Progress in Biophysics & Molecular Biology 71 (1999) 313±341
The catalytic subunit of cAMP-dependent protein kinase:
prototype for an extended network of communication
Christopher M. Smith a,*, Elzbieta Radzio-Andzelm b, Madhusudan b,
Pearl Akamine b, Susan S. Taylor b
a
San Diego Supercomputer Center, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0505,
USA
b
Howard Hughes Medical Institute, Department of Chemistry and Biochemistry, University of California, San Diego,
9500 Gilman Drive, La Jolla, CA 92093-0654, USA
Abstract
The protein kinase catalytic core in essence comprises an extended network of interactions that link
distal parts of the molecule to the active site where they facilitate phosphoryl transfer from ATP to
protein substrate. This review de®nes key sequence and structural elements, describes what is currently
known about the molecular interactions, and how they are involved in catalysis. # 1999 Elsevier
Science Ltd. All rights reserved.
1. Introduction
Protein phosphorylation is probably the most important mechanism of regulation in the
eukaryotic cell. While allosteric regulation is important in all cells and allows many proteins to
be sensitive to their immediate environment, protein phosphorylation allows cells to be
responsive to their external environment. The signal transduction pathways that allow a cell to
respond to external stimuli, whether it is a hormone, a growth factor, a photon or osmotic
shock, are all regulated at critical switch points by protein phosphorylation. The enzymes that
catalyze this transfer of a phosphate from ATP to a protein or peptide substrate, the protein
kinases, constitute a large family of enzymes. The family is comprised of two main subfamilies
* Corresponding author. Fax: +1-619-534-5113.
E-mail address: [email protected] (C.M. Smith)
0079-6107/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 7 9 - 6 1 0 7 ( 9 8 ) 0 0 0 5 9 - 5
314
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
Ð those that transfer phosphate to serine and threonine and those that transfer phosphate to
tyrosine (Hanks and Hunter, 1995). The entire family, however, has evolved from a common
origin. Based on analysis of the yeast genome, the human genome alone is predicted to encode
for over 2000 unique protein kinases each turned on and o€ in response to speci®c signals
(Hunter and Plowman, 1997). This makes the protein kinases one of the largest gene families.
When one considers as well the protein phosphatases and the myriad of regulatory and adapter
proteins that modulate protein kinase function and location, as well as the fact that over one
third of the proteins in cells are the phosphoproteins, it is clear that a signi®cant portion of the
eukaryotic cellular machinery is involved, either directly of indirectly, with protein
phosphorylation. Westheimer's elegant 1987 review of phosphates and their importance in
nature, while acknowledging their major role in nucleic acid structure and function and in
bioenergetics, ignored this equally major role that phosphates play in the regulation of cellular
processes (Westheimer, 1987).
Protein phosphorylation was ®rst recognized as a mechanism for regulation of protein
function in the 1950s by the pioneering work of Fischer and Krebs who showed that glycogen
phosphorylase was regulated by the covalent attachment of a phosphate moiety (Fischer and
Krebs, 1955). This ®nding coincided with the discovery of glycogen phosphorylase kinase, the
®rst protein kinase to be rigorously characterized (Krebs et al., 1959). The identi®cation of
glycogen phosphorylase kinase kinase, later renamed cAMP-dependent protein kinase (cAPK)
and its activation by cAMP (Walsh et al., 1968), led to two additional and fundamental
concepts. First was the idea of kinase cascades and the power of ampli®cation. We now
recognize that there are many more elaborate cascades that lead from a single event at the
plasma membrane to the eventual mediation of gene expression in the nucleus. From the
identi®cation of cAPK and its activation by cAMP, also came the recognition of the `second
messenger' concept (Sutherland and Wosilait, 1955). Binding of a molecule at the surface of a
cell can lead to the generation of a second molecule within the cell that mediates the biological
response. The three most prominent intracellular protein kinase second messengers, cAMP,
Ca++ and phospholipids, all mediate their primary responses by contributing to the activation
of speci®c protein kinases. Extracellular mediators such as growth factors mediate their
response by binding directly to a molecule having an intracellular kinase domain which is
activated upon ligand binding to the extracellular domain (Ullrich and Schlessinger, 1990).
2. cAMP-dependent protein kinase
cAMP-dependent protein kinase is not only one of the ®rst protein kinases to be discovered
(Walsh et al., 1968), it remains as one of the simplest. Its simplicity derives primarily from its
dissociative mechanism of activation (Taylor et al., 1990; Francis and Corbin, 1994). It is
comprised of two subunit types, regulatory (R) and catalytic (C). In the absence of cAMP the
enzyme exists as an inactive tetrameric holoenzyme complex, R2C2. The two main subfamilies
of cAPK are determined by the R subunit. The type I subunits are not autophosphorylated by
C and have an absolute requirement for MgATP to form a tight complex with the C subunit.
The type II R subunits, in contrast, are autophosphorylated and do not require MgATP to
form a holoenzyme complex. Depending on the cell type, cAMP can be generated by the
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
315
activation of adenylyl cyclase in response to a variety of stimuli. In all cases, the initial
receptor belongs to the 7-transmembrane spanning serpentine family of G protein coupled
receptors (Luttrell et al., 1997). The primary receptors for cAMP in eukaryotic cells are the R
subunits of cAPK. The cooperative binding of cAMP to the holoenzyme causes a ®ve order of
magnitude decrease in the anity of R for C (Granot et al., 1980) and under physiological
conditions is thought to lead to the dissociation of the holoenzyme into an R2 (cAMP)4 dimer
and two active C subunits (Gill and Garren, 1969; Brostrom et al., 1970; Tao et al., 1970). It is
this unleashed C subunit that is the active form of the enzyme.
The catalytic subunit has many known protein substrates (Zetterqvist et al., 1976), and the
speci®c targets again will vary depending on the cell type and on the particular proteins that
are expressed in that cell at any given time. In addition to phosphorylating known cytoplasmic
proteins such as glycogen phosphorylase kinase (Yeaman et al., 1977), glycogen synthase
(Parker et al., 1981), pyruvate kinase (Hjelmquist et al., 1974), phosphofructokinase II (Murray
et al., 1984), protein phosphatase inhibitor I (Cohen et al., 1977) and all proteins in the liver
that synergistically contribute to the mobilization of stored glycogen, the C subunit also leads
to increased expression of genes that are regulated in response to cAMP such as the gene for
PEPCK, a gluconeogenic enzyme (Sutherland et al., 1996). These genes are typically preceded
by a cAMP response element (CRE) (Montminy, 1997). A cAMP response element binding
protein (CREB) binds to this CRE and is activated in response to phosphorylation of Ser112.
This phosphorylation can be mediated by the C subunit. Once phosphorylated, CREB becomes
a docking site for a transcriptional activator, the CREB binding protein, CBP, which leads to
the assembly of the transcriptional complex and subsequent activation of gene expression
(Goldman et al., 1997).
All substrates, as well as the known physiological inhibitors, of the C subunit have the
general consensus recognition site summarized in Fig. 1. Important determinants are arginines
at the P-6, P-3 and P-2 positions with most substrates having either a P-6 and P-3 Arg or a P-3
and a P-2 Arg (Zetterqvist et al., 1976). A large hydrophobic residue is preferred at the P + 1
site while there are few constraints at the P-1 site (Kemp et al., 1975). This consensus site
Fig. 1. Substrates and Inhibitors of cAPK. The general features of the minimum consensus site, P-3 through P + 1,
that occupies the active site cleft, are highlighted in purple. PKI and the RI subunit share this consensus site,
although unlike the RII subunits and substrates, they are pseudosubstrates. The P-site hydroxyl group is missing.
The region required for high anity is highlighted in teal and phosphorylated residues (Serine) in blue. The essential
residues are indicated by the red diamonds. The respective binding constants for the inhibitor and hexapeptides
(Kemp et al., 1977; Whitehouse and Walsh, 1983), and Kapp (for the C subunit under physiological conditions
(Ho€man, 1980)) for the inhibitor domain of RIa and RIIa subunits are also listed.
316
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
peptide is sucient to convey low anity binding in the mM range. The regulatory subunits are
competitive inhibitors of protein substrates and have a substrate-like motif that resembles the
consensus site peptide and binds to the active site cleft.
