Selective in vivo inhibition of mitogen-activated protein kinase activation using cell-permeable peptides.

The extracellular signal-regulated kinase (ERK), a member of the mitogen-activated protein kinases (MAPKs), is essential for cellular proliferation and differentiation, and thus there exists great interest to develop specific and selective inhibitors of this enzyme. Whereas small molecule inhibitors PD098095 and U0126 have been used to study MAPK/ERK kinase (MEK), their target selectivity has been questioned recently. The cross-reactivity of ATP-directed inhibitors with other protein kinases prompted us to develop structure-based selective peptide inhibitors of ERK activation. Based on a MEK1-derived peptide, we developed inhibitors of ERK activation in vitro and in vivo. The inclusion of either an alkyl moiety or a membrane-translocating peptide sequence facilitated the cellular uptake of the peptide inhibitor and prevented ERK activation in 4-phorbol 12-myristate 13-acetate-stimulated NIH 3T3 cells or nerve growth factor-treated PC12 cells in a concentration-dependent manner. In addition, cell-permeable peptides inhibited ERK-mediated activation of the transcriptional activity of ELK1. The peptides did not have an inhibitory effect on the activity of two other closely related classes of MAPKs, c-Jun amino-terminal kinase or p38 protein kinase. Thus, these peptides may serve as valuable tools for investigating ERK activation and for selective investigation of ERK-mediated responses. With the knowledge of other kinase interacting domains, it would be possible to design cell-permeable inhibitors for investigating diverse cellular signaling mechanisms and for possible therapeutic applications.

Protein phosphorylation plays a critical role in cellular signaling in response to a variety of hormones, growth factors, neurotransmitters, and a wide range of stimuli. Mitogen-activated protein kinases (MAPKs) 1 play a pivotal role in these processes, particularly in stimulus-mediated cellular responses (1)(2)(3). The activation of these enzymes requires a cascade-like mechanism in which each MAPK is phosphorylated on two amino acid residues (Thr/Tyr) by an upstream protein kinase, MAPKK (MEK), and the latter in turn is phosphorylated on two amino acid residues (Ser/Thr) by a third protein kinase, MAPKK kinase (MEKK). There are at least three such protein kinase modules in mammalian cells as follows: extracellular signal-regulated kinases (ERKs), c-Jun amino-terminal kinases (JNKs), and the p38 MAP kinases (p38). The dual phosphorylation of MAPKs by MEKs is necessary for their activation (4) and is considered an essential step in the signaling pathways in response to growth factors and mitogenic stimuli, stress-causing agents, and cytokines.
For phosphorylation-dependent activation of MAPKs to occur, MAPK must first associate with its cognate upstream kinase, MEK. Thus, disrupting this interaction using a peptide derived from an association domain of either enzyme (5,6) would be predicted to block the activation of the downstream protein kinase. To test this hypothesis, we evaluated the ability of a peptide corresponding to the amino-terminal 13 amino acids of MEK1, which are intimately involved in the association of ERK with MEK (7,8), to inhibit selectively our target enzyme, ERK, and inhibit ERK activation in cultured mammalian cells.
For a peptide to inhibit ERK activation within cultured mammalian cells, the peptide must have access to ERK within the cells. To allow for efficient entry of peptide into cells in culture, we modified the peptide with membrane-translocating moieties. The first modification was the alkylation (myristoylation or stearation) of the inhibitor peptide in order to increase its hydrophobicity and hence its cellular uptake (9). The second modification was to link a membrane-translocating peptide (MTP) to facilitate the cellular delivery of the peptide. Several MTPs, capable of transporting peptides or even large proteins, have been described recently (10). These include peptides derived from the Drosophila melanogaster antennapedia (Antp) homeotic transcription factor (11), the human immunodeficiency virus-TAT (TAT) protein (12), the h region of the signal sequence of Kaposi fibroblast growth factor (MTS) (13), and the protein PreS2 of hepatitis B virus (HBV) (14).
