Phosphotyrosyl Peptides Block Stat3-mediated DNA Binding Activity, Gene Regulation, and Cell Transformation*

Signal transducers and activators of transcription (STATs) comprise a family of cytoplasmic signaling proteins that participates in normal cellular responses to cytokines and growth factors. Frequently, however, constitutive activation of certain STAT family members, particularly Stat3, has accompanied a wide variety of human malignancies. To identify small molecule inhibitors of Stat3, we investigated the ability of the Stat3 SH2 domain-binding peptide, PY*LKTK (where Y* represents phosphotyrosine), to disrupt Stat3 activity in vitro . The presence of PY*LKTK, but not PYLKTK or PFLKTK, in nuclear extracts results in significant reduction in the levels of DNA binding activities of Stat3, to a lesser extent of Stat1, and with no effect on that of Stat5. Analyses of alanine scanning mutagenesis and deletion derivatives of PY*LKTK reveal that the Leu residue at the Y (cid:1) 1 position nitrocellulose membranes. Probing of nitrocellulose membranes with primary antibodies and detection of horseradish peroxidase-conjugated secondary antibodies using enhanced chemiluminescence (Amersham Pharmacia Biotech) were performed as described previously (47, 49, 52, 53). The antibodies used were mouse monoclonal against amino acid residues 613–739 of Stat1 and rabbit polyclonal against the C terminus of Stat3 (Santa Cruz Biotechnology). Soft-agar Colony Formation Assays— Colony formation assays were carried out in 6-well dishes as described previously (49). In brief, each well contained 1.5 ml of 1% agarose in Dulbecco’s modified Eagle’s medium as the bottom layer and 1.5 ml of 0.5% agarose in Dulbecco’s modified Eagle’s medium containing 4000 or 6000 NIH 3T3/v-Src or NIH 3T3/v-Ras fibroblasts, respectively, as the top layer. Treatment with peptides was initiated 1 day after seeding cells by adding 75–100 (cid:1) l of medium with or without peptide, and repeated every 3 days, until large colonies were evident. Colonies were quantified by staining with 20 (cid:1) l of 1 mg/ml iodonitrotetrazolium violet, incubating at 37 °C over- night, and counting the next day.

In addition to their roles in normal cellular processes, some STATs have been observed to participate in signaling events that lead to oncogenic transformation. Constitutively active Stat3 is observed in transformation by v-Src, v-Eyk, v-Ros, v-Fps, Etk/BMX, and Lck, is associated with transformation by tumor viruses, including human T-cell lymphotrophic virus, type I, Epstein-Barr virus, and herpesvirus saimiri that activate Janus kinases or Src family tyrosine kinases (for reviews see Refs. 19 -21), and is required for oncogenic transformation (22,23). Moreover, recent observations demonstrate that a constitutively activated Stat3 mutant alone is sufficient to induce cell transformation and tumor formation in nude mice (24). In the context of human tumors, there is a high frequency of activation of Stat1, Stat3, and Stat5, with higher incidence of abnormal Stat3 activation in almost all the human tumors studied. Constitutive Stat3 activation has been detected in breast carcinomas, head and neck squamous cell carcinomas, lymphomas and leukemias, and prostate, melanoma, pancreas, ovarian, and brain tumors (reviewed in Refs. 19 -21 and 25-28). Aberrant Stat3 signaling is required for growth and survival of tumor cells, including multiple myelomas, breast carcinomas, head and neck squamous cell carcinomas, the T cell lymphoma mycosis fungoides, and large granular lymphocyte leukemia (29 -35).