In addition to the R subunits, there is another class of physiological inhibitors of C. These
are the heat stable protein kinase inhibitors (PKI's). Like the R subunits, the PKI's bind with
high anity ( < 1 nM) to free C. Like the R subunits, they also share an inhibitor site that
resembles a cAPK substrate. As seen in Fig. 1, PKI and the RI subunits have a
pseudosubstrate inhibitor site, while the RII subunits have a true phosphorylation site. While
PKI was discovered almost simultaneously with cAPK because it copuri®ed as a contaminate
(Walsh et al., 1990), the true physiological role of PKI is still unknown. It is expressed in a cell
cycle-dependent manner (Wen et al., 1995), is relatively unstructured when it is free in solution
(Hauer et al., 1999) and is multifunctional. The high anity inhibitor site in PKI is located
near the amino-terminus as indicated in Fig. 1, and the requirements for high anity binding
were elegantly mapped by Walsh and Glass (Cheng et al., 1986; Scott et al., 1986). However, in
addition to its C subunit inhibitor site, PKI contains near its carboxyl-terminus a nuclear
export signal (NES). Thus, when C is bound to PKI, the complex is rapidly exported from the
nucleus (Wen et al., 1995). This NES, ®rst discovered in PKI and HIV-rev, is now known to
be present in many proteins and contributes actively to the translocation of proteins between
the cytoplasm and the nucleus.
3. Structure of the catalytic subunit
To understand the molecular basis for the function of any protein, it is essential to have a
high resolution structure. The solution of the crystal structure of the C-subunit (Knighton et
al., 1991a, 1991b), the ®rst member of this enzyme family to be solved, was facilitated by two
things. First was the binding of PKI (5±24) which was cocrystallized with the C subunit.
Second was the expression of the C subunit in E. coli (Slice and Taylor, 1989). Furthermore,
the enzyme is readily expressed in a fully phosphorylated form in E. coli (Yonemoto et al.,
1993a), and this led to the isolation of large quantities of very pure and active enzyme.
3.1. Overall architecture
The C subunit is a bilobal enzyme with two major subdomains that are conserved
throughout the protein kinase family (Knighton et al., 1991a, 1991b). In Fig. 2(a) the structure
is correlated with the sequence motifs as de®ned by Hanks et al. (1988). The structure begins
with a myristylation motif that, in most structures of the recombinant unmyristylated enzyme,
is disordered. This is followed by a nonconserved long A-helix which spans both lobes. The
highly conserved core begins with the small domain that is dominated by a ®ve stranded
antiparallel b-sheet. The single conserved helix in the small domain is the C-helix. This domain
constitutes the nucleotide binding domain. A single linker strand joins the two lobes. Although
the larger carboxy terminal domain is dominated by helices, a single small b-sheet lies at the
active site cleft, and this region contains many of the conserved residues that are essential for
catalysis. The C-terminal tail (residues 300±350) wraps as a mostly extended chain over the
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
317
Fig. 2. Correlation of conserved sequences motifs with the structure of the catalytic subunit. The structure of the
ternary complex of C, PKI (5±24) and ATP is shown in panel a. The structure and sequence (panel b) of the C
subunit are color coordinated according to the eleven distinct sequence motifs identi®ed by Hanks et al. (1988). The
structure coordinates are from Zheng et al. (1993c) (PDB accession No.: 1ATP). Secondary structural elements: ahelices, b-sheets and loops and turns, are depicted as cylinders, planar arrows and noodle lines, respectively. The
phosphorylated Ser338 and Thr197 are indicated as white spheres, and the ATP substrate is depicted, as a ball and
stick structure, in the active site cleft.
318
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
Fig 2 (continued)
surface of both lobes and again is not conserved throughout the protein kinase family although
it is shared by all members of the cAPK, PKC and PKG families (Hanks et al., 1988). Fig.
2(b) is color-coded according to the subdomains (I±XI) de®ned by Hanks et al. (1988).
As seen in Figs. 2 and 3, the active site lies at the cleft between the two lobes with the
adenine ring of ATP deeply buried at the base of the cleft. The peptide docks to the surface of
the large lobe where the hydroxyl group of the P site Ser or Thr is poised for a direct in-line
transfer of the g-phosphate of ATP (Ho et al., 1988). The catalytic subunit of cAPK is the
only protein kinase that has been crystallized in its active form in the presence and absence of
both substrates and inhibitors. These various structures have revealed the conformational
¯exibility that is an essential part of catalysis. However, before talking about catalysis,
¯exibility and peptide recognition, we ®rst need to describe the active site more rigorously. This
review will focus in particular on the conserved motifs in the core and on our understanding of
the molecular basis for the importance of each motif. It is based on numerous structures of the
Fig. 3. Active site cleft of the catalytic subunit of cAPK is characterized by conserved loops that converge at the site
of phosphoryl transfer. The top panel (a) is a stereo view of the ternary complex, rC:PKI (5±24): MnATP. The
glycine-rich loop (residues 48±57) is in red, the linker (residues 120±127) in yellow and the catalytic loop (residues
166±171) and the magnesium positioning loop (residues 184±187) in white. The side chains of His87, phospho
Thr197, Arg280 and Glu208 are also indicated. The convergence of these three loops at the active site is shown in
the bottom panel (b).
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
Fig. 3.
319
320
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
C-subunit and on comparison with many other protein kinases that have been crystallized in
the last 4±5 years.
3.2. Conserved core
When the crystal structure of the C-subunit was ®rst solved, there was already a
considerable foundation of structure/function studies that allowed us to understand how it
might function as a catalyst. There were anity labeling and di€erential labeling studies that
had identi®ed key residues at the active site such as Lys72 and Asp184 (Zoller et al., 1981;
Buechler and Taylor, 1988). In addition, there was a growing body of genetic information that
de®ned the larger family of protein kinases and identi®ed a set of highly conserved amino
acids. Although these enzymes are very diverse in their size, in how they are regulated, and in
the substrates they recognize, all share a conserved core whose fold is described by the
structure of the C-subunit. In addition, there are a number of residues that are conserved
throughout the family. Although these residues are widely disbursed in the linear sequence of
the core, the structure revealed that most, in fact, clustered around the active site cleft and
Fig. 4. Interactions of Mg2ATP with residues in the catalytic subunit of cAPK. This ®gure is adapted from Fig. 5a
of Madhusudan et al. (1994).
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
321
appeared to be important for ATP binding and catalysis. The general convergence of active site
residues around ATP is shown in Fig. 3(b) while Fig. 4 identi®es most of the important
residues that interact directly with ATP. As summarized in Fig. 3, most of these residues are
localized speci®cally in three loops that converge at the active site cleft where the ATP is
bound. The most dynamic of these loops is contributed by the small lobe while the other two
lie in the large lobe. All of this information made interpretation of the structure feasible. We
shall de®ne these various motifs beginning ®rst with the small lobe.
The function of the small lobe, shown in Fig. 5(a), is to bind nucleotide, leaving the gphosphate poised for transfer to a peptide or protein substrate. It is dominated by a ®ve-
Fig. 5. The small domain and its interactions with Mg2ATP. The small domain and the linker segments are shown
in the upper panel (a) as a ribbon diagram with the glycine-rich loop (residues 48±57) and the C-Helix (residues 84±
98) highlighted in white. The a carbons of the highly conserved residues (Gly50, Gly52, Gly55, Lys72 and Glu91)
are indicated as spheres. Relative to the ATP substrate, the position of the glycine-rich loop is shown in the lower
panel (b) in both its open (Zheng et al., 1993b; white) and closed (Zheng et al., 1993a, 1993c; multicolor)
conformations. The hydrogen bonds between backbone amides and the b- and g-phosphates are also indicated as
dashed lines.
322
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
stranded antiparallel b-sheet with a single conserved helix. The position of the small lobe
relative to the large lobe leads to the opening and closing of the active site cleft and is an
essential feature of the catalytic process.
