Chemistry & Biology
Derivatives as Modulators of Growth
Factor Signaling and Neurite Outgrowth
Vibor Laketa,1,3Sirus Zarbakhsh,1,3Eva Morbier,2Devaraj Subramanian,1Carlo Dinkel,1Justin Brumbaugh,1
Pascale Zimmermann,2Rainer Pepperkok,1and Carsten Schultz1,*
1Cell Biology and Cell Biophysics Unit, European Molecular Biology Laboratory, Meyerhofstraße 1, 69117 Heidelberg, Germany
2Department of Human Genetics, K.U. Leuven, B-3000 Leuven, Belgium
3These authors contributed equally to this work
Phosphoinositides are important signaling mole-
cules that govern a large number of cellular pro-
cesses such as proliferation, differentiation, mem-
brane remodeling, and survival. Here we introduce
a fully synthetic membrane-permeant derivative of
a novel, easily accessible, and very potent phospha-
phate [PtdIns(3,4,5,6)P4]. The membrane-permeant
downstream of phosphatidylinositol 3-kinase (PI3K),
activated protein kinase, and protein kinase C,
more potently than similar membrane-permeant
PtdIns(3,4,5)P3and PtdIns(3,4)P2derivatives in the
absence of receptor stimulation. In addition, we
demonstrate that treatment of PC12 cells with the
membrane-permeant PtdIns(3,4)P2, PtdIns(3,4,5)P3,
number of neurites per cell in the presence of NGF.
This work establishes membrane-permeant phos-
phoinositides as powerful tools to study PI3K
signaling and directly demonstrates that 3-phos-
phorylated phosphoinositides are instrumental for
Intracellular signaling mediated by growth factors activate
various phosphoinositide 3-kinase (PI3K) isoforms (Cantley,
2002), which phosphorylate phosphoinositides like phosphati-
dylinositol 4,5-bisphosphate [PtdIns(4,5)P2] at the 3-hydroxy
group of the inositol ring, giving rise to phospholipids such
as phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3; 2]
PtdIns(3,4,5)P3 and PtdIns(3,4)P2 are recognized as second
messengers that govern many downstream events by activating
target protein kinases such as PDK1, which subsequently phos-
phorylate various signaling proteins including p70S6K, protein
kinase B (also known as Akt), and protein kinase C (PKC)
(Figure 1A) (Cantley, 2002).
So far an effective approach to manipulate phosphoinositide
levels for investigating signaling is to overexpress or knock
down phosphoinositide-forming or metabolizing enzymes. How-
ever, these methods require hours or days to develop an effect
and often the response is wiped out by a compensation mecha-
nism. Therefore, there is a necessity for methods that alter phos-
phoinositide levels within minutes, resulting in a better control of
the experiment, a more immediate onset of the signal, and there-
fore more physiological mimicking of the growth factor signaling.
modulate the activity of engineered phosphoinositide metabo-
lizing enzymes with a ‘‘chemical dimerizer’’ such as rapamycin
(Suhet al.,2006). Another, more direct, possibility uses synthetic
membrane-permeant derivatives of these phospholipids (Dinkel
et al., 2001; Jiang et al., 1998). This involves masking the
charged phosphate groups by bioactivatable acetoxymethyl
esters (AM esters). The uncharged AM esters are sufficiently
stable outside cells to be applied in regular buffers. Once the
derivatives enter cells, enzymatic cleavage of the bioactivatable
esters leads to unmasking of phosphates and the reconstitution
of negative charges and biological function (Schultz, 2003). In
addition, butyrates are frequently used to mask the hydroxy
groups (Vajanaphanich et al., 1994). Their enzymatic hydrolysis
is slower than AM ester hydrolysis (Bartsch et al., 2003), thereby
preventing phosphates from scrambling during bioactivation.
Upon addition to cells, the lipophilic derivatives pass the plasma
membrane by diffusion (Figure 1B). Once reaching the cytosol,
endogenous unspecific carboxyhydrolases are expected to first
remove the AM esters and subsequently the butyrates, thereby
releasing the phosphoinositide of interest or an active metabo-
lite. We previously applied the AM ester technique to show that
a membrane-permeant derivative of PtdIns(3,4,5)P3 (5; Fig-
ure 1C), but not PtdIns(3,4)P2(7; Figure 1C), reduced chloride
secretion of epithelial cells (Dinkel et al., 2001). An alternative
to bioactivatable masking groups are cationic polyanions for
permitting phospholipid entrance to living cells (Ozaki et al.,
The total synthesis of membrane-permeant PtdIns(3,4,5)P3(5)
and PtdIns(3,4)P2(7) derivatives required an enormous synthetic
effort (Dinkel et al., 2001). A derivative with an additional
1190 Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved
phosphate in the 6 position would be much easier to prepare.