Here we report that although the free peptide inhibited ERK activation by active MEK in vitro, it did not inhibit ERK activation in vivo presumably because of its inability to cross cellular membranes. Alkylation of the peptide inhibitor facilitated its ability to inhibit ERK activation in vivo with NGF or PMA as activating ligands. Furthermore, linking the peptide inhibitor to a membrane-translocating peptide was also effective in facilitating cellular uptake of the peptide inhibitor and inhibiting ERK activation in vivo.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-All peptides listed in Table I were synthesized using commercially available reagents and high pressure liquid chromatography purified to Ͼ90% purity by SynPep Corp. (Dublin, CA). The mass of each peptide was confirmed by electrospray mass spectrometry. Peptide 1 has the sequence of the first 13 amino acids of human MEK1, listed in Table I. Peptides 2 and 3 are synthesized with stearic acid and myristic acid, respectively, linked directly to the amino terminus of the peptide by an amide bond. Peptides 4 -6 are the combination of membrane-translocating sequences of Antp, TAT, and MTS, respectively, at the amino terminus of the sequence of peptide 1 with an intervening glycine amino acid. Peptides 7 and 8 are the combination of the membrane-translocating sequence HBV at the amino terminus and carboxyl terminus, respectively, of the sequence of peptide 1 with an intervening glycine amino acid. Peptides 1, 4, and 5-8 were synthesized with a fluorescein amide linked to aminohexanoic acid, which was in turn linked by an amide bond to the amino terminus of the peptides. Peptides 1, 4, and 5 were also synthesized without amino-terminal modification. Complete sequences of these peptides are listed in Table I.
Peptides Binding to ERK2-Fluorescence anisotropy experiments were carried out using a Beacon 2000 fluorescence polarization instrument from Panvera (Madison, WI) equipped with a 100-l sample chamber and an excitation and emission filter set optimal for fluorescein. Concentrated ERK2 was exchanged to a buffer containing HEPES, pH 7.5 (20 mM), glycerol (10%) and MgCl 2 (10 mM) by extensive dialysis. Peptide 7 was added (final concentration 5 nM) to ERK2 and incubated overnight at 4°C. An appropriate blank was made from buffer-exchanged ERK2 without peptide added. The fluorescence anisotropy of the peptide-containing sample was measured with the values of the blank subtracted. The ERK2 concentration was reduced by dilution with buffer containing peptide 7 (5 nM), and the blank was correspondingly diluted with buffer alone. The value of anisotropy of the diluted sample was measured with the values of the new blank subtracted. This dilution and measurement process was repeated until the value of anisotropy of the sample was the same as that of a free peptide solution. After all anisotropy readings were taken, the protein concentration of aliquots of the first three dilutions was determined by a modified micro-Lowry method from Sigma so that the concentration of all ERK2 samples could be calculated.
Coimmunoprecipitation of ERK2 with peptide 4 was carried out in a 200-l volume of buffer containing HEPES, pH 7.5 (50 mM), NaCl (100 mM), EDTA (1 mM), Triton X-100 (1%), leupeptin (10 g/ml), aprotinin (10 g/ml), phenylmethylsulfonyl fluoride (1 mM), fluorescein-labeled peptide 4 (50 M), and ERK2 (50 g/ml). As a control for nonspecific immunoprecipitation, unlabeled peptide 4 was used in another precipitation reaction. To these solutions, 5 l of a 1 mg/ml solution of anti-fluorescein antibody (Chemicon International; Temecula, CA) was added. These peptides, ERK2, and antibody mixtures were incubated overnight at 4°C with constant gentle shaking. Ultralink-immobilized protein G resin from Pierce was added (10 l of a 50% slurry in water) and incubated for an additional 2 h at 4°C before centrifugation to separate resin from supernatant. Resin was washed three times with 500 l of reaction buffer. Protein was removed from the resin by the addition of 200 l of 5-fold SDS-PAGE loading buffer. A sample of this loading buffer was diluted 5-fold and heated to 95°C for 10 min prior to analysis by Western blot.
Microscopy-Cells were incubated for 30 min with peptide (20 M), fixed, permeabilized, probed with anti-pan-ERK or anti-active ERK antibodies, and visualized with Cy-3-labeled secondary antibody. Cells were placed in mounting media containing 4,6-diamidino-2-phenylindole for epifluorescence microscopy or Toto-3 from Molecular Probes for confocal microscopy. Fluorescence of Cy-3 was not detected in any cells with the use of only secondary antibody (for details, see Supplemental Material).
Flow Cytometry-NIH 3T3 cells were incubated with peptide (20 M) for 30 min at 37°C and were fixed and stained with propidium iodide before passage through a 40-m filter. Fluorescence of fluoresceinlinked peptide and propidium iodide bound DNA within each cell was recorded by flow cytometry using 488 nm laser excitation detecting both fluorescein at 515 nm and propidium iodide at 617 nm (for details, see Supplemental Material).