Evidence has suggested that targeting Stat3 for cancer therapy may be clinically beneficial. Significantly, pharmacological or genetic interruption of constitutive Stat3 signaling inhibits expression of anti-apoptotic Bcl-2 family members in myeloma (30), a head and neck xenograft model (31), mycosis fungoides (34), and leukemia cells (35) and sensitizes myeloma cells to chemotherapy-induced apoptosis. 2 Importantly, using in vivo tumor models, murine B16 melanoma tumors regress on inhibition of constitutive Stat3 activity by gene therapy with a Stat3 dominant negative (37). The "cancer causing" propensity of constitutively activated Stat3 and the evidence of potential clinical benefits of blocking constitutive Stat3 signaling make strong arguments for target validity of Stat3 for drug intervention in cancer therapy, providing the rationale for developing inhibitors of STAT signaling (reviewed in Ref. 25). Among others, inhibitors of tyrosine kinases that participate in STAT signaling, such as Janus kinases, Src, and epidermal growth factor receptor, have been developed, with some of these already at different stages in clinical trials (38 -41). Others include use of small molecule mimics of ligands for receptors that couple to STAT signaling (42). These approaches target components that are upstream of STAT signaling and thus have the potential of disrupting other signaling pathways in addition to STATs.
We focused on identifying small molecules to target directly Stat3 that would serve as molecular tools for proof-of-concept testing, as well as leads for design of inhibitors of Stat3. In this regard, the ability of the Stat3 SH2 domain-binding phosphopeptide, PY*LKTK (where Y* represents phosphotyrosine), to disrupt activated Stat3 was investigated. Previous studies (43)(44)(45)(46) showed that phosphorylation of the tyrosine residue in this SH2-binding region of STATs is important for their dimerization and DNA binding activities. We show that the presence of PY*LKTK and its tripeptide derivative sequences, PY*L and AY*L, in nuclear extracts results in dose-dependent reduction in Stat3 DNA binding activity in vitro. Furthermore, in vitro protein binding studies suggest that PY*LKTK but not PYLKTK forms inactive Stat3⅐PY*LKTK complexes, thus reducing levels of active Stat3:Stat3 dimers available to bind DNA. Significantly, in vivo studies to determine the functional importance of peptide-directed inhibition of Stat3 signaling that utilize PY*LKTK-mts (mts indicates membrane translocating sequence, a set of hydrophobic amino acids for membrane permeability) show that this Stat3-derived phosphopeptide selectively and potently inhibits Stat3 activation and transcriptional activity in intact cells. Furthermore, PY*LKTK-mts peptide also suppresses Src transformation that depends on constitutive Stat3, with no effect on Stat3-independent Ras transformation. Altogether, we have identified a minimal peptide that effectively abrogates Stat3 activity in vitro and suppresses both Stat3 signaling in vivo and cell transformation. These studies provide the basis for use of the Stat3 SH2 domain-binding phosphopeptide as a lead for design of novel peptidomimetic drugs targeting Stat3.

EXPERIMENTAL PROCEDURES
Cells and Reagents-Src-transformed NIH 3T3/v-Src fibroblasts have been described previously (47,48). Cells were grown in Dulbecco's modified Eagle's medium containing 5% iron-supplemented bovine calf serum. Antibodies C-136 and E23X for Stat1 and C20 and C20X for Stat3 were all obtained from Santa Cruz Biotechnology; enhanced chemiluminescence (ECL) detection reagents were from Amersham Pharmacia Biotech. Interleukin 6 (IL-6) was obtained from BD PharMingen.
Peptides-Peptides used in studies include PY*LKTK and its mts derivative (PY*LKTK-AAVLLPVLLAAP), PYLKTK and its mts derivative (PYLKTK-AAVLLPVLLAAP), PFLKTK, bead-coupled PY*LKTK and PYLKTK, PY*L and AY*L, and their derivatives ( Table I). The mts derivatives of peptides were synthesized by the Peptide Synthesis Laboratory, Queen's University, Kingston, Ontario, Canada. Other peptides were synthesized manually using standard Fmoc (N-(9-fluorenyl)-methoxycarbonyl) solid phase chemistry. Peptides were used at concentrations up to 3 mM as indicated.
Plasmids-The Stat3 reporter, pLucTKS3, driving expression of firefly luciferase has been described previously (23,49). The pLucTKS3 plasmid harbors seven copies of a sequence corresponding to the Stat3specific binding site in the promoter of the human C-reactive protein gene (50). The plasmid, pRLSRE, contains two copies of the serum response element (SRE) from the c-fos promoter (23,51), subcloned into the Renilla luciferase reporter, pRL-null (Promega).