ships of phosphoinositides, mainly due to a lack of synthetic
chemistry efforts and solubility problems of the phospho-
inositides in physiologically relevant buffers. To facilitate the
preparation of membrane-permeant PtdIns(3,4,5)P3 (5) and
PtdIns(3,4)P2(7) derivatives and to study relationship between
phosphoinositide structure and their activity, we extended the
masking approach to a previously unknown phosphoinositide,
PtdIns(3,4,5,6)P4(3). We performed a comprehensive biochem-
ical and live cell analysis of membrane-permeant phosphoinosi-
tide derivative effects on intracellular signaling pathways
ant phosphoinositides are able to stimulate many aspects of
PI3K-dependent signaling. Remarkably, PtdIns(3,4,5,6)P4/AM
behaves like a much more potent activator of the PI3K pathway
than membrane-permeant PtdIns(3,4)P2 and PtdIns(3,4,5)P3
derivatives. In addition, we examined the role of membrane-per-
meant phosphoinositides in a more complex PI3K-dependent
process such as neurite outgrowth, where we show that addition
of membrane-permeant PI3K products induces an increase in
the number of neurites per cell, indicating their role in neurite
Synthesis of Membrane-Permeant PtdIns(3,4,5,6)P4
We prepared 2-O-butyryl-phosphatidylinositol 3,4,5,6-tetrakis-
phosphate nonakis(acetoxymethyl) ester (PtdIns(3,4,5,6)P4/AM,
PIP4/AM, 6), starting from commercially available enantio-
merically pure 1,2-O-cyclohexylidene-myo-inositol (8; Figure 2).
Phosphorylation was achieved using standard dibenzyl phos-
phoramidite chemistry in excellent yield. Deprotection of the
ketal under acidic conditions gave the tetrakisphosphate deriva-
tive 9. After regioselective butyrylation of the axial 2-OH group
over the 1-OH position (85:15) via formation of an intermediate
orthoester, 1,2-di-O-octanoyl-sn-glycerol 3-(benzyl N,N-diiso-
propyl)phosphoramidite (10) was used to introduce of the lipid
moiety. The fully protected PtdIns(3,4,5,6)P4derivative 11 was
quantitatively deprotected by catalytic hydrogenolysis. The
resulting free acid was alkylated to the final nonakis(acetoxy-
methyl) ester 6 by using 27 equivalents of acetoxymethyl
bromide and an equimolar amount of N,N-diisopropylethylamine
in acetonitrile. PIP4/AM (6) was purified to homogeneity by
preparative HPLC. This synthetic pathway involved only six
synthetic steps plus the formation of the lipid phosphoramidite
and provides the simplest access to a highly phosphorylated
membrane-permeant phosphoinositide derivative known so
far. The overall yield from 8 was 19%. In a similar way, a di-O-
myristylglyceryl derivative was prepared (data not shown).
Phosphoinositide Derivatives Activate Signaling
Networks Downstream of PI3K
First, we tested how all membrane-permeant phosphoinositide
derivatives affect different aspects of PI3K-dependent intracel-
lular signaling. Mitogen-activated protein kinases (MAPKs) pro-
vide a functional signaling ‘‘hub’’ where signaling inputs are
amplified and integrated (Cuevas et al., 2007) and thus MAPK
phosphorylation levels are a good marker of intracellular sig-
naling activity. Therefore, we first investigated p42/p44 MAPK
phosphorylation after treatment with different membrane-per-
meant phosphoinositide derivatives. All 3-phosphorylated phos-
phoinositides tested induced p42/p44 MAPK phosphorylation
(Figure 3A), while the corresponding membrane-permeant
PtdIns(4,5)P2derivative was inactive (see Figure S1A available
online). Remarkably, the membrane-permeant PtdIns(3,4,5)P3
(PIP3/AM, 5) and PtdIns(3,4)P2(PI(3,4)P2/AM) derivatives were
significantly less potent than the PtdIns(3,4,5,6)P4 derivative
(PIP4/AM, 6) (Figure 3A and see the quantification in Fig-
ure S1B). The induction of p42/p44 MAPK phosphorylation
was unchanged in the presence of PI3K inhibitor wortmannin
Figure 1. Growth Factor Signaling, Mem-
brane-Permeant Phosphoinositide Deriva-
tives, and Their Mode of Cell Delivery
(A) A simplified EGF/PI3K signaling pathway. PIP2,
PtdIns(4,5)P2; PIP3, PtdIns(3,4,5)P3.