Quantitation of Peptide Uptake-PC12 cells incubated with peptide 4 or 5 (25,50, and 100 M) or with media for 30 min at 37°C, treated with trypsin (0.05%) to degrade peptides loosely bound to the cell surface, removed with a cell scraper, and transferred to centrifuge tubes. Cells were pelleted by centrifugation, suspended in lysis buffer, and subjected to freeze-thaw. Lysis material was transferred to 96-well opaque white microtiter plates, and fluorescein fluorescence of each sample was determined using a CytoFluor II fluorescence plate reader from Perspective Biosystems (Farmingham, MA). The total protein concentration of each cell lysate was determined by a modified micro-Lowery method. The amount of peptide taken up into a given mass of cellular protein was calculated from these data and a fluorescein standard curve.
In Cell Potency of Peptide Inhibitors-Serum-starved NIH 3T3 cells were treated with peptide at indicated concentrations or the MEK inhibitor U0126 (20 M) for 30 min at 37°C and then stimulated with PMA (600 M) for an additional 30 min. Serum-starved PC12 cells were incubated with peptide at indicated concentrations for 15 or 30 min at 37°C and then treated an additional 5 min with NGF (50 ng/ml). Cells were washed with ice-cold phosphate-buffered saline, suspended in 0.5 ml of ice-cold lysis buffer, and scraped from flasks. Cell lysate was sonicated, centrifuged to remove remaining insoluble material, and flash-frozen on dry ice. Total protein concentration in the cell lysate was determined by a modified micro-Lowry method. The volume of lysate used in Western blot analysis was adjusted if there existed significant disparity in total protein concentrations between lysate samples. This adjustment was not frequently necessary and did not correspond to the peptide concentration added to the cells. The extent of ERK activation was assessed by Western blot analysis using anti-active ERK. The total ERK content was assessed by Western blot analysis using anti-pan-ERK antibody.
MTS-linked Peptide Effect on ERK Activation of the Transcription Factor ELK1-PathDetect HeLa luciferase reporter (HLR-ELK1) cells were plated at a density of 10 5 per well in 12-well plates in 1 ml of DMEM containing 10% FBS, 250 g/ml G418, and 100 g/ml hygromycin and incubated at 37°C overnight. Cells were then serum-starved by changing the medium to DMEM containing 0.5% FBS for 20 h and then treated with four different conditions as follows. Under the first condition, cells were kept in the same medium DMEM, 0.5% FBS, 0.2% Me 2 SO for 2 h to determine basal luciferase activity. Under the second condition, cells were kept in DMEM containing 100 nM PMA, 100 ng/ml EGF, and 10% FBS for 2 h to achieve maximal activation of ELK1 and of luciferase activity. Under the third condition, cells were treated with DMEM containing 0.5% FBS and 20 M U0126 for 30 min, and the medium was decanted and replaced with DMEM containing 100 nM PMA, 100 ng/ml EGF, and 10% FBS for 2 h. Under the fourth condition, cells were treated with DMEM containing 0.5% FBS and Peptide 5 (100 M) for 30 min, and the medium was decanted and replaced with DMEM containing 100 nM PMA, 100 ng/ml EGF, and 10% FBS for 2 h. Stock solutions of PMA and U0126 were made in Me 2 SO and the final concentration of Me 2 SO in the media was 0.2%, a concentration that was added in the control treatment. Media were removed from all wells; 200 l of luciferase lysing buffer was added, and plates were gently agitated for 5 min at room temperature. Lysates were assayed for the luciferase activity using 10 l of lysate and 100 l of luciferase assay buffer and measured for 15 s on a luminometer.

RESULTS
Peptide Binding to ERK2-We have demonstrated that the peptides derived from the amino terminus of MEK1 are capable of binding to ERK2 in vitro by two approaches, fluorescence anisotropy and coprecipitation. First, the fluorescence anisotropy of fluorescein-labeled peptide 7, a measure of the rotational freedom of the peptide, increases with increasing concentrations of ERK2 (Fig. 1A). This anisotropy change is consistent with the formation of a slower rotating complex of peptide 7 with ERK2 in solution. The value of the equilibrium dissociation constant (K d ) was determined by nonlinear regression analysis of the data using Equation 1 for a single-site, saturation binding event, where A min is the anisotropy of the free peptide and ⌬A is the difference in anisotropy of fully bound peptide and free peptide. By using Equation 1, we obtained a value of K d of 77 nM (see "Discussion"). Second, fluorescein-labeled peptide 5 bound to and coprecipitated ERK2 upon addition of anti-fluorescein antibodies and protein G-agarose (Fig. 1B, lane 2). The binding is specific because peptide 5 that was not labeled with fluorescein did not coprecipitate ERK2 in the presence of the antibodies and protein G-agarose (Fig. 1B, lane 3).