The ®rst two b strands, linked by a glycine-rich loop, constitute a nucleotide positioning
motif (NPM), residues 48±57. Gly50, 52 and 55 are highly conserved and are hallmark of the
protein kinase family. The NPM spans the entire length of the wedge shaped nucleotide
binding pocket and forms the ceiling of this pocket with the nucleotide ®tting snugly against
this motif. The highly conserved Val57 lies above the ribose ring while three backbone amides
at the tip of the loop position the phosphates of ATP. The glycines at the tip of the loop are
particularly important for positioning the phosphates of ATP. The backbone amide of Ser53
binds to one of the g-phosphate oxygen's while the backbone amides of Phe54 and Gly55
position one of the b phosphate oxygens (Fig. 5(b)). Gly50 lies above the ribose ring. Gly52 is
highly conserved and is also the most important for catalysis (Hemmer et al., 1997; Grant et
al., 1998). The presumed reason for its importance is that the hydrogen bonding of the Ser53
amide to the g-phosphate as shown in Fig. 5(a), is thought to be a critical step for forming the
transition state intermediate prior to phosphoryl transfer (Bossemeyer, 1994; Zheng et al.,
1993a). The tip of this loop is the most mobile part of the molecule and serves as a sensor for
what is occupying the active site cleft ( Narayana et al., 1997a). While more than a dozen
di€erent structures of the C subunit have been solved, the tip of the loop has only been
observed in a very stable form, based on temperature factors, in the ternary complex where
ATP or an ATP analog and PKI (5±24) are both bound with high anity (Zheng et al., 1993a;
Narayana et al., 1997a).
b strand 3 contains another conserved residue, Lys72. The importance of this Lys was ®rst
recognized when it was shown to be modi®ed by the ATP analog, ¯uorosulfonyl benzoyl
adenosine (Zoller et al., 1981). The structure revealed that Lys72 binds to the a and b
phosphates of ATP. b strand 3 is followed by a small B-helix. This is the only element of
secondary structure in the core of the C-subunit that is not shared by the overall protein kinase
family. The long C-helix (Fig. 6) that follows is, however, conserved and houses the third
essential motif in the small lobe, Glu91. Like the nucleotide positioning motif, this helix spans
the entire length of the wedge-shaped ATP binding pocket. In the active form of the C-subunit
this Glu is directed towards Lys72. It does not interact directly with the nucleotide but rather
positions Lys72. Through multiple interactions the C-helix actually positions the entire
molecule for catalysis.
Helical switches appear to be important modulators of protein kinase function and this Chelix switch is critical for every protein kinase. Like railroad switches connecting di€erent
tracks and determining completely di€erent fates, the twist of this helix determines the
functional mode of the enzyme. Its capability to serve as such a global switch is due to the
multiple contacts that are made by this helix to all parts of the molecule. A closer examination
of the speci®c residues in the turn of the helix that includes Glu91 is sucient to demonstrate
this global communication (Fig. 6(b)). Lys92, for example, pairs with the a-carboxylate of
Phe350. This carboxyl terminal Phe is buried in the small lobe as is the ion pair between Lys92
and Phe350. The next residue in the turn, Arg93, forms part of the deep hydrophobic pocket
where Trp30 at the end of the A-helix is docked. Replacement of Trp30 with either Tyr or Ala
introduces signi®cant instability (Herberg et al., 1998). The preceding turn of this helix
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
323
Fig. 6. The C-helix interacts with many parts of the molecule and contributes directly to ATP binding at the active
site. The C-helix is shown in white, while PKI (5±24) is in red. Panel (a) depicts the relative location of the C-helix
in the catalytic subunit. A close up view (panel (b)) indicates many of the speci®c interactions that are made by
residues in the C-helix. Panel (c) is a perpendicular view of the image depicted in panel (d).
contains His87 which under some conditions makes direct contact with the essential phosphate
on Thr197 in the large lobe (Cox and Taylor, 1995).
The correct orientation of this helix is essential, although how it is positioned appears to be
unique for every kinase. The surface of the helix that faces the b-sheet and the active site is
more-or-less conserved throughout the family and in all cases, when the kinase is in an active
324
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
conformation, this helix must be positioned such that Glu91 is oriented towards Lys72 in b
strand 3. The opposite surface of the helix is variable and is positioned di€erently in each
kinase. As indicated above, when the C subunit is in its active conformation, this surface is
correctly positioned by both the carboxyl-terminal tail and by the amino terminal A-helix Ð
two regions that lie outside the conserved core. In general, this positioning of the C-helix by
residues that lie outside the core appears to be true for most protein kinases. In some cases, it
is the segment that lies either directly amino-terminal or carboxy-terminal to the core that is
important while in other cases a separate regulatory subunit is required. In cdk2, for example,
it is cyclin binding that correctly orients this helix (De Bondt et al., 1993; Je€rey et al., 1995).
In MAP kinase or ERK2 the carboxy terminal tail, speci®cally Phe327/Phe329, ®lls this surface
(Taylor and Radzio-Andzelm, 1994; Zhang et al., 1994) and the precise orientation of this part
of the tail is dependent on phosphorylation of Thr183 and Tyr185 in the activation loop
(Canagarajah et al., 1997; Khokhlatchev et al., 1998). In the insulin receptor kinase it is the
amino-terminal segment preceding the core that complements the outer surface of the C-helix
(Hubbard et al., 1994; Hubbard, 1997). In src it is the proline-rich segment that links the SH2
domain to the core (Xu et al., 1997). In all of these cases, where structures of active and
inactive conformations are available, the orientation of the C-helix is altered as a consequence
of activation.
The loop between the C-helix and b strand 4 spans the surface of the large lobe and is
functionally part of the large lobe. b strands 4 and 5 complete the sheet and then fuse with the
linker strand (residues 120±127) that joins the two lobes. The adenine ring butts up against this
linker strand making two hydrogen bonds. The N6 nitrogen hydrogen bonds to the backbone
carbonyl of Glu121 while the N1 nitrogen in the adenine ring hydrogen bonds to the backbone
amide of Val123. The base ®ts snugly into this pocket. There are no water molecules, and the
thermostability of the enzyme is enhanced signi®cantly when this pocket is ®lled (Herberg et
al., 1998). Glu127 at the end of the linker is not invariant, but plays two roles in cAPK. It
binds to the 2' and 3 ' hydroxyls of the ATP ribose and also binds to the P-3 Arg of the
peptide. It also contributes to the binding of a highly ordered water molecule that links the
active site directly to Tyr330 in the nonconserved carboxy terminal tail (Narayana et al.,
1997b; Shaltiel et al., 1998).
The large lobe is mostly helical with the exception of a small b-sheet comprised of 4 strands
that line the active site cleft (Fig. 7(b)). The two remaining loops at the active site are
contributed by this segment. b strands 6 and 7 are joined by a conserved loop referred to as
the catalytic loop because it contains several conserved residues that contribute directly to
catalysis (Fig. 7(a)). Asp166 hydrogen bonds to the P site hydroxyl moiety of the substrate
peptide (Madhusudan et al., 1994). It is positioned to serve as a catalytic base, although the
actual rate enhancement provided by this residue appears to be small (Zhou and Adams,
1997). Lys168 is conserved in Ser/Thr protein kinases and interacts directly with one of the gphosphate oxygens of ATP. Asn171, also conserved, bridges the loop and binds to the
Fig. 7. The b-sheet in the large lobe provides most of the residues that contribute to phosphoryl transfer. Panel (a)
on the left highlights residues in the catalytic loop while panel (b) highlights residues in the Mg++ positioning loop.
The b-sheet at the active site cleft is summarized in the panel (c).
Fig. 7.
326
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
inhibitory Mg++ that interacts with the a and g-phosphates of ATP (Zheng et al., 1993a). b
strands 7 and 8 form the `¯oor' of the adenine binding pocket with the loop that joins them
being juxtapositioned up against the linker strand. The loop joining b8 and b9, referred to as
the Mg++ positioning loop, is also highly conserved (Fig. 7(c)). Asp184 in this DFG motif
binds to the activating Mg++ that bridges the b and g-phosphates of ATP. The position of the
loop is stabilized by interaction with Arg165 which binds to the backbone carbonyl of Phe187.
Arg165 which is highly conserved in most protein kinases immediately precedes the catalytic
loop and also interacts bivalently with the essential phosphate on Thr197. It is, therefore,
another important `bridging' residue. b strand 9 serves several roles: (1) it is paired with b
strand 6, (2) it contributes to stabilization of the activation loop through the interaction of
Lys189 with phospho Thr197 and (3) it helps to stabilize the A-helix through Arg190 which,
like Arg93 in the small lobe, forms part of the deep pocket between the two lobes where Trp30
lies. Both arginines form an amino-aromatic interaction with Trp30 and will be discussed later.
b strand 9 then goes into the `activation' loop.
The activation loop (Fig. 8(a)) is an extremely important segment that is highly sensitive to
regulation by phosphorylation in most protein kinases. The essential phosphate on Thr197 is
present in all the structures of the C-subunit that have been solved so far. Thus we only have
`snapshots' of the active form of the C-subunit. Like the C-helix, this loop must be oriented
correctly for the enzyme to assume a fully active conformation. This region will be discussed in
Fig. 8. The activation loop is coordinated by phospho Thr197. Panel (a) depicts the activation loop (red) relative to
the entire molecule. The C-helix and PKI (5±24) are highlighted in white and yellow, respectively. Panel (b) is a
close up view of the activation segments ¯anked by two conserved residues, Asp184 at the amino terminus and
Glu208 at the carboxyl terminus. The coordination of (P)Thr197 with Arg165, Lys189 and His87, interactions that
render it inaccessible to removal by phosphatases, are also highlighted.
