ã
Oncogene (1999) 18, 8024 ± 8032
1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00
http://www.stockton-press.co.uk/onc
Integrin-linked kinase regulates phosphorylation of serine 473 of protein
kinase B by an indirect mechanism
Danielle K Lynch1, Christine A Ellis2, Paul AW Edwards1 and Ian D Hiles*,2
1
Department of Pathology, Cambridge Unversity, Tennis Court Road, Cambridge CB2 1QP, UK; 2Molecular Pharmacology Unit,
GlaxoWellcome Research and Development, Medicines Research Centre, Gunnels Wood Road, Stevenage SG1 2NY, UK
The serine threonine kinase protein kinase B regulates
cellular activities as diverse as glycogen metabolism and
apoptosis. Full activation of protein kinase B requires 3phosphoinositides and dual phosphorylation on threonine308 and serine-473. CaM-K kinase and 3-phosphoinositide dependent-kinase-1 phosphorylate threonine-308.
Integrin-linked kinase reportedly phophorylates serine473. Consistent with this, in a model COS cell system we
show that expression of wild-type integrin-linked kinase
promotes the wortmannin sensitive phosphorylation of
serine-473 of protein kinase B and its downstream
substrates, and inhibits C2-ceramide induced apoptosis.
In contrast, integrin-linked kinase mutated in a lysine
residue critical for function in protein kinases is inactive
in these experiments, and furthermore, acts dominantly
to block serine-473 phosphorylation induced by ErbB4.
However, alignment of analogous sequences from
di€erent species demonstrates that integrin-linked kinase
is not a typical protein kinase and identi®es a conserved
serine residue which potentially regulates kinase activity
in a phosphorylation dependent manner. Mutation of this
serine to aspartate or glutamate, but not alanine, in
combination with the inactivating lysine mutation
restores integrin-linked kinase dependent phosphorylation
of serine-473 of protein kinase B. These data strongly
suggest that integrin-linked kinase does not possess
serine-473 kinase activity but functions as an adaptor
to recruit a serine-473 kinase or phosphatase.
Keywords: Akt; ErbB4; integrin-linked kinase; phosphatidylinositol 3-kinase; protein kinase B; site-directed
mutagenesis
Introduction
Protein kinase B (PKB), the cellular homologue of the
viral oncogene v-akt, is a serine-threonine protein
kinase (Bellacosa et al., 1991; Co€er and Woodgett,
1991; Jones et al., 1991). PKB regulates many diverse
pathways and responses in a cell, depending on cell
type (reviewed in Marte and Downward, 1997; Co€er
et al., 1998). For example, insulin stimulation of L6
muscle cells results in PKB-dependent phosphorylation
and inhibition of glycogen synthase kinase 3a (GSK3a) (Cross et al., 1995; van Weeren et al., 1998) and
stimulation of protein synthesis (Gingras et al., 1998;
*Correspondence: I Hiles
Received 23 June 1999; revised 27 September 1999; accepted
28 September 1999
Kitamura et al., 1998; Ueki et al., 1998). In contrast,
insulin-like growth factor-1 treatment of COS7 cells
and nerve growth factor-dependent neurons confers
PKB-mediated protection from apoptosis following
UV treatment and serum withdrawal, respectively
(Kulik et al., 1997; Dudek et al., 1997). PKB
activation also protects myc transfected Rat-1 fibroblasts (Kau€man-Zeh et al., 1997) and MDCK
epithelial cells (Khwaja et al., 1997) from the
apoptosis (termed anoikis ± Frisch and Francis, 1994)
that normally accompanies cell detachment from
extracellular matrix. PKB dependent phosphorylation
of serine-136 (S136) of BAD, a protein involved in
apoptotic signalling (Datta et al., 1997; Blume-Jensen
et al., 1998; Chao and Korsmeyer, 1998), leads to its
inactivation and may contribute to the regulation of
apoptosis by PKB. Direct phosphorylation and
inhibition of both caspase 9 and the forkhead
transcription factor FKHLR1 by PKB are also likely
to be important in the regulation of apoptosis
(Cardone et al., 1998; Brunet et al., 1999). PKB is
also activated by integrins (Frisch and Ruoslahti, 1997;
King et al., 1997; Dimmeler et al., 1998; Guilherme
and Czech, 1998) and the dual regulation of PKB by
growth factors and integrins suggests that PKB
activation represents a point where signals from
growth factors and extracellular matrix converge in a
cell.
Structurally, PKB possesses an N-terminal pleckstrin
homology (PH) domain that binds lipid products of
phosphatidylinositol 3-kinase (PI3-kinase), a central
kinase domain and a c-terminal extension containing
phosphorylation sites (Marte and Downward, 1997;
Co€er et al., 1998). The regulation of PKB kinase
activity is complex and involves several distinct steps
(Bellacosa et al., 1998; Co€er et al., 1998). The PH
domain of PKB binds to 3-phosphoinositide products
of PI3-kinase and this interaction is implicated in both
membrane targeting and kinase activation (Anjelkovic
et al., 1997; Klippel et al., 1997; Franke et al., 1997).
Dual phosphorylation on both threonine-308 (T308),
within the activation loop of the kinase domain, and
serine-473 (S473) in the c-terminal tail is also required
for full activation of kinase activity (Alessi et al., 1996).
A protein kinase that phosphorylates T308 has been
characterized and termed 3-phosphoinositide dependent-kinase 1 (PDK1) (Alessi et al., 1997a, b; Cohen et
al., 1997; Stephens et al., 1998). Like PKB, PDK1 also
contains a PH domain that binds the lipid products of
PI3-kinase (Alessi et al., 1997b; Stephens et al., 1998).