(B)Mechanism ofmembrane-permeant phosphoi-
nositide derivative cell entry and activation. The
pholipid are masked with enzyme cleavable AM
ester groups (green) while hydroxy groups are
masked by butyrates (blue). The uncharged mole-
cule diffuses across the plasma membrane. Inside
the cell the protecting groups are cleaved by
endogenous enzymes and the active compound
(C) Phosphoinositides and their membrane-per-
meant derivatives. The naturally occurring phos-
phoinositides PtdIns(4,5)P2(1) and PtdIns(3,4,5)P3
(2), the artificial tetrakisphosphate PtdIns(3,4,
5,6)P4(3), and the respective membrane-perme-
ant derivatives 4 to 6 are shown. A similar deriva-
tive of PtdIns(3,4)P2 (7) is also shown. Please
note that all lipid moieties have unnatural octanoic
acid esters unless stated otherwise. Bt, butyrate;
Chemistry & Biology
Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved 1191
phorylated phosphoinositides lasted much longer than the one
caused by EGF. However in a wash-out experiment, the pulse
application of PIP4/AM shortened the duration of MAPK phos-
phorylation, leading to a transient signaling pattern typical for
EGF treatment (Figure S1C and see the quantification in Fig-
ure S1D). This is significant because it indicates that the cells
are capable of metabolizing PIP4/AM and deal with increased
Since p42/p44 MAPK phosphorylation levels are a measure of
signaling events several stepsdownstream fromPI3K, wetested
the phosphorylation status of Akt, which is a more direct down-
Similarly to p42/p44 MAPK phosphorylation experiments, PIP4/
AM induced Akt phosphorylation with remarkable potency
(Figure 3B). The highest Akt phosphorylation was induced by
PDGF (Figure 3B), which was used as a positive control in these
experiments. Similar phosphorylation dynamics were observed
for p70S6K (Figure S1E), another direct target of PI3K/PDK1
signaling (Figure 1A) (Cantley, 2002). This demonstrates that
membrane-permeant 3-phosphorylated phosphoinositides are
able to induce phosphorylation of proteins downstream of
PI3K/PDK1, with PIP4/AM being the most potent derivative.
Given the strong responses of the PIP4/AM derivative, we
wanted to investigate its effects on intracellular signaling in a
more global fashion. We measured the phosphorylation state
of a large number of cellular proteins using commercial Kinex
antibody microarray and Kinetworks multi-immunoblot phos-
phoprotein-screening platforms (KPSS). Lysates from non-
stimulated (control) and PIP4/AM stimulated HeLa cells were
subjected to the Kinex antibody microarray assay, a screen
that uses 650 different antibodies to monitor protein expression
and phosphorylation levels. Over 40 proteins were found to
significantly change phosphorylation levels after PIP4/AM stimu-
Akt, GSK3b, p70S6K, Rac1, Bad, IRS1, FAK, JNK, etc., as well
as mediators of MAPK signaling, such as ERK1/2, MEK1/2,
MAPKAP2, etc. To confirm these results, we used two different
Kinetworks multi-immunoblot assays (KPSS 10.1 and KPSS
11.0), a more accurate western blot-based assay, which uses
30–40 phosphospecific antibodies and 20 lane multichannel
blotters. Again, the results showed an increase in the phosphor-
ylation levels of several PI3K signaling mediators such as
p70S6K, GSK3b, RSK1/2, JNK, and MAPK signaling members
such as ERK1/2 and MEK1/2 (Figures S2A and S2B; lists of all
the phosphoproteins tested are shown in Tables S2 and S3).
Moreover the phosphorylation pattern induced byPIP4/AM stim-
ulation resembled that induced by EGF stimulation both qualita-
tively and quantitatively (Figures S2A and S2B).
To investigate other aspects of PI3K signaling, such as PKC
activation, we used KCP-1, a fluorescent sensor that monitors
atypical PKC activity in real time (Schleifenbaum et al., 2004),
simultaneously with Fura red, a calcium-sensitive sensor.
Addition of PIP4/AM led to an increase in PKC-dependent phos-
phorylation, but had no effect on intracellular calcium levels
(Figure 3E). The lack of calcium response suggested that a non-
classic, calcium-independent PKC was triggered by PIP4/AM,
which is exactly the type of PKCs that are known to be activated
downstream of PI3K (Toker, 2000). In addition, the PKC sensor
used here is predominantly responding to atypical PKC phos-
phorylation (Schleifenbaum et al., 2004). Interestingly, other
3-phosphorylated phosphoinositides failed to stimulate PKC
activity. Possibly, the more powerful trigger of PIP4/AM is
required to allow KCP-1 to respond. From the above data,
PIP4/AM appears to trigger a significant subset of growth factor
signaling without the need of growth factor receptor stimulation.
Phosphoinositide Derivatives Induce PH Domain
Translocation in Live Cells
Phosphoinositides achieve direct signaling effects through the
binding of their head groups to various signaling proteins pos-
sessing phosphoinositide binding domain (Di Paolo and De
Camilli, 2006). Interaction between 3-phosphorylated phosphoi-
nositides and signaling proteins is often achieved via pleckstrin
homology (PH) domains. This interaction regularly results in
protein translocation to the plasma membrane (Di Paolo and
De Camilli, 2006). We used the GFP-tagged PH domain of
Grp1 (Grp1-PH-GFP), a commonly used PtdIns(3,4,5)P3sensor
(Halet, 2005), and imaged its dynamics after stimulation with
PIP4/AM using live cell confocal microscopy. Stimulation of
NRK cells with PIP4/AM resulted in Grp1-PH-GFP domain trans-
location to the plasma membrane. Translocation started to be
noticeable 3 min after stimulation (Figures 3C and 3D and Movie
S1). The translocation was more complete and almost instanta-
neous when the cell were stimulated with PDGF (Figures 3C and
3D and Movie S2), which was used as a positive control. The
same was observed using other PIP3sensors such as GFP-
Inhibition of endogenous PI3K by wortmannin did not have an
effect on PIP4/AM-induced PH domain translocation while it
completely inhibited growth
(Figure S3). This is in agreement with the aforementioned protein
phosphorylation results and suggests that PIP4/AM is able to
translocate relevant PH domain-bearing proteins to the correct
Figure 2. Synthesis of the Membrane-Permeant Phosphoinositide
(A) (BnO)2PNiPr2, DCI, MeCN, 1d, then ?40?C, CH3CO3H, 84%.