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
327
more detail later in Section 3.4. The activation loop is followed by the P + 1 loop which in
cAPK provides a hydrophobic docking surface for the P + 1 residue in the consensus site
peptide. This loop, discussed in more detail in Section 6, also contributes to recognition of the
P, P-2, P-3 and the P-6 sites and can thus be thought of as the peptide positioning loop.
The remainder of the large lobe consisting of the F, G, H and I helices together with the D
and E helices, form the solid core of the large domain. Residue 300 begins the nonconserved
carboxyl terminal tail. This core has been studied less extensively than the active site regions;
however, within the large lobe are two additional conserved modules or motifs that are more
distal to the active site, but, nevertheless, clearly linked. One motif is a buried ion pair between
Arg280 following the H-helix and Glu208 which follows the P + 1 loop (Fig. 8(b)). The other
conserved motif includes Asp220 and Trp222. Both are part of the F-helix (Fig. 9(a)) and are
buried. The Asp220/Trp222 motif forms a direct link, however, between the active site cleft
and the Arg280/Glu208 ion pair. Asp220 binds to the backbone amides of Arg165 and Tyr164
that precede the catalytic loop and thus most likely contributes to the positioning of the loop
(Fig. 9(b)). As indicated, Arg165 interacts bivalently with the phosphate of Thr197 and also
with the backbone carbonyl of Phe187 that ends the Mg++ positioning loop. Trp222 on the
other side of the helix provides a hydrophobic surface on which Arg280 docks. Glu208, the
partner of Arg280, forms the anchor for the P + 1 loop and probably for the entire activation
segment. Thus there is an intertwined network of conserved interactions that extends from the
active site cleft to the lower surface of the large lobe. Most of the speci®c contacts to ATP
were summarized in Fig. 4.
Fig. 9. The F-helix provides another conserved motif. The position of the F-helix (shown in white) relative to the
entire molecule is illustrated in panel (a). Depicted in panel (b) is a close up of the F-helix highlighting the speci®c
interactions it makes with the active site region and the conserved ion pair between Glu208 and Arg280.
328
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
3.3. Heads and tails
The so-called `heads' and `tails' that immediately ¯ank the kinase core vary with each kinase.
It is the surfaces they ¯ank that determine their function (Shaltiel et al., 1998). In the C
subunit the amino terminus begins with a myristylation motif, residues 1±14 Fig. 10(a). The
amino-terminal Gly is myristylated cotranslationally in mammalian cells (Carr et al., 1982) and
is also myristylated when C is coexpressed in E. coli with N-myristyl transferase (Yonemoto et
al., 1993b). The recombinant enzymes that have been crystallized so far have not been
myristylated, and the amino-terminal 10 to 15 residues are typically disordered. When the
mammalian enzyme was crystallized, the amino-terminus folded over so that the myristyl
group was buried loosely in a hydrophobic pocket on the surface of the large lobe (Zheng et
al., 1993b). The myristylation motif is followed by the long amphipathic A-helix that is
terminated at the carboxy end with Trp30. Trp30 and Phe26 from the preceding turn of the
helix form a knob that ®ts into a deep hydrophobic socket that lies precisely between the two
Fig. 10. The N-terminal segment that ¯anks the core is comprised of a myristylation motif followed by the
amphipathic A-helix. The position of the A-helix, shown in white, relative to the rest of the molecule, is shown in
panel (a). The speci®c residues in this motif and some of their partner residues in the core are shown in panel (b).
The myristyl motif is shown in yellow. The coordinates for this structure image are from Zheng et al. (1993b); PDB
accession No.: 1CMK.
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
329
lobes of the kinase core (Veron et al., 1993). It is sandwiched speci®cally between Arg93 in the
C-helix of the small lobe and Arg190 in b strand 9 of the large lobe and is positioned near the
narrow end of the wedge-shaped ATP binding site (Fig. 10(b)). The carboxyl termini of the A
and C-helices are further intertwined by a series of hydrogen bonds with each helix helping to
cap the other. The amino-terminal segment, from the myristyl motif through the amphipathic
A-helix, is coupled primarily through hydrophobic interactions to the core.
The initial portion of the carboxyl-terminal tail (residues 300±316) is anchored ®rmly to the
large lobe by a series of contacts (Narayana et al., 1997b; Fig. 11). This segment is conserved
in the cAPK, PKC and PKG subfamilies but not in other kinases such as ERK2 (Hanks and
Hunter, 1995). In ERK2 the position of the carboxyl-terminal residues cover the same surface
of the core that in cAPK is masked by the A-helix. This region changes somewhat when ERK2
is phosphorylated, and the changes that are introduced by phosphorylation of the activation
loop serve to create a new dimer interface (Canagarajah et al., 1997; Khokhlatchev et al.,
Fig. 11. The carboxyl terminal tail ¯anks both domains of the conserved core. In panel (a) the relative position of
the carboxyl tail (residues 300±350) is shown in white. Panel (b) depicts a comparison of the position of the tail in
the closed and open conformations. The closed conformation structure (purple±red multicolored ribbon) is from the
ternary complex of rC:PKI (5±24):ATP (Zheng et al., 1993c; PDB accession No.: 1ATP), and the open
conformation structure (white ribbon) is from the mammalian enzyme crystallized with a di-iodinated version of
PKI (5±22) (Zheng et al., 1993b; PDB accession No.: 1CMK).
330
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
1998). In particular, Phe329 which interfaced with the core in a manner analogous to Trp30 in
cAPK ¯ips out and now contributes directly to the dimer interface. In src the short carboxyl
terminal tail contains an inhibitory Tyr that is constitutively phosphorylated in the inactive
enzyme. In the inactive form of src this Tyr docks to its own SH2 domain thereby helping to
tether the enzyme in an inactive conformation (Xu et al., 1997). The next segment of the
carboxy-terminal tail (residues 315±335) serves as the `gate' that allows for entry and exit of
the nucleotide from the active site cleft (Chestukhin et al., 1996a; Narayana et al., 1997b;
Shaltiel et al., 1998). This segment shows considerable ¯exibility depending on what is
occupying the active site cleft. The highly acidic segment surrounding an important tyrosine,
Tyr330, also is thought to serve as a docking mechanism for basic substrates. In addition, this
region is the cleavage site for a highly speci®c kinase splitting membrane protease recently
identi®ed as meparin (Chestukhin et al., 1997, 1996b). The last segment of the tail is ®rmly
anchored to the small lobe with Phe350 deeply buried in the small lobe. As discussed
previously, the buried a-carboxylate is anchored to Lys92 in the C-helix.
3.4. Covalent modi®cations
The catalytic subunit is subject to several types of covalent modi®cation (Fig. 12). In
addition to the myristyl moiety which is added cotranslationally to the N-terminal Gly (Carr et
al., 1982), the enzyme is phosphorylated. In E. coli the C-subunit is autophosphorylated at
four sites, Ser10, Ser139, Thr197 and Ser338 (Yonemoto et al., 1997). In the mammalian
enzyme Thr197 and Ser338 are very stable phosphorylation sites and are very resistant to
removal by phosphatases (Shoji et al., 1979, 1983). Ser139 has not been observed as a
phosphorylated residue in the mammalian enzyme while Ser 10 can be autophosphorylated
slowly in vitro (Toner-Webb et al., 1992). Whether Ser10 is phosphorylated physiologically is
unknown. Based on mutational analysis, the phosphate on Ser338 is thought to contribute to
stability but does not in¯uence the kinetic properties of the enzyme (Yonemoto et al., 1997).
Thr197, in contrast, is essential for maximum activity and contributes to the correct
con®guration of residues at the active site cleft. When Thr197 was replaced with Asp, the
enzyme was still reasonably active. When replaced with Ala, however, the Km for ATP
increased 140-fold from 10 mM to 1.4 mM and the phosphoryl transfer step was reduced by
more than two orders (Adams et al., 1995).
In many other protein kinases, the conformation of the activation loop is very dynamic.