Phosphorylation of T308 in vivo is dependent on PI3kinase activity, but it is unclear if this requirement is
necessary for the unfolding of PKB to allow access of
Regulation of protein kinase B by ILK
DK Lynch et al
PDK1 to T308, direct activation of PDK1 through its
PH domain, or both of these events (Bellacosa et al.,
1998; Stephens et al., 1998). Ca2+/calmodulin-dependent protein kinase kinase (CaM-KK) can also
phosphorylate T308 or PKB in response to calcium
(Yano et al., 1998). The PI3-kinase dependent kinase
activity that phosphorylates S473 or PKB in vivo,
which has been termed 3-phosphoinositide dependent
kinase 2 (PDK2) (Alessi et al., 1997a), is less well
characterized. MAPKAP kinase 2 is capable of
phosphorylating S473 in vitro, but is unlikely to be
an important in vivo regulator of PKB (Alessi et al.,
1996). In vitro, PDK1 bound to protein kinase-C
related kinase-2 (PRK2) or a c-terminal peptide
derived from PRK2 can phosphorylate S473 of PKB
(Balendran et al., 1999). However it is unclear if this
PDK1/PRK2 complex phosphorylates S473 in vivo. A
third candidate for a S473 kinase is integrin-linked
kinase (ILK) (Delcommenne et al., 1998), originally
identi®ed as a b1-integrin binding protein in a yeast 2hybrid screen (Hannigan et al., 1996). However, the
sequence of the kinase domain of ILK di€ers from the
consensus for a protein kinase in several residues
considered critical for kinase activity.
In this paper, we assess ILK S473 kinase activity in
cells. Expression of exogenous ILK promotes phosphorylation of S473 of PKB resulting in phosphorylation of downstream e€ectors of PKB and inhibition of
C2-ceramide induced apoptosis. Using sequence alignment of ILK analogues from di€erent species
combined with site-directed mutagenesis of ILK, we
probe how ILK functions as a S473 kinase. We
conclude that ILK regulates S473 phosphorylation by
acting as an adaptor, either to recruit a distinct S473
kinase activity or to inhibit a S473 speci®c phosphatase
activity.
methionine (Carrera et al., 1993; Hanks and Hunter,
1995). Figure 1a, lane 5, shows that co-expression of
ILK and ErbB4 CYT-1 in serum starved COS cells
augments phosphorylation of S473 or PKB. Indeed,
enhanced phosphorylation of PKB on S473 is noticeable in cells expressing ILK protein alone in the
absence of ErbB4 CYT-1 (for example compare lanes 1
Results
ILK regulates S473 phosphorylation of PKB
A COS cell transient-expression system based on
expression of 2 isoforms of ErbB4 (CYT-1 and CYT2) was established to study signalling pathways
dependent on PI3-kinase activation. CYT-1 and
CYT-2 di€er by the presence or absence of 16 amino
acids containing a consensus binding site for PI3kinase (Sawyer et al., 1998; Elenius et al., 1999). As
expected, the longer isoform CYT-1 (containing the
PI3-kinase binding site), in contrast to CYT-2,
promoted the recruitment and activation of PI3-kinase
(data not shown) and this correlated with increased
phosphorylation of both S473 and T308 of PKB
(Figure 1a, compare lanes 4 and 7, Ser473 and
Thr308 panels). In all other respects, CYT-1 and
CYT-2 behaved identically. Both forms associated
equally well with SHC, NCK and GRB2 and
stimulated MAP-kinase activity (data not shown and
Figure 1c lanes 4 and 7).
Using the COS cell system, we investigated the
ability of ILK or a mutant of ILK (DN-ILK) to
promote phosphorylation of S473 of PKB. DN-ILK
was generated by changing a critical lysine residue
(lysine 220 in ILK ± see Figure 3a) conserved in most
protein kinases, and essential for kinase activity, to
Figure 1 E€ect of integrin-linked kinase on phosphorylation of
protein kinase B, glycogen synthase kinase-3a, BAD and MAPkinase in serum starved transiently transfected COS cells. COS
cells were transiently transfected with ErbB4 CYT-1, lanes 4, 5
and 6, ErbB4 CYT-2, lanes 7, 8 and 9 or mock transfected (COS),
lanes 1, 2 and 3. Certain COS cells were additionally transfected
with integrin-linked kinase (WT-ILK) (lanes 2, 5 and 8) or a
mutated integrin-linked kinase (DN-ILK) (lanes 3, 6 and 9).
RIPA bu€er extracts were fractionated by SDS ± PAGE, blotted
to PVDF membrane and probed with antisera speci®c for; (a)
ILK, ErbB4, PKB, phospho ser473 PKB, phospho thr308 PKB,
(b) GSK3 (a and b forms), phospho Ser21 GSK-3a, BAD or
phospho Ser136 BAD, (c) MAP kinase or phospho MAP kinase
as indicated. Where indicated, cells were treated with 100 nM
wortmannin for 30 min prior to cell lysis. Equal amounts of
protein were loaded per track
8025
Regulation of protein kinase B by ILK
DK Lynch et al
8026
Figure 2 E€ects of integrin-linked kinase expression on apoptosis induced by C2-ceramide in COS cells. COS cells were transiently
transfected with WT-ILK, DN-ILK or vector alone. Cells were serum starved overnight and treated with 50 mM C2-ceramide (Cera)
or C2-dihydroceramide (DHC) or ethanol vehicle for 8 h and the percentage of cells undergoing apoptosis determined by FACS
analysis (a) or TUNEL (b). Numbers in (a) and (b) represent the average of ®ve and three experiments respectively, each performed
in triplicate. Error bars represent standard deviation. (c) COS cells were serum starved overnight and treated with 50 mM C2ceramide (C2-ceramide) or C2-dihydroceramide (C2 DHC) or ethanol vehicle (mock). Alternatively COS cells were transfected with
WT-ILK or DN-ILK, serum starved overnight and treated with 50 mM C2-ceramide (WT-ILK+CER and DN-ILK+CER,
respectively). After the indicated times, cell survival was calculated by staining with trypan blue. A minimum of 600 cells were
counted per point, where possible. (d) COS cells were transiently transfected with WT-ILK, lanes 2, 5 and 8, DN-ILK, lanes 3, 6
and 9, or mock transfected, lanes 1, 4 and 7. Cells were serum starved overnight and treated with 50 mM C2-ceramide (Ceramide),
lanes 4, 5 and 6, or C2-dihydroceramide (DHC), lanes 7, 8 and 9, or ethanol vehicle, lanes 1, 2 and 3, for 8 h. RIPA extracts were
separated by SDS ± PAGE, transferred to PVDF membrane and blotted with antisera speci®c for PKB (top) and phospho-S473
PKB (bottom). Equal amounts of protein were loaded per track
and 2, and 7 and 8 in Figure 1a, S473 panel).