Transfection and Generation of Stable Clones-NIH 3T3/v-Src/ pLucTKS3 and NIH 3T3/v-Src/pRLSRE are stable clones that were generated by transfecting NIH 3T3/v-Src fibroblasts with pLucTKS3 or pRLSRE, respectively, and selecting for G418-resistant clones (49). In the case of NIH 3T3/v-Src/pLucTKS3/pRLSRE, pRLSRE was transfected into NIH 3T3/v-Src/pLucTKS3 cells, and stable Zeocin-resistant clones were selected. Transfections were carried out with Lipo-fectAMINE Plus (Life Technologies, Inc.) according to the manufacturer's protocol.
Recombinant Baculoviruses and Infection of Sf-9 Insect Cells-Stat1 and Stat3 recombinant baculoviruses and infection of Sf-9 insect cells have been described previously (52). For infection, Sf-9 cells were plated at 1 ϫ 10 6 cells/35-mm dish (6-well plates) in 2 ml of SF-900II serumfree medium (Life Technologies, Inc.) and incubated at 27°C for 1 h. Once cells were attached, the medium was replaced with 500 l of infection mixture, which is SF-900II medium containing appropriately diluted baculoviruses, and allowed to incubate at 27°C for 1 h with slow rocking. Subsequently, infection medium was replaced with fresh SF-900II, and cells were further incubated for 48 h before harvesting.
Nuclear Extract Preparation and Gel Shift Assays-Nuclear extract preparations from normal NIH 3T3 fibroblasts (treated with or without IL-6 for 30 min) or from their Src-transformed counterparts and electrophoretic mobility shift assay were carried out as described previously (23,47,53). The 32 P-radiolabeled oligonucleotide probes are high affinity sis-inducible elements (hSIE, m67 variant, 5Ј-AGCTTCATTTCCCG-TAAATCCCTA) that binds Stat1 and Stat3 (53,54) or mammary gland factor elements (MGFe from the bovine ␤-casein gene promoter, 5Ј-AGATTTCTAGGAATTCAA) for Stat1 and Stat5 binding (55,56). When peptide inhibitors were present, these were preincubated with the nuclear extract for 30 min at room temperature prior to incubation with radiolabeled probe.
Protein-Peptide Binding Assays-The Stat3-derived phosphopeptide, PY*LKTK, and non-phosphorylated PYLKTK were individually coupled to agarose beads at the C terminus to generate the bead-coupled peptides, PY*LKTK ϽϾ bead or PYLKTK ϽϾ bead. Prior to use, bead-coupled peptides were washed 5 times with 1% Nonidet P-40 lysis buffer. Following this, equivalent amounts of each of the bead-coupled peptides were incubated with recombinant baculovirus-infected Sf-9 cell lysates containing equal amounts of Stat1 or Stat3 protein at 4°C for 30 min. Mixtures were then incubated with or without free peptide, PYLKTK or PY*LKTK, at room temperature for an additional 30 min. Mixtures were washed 5 times with 1% Nonidet P-40 buffer followed by centrifugation (13,000 ϫ g, 10 s, 4°C) to remove unbound Stat1 or Stat3 protein and free peptide and obtain bead pellet. SDS sample buffer was then added to the pellet of peptide coupled to bead with bound STAT proteins, boiled, and then subjected to SDS-10% polyacrylamide gel electrophoresis.
Western Blot Analysis-Electrophoresed proteins were transferred to nitrocellulose membranes. Probing of nitrocellulose membranes with primary antibodies and detection of horseradish peroxidase-conjugated secondary antibodies using enhanced chemiluminescence (Amersham Pharmacia Biotech) were performed as described previously (47,49,52,53). The antibodies used were mouse monoclonal against amino acid residues 613-739 of Stat1 and rabbit polyclonal against the C terminus of Stat3 (Santa Cruz Biotechnology). Soft-agar Colony Formation Assays-Colony formation assays were carried out in 6-well dishes as described previously (49). In brief, each well contained 1.5 ml of 1% agarose in Dulbecco's modified Eagle's medium as the bottom layer and 1.5 ml of 0.5% agarose in Dulbecco's modified Eagle's medium containing 4000 or 6000 NIH 3T3/v-Src or NIH 3T3/v-Ras fibroblasts, respectively, as the top layer. Treatment with peptides was initiated 1 day after seeding cells by adding 75-100 l of medium with or without peptide, and repeated every 3 days, until large colonies were evident. Colonies were quantified by staining with 20 l of 1 mg/ml iodonitrotetrazolium violet, incubating at 37°C overnight, and counting the next day.

