Highly specific, membrane-permeant peptide blockers
of cGMP-dependent protein kinase I? inhibit
NO-induced cerebral dilation
Wolfgang R. G. Dostmann*†, Mark S. Taylor*, Christian K. Nickl*, Joseph E. Brayden*, Ronald Frank‡,
and Werner J. Tegge‡
*Department of Pharmacology, Department of Molecular Physiology and Biophysics, University of Vermont, College of Medicine,
Burlington, VT 05405-0068; and‡AG Molecular Recognition, Gesellschaft fu ¨r Biotechnologische Forschung,
Mascheroder Weg 1, D-38124 Braunschweig, Germany
Communicated by Susan S. Taylor, University of California, San Diego, CA, October 3, 2000 (received for review September 5, 2000)
Arrays of octameric peptide libraries on cellulose paper were
screened by using32P-autophosphorylated cGMP-dependent pro-
tein kinase I? (cGPK) to identify peptide sequences with high
binding affinity for cGPK. Iterative deconvolution of every amino
acid position in the peptides identified the sequence LRK5H (W45)
as having the highest binding affinity. Binding of W45 to cGPK
resulted in selective inhibition of the kinase with Kivalues of 0.8
?M and 560 ?M for cGPK and cAMP-dependent protein kinase
(cAPK), respectively. Fusion of W45 to membrane translocation
signals from HIV-1 tat protein (YGRKKRRQRRRPP-LRK5H, DT-2) or
LRK5H, DT-3) proved to be an efficient method for intracellular
delivery of these highly charged peptides. Rapid translocation of
the peptides into intact cerebral arteries was demonstrated by
using fluorescein-labeled DT-2 and DT-3. The inhibitory potency of
the fusion peptides was even greater than that for W45, with Ki
values of 12.5 nM and 25 nM for DT-2 and DT-3, respectively. Both
peptides were still poor inhibitors of cAPK. Selective inhibition of
cGPK by DT-2 or DT-3 in the presence of cAPK was demonstrated
in vitro. In pressurized cerebral arteries, DT-2 and DT-3 substan-
tially decreased NO-induced dilation. This study provides func-
tional characterization of a class of selective cGPK inhibitor pep-
tides in vascular smooth muscle and reveals a central role for cGPK
in the modulation of vascular contractility.
protein kinase inhibitor ? combinatorial libraries ? SPOT method ?
membrane translocation signal ? smooth muscle
way, controlling a variety of cellular responses, ranging from
smooth muscle cell relaxation to neuronal synaptic plasticity (1,
2). The structural similarity of cGPK and its closest relative, the
cAMP-dependent protein kinase (cAPK), has made it difficult
to study cGPK pathways independent of those mediated by
cAPK, primarily because of the lack of potent and selective
cGPK inhibitors. Because recent studies have suggested that
cAMP and cGMP are each able to cross-activate either cGPK or
cAPK under physiological conditions, the specific role for cGPK
within the NO?cGMP-mediated signaling pathway remains ob-
scure (for a review see ref. 1). However, recent advances have
clearly identified specific intracellular targets for the cGPK
isozymes I? and I? (3, 4). Also, inactivation of the genes for
cGPK I??I? and cGPK II showed that the cGPK isozymes
regulate distinct cellular functions by pathways separate from
those mediated by cAPK (5, 6).
Attempts to identify cGPK-selective inhibitor peptides based
on the autoinhibitory domain of the enzyme or in vivo substrates
have been tedious at best, because of the lack of a well defined
surrounding the phosphate acceptor site has been established
(7). Various synthetic peptides have been used to analyze the
he cGMP-dependent protein kinases type I? and I? (cGPK)
act directly downstream in the NO-mediated signaling path-
sequence requirements for cGPK substrates (8–11). Recently,
we developed an iterative approach using phosphorylation of
peptide libraries on cellulose paper to determine a priori the
substrate specificity of cGPK versus cAPK. Consequently, we
identified the cGPK substrate sequence TQAKRKKSLAM-
FLR, in which the serine represents the phosphate-acceptor site
(12, 13). Substitution of this serine by alanine yielded cGPK
inhibitors with Ki values of 7.5–22 ?M (13) and improved
cGPK?cAPK selectivity, as has been reported with other syn-
thetic peptide derivatives (14, 15). However, all cGPK peptide
inhibitors known so far lack satisfactory potency and selectivity.