Expression of DN-ILK leads to a drastic diminution
of phosphorylation of S473 of PKB e.g. Figure 1a,
compare lanes 4 and 6. Western blotting with an ILKspeci®c polyclonal antiserum con®rms expression of
ILK and DN-ILK in transfected cells at levels
approximately ®vefold higher than that of endogenous
protein (Figure 1a, top). DN-ILK acts speci®cally to
reduce S473 phosphorylation of PKB and not by a
general non-speci®c inhibition of kinase function, since
it had no e€ect on the phosphorylation of MAP kinase
(MAPK) (Figure 1c). Control blots (Figure 1a, PKB
panel) show that changes in PKB expression levels do
not account for the observed changes in S473
phosphorylation. Treatment with the PI3-kinase
inhibitor wortmannin (100 nM) (Ui et al., 1995) totally
abolishes the ILK-dependent increase in S473 phosphorylation consistent with ILK acting via a PI3kinase dependent mechanism (Figure 1a, bottom).
Activation of PKB requires dual phosphorylation on
T308 and S473. Since T308 is constitutively phosphorylated in serum-starved COS cells at a basal level
that is una€ected by expression of ILK or DN-ILK
(Figure 1a), changes in S473 phosphorylation induced
by expression of ILK and ILK mutants should result
in parallel changes in PKB activity towards downstream substrates.
Therefore, phosphorylation on two PKB substrate
sites, serine 21 of glycogen synthase kinase 3a (GSK3a) and serine 136 of BAD, was assessed using
antibodies speci®c for phosphorylated forms of these
proteins. As in previous studies (Blume-Jensen et al.,
1998), co-transfection of a BAD expression plasmid
proved necessary to investigate phosphorylation of
S136 of BAD as endogenous BAD levels in COS cells
are very low. Figure 1b shows that expression of ILK
enhances phosphorylation of S21 of GSK-3a (for
example, Figure 1b, phospho-S21 GSK-3a panel,
compare lanes 1 and 2, 4 and 5, 7 and 8). A similar
e€ect can also be seen on the phosphorylation of S136
of BAD (Figure 1b, phospho-S136 BAD panel,
compare lanes 1 and 2, 4 and 5, and 7 and 8). For
S136 of BAD there is a high basal level of
phosphorylation that was consistently increased two-
Regulation of protein kinase B by ILK
DK Lynch et al
8027
Figure 3 (a) An alignment of the human, Caenorhabditis elegans and Drosophila melanogaster ILK protein sequences, and human
B-Raf (residues 445 ± 719) using Clusta1W (Thompson et al., 1994). #### indicates the position of the potential FSF
phosphorylation site sequence in the activation loop of ILK. The three conserved ankyrin repeats are indicated. A* marks the
position of the lysine residue changed to methionine to create DN-ILK. The DFG motif, only present in B-Raf, is underlined. (b)
Calculation of percentage identi®es between the three ILK homologous sequences using 1Align (Huang and Miller, 1991). (c) An
alignment of the predicted catalytic loop sequences of the three ILK sequences and B-Raf with the protein kinase consensus (Hanks
and Hunter, 1995). A* marks the position of the catalytic aspartate residue critical for kinase function
fold by expression of ILK (measured by densitometry ±
data not shown). S136 of BAD is phosphorylated by
other protein kinases and these are likely to contribute
to the phosphorylation on this site seen in this
experiment (Scheid and Duronio, 1998). Expression
of DN-ILK leads to signi®cant inhibition of phosphorylation of S21 of GSK-3a and S136 of BAD (Figure
1b, compare lanes 4 and 6, phospho-21GSK-3a and
phospho-S136 BAD panels, respectively). Wortmannin
treatment (100 nM) abolished both the S21 phosphorylation of GSK-3a and the increase in the S136
phosphorylation of BAD (Figure 1b).
ILK expression rescues C2-ceramide induced apoptosis
Having demonstrated the ability of ILK to mediate
phosphorylation of PKB and its substrates, we next
tested whether ILK expression modulates PKB
Regulation of protein kinase B by ILK
DK Lynch et al
8028
biological function. Treatment of COS cells with C2ceramide induces apoptosis and this can be inhibited
by the activation of PKB (Zhou et al., 1998). We
examined the e€ects of expressing ILK or DN-ILK on
the percentage of cells undergoing apoptosis in this
assay. C2-ceramide treatment of cells caused several
morphological changes indicative of apoptosis. These
included, loss of viability, cell shrinkage, decreased
adherence to tissue culture treated plastics, cell shape
changes, the increased appearance of round phase
bright cells and membrane blebbing (data not shown).
The proportion of apoptotic cells was quantitated by
two di€erent methods (1) FACS analysis of the sub-G1
peak following staining with propidium iodide
(indicative of DNA fragmentation); (2) TUNEL (a
measure of DNA fragmentation and strand breakage).