(B) TFA/CHCl3, 78%.
(C) C3H7C(OMe)3, PPTS-resin, CH2Cl2, H2O, 80%.
(D) 10, DCI, DCM, then ?40?C, CH3CO3H, 80%.
(E) Pd/C, H2, 99%.
(F) AcOCH2Br, iPr2NEt, MeCN, 48%.
Chemistry & Biology
1192 Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved
membrane and activate signaling downstream of PI3K in the
presence of PI3K inhibitor. The delay of 3 min in PH domain
translocation between growth factor and PIP4/AM stimulations
(Figures 3C and 3D) probably reflects the time needed for the
enzymatic reaction required to convert PIP4/AM to a biologically
active compound. We were not able to detect significant PH
domain translocation after stimulation with PIP3/AM, which
might explain its lower potency.
we used YFP-tagged p42 MAPK and observed its translocation
from the cytosol to the nucleus. In HeLa cells, PIP4/AM led to
YFP-p42 MAPK translocation from the cytosol to the nucleus
within several minutes (Figure S4A and Movie S3). Similar to PH
started 5 min after the stimulation (Figure S4B). EGF stimulation,
used as a control, induced almost instantaneous YFP-p42
MAPK translocation (Figures S4A and S4B and Movie S4).
From the present data, it is difficult to determine whether the
butyrates of membrane-permeant phosphoinositide derivatives
Figure 3. Effects of Membrane-Permeant
Phosphoinositide Stimulation on Intracel-
lular Signaling Downstream of PI3K
(A) Time course of p42/44 MAPK phosphorylation
after stimulation of HeLa cells with 10 mM
(100 ng/ml). p42/44 MAPK phosphorylation on
residues T202/Y204 is shown in green and
a-tubulin control is shown in red. Duration of stim-
ulation in minutes is indicated on the top of the
(B) Time course of Akt phosphorylation after stim-
ulation of NRK cells with PIP3/AM (10 mM), PIP4/
AM (10 mM), and PDGF (100 ng/ml). Akt phosphor-
ylation on S473 is shown in green and a-tubulin
control is shown in red. The time course is shown
on top of the panel.
(C) Translocation of the PH-GFP domain of Grp1
(PIP3 sensor) to the plasma membrane after
PDGF (100 ng/ml) or PIP4/AM (10 mM) treatment
of NRK cells. Arrows point to areas where the
translocation is the most pronounced. Scale
bars, 15 mm.
(D) Quantification of PH-GFP Grp1 domain trans-
location. Shown is the loss of cytosolic fluores-
cence as the GFP fusion translocates to the
plasma membrane. Error bars represent the SD
of the mean of seven cells from three independent
(E) Multiparameter imaging of events downstream
from PI3K. PKC activity is monitored by the FRET-
measured by the indicator Fura Red (red) (as
described in Experimental Procedures).
need to be removed for exhibiting biolog-
ical activity in cells. We therefore incu-
bated the 2-O-butyrylated PIP4derivative
(BtPIP4) and the 2,6-di-O-butyrylated
PIP3 derivative (Bt2PIP3) with the two
most prominent metabolizing enzymes
for PIP3, PTEN and SHIP2. Both enzymes
suggesting that the butyrate is required to be removed for rapid
dephosphorylation after the washout. In addition, we investi-
gated the recognition of BtPIP4by the Grb-1 PH domain using
surface plasmon resonance experiments. As is shown in Fig-
ure S6, only weak binding of the Grp-1 PH domain to BtPIP4-
containing vesicles was observed. Again, this implies that the
butyrate needs to be removed for the substantial Grp-1 PH
domain translocation observed in NRK cells after incubation
with PIP4/AM (Figure 3C).
Regulation of Neurite Formation by Different
After investigating the effects of membrane-permeant phosphoi-
nositides 5, 6, and 7 in rapidly occurring PI3K-dependent
events such as protein phosphorylation and translocation, we
were interested in observing more complex, PI3K-regulated
physiological processes, such as neurite formation. By using
PI3K inhibitors and by overexpressing constitutively active
Chemistry & Biology
Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved 1193
PI3K, it has been very well established that neurite outgrowth is
a PI3K-dependent process (Kimura etal., 1994; Kobayashi etal.,
2005, 2007), a very early step during neurite outgrowth.
