct. Bid. 1,
369-383
KINASE
1265
SERIAL REVIEW. 25.
26.
Orellana, S. A., and McKnight, C. S. (1992)
anhydride: 27. 28.
Mutations in the catalytic subunit
of cAMP-dependent protein kinase result in unregulated Proc. Nat!. Acad. Sci. USA 89,4726-4730 Buechler, J. A., Vedvick, T. A., and Taylor, S. S. (1989) of the catalytic subunit of cAMP-dependent protein
substrate-induced
biological
Differential labeling kinase with acetic changes. Biochemistry 28,
conformational
Bossemeyer, D. (1994) The glycine-rich sequence of protein kinases: a multifunctional element. Trends Biochem. Sci. 19, 201-205 Songyang, Z.. Carraway, K. L, Eck, M. J., Harrison, S. C., Feldman, R. A., Mohammadi, M., Schiessinger, J., Hubbard, S. R., Smith, D. P., and Eng, C.
36.
32.
receptor kinase domain autophosphorylation regulates receptor enzymatic function.). Biol. C/tern. 267, 16660-16608 Yonezawa, K., and Roth, K. A. (1991) Assessment of the in situ tyrosine kinase activity of mutant insulin receptors lacking tyrosine lation sites 1162 and 1163. Mol. End ocri no!. 5, 194-200
Zhang, B., Tavare, J. M., Ellis, L., and Roth, K. A. (1991) The regulatory role of known tyrosine autophosphorylation sites of the insulin receptor kinase domain. J. Biol. C/tern. 266,990-996 Murakami, M. S., and Rosen, 0. M. (1991) The role of insulin receptor autophosphorylation in signal transduction. J. Biol. Chem. 266, 22653-22660
35.
31.
30.
34.
3018-3024
(1995) Catalytic specificity of protein-tymsine kinases is critical for selective signalling. Nature (London) 373, 536-539 Wilden, P. A., Backer, J. M., Kahn, C. R., Cahill, D. A., Schroeder, C. J., and White, M. F. (1990) The insulin receptor with phenylalanine replacing tyrosine-1 146 provides evidence for separate signals regulating cellular metabolism and growth. Proc. Nail. Acad. Sci. USA 87,3358-3362 Wilden, P. A., Siddle, K., Haring, E., Backer, J. M., White. M. F., and Kahn. C. R. (1992) The role of insulin receptor kinase domain autophosphorylation in receptor-mediated activities.). Bid. C/tern. 267, 13719-13727 Wilden, P. A., Kathn, C. R., Siddle, K., and White, M. F. (1992) Insulin
29.
33.
activity.
37. 38.
Rafaeloff, K., Maddux, B. A., Brunetti, A., Sbraccia, P.. Sung, C. K., Patel. R.. Hawley, D. M., and Coldflne, I. D. (1991) Transmembrane signalling by insulin via an insulin receptor mutated at tyrosines 1158, 1162. and 1163. Biochem. Biophys. Res. Commun. 179,912-918 White, M. F. (1994) The 1KS-l signaling system. Curr. fsin. Gets. Dee. 4, 47-54 van Heyningen, V. (1994) One gene-four syndromes. Nature (London) 367, 319-320 Hsuan, J. J., Downward, J., Clark, S., and Waterfield, M. D. (1989) Proteolytic
39.
generation
of constitutive
tyrosine
kinase
of the human
insulin
receptor. Biochem. J. 259, 519-527 Yu, K. T., Werth, D. K., Pastan, I. H., and Czech, M. P. (1985)
src kjnase the phosphorylation and activation of the insulin receptor kinase. J. Biol. Chem. 260,5838-5846 Rosen, 0. M., Herrera, R., Olowe, Y., Petruzzelli, L M., and Cobb, M. H. (1983) Phosphorylation activates the insulin receptor tyrosine protein kinase. Proc. Nail. Acad. Sci. USA 80, 3237-3246 Dutil, E. M., Keranen, L. M., DePaoli-Koach, A. A., and Newton, A. C. (1994) catalyzes
40.
41.
In vivo regulation of protein kinase C by trans-phosphorylation autophosphorylat ion.). Biol. C/mm. 269, 29359-29362
autophosphory-
The FASEB Journal
-
followed by
1996
A Series of Reviews The Hepatic Canalicular Membrane Coordinated Scheduled
by Irwin M. Arias and Dietrich Keppler
Reviews:
I. M. Arias
hepatocyte
canalicular
Bile canalicular
I. M. Arias. P. Gros.
Structure-function Canalicular
N. Kaplowilz. D. Keppler. conjugate
Introduction membrane
and D. Keppler.
Expression export
R. Oude Elferink.
ecto-enzymes relationship glutathione
across the
and the purine transporter of MDR transporters
transport
and localization
pump
to transport
of the MRP gene-encoded
in liver
Phospholipid
transport
mediated
by the mdr2
gene product F. Suchy.
Canalicular
S. S. Thorgeirsson.
1266
Vol. 9
October
1995
bile salt transport Expression
of mdr genes in liver
The FASEB Journal
TAYLOR ET AL.