Disruption of Stat3 DNA Binding Activity by Stat3-derived
Phosphopeptide-The phosphopeptide, PY*LKTK (where Y* represents phospho-Tyr, Tyr(P)), is derived from the sequence of amino acids in the vicinity of Tyr(P) in the SH2-binding region of Stat3. The Tyr(P) contained in the PY*LKTK sequence of one STAT monomer interacts with the SH2 domain of another STAT monomer to form stable dimers that can bind DNA. The prediction is that PY*LKTK peptide will bind to Stat3 by way of the same Tyr(P)-SH2 interaction and disrupt dimerization and DNA binding activity, which can be measured by an electrophoretic mobility shift assay (EMSA). To test this prediction, nuclear extracts of equal total protein containing activated Stat3 were incubated with or without different concentrations of peptides, PY*LKTK, PYLKTK, or PFLKTK, prior to incubation with 32 P-labeled oligonucleotide probe, hSIE, for EMSA. The presence of PY*LKTK results in a dosedependent decrease in the level of Stat3 binding to the hSIE probe (Fig. 1A), with 50% reduction at peptide concentration of 235 M. In contrast, preincubation with non-phosphorylated PYLKTK or PFLKTK has no effect on Stat3 DNA binding activity, up to concentrations of 1 mM (Fig. 1, A and B, and data not shown). Together, these results show that PY*LKTK phosphopeptide disrupts Stat3 activity in vitro. Furthermore, the results also demonstrate that the Tyr(P) residue of PY*LKTK is necessary for the observable reduction in Stat3 DNA binding activity. This finding is in agreement with previously reported data that the Tyr(P)-containing peptide of Stat1, GY*IKTE, disrupts Stat1 (44).
Alanine Scanning Mutagenesis of PY*LKTK and Determination of Critical Amino Acids Required to Disrupt Stat3 Activity-To determine the importance of each amino acid residue, Ala substitution derivatives of PY*LKTK were analyzed for their abilities to disrupt active Stat3 as measured by reduction in DNA binding activity. Nuclear extracts of equal total protein containing active Stat3 were preincubated with different concentrations of Ala-substituted derivatives of PY*LKTK prior to incubation with 32 P-labeled hSIE probe. Samples were then subjected to EMSA, and the results are shown in Fig. 2, A and B. Similar to observations in Fig. 1, preincubation of nuclear extracts with PY*LKTK or its Ala-substituted derivatives, AY*LKTK, PY*LATK, PY*LKAK, or PY*LKTA, results in significantly decreased levels of Stat3 DNA binding activities (Fig.  2, A and B). That these Ala-substituted derivatives effectively reduced Stat3 DNA binding activity suggests that the substituted amino acid residues, Pro (position Ϫ1 relative to Y), Lys (Yϩ2 and Yϩ4) and Thr (Yϩ3), are not required for the inhibitory effect of PY*LKTK. On the other hand, preincubation of nuclear extracts with the Ala-substituted derivative PY*AKTK has no significant effect on Stat3 DNA binding activity ( Fig.  2A), indicating a loss of activity of PY*AKTK following substitution of Leu at Yϩ1. Together, these results suggest that in addition to Y*, the Leu residue at Yϩ1 position is critical and that amino acid residues at positions Yϩ2 and C-terminal are not essential for the ability of PY*LKTK to disrupt active Stat3. Phosphotyrosyl Tripeptide, XY*L, Is Sufficient to Decrease Stat3 DNA Binding Activity-Based on the Ala substitution analyses of PY*LKTK, we inferred that a tripeptide of primary structure, XY*L (where X represents any amino acid), might be sufficient to disrupt activated Stat3. We therefore examined the effects of tripeptides, PY*L and AY*L, on Stat3 DNA binding activity in vitro. As predicted, following preincubation of nuclear extracts containing active Stat3 with PY*L or AY*L, significant disruption of Stat3 DNA binding activity was observed in a dose-dependent manner (Fig. 3). Moreover, both PY*L and AY*L were as effective in disrupting Stat3 as the parent peptide, PY*LKTK. Thus, we deduce that tripeptides of the structure, XY*L, encompass the critical amino acid sequence required for disruption of active Stat3. We further infer from these and the Ala scanning mutagenesis studies that Pro at the YϪ1 position is not essential for the inhibitory effect of PY*LKTK.