While we have yet to see the conformation in the inactive, dephosphorylated form of cAPK, in
the active form of the enzyme this loop is very stable. As was seen in Fig. 8, it provides part of
the surface on which PKI (5±24) docks, and it does not appear to undergo any major
conformational changes as a consequence of substrate or inhibitor binding. A structure of an
adenosine binary complex represented the ®rst structure of the C subunit that had no peptide
bound, and there were no changes in this region (Narayana et al., 1997b). It does indeed
appear to be a stable surface on which the substrate protein or inhibitor docks.
The catalytic subunit is typically assembled as a fully active enzyme phosphorylated on
Ser338 and Thr197. The enzyme is then kept in an inactive state by its association with
regulatory subunits. The active enzyme is then unleashed in response to cAMP. Although the
phosphate on Thr197 is not thought to turnover rapidly in cAPK, unlike src and MAP kinase
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
331
Fig. 12. The catalytic subunit of cAPK is myristylated and phosphorylated. The myristyl moiety is shown in yellow
(spheres) and key phosphorylated residues as light blue spheres. Adapted from (Zheng et al., 1993b).
and many other members of the protein kinase family, the enzyme does go through a
processing procedure following synthesis and assembly into a holoenzyme complex. Based on a
kinase minus mutant of S49 mouse lymphoma cells, Steinberg (1991) identi®ed an inactive
particulate form of C that was not phosphorylated on Thr197. A kinase activity was
subsequently puri®ed that phosphorylated Thr 197 in the C-subunit (Cauthron et al., 1998).
More recently, it has been shown that the 3-phosphoinositide-dependent protein kinase, PDK1,
can serve in vitro as a cAPK kinase that speci®cally phosphorylates Thr197 in the activation
loop of the C subunit (Cheng et al., 1998a). PDK1 is a newly isolated member of the cAPK,
PKC, PKG subfamily. It is also related to PKB(Akt) (Downward, 1998). The kinase core is
332
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
closely related to PKC, PKG and cAPK; however, PDK1 as well as PKB (Akt) also contains a
plextrin homology domain that is thought to bind to phosphoinositide lipids. PDK1 was
shown to phosphorylate the activation loop of Akt (Alessi et al., 1997a, 1997b) and of p60 S6
kinase (Pullen et al., 1998). Based on sequence similarities in the activation loops of these two
enzymes and the activation loop of cAPK, its ability to phosphorylate the activation loop of
the C-subunit was tested. A dephosphorylated form of the C-subunit was isolated by
expressing the enzyme in E. coli in the presence of H89, a speci®c inhibitor of cAPK. This Csubunit was readily phosphorylated and activated by PDK1 (Cheng et al., 1998a). PKC z
(Chou et al., 1998; Le Good et al., 1998), like MAP kinase appears to be activated by the
dimeric turnover of the activation loop phosphate, whereas PKCa and PKCb more closely
resemble cAPK in that they are assembled as an active phosphorylated protein and then
activated in response to second messengers including both Ca++ and lipids (Dutil et al., 1998).
A new aspect of the processing of the C-subunit has thus been opened up by these ®ndings.
This will be particularly intriguing in light of recent suggestions that the catalytic subunit may
function in other ways that are completely independent of R subunits and cAMP (Zhong et al.,
1997).
The myristyl group is loosely anchored to the core and may play a role in phosphorylating
membrane proteins; however, as yet there is no convincing evidence to support this. Certainly
in other proteins such as MARCKS, src and recoverin, the myristyl group provides a loose
membrane anchor that is joined by patches of basic residues to create a tight membrane
anchor (Murray et al., 1997). In the case of MARCKS and recoverin, the membrane anchored
conformation is regulated by a switch mechanism. Membrane anchoring of MARCKS is
released by PKC phosphorylation while Ca++ regulates the accessibility of the myristyl group
in recoverin (McLaughlin and Aderem, 1995; Ames et al., 1997). Although the potential for
such a switch mechanism exists for the C-subunit, it has not yet been demonstrated.
A novel type of covalent modi®cation was also identi®ed recently at the amino-terminus of
the C subunit, deamidation of Asn2 (Jedrzejewski et al., 1998). This modi®cation appears to
account for the two major isoforms of the C subunit, A and B, that were identi®ed in
mammalian tissues (Van Patten et al., 1986). When Asn2 is deamidated, the amino-terminus of
the protein (residues 1±14) are disordered in the crystal structure as they are in the crystal
structures of the recombinant C-subunit that is not myristylated and is phosphorylated on
Ser10. The crystal structure of the mammalian C-subunit that is not deamidated shows the
amino-terminal mytristyl moiety folded into a hydrophobic pocket on the surface of the Csubunit. Although the chain can be traced, the temperature factors for residues 1±10 remain
high suggesting that the acyl group may be loosely anchored. The amino-terminus thus appears
to be sensitive to a number of modi®cations that may contribute in ways that are still
unrecognizable to physiological function.
Fig. 13. Stereo views of the open and closed conformations of the catalytic subunit. The closed conformation (light
blue trace) is from the ternary complex C:PKI(5±24):ATP (PDB accession No.: 1ATP) (Zheng et al., 1993c), and
the open conformation (pink trace) is the mammalian binary complex (Zheng et al., 1993b; PDB accession No.:
1CMK). The image in panel (b) is a perpendicular view of the image depicted in panel (a).
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
Fig ?? (continued)
Fig. 13.
333
334
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
4. Conformational ¯exibility
4.1. Crystallographic evidence
The conformational ¯exibility of the active C subunit is revealed, in part, by the various
crystal structures that have been solved. The two ternary complexes, C:PKI (5±24):MgATP
and the C:PKI:MgAMPPNP, both show a tightly closed conformation (Bossemeyer et al.,
1993; Zheng et al., 1993a; Fig. 13). In this conformation the small lobe is oriented so that
His87 at the beginning of the C-helix forms an ion pair with the phosphate on Thr197 in the
activation loop. The glycine-rich loop is folded over the phosphates of ATP so that the
backbone amide of Ser53 is close enough to hydrogen bond to the g-phosphate as was shown
in Fig. 5. In contrast to all of the other crystal structures of the C-subunit, the temperature
factors for the tip of the loop are very low for this ternary complex. This structure very likely
resembles the transition state in catalysis. Although the side chain of Ser53 also interacts with
the backbone carbonyl of the P site residue in the inhibitor peptide (Bossemeyer, 1994),
replacement of Ser53 with Ala does not hinder catalysis (Aimes and Taylor, 1998). This side
chain interaction is thus not essential. Replacement of Gly52 with Ala does, however, lead to
impaired catalysis, most likely because it sterically interferes with formation of the hydrogen
bond between the Ser53 amide and the g-phosphate of ATP thus reinforcing the importance of
that hydrogen bond for catalysis.
While the fully closed conformation is essential for ecient phosphoryl transfer, the
nucleotide cannot be released when the enzyme is in this state. Several di€erent more open
conformations have been observed. The structure of the mammalian enzyme that is bound to a
di-iodinated form of PKI (5±22) revealed a much more open conformation (Karlsson et al.,
1993; Zheng et al., 1993b; Fig. 13), whereas other complexes such as the rC:adenosine binary
complex reveal an intermediate conformation (Narayana et al., 1997b). A comparison of these
structures has allowed us to de®ne the features that are associated with opening of the cleft as
well as the rigid body motions that comprise the enzyme (Shaltiel et al., 1998). The general
degree of openness can be approximated by three measurements: (1) the distance between
His87 and Thr197, (2) the distance between Ser53 and Asp166 and (3) the distance between
Tyr330 and Glu127. In the most open conformation the C-helix has pivoted with the
positioning of the carboxyl terminal portion unchanged but with the amino terminal portion
now lifted away from the large lobe. His87 which interacts with the phosphate of Thr197 in
the closed conformation is now more than 6 AÊ away (Zheng et al., 1993b). This is the primary
electrostatic or hydrogen bond contact between the two lobes. The Ser53 distance re¯ects the
tip of the glycine loop. In the open conformation this distance is now >7.1 AÊ. This opening of
the cleft and lifting of the glycine-rich lid is contributed by two kinds of motions Ð a lifting of
the tip of the loop and a shearing of the small domain (Narayana et al., 1997b). The third
distance re¯ects the movement of the carboxyl terminal tail which serves as a `gate' that allows
for entry and exit of the nucleotide. This segment mostly stays linked to the surface of the
small lobe. In the closed conformation Tyr330 contacts a primary structured water molecule
and the P-3 Arg and thus contributes directly to the con®guration of residues at the active site
cleft. Replacement of Tyr330 leads to a decrease in catalytic eciency (Chestukhin et al.,
1996b). This segment is very stable with relatively low temperature factors when the nucleotide
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
335
gate is `closed'. In the open conformation this distance between Tyr330 and Asp127 is greater
than 7 AÊ, and the temperature factors of the tail are very high, indicating a signi®cant degree
of disorder. The `malleability' of this tail, predicted ®rst on the basis of di€erential reactivity of
the endogenous cysteines, Cys199 and Cys343 (Jimenez et al., 1982), was subsequently
con®rmed by the susceptibility of the free subunit to cleavage at this site by the kinase speci®c
membrane protease, KSMP (Chestukhin et al., 1996b). That malleability has been con®rmed
dramatically by the crystal structures. The intermediate structures reveal di€erences primarily
in the opening and ¯exibility of the glycine-rich loop.