In addition, trypan blue exclusion from the nucleus (a
measure of the integrity of the nuclear membrane) was
used to measure cell viability. Figures 2a (propidium
iodide staining) and 2b (TUNEL) show that markedly
increased apoptosis occurs in COS cells treated with
C2-ceramide but not the biologically inactive analogue,
C2-dihydroceramide. Cell viability is also decreased by
C2-ceramide (Figure 2c). In all three assays, ILK
expression signi®cantly reduced the number of cells
undergoing apoptosis/loss of viability following treatment with C2-ceramide (Figure 2a, b and c). Expression
of DN-ILK has no e€ect. Morphologically, C2ceramide treated cells expressing exogenous ILK
resemble those treated with C2-dihydroceramide or
ethanol vehicle alone, whereas those expressing DNILK were rounded up, shrunken and loosely attached
to the dish (data not shown). The number of apoptotic
events recorded by each method of assessment varies
owing to di€erences in sensitivity of the detection
method and the di€ering end points i.e. the stage of
apoptosis measured by each technique. In each case,
however, expression of ILK results in a twofold
decrease in C2-ceramide induced-apoptosis compared
to control treated cells and cells expressing DN-ILK
(Figure 2a and b). The extent of rescue we observe with
ILK may be an underestimate of the real e€ect as our
transfection eciency of 50 ± 80% means that 80%
rescue rather than 100% is the maximal response we
could anticipate. Increased PKB activity also rescues
COS cells from C2-ceramide induced apoptosis (Zhou
et al., 1998). Since we have shown ILK regulates PKB
activity, we propose ILK regulates COS cell apoptosis
through a PKB-dependent pathway. Interestingly, ILK
induced S473 phosphorylation is reduced by C2ceramide treatment compared to treatment with the
inactive analogue C2-dihydroceramide (Figure 2d)
suggesting in COS cells C2-ceramide exerts its e€ects
by down-regulation of PKB activity.
ILK is conserved through evolution and lacks key
catalytic residues
Since ILK lacks both HRDLXXN (containing the
catalytic aspartate residue) and DFG (involved in the
coordination of a magnesium ion) motifs that are
considered to be critical for kinase function. (Hannigan
et al., 1996; Hanks and Hunter, 1995), we were
intrigued by the mechanism of ILK function. We
sequenced ILK analogues from C. elegans and D.
melanogaster, arguing that functionally important
residues should be conserved during evolution. Figure
3a shows an alignment of these three ILK sequences
together with the B-Raf sequence as a reference kinase.
The overall degree of sequence conservation (Figure
3b) suggests the predicted C. elegans and D.
melanogaster proteins are likely to have homologous
functions to human ILK. Near the N-terminus all three
ILK homologues possess three ankyrin repeats
(underlined in Figure 3a), followed by a linker region
that may bind lipids (Delcommenne et al., 1998) and
then a kinase homology domain. A fourth ankyrin
repeat has been noted in human ILK (Hannigan et al.,
1996) but the sequence identi®ed is divergent from the
consensus for the ankyrin repeat motif and it is not
conserved through evolution.
Comparison of the kinase domain of ILK to that of
other protein kinases e.g. B-Raf (Figure 3a) shows that
residues important in maintaining kinase structure are
conserved, but that residues involved in catalytic
function are more divergent (Hubbard et al., 1994;
Hanks and Hunter, 1995). In particular, ILK sequences
corresponding to the catalytic loop show little
similarity, implying they have not been constrained
by function during evolution. None of the ILK
sequences have a DFG motif or a conserved substitute
for the catalytic aspartate residue found in other
kinases. A phe-ser-phe (FSF) motif is conserved in all
three ILK analogues within a sequence corresponding
to the activation loop of protein kinases (Hubbard et
al., 1994; Hanks and Hunter, 1995 ± marked with ###
in Figure 3a. In many kinases including insulin
receptor, PKB and MAP kinase, residues in the
activation loop are subject to reversible phosphorylations which control kinase activity (Her et al., 1993;
Hubbard et al., 1994; Co€er et al., 1998). Substitution
of a negatively charged residue e.g. aspartate or
glutamate for the phosphorylated residue mimics the
change in charge that accompanies phosphorylation
and results in constitutive activation of the kinase. In
contrast, substitution of alanine prevents phosphorylation consequently inhibiting kinase activation. FSF is
similar to the consensus sequence de®ned for
phosphorylation by PDK2, phe-X-X-phe-ser-phe/tyr
(Pearson et al., 1995; Alessi et al., 1996). If ILK
possesses intrinsic PDK2 activity, this site may be
subject to autophosphorylation and possibly regulate
ILK activity. To investigate this we changed the serine
residue of the FSF motif of either wild-type ILK or
DN-ILK to alanine, aspartate or glutamate, and
assessed the e€ect on the phosphorylation of S473 of
PKB.
E€ects of changes in the FSF motif on PKB-S473
phosphorylation
Figure 4 shows that expression of either FDF or FEF
point-mutated ILK induce phosphorylation of S473 of
PKB. In contrast ILK-FAF is inactive. These data
support the idea that the charge on the serine residue
of the FSF motif is important in regulating ILK
function and are consistent with this being a site for
phosphorylation. Changes in expression levels of ILK
and PKB do not account for the large changes
observed in S473 phosphorylation (Figure 4, ILK and
PKB panels). DN-ILK-FAF was inactive in promoting
S473 phosphorylation. However, surprisingly, both
Regulation of protein kinase B by ILK
DK Lynch et al
DN-ILK-FDF and DN-ILK-FEF promote phosphorylation of S473 of PKB at a third of the wild type level,
even though from many previous studies the lysine to
methionine mutation is expected to eciently inhibit
protein-kinase activity (Figure 4, S473 panel) (Carrera
et al., 1993; Hanks and Hunter, 1995). None of the
mutations had a signi®cant e€ect on the phosphorylation state of T308 (Figure 4). We next examined the
ability of the PI3-kinase inhibitor wortmannin to
inhibit the phosphorylation of S473 of PKB induced
by expression of the various mutant ILK proteins.