Structure-Activity Analysis of PY*LKTK-To further characterize PY*LKTK in relation to its ability to disrupt Stat3 DNA binding activity, we carried out peptide "structure-activity relationship" studies. Various derivatives of PY*LKTK were designed and tested on Stat3 DNA binding activity in vitro. Nuclear extracts containing active Stat3 were preincubated with different concentrations of each phosphopeptide derivative prior to incubation with radiolabeled hSIE probe and analysis by EMSA. Bands corresponding to DNA binding activities were scanned and quantified by densitometric analysis. For each peptide derivative, quantified amounts corresponding to DNA binding activities were represented as percentages of control (no peptide added). Percent of DNA binding against concentration of peptide was plotted, from which the concentrations of peptides or derivatives corresponding to 50% reduction in Stat3 DNA binding activity (DB 50 ) were determined and shown in Table I.
The DB 50 value for PY*LKTK was determined to be 235 M, and the values for PY*L and AY*L are 182 and 217 M, respectively (Table I). As stated previously, the control peptides, PYLKTK or PFLKTK, have no effect on Stat3 DNA binding activity when preincubated with nuclear extract containing active Stat3 prior to EMSA. Consistent with the Ala scanning analysis (Fig. 2), except for Ala substitution of Leu at the Yϩ1 position, Ala-substituted derivatives of residues at Yϩ2 and downstream positions have comparable DB 50 values to PY*LKTK (Table I). Results also show that the tetrapeptide, PY*FK, has negligible interaction with Stat3 and causes no significant disruption of Stat3 DNA binding activity, consistent with the requirement for a Leu residue at the Yϩ1 position (Fig. 2). This also suggests that a hydrophobic aliphatic amino acid is preferred to a hydrophobic aromatic amino acid at the Yϩ1 position.
We have already shown that the Pro residue at the YϪ1 position is not critical, as Ala-substituted AY*LKTK and the tripeptide, AY*L, both disrupt Stat3 DNA binding activity (Figs. 2 and 3). This observation, however, does not indicate whether the YϪ1 position has an influential role on the activity of peptide. To assess this, we tested the effect of tripeptide, Y*LK, on Stat3 DNA binding activity. Surprisingly, results show that Y*LK has no effect on active Stat3 (Table I). This finding suggests that although the Pro residue is not essential, occupation of the YϪ1 position is critical for effective disruption of Stat3 DNA binding levels. To test this possibility, we made new derivatives, AY*LK, PY*LK, and AcY*LK, with either Ala, Pro, or an acetyl group (Ac) at the YϪ1 position, respectively. Significantly, incubation of nuclear extracts with AY*LK, PY*LK, or AcY*LK disrupts Stat3 DNA binding, with nearly identical DB 50 values as obtained for PY*LKTK (Table I).
These results indicate that either the YϪ1 amide group strongly contributes to the overall activity of the phosphopeptide in the context of its interaction with Stat3 or a free ammonium group on the Y* residue is disruptive. Altogether, our results show that in addition to PY*LKTK, the tripeptide with sequence XY*L (where X represents a group that forms amide bond with Y*) is sufficient to significantly diminish Stat3 DNA binding activity.