Here we report a peptide library screen specifically designed
First, we took advantage of the autophosphorylation properties
of cGPK, which provides the means to study the transient
enzyme–peptide interactions. Second, we used peptide libraries
that lack the phosphate acceptor residues serine and threonine
to select for peptide binding over phosphorylation. Linking the
best sequence from this screen to membrane translocation
signals (MTS) for intracellular delivery resulted in the highly
effective cGPK I? inhibitors DT-2 and DT-3. Finally, we have
demonstrated that both peptides are powerful tools for studying
the specific functional roles of cGPK in smooth muscle.
Enzyme Preparations. cAPK-C? was expressed and purified from
Escherichia coli (16), and cGPK type I? was expressed and
purified from SF9-insect cells as described in ref. 13.32P-labeling
of cGPK was accomplished by incubating the enzyme in buffer
A (50 mM Mops, pH 6.9?0.4 mM EDTA?1 mM Mg-
acetate?200 mM NaCl?1 mg/ml BSA?10 mM DTT) and
[?-32P]-ATP (0.1 mM; specific activity 1,600 cpm?pmol) for 2 h
at room temperature. Excess label was removed by using G-25
Sepharose chromatography, and SDS?PAGE and Western blot-
ting were performed (17).
Synthesis of Peptide Libraries and cGPK Screening. The peptide
arrays on paper (SPOTs) were generated as described in refs. 12,
13, and 18. Ser and Thr were omitted from the defined and the
randomized positions resulting in arrays of 18 ? 18 sublibrary
spots. The following arrays were generated and used successively
in the iterative approach: 1, XXX12XXX; 2, XXXRK12X;
3, XRKKK12X; 4, 1RKKKKK2; 5, LRKKKKKH12; 6,
Abbreviations: cAPK, cAMP-dependent protein kinase; cGPK, cGMP-dependent protein
kinase; MTS, membrane translocation signal; PKI, protein kinase inhibitor.
†To whom reprint requests should be addressed. E-mail: email@example.com.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
December 19, 2000 ?
vol. 97 ?
12LRKKKKKH (with ‘‘1’’ and ‘‘2’’ denoting defined amino acids
and ‘‘X’’ denoting the mixture of natural amino acids excluding
Ser and Thr). Sublibraries from each generation with the highest
binding affinity were assembled in 5 to 10 copies each, which
were then used to verify the previous results. The library
membranes were blocked with BSA (13). After addition of 1
nmol32P-labeled cGPK, the libraries were agitated for 12 h at
room temperature in buffer A and 0.1 mM ATP. The papers
were washed 5–10 times with buffer A and analyzed by using a
32P-imaging system (12, 13).