Surprisingly, S473 phosphorylation induced by expression of ILK-FDF, ILK-FEF, DN-ILK-FDF and DNILK-FEF was partially resistant to wortmannin
treatment at concentrations that completely abolished
S473 phosphorylation promoted by wild type ILK
(Figure 4, bottom). In all cases phosphorylation on S21
of the PKB substrate GSK3-a was observed to change
in parallel with changes in phosphorylation of S473 of
PKB (data not shown). Thus by substituting a
negatively charged residue in place of the serine
residue of the FSF motif, we have changed the ILK
induced-phosphorylation of S473 of PKB from a PI3kinase dependent, to a partially PI3-kinase independent
mechanism. Furthermore, substitution of FEF or FDF
in the DN-ILK background restores function to the
otherwise defective DN-ILK. This ability to restore
activity to DN-ILK argues strongly against ILK being
PDK2.
Figure 4 E€ect of mutations in the FSF motif of integrin-linked
kinase on phosphorylation of protein kinase B in serum starved
transiently transfected COS cells. COS cells were transiently
transfected with wild-type ILK (lane 1) or the ILK mutant
plasmids, DN-ILK (lane 2), ILK-FDF (lane 3), ILK-FEF (lane
4), ILK-FAF (lane 5), DN-ILK-FDF (lane 6), DN-ILK-FEF
(lane 7), DN-ILK-FAF (lane 8), or left untransfected (lane 9).
RIPA bu€er extracts were fractionated by SDS ± PAGE, blotted
to PVDF membrane and probed with antisera speci®c for; PKB,
phospho-Ser 473 PKB, phospho thr-308 PKB or ILK. Where
indicated, cells were treated with 100 nM wortmannin for 30 min
prior to cell lysis. Equal amounts of protein were loaded per track
Discussion
Our results clearly demonstrate that ILK induces
phosphorylation of S473 of PKB in COS cells. PKB
S473 phosphorylation correlates with the phosphorylation of PKB substrates, GSK3a and BAD, and the
inhibition of apoptosis induced by treatment of COS
cells with C2-ceramide. Interestingly, the same apparent
degree of phosphorylation of S473 on PKB leads to
di€erent e€ects on substrate phosphorylation depending on whether ErbB4 is expressed or not. For
example, in Figure 1a and b, the extent of phosphorylation of S473 of PKB is similar in lanes 2 and 8, but
the phosphorylation on S21 of GSK-3a is reduced in
lanes 8 compared to lane 2. Similarly, co-expression of
DN-ILK and CYT-1 results in a large but partial loss
of S473 phosphorylation of PKB, yet a complete loss
of phosphorylation of S21 of GSK-3a (Figure 1a and
b, lane 6, phospho S473 PKB and phospho S21 GSK3a blots). Although the reason for this is unclear, other
kinases such as PKC can also phosphorylate S21 of
GSK-3a (Cook et al., 1996). Constitutive activation of
ErbB4 could result in down-regulation of the activity
of these kinases or up-regulation of an S21 phosphatase activity. Delcommenne et al., 1998 concluded it
most likely that ILK directly regulates S473 phosphorylation of PKB. However, our data suggest this is
improbable for three reasons. (1) We can detect no
signi®cant protein kinase activity in immuneprecipitates of ILK using a variety of di€erent substrates,
including PKB (data not shown). Others have also
failed to detect kinase activity in ILK immuneprecipitates (Balendran et al., 1999). (2) ILK lacks both a
DFG motif and the catalytic aspartate residue
considered essential for ecient catalytic activity
(Figure 3). In protein kinase A (PKA), mutation of
the catalytic aspartate residue to alanine results in a
signi®cant reduction of catalysis to 0.4% of wild type
activity (Gibbs and Zoller, 1991). Thus, although ILK
may possess a low intrinsic kinase activity that explains
the previously reported ability of ILK to phosphorylate
b1 integrin tail, PKB and GSK-3a in vitro (Hannigan
et al., 1996; Delcommenne et al., 1998), this kinase
activity is unlikely to be sucient to account for the
rapid phosphorylation of S473 of PKB that is noted in
vivo (Cohen et al., 1997; van Weeren et al., 1998). (3)
Substitution of an FSF motif in ILK by either FDF or
FEF restores activity to a kinase-dead ILK protein
(DN-ILK) arguing strongly against direct phosphorylation of PKB by ILK (Figure 4).
If ILK isn't responsible for the direct phosphorylation of S473 of PKB then this raises the questions of
how does ILK function and what is the identity of
PDK2? Two related models can be proposed for ILK
function. In both models, ILK initially binds ATP in
its ATP binding pocket and is then able to become
phosphorylated on the FSF motif. If this phosphorylation is mediated by a low intrinsic kinase activity this
then explains the ability of DN-ILK to act as a
dominant negative inhibitor, since autophosphorylation on the FSF motif will be blocked in this instance.