Selective Disruption of Activated Stat3 and to a Lesser Extent Stat1-Because Stat3 forms homodimers with itself and heterodimers with Stat1, we sought to determine whether PY*LKTK or its derivatives can interact with and disrupt DNA binding activities of other STATs, particularly Stat1 and Stat5. For this purpose, we used nuclear extracts containing active Stat1, Stat3, and Stat5. Nuclear extracts of equal total proteins were preincubated with different concentrations of each of the three peptides, followed by incubation with radiolabeled hSIE probe that binds Stat1 and Stat3 (hSIE) (53,54) or MGFe probe that binds Stat1 and Stat5 (MGFe) (55,56). In EMSAs using hSIE probe, activated STATs show three distinct bands, corre- sponding to Stat3:Stat3 (upper band), Stat1:Stat3 (intermediate band), and Stat1:Stat1 dimers (lower band) (Fig. 4A) (29,53). Preincubation of nuclear extracts with different concentrations of peptides, PY*LKTK, PY*L, or AY*L, results in significant decreases in DNA binding activities of dimers containing Stat3 and Stat1 (Fig. 4A). Disruption of Stat3:Stat3 dimers is more pronounced, occurring at relatively lower concentrations of peptides than concentrations that disrupt Stat1:Stat3 or Stat1:Stat1 dimers (Fig. 4A). This suggests that all three peptides may have higher affinities for Stat3 than for Stat1. Moreover, disruption of Stat1:Stat1 dimer is comparable with that of Stat1:Stat3 dimer, perhaps reflecting the commonality of Stat1  a Data shown are means and S.D. of at least 5 independent assays; NE, no effect; DB 50 , concentration of peptide at which DNA binding activity is reduced by 50%. in both complexes. Data also suggest that both PY*L and AY*L do not inhibit Stat1:Stat3 or Stat1:Stat1 dimers as effectively as PY*LKTK. These results demonstrate that PY*LKTK, PY*L, and AY*L disrupt DNA binding activities of dimers containing either Stat3 or Stat1.
EMSA of nuclear extracts containing active Stat1 and Stat5 following incubation with MGFe probe reveals two distinct bands corresponding to Stat5:Stat5 (upper band) and Stat1: Stat1 dimers (lower band) (Fig. 4B) (53). Results of incubation with phosphopeptides of nuclear extracts prior to incubation with labeled MGFe probe and EMSA show that PY*LKTK, PY*L, and AY*L did not substantially disrupt Stat5:Stat5 DNA binding activity (Fig. 4B, upper band). Together, these results indicate preferential affinity of Stat3-derived phosphopeptides for Stat3, and to a lesser extent Stat1, over Stat5.
The DNA binding activities of STATs were quantified, and the changes induced by the disruption of active STATs with peptides were calculated as a percent of control (no peptide added) and plotted against concentration of peptide. Peptide concentrations where there is a 50% reduction in DNA binding (DB 50 ) activity of Stat1, Stat3, or Stat5 were determined and are presented in Table II. As shown above, all three phosphopeptides, PY*LKTK, PY*L, and AY*L, reduce DNA binding activities of Stat1:Stat1, Stat1:Stat3, and Stat3:Stat3 dimers. However, the inhibitory constant for suppression of Stat3:Stat3 DNA binding activity is almost two times lower than that for reduction of Stat1:Stat3 or Stat1:Stat1 activities (Table II) for all three peptides, reflecting a relatively higher affinity for Stat3 over Stat1. None of the peptides had any appreciable effect on active Stat5, with no changes in the levels of Stat5 DNA binding activity at concentrations lower than 1 mM (Table II).