In de®ning the kinase core and the conserved active site residues, it is clear that a number of
highly conserved loops converge at the active site cleft. These include the glycine-rich loop in
the small lobe and the catalytic and magnesium positioning loops in the large lobe. In
comparing these loops in the various binary and ternary complexes observed so far, it is
striking that only the glycine-rich loop is subject to major conformational changes.
4.2. Solution evidence
While the crystal structures reveal the detailed molecular structures of the enzyme, the
structures represent an ensemble of rigid `snapshots'. Other methods are required to probe the
dynamics of the enzyme in solution. Recent footprinting using Fe-EDTA free radicals as a
cleavage reagent revealed distinct conformations for the free enzyme and the C:PKI (5±
24):ATP ternary complex (Cheng et al., 1998b). The pattern of footprinting when correlated
with the crystal structures is consistent with the open and closed conformations and
furthermore demonstrates that the free enzyme adopts a much `looser' conformation.
Surprisingly, when the e€ects of PKI(5±24) and MgATP were characterized separately, it was
found the peptide had little e€ect on global conformation. In contrast, the footprinting
observed in the presence of MgATP was indistinguishable from the ternary complex indicating
that the nucleotide is primarily responsible for inducing the changes that lead to eventual
catalysis (Cheng et al., 1998b).
Fluorescence anisotropy provides an alternative strategy to characterize local conformational
changes. The catalytic subunit contains only two endogenous cysteines, Cys199 and Cys343.
Since both can be protected against covalent modi®cation by MgATP, it is possible to
introduce single cysteines into various sites on the C subunit and then label them with a
¯uorescent probe such as ¯uoresceine. The conformational ¯exibility of ®ve such C-subunits
(Cys16, Cys81, Cys244, Cys327 and Cys343) have now been characterized. The ¯uorescence
anisotropy measurements reveal a range of backbone ¯exibility (Gangal et al., 1998), and the
in¯uence of binding of the regulatory subunits is now being evaluated.
5. Catalysis
The early studies of Cook and Walsh demonstrated that the reaction assumes a preferred
ordered mechanism with ATP most likely binding ®rst (Cook et al., 1982; Whitehouse and
Walsh, 1983; Yoon and Cook, 1987). The kinetic properties of the enzyme are highly sensitive
to Mg++ (Cook et al., 1982; Adams and Taylor, 1993). In trying to de®ne the individual steps
336
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
associated with catalysis and to evaluate mutants, it was essential to ®rst characterize the rate
of phosphoryl transfer more carefully. The Km's for a heptapeptide substrate and ATP are
approximately 10±20 mM; the kcat is typically about 20 sÿ1. When the e€ect of viscogens on the
kcat was measured, however, it was found that the kcat was highly dependent on viscosity
(Adams and Taylor, 1992). This suggested that the kcat did not re¯ect the true chemical
transfer step but instead re¯ected the rate of product release. To con®rm this, Grant and
Adams used a rapid quench assay to measure the presteady state rate constants for the
formation of the phosphopeptide. The rapid burst (500 sÿ1) was predicted to correlate with the
phospohoryl transfer step while the slower rate (20 sÿ1) corresponded to the kcat (Grant and
Adams, 1996). Based on the analysis of a ¯uorescently labeled form of the subunit, a rapid
burst was also observed that correlated well with the 500 sÿ1 seen in the rapid quench assay
(Lew et al., 1997). The slower rate was thought to be comprised of the ADP o€ rate plus
conformational changes that are associated with the release of ADP. The reaction pathway for
the C-subunit can thus be described by the following reaction pathway:
k2
k
k4
kÿ2
500 sÿ1
20 sÿ1
3
ÿ
*
ÿ
*
E ‡ ATP ÿ
)
ÿ
ÿ E:
)
ÿ
ÿ E ‡ ATP ‡ S ÿÿ4 E ‡ ADP ‡ P ÿ
This reaction pathway with a very rapid phosphoryl transfer step and a relatively slow product
release step also demonstrates that the Km for peptide (approximately 20 mM for Kemptide:
Leu±Arg±Arg±Ala±Ser±Leu±Gly) does not re¯ect the true (Adams and Taylor, 1993). The Kd
for peptide binding is actually much greater, approximately 200 mM for Kemptide.
In considering the reaction pathway, it is also important to recognize that the
conformational changes described in the preceding section are all associated with the active
form of the enzyme. Traversing that reaction pathway requires conformational changes that
allow for opening and closing of the cleft between the two lobes. Furthermore, it is the
opening of the cleft allowing for release of the nucleotide that is rate-limiting. The protein
kinases thus appear to be remarkably dynamic proteins. Tethering their intrinsic mobility in
any way will inevitably limit their catalytic eciency.
6. Extended network of interactions
The C-subunit is remarkable for the extended network of interactions that link distal parts
of the molecule to the active site where the primary mission of the enzyme is to transfer the gphosphate of ATP to a protein substrate. In discussing the activation loop and the carboxy
terminal tail, several speci®c residues such as Thr197 and Tyr330 have already been discussed
that demonstrate the extent of this network. Mutation of these residues that lie quite a distance
from the phosphoryl transfer site signi®cantly reduced catalytic eciency not only by
increasing the Km for ATP but also by reducing k3. The peptide binding site is also remarkable
not only for the way in which di€erent parts of the molecule, relatively far apart in linear
sequence come together to recognize speci®c residues, but also for the network that links
peptide recognition to the site of phosphoryl transfer. Glu230, one of the residues associated
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
337
Fig. 14. Recognition of the consensus site peptide. Panel (a) illustrates the binding site for the P-2 and P-3 arginines, while the multiple interactions
of the P + 1 loop with the peptide are shown in panel (c). Panel (b) summarizes the extended network that links distal parts of the active site.
338
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
with recognition of the P-2 Arg in the peptide substrate, is another excellent example to
demonstrate this networking (Grant et al., 1996).
The minimum peptide recognition site can be de®ned as the region at the active site cleft
where the consensus site peptide docks. This region is sucient to convey low anity binding
typical of most substrates. As was seen in Fig. 1, the arginines at the P-3 and the P-2 positions
are key determinants for peptide recognition of the consensus peptide by the catalytic subunit.
Each site for recognition of these arginines is comprised of several acidic residues as
summarized in Fig. 14(a). The P-2 Arg interacts with two carboxylates, Glu230 and Glu170.
Replacement of Glu230 with Gln had two consequences (Grant et al., 1996). As anticipated the
Km was increased from 6.9 mM to approximately 1.4 mM. However, the k3 also was reduced
from 500 to 20 sÿ1. While the change in the electrostatic properties are signi®cant (Tsigelny et
al., 1995), an examination of this residue and its links to other parts of the molecule is
revealing. As seen in Fig. 14, the other residue that forms the P-2 recognition site is Glu170
which is located in the middle of the catalytic loop. The adjacent P-3 Arg interacts directly
with Glu127 in the linker segment that joins the two lobes. Glu127 also interacts with the 2'
and 3 ' -hydroxyl of the ATP ribose. The P-3 Arg also interacts with Tyr330 in the carboxy
terminal tail which coordinates a highly ordered water molecule that was discussed earlier. The
hydrophobic P + 1 residue in the consensus peptide is bound to the P + 1 loop that follows
phospho Thr197. The speci®c hydrophobic residues that form the binding pocket are Leu198,
Phe202, and Leu205. As seen in Fig. 14 (right), in addition to contributing to recognition of
the P + 1 residue, this loop communicates with many other parts of the protein. Gly200
hydrogen bonds to the backbone amide of the P site residue while Thr201 interacts directly
with Lys168 and, in some cases, Asp166 in the catalytic loop where it appears to contribute to
the actual phosphoryl transfer step. Glu203 forms the site for recognition of the P-6 Arg, while
Tyr204 hydrogen bonds to Glu230, a primary determinant of the P-2 Arg recognition site. This
highly interactive network that allows the kinase to recognize peptides and inhibitors as well as
ATP extends in all directions.