3-phosphoinositides regulate access of ATP into the
ILK ATP binding pocket in an analogous way to that
proposed for the regulation of ATP access into the
ATP binding site by phosphorylation in the atrial
natriuretic receptor (Foster and Garbers, 1998; Potter
8029
Regulation of protein kinase B by ILK
DK Lynch et al
8030
and Hunter, 1998). Indeed the glycine rich loop of the
ILK ATP binding pocket overlaps with sequences
predicted by Delcommenne et al., 1998 to be
responsible for the binding of 3-phosphoinosides by
ILK. Subsequently, ILK phosphorylated on the FSF
motif either (a) activates PDK2 activity or (b) inhibits
a S473 phosphatase activity. Both scenarios result in
increased phosphorylation of S473 of PKB. Although
this model predicts the ILK-FDF and ILK-FEF
mutants no-longer require 3-phosphoinositides for
function, wortmannin does partially inhibit the ability
of these mutants to induce phosphorylation on S473 of
PKB (Figure 4). We suspect this is because 3phosphoinositides are still required for proper membrane targeting and unfolding of PKB. Current
experiments are aimed at distinguishing between these
two possible methods of action of ILK. Model A
predicts that PDK2 activity lies in a wortmannin
insensitive kinase that associates with ILK phosphorylated on the FSF motif. However, this model is still
consistent with PDK2 activity being 3-phosphoinositide dependent in vivo, since there is a requirement for
3-phosphoinositides for the initial phosphorylation of
ILK on the FSF motif. In vitro, PDK1 gains PDK2
activity when complexed with PRK2 or c-terminal
peptides from PRK2 containing an FEF sequence
motif (Balendran et al., 1999). However, it is unlikely
that the phosphorylated FSF motif of ILK functions in
a similar manner, as, unlike the PRK2/PDK1 complex,
which retains kinase activity towards T308, we see no
increase in T308 phosphorylation when ILK is
expressed and phosphorylation on S473 of PKB is
increased. In addition, we are unable to detect PDK1
in ILK-containing immuneprecipitates (data not
shown). Model B, where ILK inhibits a S473phosphatase activity may be more likely given the
large increase in S473 phosphorylation promoted by
ILK expression. Whilst serum-starvation of COS cells
only decreases PKB activity twofold (Meier et al.,
1997), treatment with phosphatase inhibitors increases
activity several-fold (Andjelkovic et al., 1996; Meier et
al., 1997; Millward et al., 1999).
In this paper, we show that full activation of PKB in
a model COS cell system requires both the recruitment
and activation of PI3-kinase by ErbB4 and ILK. Since
ILK is regulated by integrins (Hannigan et al., 1996),
PKB activation is therefore dependent on integrin- and
growth factor-mediated signals. Dual regulation of
PKB activity in this way may be required to drive cellcycle progression, and/or cell survival, both biological
responses known to require joint input from growth
factor- and extracellular matrix-mediated signals. The
mechanism by which ILK regulates S473 phosphorylation on PKB is unclear, although our results strongly
suggest it is unlikely to be due to direct phosphorylation of PKB by ILK. Rather, our results suggest that
PDK2 activity residues in a distinct molecule and ILK
functions as an adaptor protein that either activates
PDK2 or inhibits a S473 phosphatase.
Materials and methods
Antibodies
ErbB4, Santa Cruz C-18; phosphoserine 473 Akt (PKB),
phosphothreonine 308 Akt (PKB), Akt (PKB), BAD,
phosphoserine 136 BAD (New England Biolabs); GSK-3 (a
and b), phosphoserine 21 GSK-3a (Upstate Biotechnologies
Inc.); phospho-Map kinase, Map kinase (Promega). ILK
antiserum was raised against a TrpE-ILK fusion protein
encompassing the N-terminal ankyrin repeat region of ILK
(residues 1 ± 171).
Clones
Expression constructs were made by subcloning cDNAs for
ErbB4 CYT-1 and CYT-2 and ILK from pBluescript into the
expression vector pMT2 (Sambrook et al., 1987) with
modi®ed cloning sites including NotI. DN-ILK was
constructed by PCR mutagenesis. Oligonucleotides including
restriction sites for BsrD1 5'-CACAGAATTCGCAATGACATCGTTGTGATGGTGCTGAAGGTTCGAG (5 prime)
and NdeI 5'-TAGAGGGATCCATATGGCATCCAG (3
prime) were used to amplify a 188 bp fragment containing
the desired lysine to methionine change (residue 220 ± in
bold). This was cloned into pBluescript as a EcoRI-BamHI
fragment and the change con®rmed by DNA sequencing. A
BsrDI-NdeI fragment (a silent G-A polymorphism creates a
NdeI site in our ILK cDNA) was re-cloned into pBluescriptILK and the ILK cDNA with the lysine to methionine
mutation subcloned into pMT2 as a NotI fragment.
Mutation of the FSF motif was performed by PCR
directed mutagenesis. PCR primers were constructed covering
and including the NdeI (5'-CTCTAAGCTTCACACACTGGATGCCATATGGAT) and BstXI sites. Primers covering
the BstXI site included the following amino-acid substitutions
FSF to FEF: (5'-TGTGTCTAGACCAGGACATTGGAATTCGAACTTGACATCAG); FSF to FDF (5'TGTGAATTCCAGGACATTGGAAATCGAACTTGACATCAG); FSF to FAF (5'-TGTGAATTCCAGGACATTGGAACGCGAACTTGACATCAG). Fragments were ampli®ed by PCR using Pfu DNA polymerase (Stratagene) and
subcloned into pMT2-ILK or pMT2 DN-ILK as NdeI-BstXI
fragments. Insert segments of all vectors were con®rmed by
DNA sequence analysis.
COS Transfections
COS1 cells were grown in Glasgow's MEM (Life Technologies) supplemented with 10% v/v newborn calf serum and
2 mM L-glutamine. Transfections were performed with
Fugene 6 (Boehringer) and cells left for 72 h before harvest.
For analysis of BAD, pEBG-mBAD (New England Biolabs),
encoding a 49KD murine-BAD fusion protein, was
transiently co-transformed with other plasmids. Cells were
serum starved for 16 h, washed twice in ice-cold phosphate
bu€ered saline (PBS) and lysed for 15 min on ice in
radioimmune precipitation assay (RIPA) bu€er 0.5% w/v
sodium deoxycholate, 150 mM sodium chloride, 1% v/v
NP40, 50 mM Tris-HCL pH 8.0, 0.1% w/v SDS, 10% v/v
glycerol, 5 mM EDTA, 1 mM sodium orthovanadate, 1 mM
PMSF, 10 mg ml71 aprotinin, 10 mg ml71 leupeptin.
Insoluble material was removed by centrifugation (12 000 g
for 10 min at 48C).
Western blots
Cells were lysed in 200 ml RIPA lysis bu€er and protein
content estimated using Bicinchoninic acid Protein Assay
Reagent (Pierce). 20 mg protein was boiled in SDS sample
bu€er and resolved on a 7.5% SDS-polyacrylamide gel.