In Vitro Inhibition of ERK2 Activation-The in vitro inhibitory potency of each peptide was determined by monitoring ERK2 activation using antibodies specific for the activated (dually phosphorylated) form of ERK. As shown in Fig. 2A, the addition of NGF stimulated ERK activation without alteration in the total ERK protein (lane 2 versus lane 1). It is also evident that peptide 2 inhibited the activation of ERK in a concentrationdependent manner without noticeable change in the amount of total ERK (compare lanes 7, 6, 5, 4, and 3 with lane 2). The bands in Fig. 2A were quantitated and plotted against the peptide concentration added to the medium. The value of IC 50 for each peptide was determined from these data by non-linear least squares regression analysis using Equation 2 for a singlesite binding inhibitor, where I band is the value of intensity of each band; I Inhibited is the value of band intensity for inactive ERK (background); ⌬I is the difference of the values of the band intensity for fully activated ERK and inactive ERK, and IC 50 is the peptide concentration required to decrease band intensity by 50%. An example using peptide 2 is shown in Fig. 2B, where an IC 50 value of 2.5 M was obtained. The values of IC 50 for the other peptides we tested are listed in Table I. The lack of in vitro potency of these same peptides for inhibition of JNK and p38 activation was demonstrated in an activation assay analogous to the one performed for ERK2 but using the appropriate MAPKK and anti-active antibodies for the activated forms of JNK and p38 protein kinases. Neither the stearated peptide 2 ( show the peptide appears to have a localization pattern that is similar to that of ERK but different from that of active ERK. The peptide was also absent from the nucleus. The cellular uptake of the peptides was also independent of the DNA content indicating that cellular uptake was independent of phases of cell cycle (see Supplemental Material Fig. S1C and S1D). Whereas cells treated with peptides 4 and 5 were 50 -100-fold more fluorescent than untreated cells, no increase in relative fluorescence of cells treated with peptides 1 and 6 -8 was observed indicating lack of cellular uptake of the latter peptides.
The extent of cellular uptake of peptide 4 by PC12 corresponded to the concentration of the peptide in the medium and reached saturation at 50 -100 M (see Supplemental Material Fig. S2A) which represents about 2-4 pmol of peptide per g of cellular proteins. Similar results were obtained with peptide 5. The binding was specific because pretreatment of the cells with non-fluorescent peptides 5 before the fluorescently labeled peptide 5 hindered the cellular uptake of the latter (see Supplemental Material Fig. S2A and S2B). In Vivo Inhibition of ERK2 Activation-To demonstrate the potency of the peptides in inhibiting ERK activation in vivo, cells were treated with or without peptides prior to stimulation with the appropriate ligand, NGF for PC12 cells and PMA for NIH 3T3 cells. Cells were then lysed, and the lysates were applied to SDS-PAGE followed by Western blotting to determine the extent of ERK activation. As shown in Fig. 3A, ERK activation was robust when NGF was added for 5 min compared with serum-starved PC12 (lanes 2 and 1). The addition of the stearated peptide inhibitor peptide 2 attenuated the activation of ERK in a concentration-dependent manner without altering the total amount of ERK (lanes 7, 6, 5, 4, and 3  compared with lane 2). The bands corresponding to the various peptide concentrations were quantitated and plotted as shown in Fig. 3B, and a value of IC 50 was calculated in the same manner as the in vitro analysis using Equation 2 (Table I). It is noteworthy that the addition of stearated Ht 31-derived peptide, which is shown to inhibit selectively cAMP-dependent protein kinase anchoring in cells (9), had no effect on the activation of ERK in vivo (results not shown). Thus the inhibition of ERK activation by peptide 2 and 3 is specific, and stearation of the peptide had no effect by itself on ERK activation but facilitated the cellular uptake of the MEK-derived peptide, resulting in inhibition of ERK activation. Similar studies on the effect of the MTP-linked peptide inhibitor (peptide 4) in PMA-treated NIH 3T3 cells were carried out (Fig. 3C). As

TABLE I Amino acid sequences and IC 50 values of peptides tested for their inhibition of ERK activation in vitro and in vivo