Synthesis of Soluble Peptides and Determination of Inhibition Con-
stants. Synthesis, purification, and characterization of soluble
peptides for kinetic determinations were carried out as described
in refs. 12 and 13. Cysteine-containing peptides were prepared
by using fluorenylmethoxycarbonyl-Cys(Trt)-OH. All peptides
were prepared as C-terminal amides. The concentrations of the
stock solutions were determined by quantitative amino acid
analyses. Fluorescein peptide labeling was carried out by incu-
bating 10 mg of peptide, containing an extra Cys followed by a
?-Ala at the N terminus, in 1 ml of 1 M phosphate buffer (pH
7.4) with 60 ?l of a 0.1 M stock solution of fluorescein-5-
maleimide (Molecular Probes) in DMSO at 4°C overnight in the
dark. The mono-labeled products were purified from excess
reagent, and multiple labeled compounds were separated by
preparative HPLC (acetonitrile?0.5% trifluoroacetic acid gra-
dients). Peptide analysis was carried out by matrix-assisted laser
desorption ionization-MS. Kinetic constants (Km, Vmax, Ki) were
determined by an adapted phospho-cellulose assay (19) as
cGPK Assays in Cultured Vascular Smooth Muscle Cells.Humanaortic
smooth muscle cells (HuASM; Clonetics, San Diego) were
cultured in smooth muscle cell basal medium (SmBM; Clonetics)
containing 0.5 ?g?ml hEGF, 2.5 ?g?ml insulin, 0.5 ng?ml
hFGF, 50 ?g?ml gentamicin, 50 ?g?ml amphotericin-B, 1%
FBS. Cells were used at passage 2–5. To increase endogenous
cGPK activity, HuASM cells (75% confluent) were transfected
by using 5 ?g pMT3-cGPK I? vector and 15 ?l of Fugene (Roche
Molecular Biochemicals) in 800 ?l of serum-free medium per
10-cm plate for 15–30 min. Cells were harvested after 48 h. For
each experiment, two dishes were incubated with 50 ?M W45,
DT-2, DT-3, DT-5, DT-6, or buffer for 60 min; the cells washed,
trypsinized, pelleted, and homogenized in cold lysis buffer (50
mM KPO4, pH 6.5?10 mM DTT?10 mM benzamidin?5 mM
EDTA?5 mM EGTA?5 mg/ml N?-(p-tosyl)lysine chloromethyl
ketone?10 mg/ml l-1-tosylamido-2-phenylethyl chloromethyl ke-
tone?17 mg/ml PMSF?2 mg/ml soybean trypsin inhibitor?25
?g/ml antipain). The homogenate was then centrifuged and
kinase activity in the supernatant was quickly determined.
Endogenous cAPK activity was suppressed by the addition of 70
nM protein kinase inhibitor (PKI)(15–24)in all assays and assay
cyclic nucleotide concentrations were 1 ?M.
Fluorescence Imaging of Vascular Smooth Muscle Cells in Intact
Arteries. Isolated intact cerebral arteries were incubated with
fluorescein-labeled peptides (2 ?M) in Hepes buffer (10 mM
binding sequences. A, B, C, and F show phosphorescence images of consecu-
tive generations of the library screens after binding of32P-labeled cGPK. Each
library membrane carries 18 ? 18 spots, which resulted from substitutions of
code. The best combinations are indicated with arrows. (D) Western blot of
material eluted from excised sublibrary spots (red numbers in C) by using a
cGPK I-specific antibody. (E) PhosphorImage of the blot shown in D.
Evolution of peptide libraries for the iterative screening of cGPK
Dostmann et al.
December 19, 2000 ?
vol. 97 ?
no. 26 ?
Hepes, pH 7.4?6 mM KCl?140 mM NaCl?2 mM CaCl2?1 mM
MgCl2?10 mM glucose) for 3–30 min at room temperature.
Arteries were washed 3 to 5 times and examined with a Bio-Rad
MRC 1024ES confocal scanning laser microscopy system. Flu-
orescein excitation was imaged with the 488-nm line of a
krypton–argon laser. All images were captured with a 100? oil
immersion objective lense (NA ? 1.3) mounted on an Olympus
BX50 (New Hyde Park, NJ) upright microscope. Optical thick-
ness was approximately 1.5 ?m. Single optical sections were
acquired with seven Kalman averages.
Diameter Measurements in Isolated Pressurized Cerebral Arteries.
Small posterior cerebral arteries (internal diameter ?150 ?m)
from euthanized rats were isolated in cold bicarbonate buffered
physiological salt solution (PSS) and denuded of endothelium by
placing an air bubble in the lumen for 2 min. Artery segments
were cannulated on glass pipettes, pressurized to 20 mmHg (1
mmHg ? 133 Pa) and superfused with warmed (37°C), gassed
(95% O2?5% CO2) PSS. Arterial diameter was measured by
using a video dimension analyzer as described in refs. 20 and 21.