Proteins were transferred to Immobilon PVDF membrane
(Millipore) by electro-blotting and probed with appropriate
antibodies at the recommended dilutions. Protein detection
was performed using ECL (Amersham Life Science) followed
by autoradiography.
Regulation of protein kinase B by ILK
DK Lynch et al
C2-ceramide-induced apoptosis
ILK sequences
This was performed as described in Zhou et al. (1998).
Partial C. elegans sequences with homology to human ILK
were identi®ed in EST databases (GenBank). PCR primers
were constructed and RT ± PCR carried out using mRNA
from whole embryos to obtain full-length cDNAs. Screening
of EST databases (GenBank) identi®ed EST clones covering
the N-terminus of ILK from D. melanogaster. Two of these
(LD02317 and LD08711: Berkeley Drosophila Genome
Project) were obtained from Research Genetics Inc and
sequenced. DNA sequence analysis was performed using
Perkin-Elmer ABI Prism 377 Sequencers and BigDye
Terminator Chemistry (Perkin-Elmer, Foster City, CA.,
USA).
Accession numbers for sequences are C. elegans
(AJ249344) and D. melanogaster (AJ249345).
Detection of apoptosis
TUNEL staining was performed using the Apoptosis
Detection System, Fluorescein (Promega) Flow-cytometry/
FACS analysis with propidium iodide. Cells (106) were
harvested by mild trypsinisation and pelleted (1500 r.p.m.,
5 min). Cell pellets were permeabilized with 1 ml 3.7%
paraformaldehyde (5 min 228C), pelleted, washed with PBS
and resuspended in 1 ml 0.1% Triton X-100, 1% BSA in
PBS, incubated for 5 min at 48C, repelleted (3000 r.p.m.,
5 min) and resuspended in 1 ml PBS containing 5 mg ml71
propidium iodide and 100 mg ml71 ribonuclease. Cells were
incubated for 1 h at 228C in the dark and cell ¯uorescence
measured at 620 nm (excitation 488 nm) using a FACscan
¯ow cytometer and CELLQuest acquisition and analysis
software (Becton Dickinson).
Trypan Blue exclusion assay. Cells were harvested by mild
trypsinization, pelleted (1500 r.p.m., 5 min) and resuspended
in 1 ml growth medium. 50 ml cells was added to 50 ml 0.4%
Trypan Blue in PBS with mixing. After 2 min, stained and
unstained cells were quantitated by counting with a
haemocytometer. A minimum of 600 cells were counted per
point.
Acknowledgements
We thank Simon Barry, Harren Jhoti, Rob Harris and
Peter Soden for helpful discussions and technical advice.
This work was supported by a MRC collaborative
studentship award between GlaxoWellcome and Cambridge University (DK Lynch) and a grant from the
Association for International Cancer Research (PAW
Edwards).
References
Andjelkovic M, Jakubowicz T, Cron P, Ming XF, Han JW
and Hemmings BA. (1996). Proc. Natl. Acad. Sci., 93,
5699 ± 5704.
Andjelkovic M, Alessi DR, Meier R, Fernandez A, Lamb
NJC, Frech M, Cron P, Cohen P, Lucocq JM and
Hemmings BA. (1997). J. Biol. Chem., 272, 31515 ± 31524.
Alessi DR, Andjelkovic M, Caudwell B, Cron P, Morrice N,
Cohen P and Hemmings BA. (1996). EMBO J., 15, 6541 ±
6551.
Alessi DR, James SR, Downes CP, Holmes AB, Ga€ney PR,
Reese CB and Cohen P. (1997a). Curr. Biol., 7, 261 ± 269.
Alessi DR, Deak M, Casamayoy A, Caudwell FB, Morrice
N, Norman DG, Ga€ney P, Reece CB, MacDougal CN,
Harbison D, Ashworth A and Bownes M. (1997b). Curr.
Biol., 7, 776 ± 789.
Balendran A, Casamayor C, Deak M, Paterson A, Ga€ney
P, Currie R, Downes CP and Alessi D. (1999). Curr. Biol.,
9, 393 ± 404.
Bellacosa A, Testa JR, Staal SP and Tsichlis PN. (1991).
Science, 254, 274 ± 277.
Bellacosa A, Chan TO, Ahmed NN, Datta K, Malstrom S,
Stokoe D, McCormick F, Feng J and Tsichlis P. (1998).
Oncogene, 17, 313 ± 325.
Blume-Jensen P, Janknecht R and Hunter T. (1998). Curr.
Biol., 8, 779 ± 782.
Brunet A, Bonni A, Zigmond M, Lin M, Juo P, Hu L,
Anderson M, Arden K, Blenis J and Greenberg M. (1999).
Cell, 96, 857 ± 868.
Cardone MH, Roy N, Stennicke HR, Salvesen GS, Franke
TF, Stanbridge E, Frisch S and Reed JC. (1998). Science,
282, 1318 ± 1321.
Carrera AC, Alexandrov K and Roberts TM. (1993). Proc.
Natl. Acad. Sci., 90, 442 ± 446.
Chao DT and Korsmeyer SJ. (1998). Annu. Rev. Immunol.,
16, 395 ± 419.
Co€er PJ and Woodgett JR. (1991). Eur. J. Biochem., 201,
475 ± 481.
Co€er PJ, Jin J and Woodgett JR. (1998). Biochem. J., 335,
1 ± 13.
Cohen P, Alessi DR and Cross DAE. (1997). FEBS Lett.,
410, 3 ± 10.
Cook D, Fry MJ, Hughes K, Sumathipala R, Woodgett JR
and Dale TC. (1996). EMBO J., 15, 4526 ± 4536.
Cross DAE, Alessi DR, Cohen P, Andjelkovich M and
Hemmings BA. (1995). Nature, 378, 785 ± 789.
Datta SR, Dudek H, Tao X, Masters S, Fu H, Gotoh Y and
Greenberg ME. (1997). Cell, 91, 231 ± 241.