The ability of the peptides to enter cells was evaluated both by fluorescence microscopy (M) and fluorescence-activated flow cytometry (F.C.) using peptides labeled at the amino terminus with fluorescein via an aminohexanoic acid linker. Peptides 2 and 3 are the stearated (Steϳ) and myristoylated (Myrϳ) forms of the MEK-1 derived peptide inhibitor 1. Peptides 4 and 5 are the MEK1 peptide conjugated to the Antp and TAT MTP, respectively, and both peptides were made with and without fluorescein at the amino terminus. The in vitro inhibitory potency of the peptides was measured based on their ability to inhibit MEK1-mediated phosphorylation of ERK2. The in vivo inhibitory potency was determined in both PMA-treated NIH 3T3 cells and NGF-stimulated PC12 cells. ND, not determined.   Fig. 3C). To determine the time required for the peptide to manifest its inhibitory effect in vivo, we treated cells with peptide for varying times before the addition of NGF to PC12 cells. (Fig. 3, D and E). As shown in Fig. 3D, peptide 4 (100 M) inhibited the activation of ERK in NGF-stimulated cells to a greater extent when preincubated with the cells for 20 min or longer prior to the addition of NGF. The quantitated band intensities of each incubation time were fitted by non-linear least squares regression analysis to Equation 3 for single exponential decay (Fig. 3E), where I band is the value of band intensities of the measured bands; I active is the value of the fully active ERK; k is the value of the rate at which ERK activation is inhibited by the peptides; and t is the incubation time of cells with the inhibitor peptide. By using Equation 3, we obtained values of k for peptides 4 and 5 of 0.036 and 0.052 min Ϫ1 , respectively. Therefore, the time required to achieve half-inhibition (t1 ⁄2 ) for peptides 4 and 5 are 19 and 13 min, respectively.

MTS-linked Peptide Effect on Activation of Transcription
Factor ELK1-Because we successfully demonstrated the inhibition of MEK phosphorylation of ERK by the cell-permeable peptides, we hypothesized those downstream effectors of ERK such as transcription factor ELK1 would therefore be affected by these inhibitors. To test this hypothesis, we used the Path-Detect HeLa luciferase reporter (HLR-ELK1) cell line. This stably transfected cell line features constitutive expression of the GAL4-ELK1 fusion protein and the firefly luciferase gene that is controlled by a promoter that responds to GAL4 fusion. When activated by phosphorylation, the fusion protein binds to the promoter and induces luciferase expression. Therefore, luciferase activity reflects the activation status of the ERK signaling pathway, and thus the effect of extracellular stimuli such as PMA, EGF, serum, etc. that converge on ERK activa- tion can be studied. Studies demonstrating the successful application of this approach to study the effect of ERK activation on ELK1 phosphorylation and ELK1 transcriptional activity were reported earlier (15).
Serum-starved cells were treated with U0126, peptide 5, or mock-treated for 30 min before they were stimulated with a combination of PMA, EGF, and FBS. Basal level of luciferase activity was determined by keeping cells in a serum-starving medium. Media were removed, and cells were lysed and tested for luciferase activity. As shown in Fig. 4, the addition of extracellular stimuli resulted in 15-fold increase in luciferase activity, whereas the pretreatment with U0126 or peptide 5 for 30 min prior to stimulation abrogated this increase resulting in only 2.2-and 1.5-fold increase in luciferase activity over the basal control. Peptide 5 was as effective at 5 h of induction with EGF, PMA, and FBS after peptide treatment indicating the stability of peptide for at least that period of time (data not shown). Thus peptide 5 was an effective inhibitor of ERK activation and subsequent phosphorylation and activation of the ELK1 transcriptional activity. DISCUSSION Because MAPKs play a pivotal role in cell signaling and are implicated in a variety of cellular functions, a need for selective and specific inhibitors is warranted to dissect the complex nature of the MAPKs components in the cell. Because most of the available protein kinase inhibitors are directed toward the ATP-binding site, the selectivity of many of such inhibitors has been questioned. The recent advances made in understanding protein-protein interactions and the discovery of protein motifs that are selective in their binding to their targets have spurred interest in the cell signaling field. The use of such specific recognition motifs may be useful in developing specific inhibitors for protein kinases. Furthermore, if such inhibitors can be delivered to their cellular targets in a controllable and direct manner, it will facilitate their use in cell signaling research. We have shown that a peptide derived from the amino terminus of MEK1 (peptide 1) inhibits the in vitro activation of ERK2 by MEK1 with an IC 50 value of 30 M (Table I). Equilibriumbinding assay using fluorescence anisotropy and a fluoresceinlabeled peptide demonstrated direct binding of the peptide to ERK2 with high affinity (77 nM). Because the buffer conditions of this binding assay were dissimilar from buffer conditions used for the inhibition of ERK activation by MEK in vitro and in vivo, there was a much tighter binding affinity determined biophysically (K d ) than IC 50 value determined from kinetic inhibition assays. Because determination of the value of K d is dependent on anisotropy values measured at saturating protein concentrations, which we could not achieve in our protocol for the enzyme, this value of K d should be regarded as an estimation of the in vitro binding affinity. To demonstrate further the interaction of ERK2 with these peptides, we show that these peptides were able to coimmunoprecipitate ERK2 with fluorescein-labeled peptide using an anti-fluorescein antibody.