The vessels were subsequently pressurized to 80 mmHg, which
induced sustained myogenic tone. Additional agents (peptides,
these experiments, endothelium removal was confirmed by the
lack of 10 ?M acetylcholine-mediated vasodilation, and maximal
arterial diameters were obtained via treatment with the Ca2?
channel blocker nisoldipine (10 ?M).
Affinity Selection of Inhibitors for cGPK from Peptide Library Arrays.
We constructed peptide libraries without Ser and Thr and
monitored the binding of32P-autophosphorylated cGPK I? as
described in Methods (Fig. 1). By using this approach, we took
advantage of the autophosphorylation properties of cGPK,
which does not alter the catalytic constant (kcat) of the enzyme
(22). The octameric library array XXX12XXX revealed strong
binding to the kinase with the amino acid combinations RR, KR,
and RK at positions 1 and 2. The RK motif gave the strongest
signal (Fig. 1A). For this library and all subsequent libraries, we
reexamined and verified the strongest binding motifs by synthe-
sizing 5–10 copies for each combination (data not shown). The
second-generation library with the structure XXXRK12X iden-
tified unambiguously the combination KK as the strongest
binding motif (Fig. 1B). In the third library XRKKK12X, again
KK was favored, although other lysine-containing combinations
(AK, WK, KF) were also selected by cGPK (Fig. 1C). To
demonstrate that the phosphorescence imaging signals we ob-
served were caused by
contamination, individual sublibrary spots (red numbers in Fig.
with SDS?PAGE sample buffer. Western blot analysis of this
material by using an anti-cGPK I??I? antibody (ref. 23; Fig. 1D)
confirmed the presence of the enzyme. In addition, phospho-
rescence imaging of the same blot identified the kinase as
32P-labeled (Fig. 1E). In the fourth library, 1RKKKKK2, cGPK
C- and N-terminal extended libraries identified more hydropho-
bic residues surrounding the cluster of positive residues (data not
shown), with the dodecamer FLLRKKKKKHHK as the longest
peptide included in our search.
32P-labeled cGPK and not
Kinetic Analysis of Synthetic Peptide Inhibitors and Fusion to Mem-
brane Translocation Sequences. Dixon plot analysis for cGPK
inhibition by the synthetic peptides was performed by using the
peptide TQARKKSLAMA as substrate (13). A summary of the
inhibition constants is shown in Table 1. The octamer LRK5H
(W45) derived from the fourth-generation library (Fig. 1E) was
a potent and selective competitive inhibitor of cGPK. Further-
more, W45 showed clearly enhanced inhibitory potency when
compared with WW21, a dodecameric peptide derived from
cGPK substrate libraries (13). N- and C-terminal extensions in
fifth- and sixth-generation libraries did not provide peptides with
improved Ki values (W74 and W75, Table 1). Ala scanning
optimal inhibition. However, the peptide LRKKKAKH (9W31)
showed the smallest reduction of inhibitory potency (Ki? 5.3
?M), suggesting that the fourth Lys in W45 occupies the
phosphorylation site of cGPK. To probe further the apparent
need of cGPK for lysines, we synthesized the peptide polylysine
(K9) and found that it still inhibits cGPK but with diminished
inhibitory potency (Ki? 10 ?M). Also, we tested polycations
such as spermine and spermidine and found that they did not
To allow internalization of the highly charged peptide W45
into live cells, we used two MTS sequences, one from the HIV-1
Tat protein (amino acids 47–59; refs. 24 and 25) and the other
from the Drosophila Antennapedia homeodomain (amino acids
43–58; ref. 26). N-terminal fusion of either MTS sequences to
W45 resulted in the competitive inhibitors DT-2 and DT-3. Both
peptides showed potentiated inhibitory potencies with Kivalues
40 to 80 fold lower than W45 (Table 1). Interestingly, the
fusion peptides alone had significant cGPK inhibitory activity
Table 1. Inhibition constants (Ki) of synthetic peptides for cAPK and cGPK
No.Sequence cGPK Ki, ?McAPK Ki, ?M cAPK?cGPK
0.82 ? 0.33 (10)
14.5 ? 4.9 (3)
11.3 ? 2.6 (3)
9.9 ? 2.6 (3)
5.3 ? 1.9 (3)
46.1 ? 5.5 (3)
1.2 ? 0.16 (6)
1.2 ? 0.38 (5)
0.025 ? 0.009 (8)
0.97 ? 0.24 (4)
0.0125 ? 0.003 (5)
1.1 ? 0.22 (3)
559 ? 108 (4)
493 ? 86 (3)
107 ? 32 (4)
16.5 ? 3.8 (4)
26 ? 4 (4)
1.0 ? 102
6.8 ? 102
1.97 ? 104
1.10 ? 102
1.32 ? 103
2.36 ? 101
The number of experiments for each Kiare given in brackets. ND, not determined.