Delcommenne M, Tan C, Gray V, Rue L, Woodgett J and
Dedhar S. (1998). Proc. Natl. Acad. Sci., 95, 11211 ±
11216.
Dimmeler S, Assmus B, Hermann C, Haedeler J and Zeier
AM. (1998). Circ. Res., 83, 334 ± 341.
Dudek H, Datta SR, Franke TF, Birnbaum MJ, Yao R,
Cooper GM, Segal RA, Kaplan DR and Greenberg ME.
(1997). Science, 275, 661 ± 665.
Elenius K, Choi CJ, Paul S, Santiestevan EE and Klagsbrun
M. (1999). Oncogene, 18, 2607 ± 2615.
Foster DC and Garbers DL. (1998). J. Biol. Chem., 273,
16311 ± 16318.
Franke TF, Kaplan DR, Cantley LC and Toker A. (1997).
Science, 275, 665 ± 668.
Frisch SM and Francis H. (1994). J. Cell. Biol., 124, 619 ±
626.
Frisch SM and Ruoslahti E. (1997). Curr. Opin. Cell Biol., 9,
701 ± 706.
Gibbs CS and Zoller MJ. (1991). J. Biol. Chem., 266, 8923 ±
8931.
Gingras AC, Kennedy SG, Oleary MA, Sonenberg N and
Hay N. (1998). Genes. Dev., 12, 502 ± 513.
Guilherme A and Czech MP. (1998). J. Biol. Chem., 273,
33119 ± 33122.
Hanks SK and Hunter T. (1995). FASEB J., 9, 576 ± 596.
Hannigan GE, Leung-Hagesteijn C, Fitz-Gibbon L, Coppolino MG, Radeva G, Filmus J, Bell JC and Dedhar S.
(1996). Nature, 379, 91 ± 96.
Her JH, Lakhani S, Zu K, Vila J, Dent P, Sturgill TW and
Weber MJ. (1993). Biochem. J., 296, 25 ± 31.
8031
Regulation of protein kinase B by ILK
DK Lynch et al
8032
Huang X and Miller W. (1991). Adv. Appl. Math., 12, 337 ±
357.
Hubbard SR, Wei L, Ellis L and Henderson WA. (1994).
Nature, 372, 746 ± 754.
Jones PJ, Jakubowicz T, Pitossi FJ, Maurer F and
Hemmings BA. (1991). Proc. Natl. Acad. Sci., 88, 4171 ±
4175.
Kau€man-Zeh A, Rodriguez-Viciana P, Ulrich E, Gilbert C,
Co€er P, Downward J and Evan G. (1997). Nature, 385,
544 ± 548.
Khwaja A, Rodriguez-Viciana P, Wennstrom S, Warne PH
and Downward J. (1997). EMBO J., 16, 2783 ± 2793.
King WG, Mattaliano MD, Chan TO, Tsichlis PN and
Brugge JS. (1997). Mol. Cell Biol., 17, 4406 ± 4418.
Klippel A, Kavanaugh WM, Pot D and Williams LT. (1997).
Mol. Cell Biol., 17, 338 ± 344.
Kitamura T, Ogawa W, Sakaue H, Hino Y, Kuroda S,
Takata M, Matsumoto M, Maeda T, Konishi H, Kikkawa
U and Kasuga M. (1998). Mol. Cell Biol., 18, 3708 ± 3717.
Kulik G, Klippel A and Weber MJ. (1997). Mol. Cell Biol.,
17, 1595 ± 1606.
Marte BM and Downward J. (1997). Trends Biochem. Sci.,
22, 355 ± 358.
Meier R, Alessi DR, Cron P, Andjelkovic M and Hemmings
BA. (1997). J. Biol. Chem., 272, 30491 ± 30497.
Millward TA, Zolnierowicz S and Hemmings BA. (1999).
Trends Biochem. Sci., 24, 186 ± 191.
Pearson RB, Dennis PB, Han JW, Williamson NA, Kozma
SC, Wettenhall REH and Thomas G. (1995). EMBO J.,
14, 5278 ± 5287.
Potter LR and Hunter T. (1998). Mol. Cell. Biol., 18, 2164 ±
2172.
Sambrook J, Fritsch E and Maniatis T. (1987). Molecular
Cloning: A laboratory Manual, 2nd edn. Cold Spring
Harbor Laboratory Press, 16.20 ± 16.22.
Sawyer C, Hiles I, Page M, Crompton M and Dean C. (1998).
Oncogene, 17, 919 ± 924.
Scheid MP and Duronio V. (1998). Proc. Natl. Acad. Sci., 95,
7439 ± 7444.
Stephens L, Anderson K, Stokoe D, Erdjument-Bromage H,
Painter GF, Holmes AB, Ga€ney PR, Reese CB,
McCormick F, Tempst P, Coadwell J and Hawkins PT.
(1998). Science, 279, 710 ± 714.
Thompson JD, Higgins DG and Gibson TJ. (1994). Nucleic
Acids Res., 22, 4673 ± 4680.
Ueki K, Yamamoto-honda R, Kaburagi Y, Yamauchi T,
Tobe K, Burgering BMT, Co€er PJ, Komuro I, Akanuma
Y, Yazaki Y and Kadowaki T. (1998). J. Biol. Chem., 273,
5315 ± 5322.
Ui M, Okada T and Hazeki O. (1995). Trends Biochem. Sci.,
20, 303 ± 307.
Van Weeren PC, Debruyn KMT, Devriessmits AMM,
Vanlint J and Burgering BMT. (1998). J. Biol. Chem.,
273, 13150 ± 13156.
Yano S, Tokumitsu H and Soderling TR. (1998). Nature,
396, 584 ± 587.
Zhou H, Summers SA, Birnbaum MJ and Pittman RN.
(1998). J. Biol. Chem., 273, 16568 ± 16575.