However, this has not been directly demonstrated in living cells.
To test this directly, we chose PC12 cells stably transfected with
human NGF receptor (PC12 6–15), a model cell line for studying
early events during neurite outgrowth including neurite initiation
(Laketa et al., 2007). This cell line shows rapid responses to NGF
and develops neurites on the time scale of hours rather than
days, as is the case for standard PC12 cells (Hempstead et al.,
1992). When these cells were incubated with 10 mM PIP4/AM
(6) in the presence of NGF, they produced dramatically more
neurites per cell compared to cells treated with NGF alone
(Figures 4A and 4B). While NGF-treated control cells produced
an average of two neurites per cell, cells treated with PIP4/AM
(6) produced mostly four or more neurites per cell (Figures 3A
and 3B). Ten micromoles of PIP3/AM and PI(3,4)P2/AM had no
effect on neurite numbers, again confirming their lower potency
(data not shown). However, when the cells were treated with five
times higher concentration of PIP3/AM and PI(3,4)P2/AM, we
observed similar effects on neurite number as with PIP4/AM
(Figures 4A and 4B). No effect on the number of neurites per cell
was observed when the cells were treated with membrane-
permeant PtdIns(4,5)P2(4) or membrane-permeant PtdIns(3)P
derivatives (Figures 4A and 4B), demonstrating specificity. This
suggests that the number of neurites is under the direct control
of 3-O-phosphorylated lipids as products of PI3K.
When cells were incubated with PIP4/AM in the absence of
NGF, a large number of short extensions resembling neurite initi-
ation events were formed within 24 hr (Figure 4C). To be sure
that this morphological observation resembles the beginning of
a differentiation process we used two additional biochemical
indicators. First, suppression of proliferation was measured by
the extent of BrdU incorporation into DNA of proliferating cells.
Both NGF and PIP4/AM almost completely abolished BrdU
incorporation, indicating suppression of PC12 cell proliferation
(Figure S7A). Second, we used a MAPK phosphorylation profile
as an indicator of differentiation. It was previously shown that
sustained MAPK phosphorylation leads to PC12 cell differentia-
tion, while transient MAPK phosphorylation results in cell prolif-
eration (Marshall, 1995). Both NGF and derivative PIP4/AM
induced a sustained MAPK phosphorylation over 2 hr, which is
typical for differentiation processes, while EGF stimulation pro-
duced only transient MAPK phosphorylation and failed to trigger
neurite initiation (Figure S7B). Taken together, both morpholog-
ical and biochemical observations indicate that cells undergo
a differentiation process in response to the PtdIns(3,4,5)P3-
mimicking derivatives such as PIP4/AM. However, the neurites
induced by PIP4/AM remained short even after 48 hr, suggesting
that longitudinal neurite growth depends on additional sig-
nals, not triggered by the 3-phosphorylated phosphoinositide
We have demonstrated, in a comprehensive set of biochemical
and live cell assays, that membrane-permeant phosphoinositi-
des are able to stimulate many aspects of PI3K-dependent
signaling. Remarkably, the ‘‘unnatural’’ PtdIns(3,4,5)P3deriva-
tive with a phosphate on 6-OH position, PIP4/AM (6), is a much
more powerful activator of PI3K signaling than the other 3-O
phosphorylated membrane-permeant compounds, also in the
presence of PI3K inhibitor. Although our results suggest that
endogenous PtdIns(3,4,5)P3levels are not altered, the precise
nature of the biologically active entity produced in cells and the
mechanism of action remains to be elucidated. As estimated
Figure 4. Effects of Membrane-Permeant
Phosphoinositides on NGF-Induced Neurite
Outgrowth in PC12 Cells
(A) Phenotypes of PC12 cells after 24 hr of stimu-
lation with NGF alone (control) and NGF together
with PI(3,4)P2/AM (50 mM), PI(3,4,5)P3/AM (50 mM),
PIP4/AM (10 mM), PI(3)P/AM (100 mM), and
PI(4,5)P2/AM (100 mM), respectively. Arrows point
to individual neurites. Note that PI(3,4)P2/AM,
PI(3,4,5)P3/AM, and PIP4/AM induced a dramatic
increase in the number of neurites per cell, albeit
with differences in the required doses, whereas
PI(3)P/AM and PI(4,5)P2/AM had no effect on neu-
rite number. Scale bar, 10 mm.
(B) Quantification of the effect of synthetic phos-
phoinositide derivatives on neurite numbers.
Shown are changes in the percentage of cells
with 1, 2, 3, and 4 or more neurites per cell in
response to treatment with the derivative indi-
cated. Error bars represent SD of the mean of at
least 300 cells from two or more independent
(C) PC12 cell phenotypes after stimulation with
serum, PIP4/AM (10 mM), or NGF (100 ng/ml).
Note the appearance of short neurite-like struc-
tures (arrows) on PIP4/AM-stimulated PC12 cells.