*Taken from ref. 13.
www.pnas.org Dostmann et al.
Intracellular Delivery into Smooth Muscle Cells and Intact Arteries.
The internalization of DT-2 and DT-3 into living cells was
monitored by using fluorescein-labeled peptides (Fluo-DT-
2?3). Fig. 2 shows segments of a cerebral artery from rat, stained
for 30 min with Fluo-DT-2 (Fig. 2A), Fluo-DT-3 (Fig. 2B), or
Fluo-W45 (Fig. 2C). The labeled peptide analogs of DT-2 and
DT-3 were rapidly internalized and distributed through the
cytosol and nuclei in a time-dependent manner, whereas
Fluo-W45 did not noticeably penetrate through the plasma
Selective Inhibition of cGPK Over cAPK in Vitro.To demonstrate that
DT-2 and DT-3 are both capable of inhibiting cGPK under
conditions where the cAPK-selective inhibitor PKI(5–24)and?or
cAPK are present, we established an in vitro reconstitution assay
(Fig. 3). In this assay, we used purified recombinant cAPK and
cGPK at low concentrations (1 nM) and chose cyclic nucleotide
concentrations of 1 ?M, conditions under which cAMP will
activate only cAPK and cGMP will activate only cGPK. Also, we
selected concentrations for the inhibitors DT-2, DT-3, and PKI
that should selectively inhibit only cGPK or cAPK, respectively.
Fig. 3 shows that cGPK and cAPK are stimulated only by their
specific agonists (cGMP or cAMP) and inhibited only by their
specific inhibitors (DT-2?3 or PKI). In a mixture of both
enzymes, this result could be verified, which means that cGMP-
stimulated cGPK was inhibited only by DT-2?3 and cAMP
stimulated cAPK only by PKI. When both enzymes were acti-
vated with cGMP and cAMP in the mixture, a differential
response of approximately 50% inhibition was observed with
either DT-2?3 or PKI.
To establish the ability of DT-2 and DT-3 to inhibit cGPK in
intact cells, human aortic smooth muscle cells were incubated
with DT-2, DT-3, or control peptides W45, DT-5, and DT-6 for
60 min. Cells were then harvested, washed, and homogenized,
and a phosphoryltransferase assay was performed. Fig. 4 shows
that only preincubation with DT-3 or DT-2 caused inhibition of
cGMP-stimulated cGPK activity. Preincubation with W45 or the
control peptides DT-5 and DT-6 showed no significant effect,
and PKI(5–24). The assays contained 1 nM enzyme, 16 ?M substrate peptide
volume of 100 ?l as described (13).
Differential inhibition of recombinant cGPK and cAPK by DT-3, DT-2,
isolated from rat and incubated for 30 min with Fluo-DT-2 (A), Fluo-DT-3 (B),
or Fluo-W45 (C). Confocal images (100?) of tissue samples were taken after
washing arteries with physiological salt solution. Fluorescein excitation was
imaged at 488 nm. Images show the smooth muscle layer of the arteries
(optical thickness ? 1.5 ?m).