Although peptide 1 inhibited ERK2 activation by MEK1, this peptide was not capable of inhibiting ERK activation in cultured mammalian cells (Table I), presumably because of its inability to cross the cellular membrane. Fluorescein-labeled peptide 1 was not detected (by either microscopy or flow cytometry) in cells treated with 100 M labeled peptide 1 (Table I). To investigate further the effect of this peptide on ERK activation in vivo, it was necessary to modify the peptide to facilitate its cellular uptake.
Alkylated Peptides-Conjugation of the peptide to stearic acid did not adversely affect the inhibitory potency of the peptide in vitro ( Fig. 2A). The stearated peptide 2 inhibited ERK2 phosphorylation in a concentration-dependent manner with an IC 50 value of 2.5 M (Table I and Fig. 2B), which is equally effective or more effective than the MEK1 inhibitors PD098059 or U0126. Similar results were obtained when the peptide was myristoylated (data not shown). The alkylated forms of the peptide were also selective, as no inhibition of p38 and JNK was observed, even at 100 M of peptide 2 (Fig. 2C).
Alkylated MEK-derived peptides 2 and 3 entered cells and potently inhibited the activation of ERK1 and ERK2 with IC 50 value of 13 M; the non-alkylated peptide 1 had no effect on the activation of ERK1 and ERK2 in NGF-stimulated PC12 cells (Table I). When the alkylated peptides were incubated for periods of 15 and 30 min before the addition of NGF, both the stearated peptide 2 and myristoylated peptide 3 inhibited the activation of ERK1 and ERK2; however, the stearated peptide was more effective than the myristoylated peptide at shorter duration (15 min) (data not shown). It is noteworthy that stearation of unrelated peptide such as a cAMP-dependent protein kinase anchoring protein-derived peptide (9) did not inhibit ERK activation in vivo.
MTP-linked Peptides-Because MTPs have been successfully used as carrier of other cargo peptides (10 -14, 16 -18), it was of interest to compare the effectiveness of alkylation as a mean to deliver peptide inhibitors to cellular targets with the effectiveness of ligation to MTPs. Toward this goal, we synthesized peptides containing four different MTPs combined with the sequence of peptide 1, and we tested these peptides for their ability to enter cells as well as for their inhibitory potency in vitro and in vivo. All MTP-linked peptides were inhibitory in vitro with IC 50 values between 0.21 and 30 M, except peptide 8 (Table I), indicating that the fusion of the carrier peptides did not adversely affect its potency to inhibit ERK activation by MEK. The latter peptide (peptide 8) consists of the HBV-derived MTP peptide linked to the carboxyl terminus of the inhibitory peptide and had an IC 50 value of over 100 M. Interestingly, when this HBV-derived MTP is linked to the amino terminus of the inhibitory peptide, the resulting peptide 7 is a 4-fold more effective inhibitor (7 M) than the free peptide 1, suggesting that the orientation of the HBV-derived carrier peptide in relation to the peptide inhibitor can alter its potency. Peptide 7 is a potent inhibitor of ERK activation in vitro but does not inhibit JNK or p38 activation (Fig. 2D), indicating its specificity for ERK.