The interactiveness of these listed residues at the active site is apparent from several
mutational studies. For example, replacing Glu230 with Gln had two major consequences.
First it reduced the Km for Kemptide as expected, to 6 mM, however, it also signi®cantly
reduced the rate of phosphoryl transfer to 20 sÿ1. Thus there appears to be quite an extended
network that contributes to catalytic eciency.
7. Summary
Our understanding of the protein kinase family has increased considerably since the ®rst
structure was solved in 1991 (Knighton et al., 1991a, 1991b). Not only have many new protein
kinases been discovered making this one of the largest enzyme families, but structures are now
available for quite a few di€erent protein kinases. Several things are striking. First is the
remarkable rearmation of the original prediction that the core would be conserved. Although
there are some di€erences between the protein kinases that phosphorylate Ser/Thr and those
that phosphorylate Tyr, the di€erences are subtle. The conserved residues that we have
reviewed here are more-or-less a constellation of ®xed points that can be superimposed readily
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
339
for all members of the family. Second is the incredible diversity in how each core is activated
and regulated. The mechanisms by which the core can be tethered or locked into an inactive
conformation are limitless. Finally is the dynamic nature of these enzymes. We have focused
here on the dynamics of the active enzyme as it undergoes catalysis. Equally dynamic are the
modules that ¯ank the core and the inhibitory molecules that latch onto the core.
Understanding these dynamics will be a major challenge Ð one that will require a diverse set
of approaches to fully understood at the molecular level.
Acknowledgements
This research was supported by the National Institutes of Health (GM19301). Other support
was provided by the National Biomedical Computation Resource (NIH P41 RR08605-05) and
the National Partnership for Advanced Computation Infrastructure (NPACI; http://
www.npaci.edu). Structure images were prepared using InsightII (Molecular Simulations, Inc.,
San Diego, CA) and Molscript v2.1 (Kraulis, 1991; Avatar Software AB [http://
www.avatar.se]). The authors would also like to acknowledge the assistance of David
Baraclough in the preparation of this manuscript.
References
Adams, J.A., McGlone, M.L., Gibson, R., Taylor, S.S., 1995. Biochem. 34, 2447±2454.
Adams, J.A., Taylor, S.S., 1992. Biochem. 31, 8516±8522.
Adams, J.A., Taylor, S.S., 1993. Prot. Sci. 2, 2177±2186.
Alessi, D., James, S., Downes, C., Holmes, A., Ga€ney, P., Reese, C., Cohen, P., 1997a. Curr. Biol. 7, 261±269.
Alessi, D.R., Deak, M., Casamayor, A., Caudwell, F.B., Morrice, N., Norman, D.G., Ga€ney, P., Reese, C.B.,
MacDougall, C.N., Harbison, D., Ashworth, A., Bownes, M., 1997b. Curr. Biol. 7, 776±789.
Aimes, R., Taylor, S., 1998. Personal communication.
Ames, J.B., Ishima, R., Tanaka, T., Gordon, J.I., Stryer, L., Ikura, M., 1997. Nature 389, 198±202.
Bossemeyer, D., 1994. TIBS Rev. 19, 201±205.
Bossemeyer, D., Engh, R.A., Kinzel, V., Ponstingl, H., Huber, R., 1993. EMBO J. 12, 849±859.
Brostrom, M.A., Reimann, E.M., Walsh, D.A., Krebs, E.G., 1970. Adv. Enzyme Regul. 8, 191±203.
Buechler, J.A., Taylor, S.S., 1988. Biochem. 27, 7356±7361.
Canagarajah, B.J., Khokhlatchev, A., Cobb, M.H., Goldsmith, E.J., 1997. Cell 5, 859±869.
Carr, S.A., Biemann, K., Shoji, S., Parmalee, D.C., Titani, K., 1982. Proc. Natl. Acad. Sci. USA 79, 6128±6131.
Cauthron, R.D., Carter, K.B., Liauw, S., Steinberg, R.A., 1998. Mol. Cell. Biol. 18, 1416±1423.
Cheng, H.-C., van Patten, S.M., Smith, A.J., Walsh, D.A., 1986. Biochem. J. 231, 655±661.
Cheng, X., Ma, Y., Moore, M., Hemmings, B.A., Taylor, S.S., 1998a. Proc. Nat. Acad. Sci. USA 95, 9849±9854.
Cheng, X., Shaltiel, S., Taylor, S.S., 1998b. Biochem. 37, 14005±14013.
Chestukhin, A., Litovchick, L., Schourov, D., Cox, S., Taylor, S.S., Shaltiel, S., 1996a. J. Biol. Chem. 271, 10175±
10182.
Chestukhin, A., Muradov, K., Litovchick, L., Shaltiel, S., 1996b. J. Biol. Chem. 271, 30272±30280.
Chestukhin, A., Litovchick, L., Muradov, K., Batkin, M., Shaltiel, S., 1997. J. Biol. Chem. 272, 3153±3160.
Chou, M.M., Hou, W., Johnson, J., Graham, L.K., Lee, M.H., Chen, C.-S., 1998. Curr. Biol. 8, 1069±1077.
Cohen, P., Rylatt, D.B., Nimmo, G.A., 1977. FEBS Lett. 76, 182.
Cook, P.F., Neville, M.E., Vrana, K.E., Hartl, F.T., Roskoski, R., 1982. Biochem. 21, 5794±5799.
Cox, S., Taylor, S.S., 1995. Biochem. 34, 16203±16209.
340
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
De Bondt, H.L., Rosenblatt, J., Jancarik, J., Jones, H.D., Morgan, D.O., Kim, S.H., 1993. Nature 363, 595±602.
Downward, J., 1998. Curr. Opin. Cell. Biol. 10, 262±267.
Dutil, E.M., Toker, A., Newton, A., 1998. Curr. Biol. 8, 1366±1375.
Fischer, E.H., Krebs, E.G., 1955. J. Biol. Chem. 216, 121±132.
Francis, S., Corbin, J., 1994. Ann. Rev. Physiol. 56, 237±272.
Gangal, M., Cox, S., Lew, J., Cli€ord, T., Garrod, S., Aschbaher, M., Taylor, S.S., Johnson, D.A., 1998. Biochem.
37, 13728±13735.
Gill, G.N., Garren, L.D., 1969. Proc. Natl. Acad. Sci. USA 63, 512±519.
Goldman, P.S., Tran, V.K., Goodman, R.H., 1997. Recent Prog. Hormone Res. 52, 103±120.
Granot, J., Mildvan, A.S., Hiyama, K., Kondo, H., Kaiser, E.T., 1980. J. Biol. Chem. 255, 4569±4573.
Grant, B.D., Adams, J.A., 1996. Biochem. 35, 1533±1539.
Grant, B.D., Tsigelny, I., Adams, J.A., Taylor, S.S., 1996. Prot. Sci. 5, 1316±1324.
Grant, B.D., Hemmer, W., Tsigelny, I., Adams, J.A., Taylor, S.S., 1998. Biochem. 37, 7708±7715.
Hanks, S.K., Hunter, T., 1995. FASEB J. 9, 576±596.
Hanks, S.K., Quinn, A.M., Hunter, T., 1988. Science 241, 42±52.
Hauer, J.A., Barthe, P., Taylor, S.S., Parello, J., Padilla, A., 1999. Pro. Sci., in press.
Hemmer, W., McGlone, M.L., Taylor, S.S., 1997. J. Biol. Chem. 272, 16946±16954.
Herberg, F.W., Doyle, M., Cox, S., Taylor, S.S., 1998. Biochem., submitted for publication.
Hjelmquist, G., Anderson, J., Edlund, B., Engstrom, L., 1974. Biochem. Biophys. Res. Comm. 61, 559±563.
Ho€man, F., 1980. J. Biol. Chem. 255, 1559±1564.
Ho, M.-f., Bramson, H.N., Hansen, D.E., Knowles, J.R., Kaiser, E.T., 1988. J. Am. Chem. Soc. 110, 2680±2681.
Hubbard, S.R., 1997. EMBO J. 16, 5572±5581.
Hubbard, S.R., Wei, L., Ellis, L., Hendrickson, W.A., 1994. Nature 372, 746±754.
Hunter, T., Plowman, G.D., 1997. Trends Biochem. Sci. 22, 18±22.