Translocation of fluorescein-labeled MTS fusion peptides (Fluo-DT-2
Dostmann et al.
December 19, 2000 ?
vol. 97 ?
no. 26 ?
uptake into the cells leading to inhibition of endogenous cGPK
activity. To observe the effects shown in Fig. 4, we had to use
high concentrations (50 ?M) of peptides, because the cytosolic
fractions were diluted 500-fold in the enzyme assay buffer. In
addition, endogenous cAPK activity was inhibited by the addi-
tion of 70 nM PKI(5–24).
Functional Antagonism of NO-Mediated Vasodilation. To evaluate
the physiological effects of DT-2 and DT-3 as selective cGPK
inhibitors in smooth muscle further, we studied their effects on
NO-induced vasodilation in intact cerebral arteries. The NO
donor NONOate elicited concentration-dependent vasodilation.
Pretreatment of arteries with the cGPK inhibitors DT-2 or DT-3
for 20 min substantially impaired NO-mediated vasodilation
(Fig. 5). DT-2 (1 ?M) significantly (P ? 0.05) increased the EC50
for NONOate (47 ? 9 ?M, n ? 6) compared with untreated
vessels (0.43 ? 0.15 ?M, n ? 11) or vessels treated with the HIV
tat carrier sequence DT-6 alone (0.92 ? 0.67 ?M, n ? 5).
Similarly, DT-3 (0.1 ?M) significantly (P ? 0.05) increased the
NONOate EC50(34 ? 18 ?M, n ? 6) compared with untreated
vessels or vessels treated with the Antennapedia carrier se-
quence DT-5 alone (0.13 ? 0.09 ?M, n ? 5). NO responses in
arteries treated with the carrier sequences (DT-6 and DT-5)
were not different from those in untreated arteries. In all cases,
the fusion peptides or carrier peptides alone caused slight to
moderate increases (?10%) in vascular tone. Table 2 summa-
rizes the effects of peptide treatment on NO-induced vasodila-
tion. Collectively, these data indicate that the combination
peptides DT-2 and DT-3 translocate into smooth muscle cells in
intact arteries and effectively inhibit NO-induced vasodilation,
presumably by inhibiting cGPK.
The SPOT method has been used successfully to identify sub-
strate sequence motifs for a number of protein kinases (27, 28).
In this study, we have expanded the practical utility of the SPOT
method and demonstrated that peptide libraries specifically
are potent inhibitors of cGPK. Furthermore, the peptide with
the highest inhibitory potency (W45) showed unprecedented
I? was inhibited by W45 in much the same way as the I? isoform,
whereas cGPK II was relatively insensitive (?100 fold) to these
inhibitors (data not shown). These results clearly demonstrate
the power of the strategies and approaches we have used to
develop novel cGPK inhibitors.
muscle cells by internalization of exogenous MTS-fusion peptides DT-2 and
DT-3. Cells were preincubated with either W45, DT-5, DT-3, DT-6, or DT-2, and
activity was blocked with PKI(5–24).*and ? indicate significant differences
(P ? 0.05, ANOVA followed by Bonferroni post hoc test) from the untreated
and W45 treated control groups, respectively.
Inhibition of endogenous cGPK activity in human aortic smooth
DT-2 and DT-3. Pressurized segments of rat posterior cerebral artery were
dilated with the NO donor NONOate (1 nM to 1 mM, indicated by arrows) in
the presence or absence of the fusion peptides DT-2 and DT-3 for 25 min. (A)
Continuous diameter tracings are shown for untreated, DT-2 treated, and
DT-3 treated vessels. (B) NONOate dose–response curves in untreated arteries
(n ? 11) or arteries exposed to DT-2 (n ? 6) or DT-3 (n ? 6).
Inhibition of NO-mediated vasodilation of intact cerebral arteries by
www.pnas.orgDostmann et al.