Scale bar, 10 mm.
Chemistry & Biology
1194 Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved
by reversed-phase chromatography the hydrophobicity of
all membrane-permeant derivatives seems to be very sim-
ilar. Therefore, large differences in cell entry between the
compounds are not expected. Very little is known about struc-
ture-activity relationships of phosphoinositides; however, one
ination of the PDK1 PH domain crystal structures. The structure
shows that the phosphoinositide-binding site of PDK1 is unusu-
ally spacious (Komander et al., 2004). In contrast to other
PtdIns(3,4,5)P3-binding PH domains such as those from Akt or
Grp1 where the 6-OH position is exposed to the solvent (Lietzke
et al., 2000), in the PDK1 PH domain additional binding sites are
present around the 6-OH position, which could interact with
another phosphate group. In fact, it was demonstrated that
PDK1 binds 6-OH phosphorylated inositol phosphates, such
as Ins(1,3,4,5,6)P5 and InsP6, with submicromolar affinity,
whereas binding to inositol phosphates without the phosphate
on the 6-OH position, such as Ins(1,4,5)P3, is virtually undetect-
able (Komander et al., 2004). It is therefore possible that PIP4/
AM-derived phosphoinositides, by having an additional phos-
phate on 6-OH position, may bind to PDK1 with higher affinity
than those from the similar compound PIP3/AM and conse-
quently induce the PI3K signaling pathways more effectively.
Alternatively, the higher activity could be assigned to a slower
metabolism of the biologically active product in cells by endog-
enous phosphatases and/or the lack of butyrate hydrolysis
required in the 6-OH position. Finally, we cannot entirely exclude
that PtdIns(3,4,5,6)P4occurs in living cells as a putative mes-
senger lipid and that its very low abundance has prevented
detection so far. Despite our current lack of understanding of
theprecisemodeofaction, PIP4/AMwilllikely becomeanattrac-
tive tool to study growth factor signaling in the future, especially
considering its comparably facile synthesis and potency. A tech-
nical problem to be considered in this respect is the low shelf-life
of about six months for this type of compound.
Application of PI(3,4)P2/AM, PIP3/AM, and PIP4/AM together
per cell, which is consistent with their role in neurite initiation dis-
cussed in the literature. Aoki et al. (2005, 2007), using FRET
based PtdIns(3,4,5)P3sensors, showed that local accumulation
brane and constructed a model for neurite initiation in silico
based on regulation of PIP3levels. Furthermore, Fivaz et al.
(2008) have shown that the local PI3K-HRas-positive feedback
loop is instrumental for breaking the initial neuronal symmetry
and for neurite initiation. These studies failed to directly test
the hypothesis that more PtdIns(3,4,5)P3will lead to more neu-
rites in vivo. Our results are thus complementing these studies
by showing that application of membrane-permeant PI3K prod-
ucts leads to a dramatic increase in the number of neurites per
cell. Also, we provide a direct demonstration of the involvement
of these phosphoinositides in neurite initiation.
Together with the biochemical and live cell data on signaling
network activation downstream of PI3K/PDK1, we show that
membrane-permeant phosphoinositide derivatives are useful
to studyboth immediate PI3K-dependent events suchasprotein
phosphorylation and translocation, as well as complex PI3K-
dependent cellular processes, which take many hours to
develop, such as neurite outgrowth.
In order to facilitate the investigation of PI3K signaling and
the function of the various phosphoinositides, methods
are needed that artificially, acutely, but nondisruptively,
increase the concentration of a specific phospholipid. This
work demonstrates that membrane-permeable phosphoi-
nositide derivatives are suitable for rapidly activating intra-
cellular signaling pathways downstream of PI3K. This
makes them important tools to investigate the contribution
sized a phosphoinositide analog with a phosphate on the
6-OH position, which showed remarkable potency in acti-
vating PI3K-dependent signaling. Although phosphoinositi-
des with a phosphate on this position have not been
detected in vivo, we cannot entirely exclude that PtdIns(3,4,
escaped detection due to its very low abundance. In addi-
tion, we show that membrane-permeant phosphoinositides
are able to modulate complex PI3K-dependent processes
such as neurite initiation and outgrowth. Previous work
has suggested that the PI3K product PtdIns(3,4,5)P3plays
a crucial role in neurite initiation but the hypothesis that
more PtdIns(3,4,5)P3will lead to more neurites in vivo was
never directly tested. The increase in the number of neurites
per cell after application of membrane-permeant PI3K prod-
ucts is complementing earlier neurite initiation studies from
other labs and provides a direct demonstration of the
involvement of these phosphoinositides in neurite initiation.
In the future, membrane-permeant derivatives that are
limited in their metabolism or photoactivatable varieties
might pave the way for even more exciting possibilities.