Subcellular Localization of MTP-linked Inhibitory Peptide-When peptides derived from Antp or TAT were used as carriers of the inhibitory peptide, the resulting fusion peptides (4 and 5, respectively) were effective at penetrating intact cells and acting as inhibitors of ERK activation. Cellular uptake of the peptides was first assessed by epifluorescence and confocal microscopy (see Supplemental Material Fig. S1A). Both the PC12 and NIH 3T3 cells took up the fluorescein-labeled peptides 4 and 5 from the media and remained fluorescent for up to 30 min after washing away excess peptide with the culture medium (data not shown).
For the inhibitory peptide to act in vivo, the peptide must colocalize with ERK in the cytosol. Thus, we compared the cellular distribution of ERK to that of peptides 4 and 5 using confocal microscopy. Peptide 4 was internalized by the cell and was present in the cytosolic compartment and was not confined to the surface of the plasma membrane (see Supplemental Material Fig. S1A). We also showed that fluorescein fluorescence of labeled peptide was in the same focal plane as fluorescence associated with antibodies reactive with total ERK, and the distribution of ERK within the cells was similar to that of peptides. However, the activated ERK localizes to the nucleus more than does peptide 4 (see Supplemental Material To confirm the relative abilities of peptides to enter cells, we analyzed peptide-treated cells with fluorescence-activated flow cytometry (see Supplemental Material Fig. S1C and S1D). Fluorescently labeled peptides 4 or 5 were effectively taken up by NIH 3T3 cells, and cells become highly fluorescent (100-fold more fluorescent than untreated cells) when incubated with peptides 4 and 5. Because the cells were permeabilized after fixation prior to observation, this increase in fluorescence could not be due to free fluorescein trapped within the cells. Peptides 4 and 5 were retained within the cells during the wash steps to remove excess peptide from the cellular media (ϳ20 min). The other MTP-conjugated peptides 6 -8 were not taken up by cells as judged by flow cytometry. Thus, even though peptides 6 and 7 were effective in inhibiting ERK2 activation by MEK1 in vitro, they were not useful inhibitors in intact cells due to their inability to enter cells (Table I). We also show that the ability of peptides 4 and 5 to penetrate cells was not dependent on the cell cycle because all cells, regardless of DNA content, were able to take up peptides 4 and 5.
Two lines of evidence indicate that cellular uptake of peptides is saturable. First, the uptake of the Antp-conjugated peptide 4 by PC12 cells saturates at an extracellular concentration of 100 M, and at saturating concentrations, peptide 4 entered cells to about 4 pmol per 1 g of cellular protein. This value corresponds to a 400 M internal concentration assuming 110 pg of protein per cell and a volume of 1.4 pl per cell. The 4-fold enhancement over the external peptide concentration of 100 M is similar to the level of peptide enrichment (ϳ5-fold) reported earlier for other cell-permeable peptides (17,18). Second, when NIH 3T3 cells were pretreated with unlabeled peptide 5, followed by the addition of the fluorescein-labeled peptide 5, the labeled peptide was taken up significantly less by cells. Because the labeled peptide was shown to enter the cells at this concentration (1.0 M), it is apparent that the peptide entered the cells in a selective and saturable manner. Dowdy and co-workers (18) have observed a free diffusion of human immunodeficiency virus-TAT peptide into and out of cells with-out apparent saturation. We suspect that the observed saturation phenomenon and retention within cells of our peptide is a result of peptide binding to ERK within cells. Because ERK concentration is less than that of the peptide retained within cells, it is possible that the peptide may bind to other cellular proteins.
Inhibition of ERK Activation by MTP-linked Peptides in Vivo-To demonstrate that peptides 2, 4, and 5 inhibit ERK activation in both NIH 3T3 and PC12 cell lines, cells were treated with a range of peptide concentrations prior to stimulation. Peptide 2 inhibited the activation of ERK in NGF-stimulated PC12 cells in a concentration-dependent manner with an IC 50 value of 13 M, with no effect on the total amount of ERK. Similarly, peptides 4 and 5 inhibited ERK activation in cells with IC 50 values of 45 and 29 M, respectively; however, other peptide conjugates 6 and 7 that did not permeate cells (see above) were not able to inhibit MAPK activation (Table I). Thus, the effectiveness of these peptides in inhibiting ERK activation in vivo correlates well with their ability to enter the cells.