Jedrzejewski, P.T., Girod, A., Tholey, A., Keonig, N., Thiller, S., Kinzel, B., Bossemeyer, D., 1998. Prot. Sci. 7,
457±469.
Je€rey, P.D., Russo, A., Polyak, K., Gibbs, E., Hurwitz, J., Massague, J., Pavletich, N.P., 1995. Nature 376, 313±
320.
Jimenez, J.S., Kupfer, A., Gani, V., Shaltiel, S., 1982. Biochem. 21, 1623±1630.
Karlsson, R., Zheng, J., Xuong, N.-h., Taylor, S.S., Sowadski, J.M., 1993. Acta Crystal. D49, 381±388.
Kemp, B.E., Bylund, D.B., Huang, T.-S., Krebs, E.G., 1975. Proc. Natl. Acad. Sci. USA 72, 3448.
Kemp, B.F., Graves, D.J., Benjamini, E., Krebs, E.G., 1977. J. Biol. Chem. 252, 4888±4894.
Knighton, D.R., Zheng, J., Ten Eyck, L.F., Ashford, V.A., Xuong, N.-h., Taylor, S.S., Sowadski, J.M., 1991a.
Science 253, 407±414.
Knighton, D.R., Zheng, J., Ten Eyck, L.F., Xuong, N.-h., Taylor, S.S., Sowadski, J.M., 1991b. Science 253, 414±
420.
Khokhlatchev, A.V., Canagarajah, B., Wilsbacher, J., Robinson, M., Atkinson, M., Goldsmith, E., Cobb, M.H.,
1998. Cell 93, 605±615.
Kraulis, P.J., 1991. J. Appl. Crystal. 24, 946±950.
Krebs, E.G., Graves, D.J., Fischer, E.H., 1959. J. Biol. Chem. 234, 2867±2873.
Lew, J., Taylor, S.S., Adams, J.A., 1997. Biochem. 36, 6717±6724.
Luttrell, L.M., van Biesen, T., Hawes, B.E., Koch, W.J., Krueger, K.M., Touhara, K., Lefkowitz, R.J., 1997. Adv.
Second Messenger Phosphoprotein Res. 31, 263±277.
Le Good, J.A., Ziegler, W.H., Parekh, D.B., Alessi, D.R., Cohen, P., Parker, P.J., 1998. Science 281, 2042±2045.
Madhusudan, K.R., Trafny, E.A., Xuong, N.-h., Adams, J.A., Ten Eyck, L.F., Taylor, S.S., Sowadski, J.M., 1994.
Prot. Sci. 3, 176±187.
McLaughlin, S., Aderem, A., 1995. Trends Biochem. Sci. 20, 272±276.
Montminy, M., 1997. Annu. Rev. Biochem. 66, 807±822.
Murray, K.J., El-Maghrabi, M.R., Kountz, P.D., Lukas, T.J., Soderling, T.R., Pilkis, S.J., 1984. J. Biol. Chem. 259,
7673±7681.
Murray, D., Ben-Tal, N., Honig, B., McLaughlin, S., 1997. Structure 5, 985±989.
Narayana, N., Cox, S., Shaltiel, S., Taylor, S.S., Xuong, N., 1997a. Biochem. 36, 4438±4448.
C.M. Smith et al. / Progress in Biophysics & Molecular Biology 71 (1999) 313±341
341
Narayana, N., Cox, S., Xuong, N.-h., TenEyck, L.F., Taylor, S.S., 1997b. Structure 5, 921±935.
Parker, P.J., Aitken, A., Bilham, T., Embi, N., Cohen, P., 1981. FEBS Lett. 123, 332±335.
Pullen, N., Dennes, P.B., Andjelkovic, M., Dufner, A., Kozma, S.C., Hemmings, B.A., Thomas, G., 1998. Science
279, 707±710.
Scott, J.D., Glaccum, M.B., Fischer, E.H., Krebs, E.G., 1986. Proc. Natl. Acad. Sci. USA 83, 1613±1616.
Shaltiel, S., Cox, S., Taylor, S.S., 1998. Proc. Natl. Acad. Sci. USA 95, 484±491.
Shoji, S., Titani, K., Demaille, J.G., Fischer, E.H., 1979. J. Biol. Chem. 254, 6211±6214.
Shoji, S., Ericsson, L.H., Walsh, D.A., Fischer, E.H., Titani, K., 1983. Biochem. 22, 3702±3709.
Slice, L.W., Taylor, S.S., 1989. J. Biol. Chem. 264, 20940±20946.
Steinberg, R.A., 1991. Mol. Cell. Biol. 11, 705±712.
Sutherland, C., O'Brien, R.M., Granner, D.K., 1996. Phil. Trans. R. Soc. London. Series B: Biol. Sci. 351, 191±199.
Sutherland, E.W., Wosilait, W.D., 1955. Nature 175, 169±170.
Tao, M., Salas, M.L., Lipmann, F., 1970. Proc. Natl. Acad. Sci. USA 67, 408±414.
Taylor, S.S., Buechler, J.A., Yonemoto, W., 1990. Annu. Rev. Biochem. 59, 971±1005.
Taylor, S.S., Radzio-Andzelm, E., 1994. Structure 2, 345±355.
Toner-Webb, J., van Patten, S.M., Walsh, D.A., Taylor, S.S., 1992. J. Biol. Chem. 267, 25174±25180.
Tsigelny, I., Grant, B.D., Taylor, S.S., Ten Eyck, L.F., 1995. Biopolymers 39, 353±365.
Ullrich, A., Schlessinger, J., 1990. Cell 61, 203±212.
Van Patten, S.M., Fletcher, W.H., Walsh, D.A., 1986. J. Biol. Chem. 261, 5514±5523.
Veron, M., Radzio-Andzelm, E., Tsigelny, I., Ten Eyck, L.F., Taylor, S.S., 1993. Proc. Nat. Acad. Sci. USA 90,
10618±10622.
Walsh, D.A., Perkins, J.P., Krebs, E.G., 1968. J. Biol. Chem. 243, 3763±3765.
Walsh, D.A., Angelos, K.L., Van Patten, S.M., Glass, D.B., Garetto, L.P., 1990. In: Kemp, B.E. (Ed.), Peptides
and Protein Phosphorylation CRC Press, Boca Raton, pp. 43±84.
Wen, W., Taylor, S.S., Meinkoth, J.L., 1995. J. Biol. Chem. 270, 2041±2046.
Westheimer, F.H., 1987. Science 235, 1173±1178.
Whitehouse, S., Walsh, D.A., 1983. J. Biol. Chem. 258, 3682±3692.
Xu, W., Harrison, S.C., Eck, M.J., 1997. Nature 385, 595±599.
Yeaman, S.J., Cohen, P., Watson, D.C., Dixon, G.H., 1977. Biochem. J. 162, 411.
Yonemoto, W., Garrod, S.M., Bell, S.M., Taylor, S.S., 1993a. J. Biol. Chem. 268, 18626±18632.
Yonemoto, W., McGlone, M.L., Taylor, S.S., 1993b. J. Biol. Chem. 268, 2348±2352.
Yonemoto, W., McGlone, M.L., Taylor, S.S., 1997. Prot. Eng. 10, 915±925.
Yoon, M.-Y., Cook, P.F., 1987. Biochem. 26, 4118±4125.
Zetterqvist, O., Ragnarsson, U., Humble, E., Berglund, L., Engstrom, L.T., 1976. Biochem. Biophys. Res. Comm.
70, 696±703.
Zhang, F., Strand, A., Robbins, D., Cobb, M.H., Goldsmith, E.J., 1994. Nature 367, 704±711.
Zheng, J., Knighton, D.R., Ten Eyck, L.F., Karlsson, R., Xuong, N.-h., Taylor, S.S., Sowadski, J.M., 1993a.
Biochem. 32, 2154±2161.
Zheng, J., Knighton, D.R., Xuong, N.-h., Taylor, S.S., Sowadski, J.M., Ten Eyck, L.F., 1993b. Prot. Sci. 2, 1559±
1573.
Zheng, J., Trafny, E.A., Knighton, D.R., Xuong, N.-h., Taylor, S.S., Ten Eyck, L.F., Sowadski, J.M., 1993c. Acta
Cryst. D49, 362±365.
Zhong, H., SuYang, H., Erdjument-Bromage, H., Tempst, P., Ghosh, S., 1997. Cell 89, 413±424.
Zhou, J., Adams, J., 1997. Biochem. 36, 2977±2984.
Zoller, M.J., Nelson, N.C., Taylor, S.S., 1981. J. Biol. Chem. 256, 10837±10842.