Cell Lines and Antibodies
PC12 6–15 cells (stably transfected with NGF receptor TrkA) were a gift from
D. Martin Zanca (University of Salamanca). NRK cells were a gift from J. Ellen-
berg (European Molecular Biology Laboratory). U2OS cells were a gift from
A. Dinarina (Spanish National Cancer Research Centre). HeLa CCL-2 cells
were purchased from LGC Promochem GmbH. Anti-phosphoERK1/2 (T202/
Y204), anti-phosphoAkt (S473), and anti-phosphop70S6K (T389) were
purchased from Cell Signaling Technology. Anti-a-tubulin antibody was
obtained from NeoMarkers. Secondary antibodies for western blots were
goat anti-rabbit IRDye 800CW (Rockland Immunologicals) and goat anti-
mouse Alexa-680 (Molecular Probes).
Imaging of PH Domain and MAPK Translocation
NRK or HeLa cells were grown on 35 mm glass-bottom dishes (MatTek). The
NRK were transfected with GFP-tagged PH domain of Grp1 while HeLa cells
were transfected with YFP-tagged ERK1 using Fugene6 (Roche) according
to the manufacturer’s protocol. After the 12–16 hr starvation in serum-free
DMEM, the cells were imaged using a Leica SP2 AOBS confocal microscope
with a 633 oil objective in PH domain translocation experiments or using a
Leica AF6000 LX widefield microscope with a 603 glycerol objective for
studying MAPK translocation.
Ca2+and PKC Activity Imaging
HeLa CCL-2 cells (LGC Promochem GmbH) were transfected with KCP-1
using Fugene6 (Roche). After 12 hr the cells were incubated for 30 min with
15 mM Fura Red/AM (Molecular Probes) and imaged using a Leica SP2 AOBS
confocal microscope with a 633 oil immersion objective (Leica). FRET
Chemistry & Biology
Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved 1195
measurements with KCP-1 were carried out as previously described (Schlei-
fenbaum et al., 2004).
Neurite Outgrowth Assay
Neuriteoutgrowth assaywasperformed aspreviouslydescribed (Laketa etal.,
2007; also available in the Supplemental Experimental Procedures).
Western Blot Assay
HeLa or NRK cells were grown in a 6 well dish in DMEM and 10% fetal calf
serum. After 12–16 hr starvation in serum-free DMEM, the cells were treated
with EGF (100 ng/ml) and PDGF (100 ng/ml) (Sigma-Aldrich) or with various
concentrations of membrane-permeant phosphoinositide derivatives 4–7.
Obtained lysates were analyzed using standard western blot protocols. Anti-
body signal detection was performed with an Odyssey Infrared Imaging
System (LI-Cor Biosystems).
Protein Phosphorylation Screens
Detergent-solubilized extracts from non-stimulated (control) and PIP4/AM
(10 mM)-stimulated HeLa cells were subjected to Kinex antibody microarray
screen. Also, extracts from non-stimulated (control), PIP4/AM (10 mM)-, and
EGF (100 ng/ml)-stimulated HeLa cells were subjected to KPSS 10.1 and
KPSS 11.0 as described on the Kinexus Bioinformatics Corp. website
(http://www.kinexus.ca). Kinex antibody microarray screen uses a panel of
650 antibodies to track the differential binding of dye-labeled proteins in
lysates prepared from stimulated versus non-stimulated cells. This screen
was performed twice. KPSS 10.1 and KPSS 11.0 screens use panels of
30–40 highly validated commercial phosphosite-specific antibodies and 20
lane multichannel blotters. The intensity of the ECL signals for the target
protein bands on the Kinetworks immunoblots were quantified with a FluorS
Max Multi-Imager and Quantity One software (Bio-Rad).
The extended description of the experimental procedures and chemical
materials is available in the Supplemental Experimental Procedures.
Supplemental Data include Supplemental Experimental Procedures, seven fig-
ures, three tables, and four movies and can be found with this article online at
WethankH.Stichnothfor cultured cells and J.Gross (University ofHeidelberg)
for high resolution massdetermination. Funding was provided by the VWfoun-
dation (I/81 597) and the Helmholtz Initiative for Systems Biology (SBCancer)
to C.S. R.P. is supported by a grant from the German Federal Ministry of
Education and Research within the framework of NGFN2 SMP Cell
Received: July 6, 2009
Revised: October 6, 2009
Accepted: October 7, 2009
Published: November 24, 2009
Aoki, K., Nakamura, T., Fujikawa, K., and Matsuda, M. (2005). Local phospha-
tidylinositol 3,4,5-trisphosphate accumulation recruits Vav2 and Vav3 to
activate Rac1/Cdc42 and initiate neurite outgrowth in nerve growth factor-
stimulated PC12 cells. Mol. Biol. Cell 16, 2207–2217.
tial role for the SHIP2-dependent negative feedback loop in neuritogenesis of
nerve growth factor-stimulated PC12 cells. J. Cell Biol. 177, 817–827.