Similar to the alkylated peptides 2 and 3, inhibition of ERK activation by the MTP-linked peptide 4 or 5 shows a time dependence. Inhibition of ERK activation is more significant with longer preincubation times. PC12 cells were incubated with peptide 4 for varying times prior to a 5-min stimulation with NGF. Inhibition by peptide 4 or 5 gradually increases with time reaching a maximum after 40 min, with a half-inhibition time of ϳ15 min. However, the increase in fluorescence in cells treated with peptides 4 or 5 occurred under 5 min. This lag period before inhibition of ERK activation may be required for the diffusion of the peptide to ERK and/or the disruption of pre-existing MEK⅐ERK complexes.
MTS-linked Peptide Effect on the Activation of Transcription Factor ELK1 in Vivo-Finally, we have shown that the MTP not only inhibited ERK activation in vivo, but it also inhibited ERK-mediated effects such as the phosphorylation of the transcription factor ELK1. The doubly transfected cells contain GAL4 DNA binding domain fused to ELK1 transactivation domain, and firefly luciferase as reporter that is controlled by a promoter that responds to GAL4/ELK1 fusion protein. These cells respond to extracellular stimuli such as PMA, EGF, and serum that affects the ERK activation by increasing the phosphorylation of ELK1 and luciferase activity. Elegant studies using this approach demonstrated a strong correlation between phosphorylation of ELK1 by ERK and ELK1 transcriptional activity (15). The addition of peptide 5 to cells before stimulation with a combination of PMA, EGF, and FBS abrogated the effect of these stimuli to activate ERK and luciferase expression (Fig. 4). In fact, peptide 5 was as effective as the commonly used inhibitor U0126. It was proposed that MEK functions not only as a direct activator of ERK but also as a cytoplasmic anchoring protein for ERK (19,20). The phosphorylation of ERK by MEK leads to the weakening of ERK/MEK interaction and dissociation of the two enzymes followed by translocation of ERK to the nucleus (19,20). Based on our data (Supplemental Material Fig. S1A and S1B and Fig. 4), it is likely that the MEK-derived peptide binds to ERK preventing its association with MEK and thus inhibiting ERK phosphorylation. Our results also show that phosphorylation of ERK, which is a prerequisite for its activity is necessary for phosphorylation of ELK1 and ELK1 transcriptional activity. It appears that the peptide not only inhibited ERK phosphorylation in vitro and in vivo but it also inhibited its translocation to the nucleus to activate ELK1 transcriptional activity. Thus, the peptide is not only specific for inhibition of ERK activation by MEK (as shown by lack of inhibition of JNK and p38), but it also inhibits downstream effectors that are dependent on activated ERK.
Because commercially available organic inhibitors are directed toward the ATP-binding site of protein kinases, the specificity of these inhibitors has been questioned recently (21). Peptide 5, however, was designed to inhibit specifically the interaction between ERK and MEK, and thus it can be used to study selectively the effect of downstream effects of ERK activation. We are currently pursuing similar studies where ERK activation is required. These peptide inhibitors can be used as a reliable diagnostic tool to study cellular signaling pathways where ERK activation is involved.
Conclusions-we have demonstrated that peptide 1, derived from the amino terminus of MEK1 in the ERK-interacting domain, served as a basis for the design of selective inhibitors of ERK activation in vivo. Peptide inhibitors have the added benefit of providing information about biologically relevant interactions. Our approach of designing inhibitors of ERK activation using cell-permeable, domain-specific interacting peptides offers several advantages over existing approaches. The selectivity of small molecule inhibitors has been questionable, and some are found not to be useful when tested rigorously (21,22). The use of in vivo expression of transfected genes as a tool to interfere with signaling pathways has been severely limited by inability to quantitatively deliver the exact amount of inhibitor. Furthermore, the cell-permeable peptide inhibitor approach makes it feasible for applications in large scale studies when compared with microinjection of peptide inhibitors into cells individually.
The cell-permeable peptide inhibitors we have developed can be used to dissect cellular responses that involve ERK activation in response to extracellular stimuli. It is noteworthy that other protein kinase inhibitors that are directed to the ATPbinding site do not have the desired specificity (21,22). Our approach can be used to develop similar cell-permeable inhibitors of other MAPK pathways such as JNK and p38. In fact, during the preparation of this work for publication, others (23) used a similar approach to develop an inhibitor of JNK activation composed of a TAT-linked peptide derived from JNKinteracting protein. Several anchoring proteins that have specific recognition motifs to protein kinases in the MAPKsignaling pathways have been reported (24,25). Such motifs can be used to design and develop selective and specific inhibitors for individual MAPK pathways. We are currently undertaking efforts to screen peptide libraries for a more potent inhibitor based on this inhibitor prototype.