Bartsch,M., Zorn-Kruppa, M.,Kuhl, N., Genieser, H.G., Schwede, F.,and Jas-
torff, B. (2003). Bioactivatable, membrane-permeant analogs of cyclic nucleo-
tides as biological tools for growth control of C6 glioma cells. Biol. Chem. 384,
Cantley, L.C. (2002). The phosphoinositide 3-kinase pathway. Science 296,
Cuevas, B.D., Abell,A.N., and Johnson, G.L.(2007).Roleof mitogen-activated
protein kinase kinase kinases in signal integration. Oncogene 26, 3159–3171.
Di Paolo, G., and De Camilli, P. (2006). Phosphoinositides in cell regulation and
membrane dynamics. Nature 443, 651–657.
Dinkel, C., Moody, M., Traynor-Kaplan, A., and Schultz, C. (2001). Membrane-
permeant 3-OH-phosphorylated phosphoinositide derivatives. Angew. Chem.
Int. Ed. Engl. 40, 3004–3008.
Fivaz, M., Bandara, S., Inoue, T., and Meyer, T. (2008). Robust neuronal
symmetry breaking by Ras-triggered local positive feedback. Curr. Biol. 18,
Halet, G. (2005). Imaging phosphoinositide dynamics using GFP-tagged
protein domains. Biol. Cell 97, 501–518.
Hempstead, B.L., Rabin, S.J., Kaplan, L., Reid, S., Parada, L.F., and Kaplan,
D.R. (1992). Overexpression of the trk tyrosine kinase rapidly accelerates
nerve growth factor-induced differentiation. Neuron 9, 883–896.
Jiang, T., Sweeney, G., Rudolf, M.T., Klip, A., Traynor-Kaplan, A., and Tsien,
R.Y. (1998). Membrane-permeant esters of phosphatidylinositol 3,4,5-tri-
sphosphate. J. Biol. Chem. 273, 11017–11024.
Kimura, K., Hattori, S., Kabuyama, Y., Shizawa, Y., Takayanagi, J., Nakamura,
S., Toki, S., Matsuda, Y., Onodera, K., and Fukui, Y. (1994). Neurite outgrowth
of PC12 cells is suppressed by wortmannin, a specific inhibitor of phosphati-
dylinositol 3-kinase. J. Biol. Chem. 269, 18961–18967.
S.,Ui, M., Iba, H., Konishi, H., et al. (1997). Expression of aconstitutively active
phosphatidylinositol 3-kinase induces process formation in rat PC12 cells.
Use of Cre/loxP recombination system. J. Biol. Chem. 272, 16089–16092.
Komander, D., Fairservice, A., Deak, M., Kular, G.S., Prescott, A.R., Peter
Downes, C., Safrany, S.T., Alessi, D.R., and van Aalten, D.M. (2004). Structural
insights into the regulation of PDK1 by phosphoinositides and inositol phos-
phates. EMBO J. 23, 3918–3928.
Laketa, V.,Simpson,J.C.,Bechtel,S.,Wiemann, S.,andPepperkok,R.(2007).
High-content microscopy identifies new neurite outgrowth regulators. Mol.
Biol. Cell 18, 242–252.
Lietzke, S.E., Bose, S., Cronin, T., Klarlund, J., Chawla, A., Czech, M.P., and
Lambright, D.G. (2000). Structural basis of 3-phosphoinositide recognition
by pleckstrin homology domains. Mol. Cell 6, 385–394.
Marshall, C.J. (1995). Specificity of receptor tyrosine kinase signaling: tran-
sient versus sustained extracellular signal-regulated kinase activation. Cell
Ozaki, S., DeWald, D.B., Shope, J.C., Chen, J., and Prestwich, G.D. (2000).
Intracellular delivery of phosphoinositides and inositol phosphates using poly-
amine carriers. Proc. Natl. Acad. Sci. USA 97, 11286–11291.
Schleifenbaum, A., Stier, G., Gasch, A., Sattler, M., and Schultz, C. (2004).
Genetically encoded FRET probe for PKC activity based on pleckstrin.
J. Am. Chem. Soc. 126, 11786–11787.
Schultz, C. (2003). Prodrugs of biologically active phosphate esters. Bioorg.
Med. Chem. 11, 885–898.
Suh, B.C., Inoue, T., Meyer, T., and Hille, B. (2006). Rapid chemically induced
changes of PtdIns(4,5)P2 gate KCNQ ion channels. Science 314, 1454–1457.
Toker, A. (2000). Protein kinases as mediators of phosphoinositide 3-kinase
signaling. Mol. Pharmacol. 57, 652–658.
Vajanaphanich, M., Schultz, C., Rudolf, M.T., Wasserman, M., Enyedi, P.,
Craxton, A., Shears, S.B., Tsien, R.Y., Barrett, K.E., and Traynor-Kaplan, A.
(1994). Long-term uncoupling of chloride secretion from intracellular calcium
levels by Ins(3,4,5,6)P4. Nature 371, 711–714.
Chemistry & Biology
1196 Chemistry & Biology 16, 1190–1196, November 25, 2009 ª2009 Elsevier Ltd All rights reserved