Evidence That PY*LKTK Can Bind Stat3 and Disrupt Tyr(P)-SH2 Interactions-STAT dimerization is critical for their DNA binding activity (3,44,46). Because the PY*LKTK sequence is derived from Stat3, the conjecture is that PY*LKTK and its tripeptide derivatives can physically interact with the SH2 domain of Stat3. In the case of preformed Stat3: Stat3 dimers, this interaction will disrupt existing dimers, creating complexes of Stat3⅐peptide. To confirm and study this interaction, we coupled the C terminus of PY*LKTK to beads and generated PY*LKTK ϽϾ bead. We incubated PY*LKTK ϽϾ bead with extracts containing monomeric Stat1, Stat3, or Stat5 protein prepared from Sf-9 cells infected with Stat1, Stat3, or Stat5-encoding recombinant baculoviruses, respectively. During this incubation, we expected to form Stat1⅐PY*LKTK ϽϾ bead or Stat3⅐PY*LKTK ϽϾ bead complexes through Tyr(P)-SH2 interactions (SH2 domain present in Stat1 or Stat3). Non-phosphorylated PYLKTK ϽϾ bead was also incubated with each of the STATs as a control. Sample mixtures containing complexes of STAT⅐PY*LKTK ϽϾ bead were then incubated with different concentrations of soluble PY*LKTK or PYLKTK peptides as competitors. Following extensive washing to remove unbound proteins, the remaining STAT proteins bound to beads were subjected to SDS-polyacrylamide gel electrophoresis and blotted with anti-STAT antibodies.
Results show that incubation of Stat1 or Stat3 protein with PY*LKTK ϽϾ bead results in physical binding of these proteins to peptide (Fig. 5, A and B, compare lanes 3 with 4). In contrast, incubation of Stat1 or Stat3 with the non-phosphorylated PYLKTK ϽϾ bead did not result in physical association of either STAT protein with peptide (Fig. 5, A and B, lanes 1 and  2). Together, these results indicate that Tyr(P) is required for binding of peptide to Stat1 or Stat3. Incubation of Stat1: PY*LKTK ϽϾ bead or Stat3:PY*LKTK ϽϾ Bead with increasing concentrations of soluble PY*LKTK peptide resulted in dose-dependent reduction of Stat1 or Stat3 protein bound to bead-coupled peptide (Fig. 5, A and B, upper panels, compare  lanes 4 with 9 -12). These findings together demonstrate that PY*LKTK disrupts preformed complexes of STAT⅐peptide. In contrast, incubation of Stat1:PY*LKTK ϽϾ bead or Stat3: PY*LKTK ϽϾ bead with increasing concentrations of nonphosphorylated PYLKTK peptide had no effect on the bound Stat1 or Stat3 (Fig. 5, A and B, upper panels, compare lanes 4  with 5-8). The lower panels of Fig. 5, A and B, represent total Stat1 or Stat3 protein bound to bead-coupled peptide prior to addition of soluble PY*LKTK or PYLKTK peptides.
Together, these results provide strong evidence that PY*LKTK engages in direct Tyr(P)-SH2 interactions with Stat1 or Stat3 and suggest that the phosphopeptide disrupts preformed dimers containing either STAT protein. Moreover, these findings are consistent with the EMSA data showing that preincubation of PY*LKTK or its tripeptide derivatives with active Stat1 or Stat3 dimers results in a decline in the levels of STAT DNA binding activity (Figs. 1-4). As a control for specificity, incubation of the PY*LKTK ϽϾ bead with Stat5 did not result in any binding of Stat5 to peptide (data not shown). This is consistent with the finding that incubation of PY*LKTK with active Stat5 dimers has no effect on the levels of Stat5 DNA binding activity (Fig. 4B and Table II). By densitometric analysis, we compared the extent of disruption by PY*LKTK of complexed Stat1⅐PY*LKTK ϽϾ bead to that of Stat3⅐PY*LKTK ϽϾ bead. Bands (in Fig. 5, A and B) were scanned and quantified. For each complex, the percent of band intensity was plotted against concentration of PY*LKTK and is shown in Fig.  5C. The concentrations of PY*LKTK corresponding to 50% reduction of band intensity (BI 50 ) for each complex were determined. The BI 50 values obtained were 245 M for Stat1: PY*LKTK ϽϾ bead (Fig. 5C, right panel) and 104 M for Stat3:PY*LKTK ϽϾ bead (Fig. 5C, left panel). These BI 50 values suggest that this phosphopeptide has a relative preference for Stat3 over Stat1. The BI 50 values obtained here are ϳ2-fold lower than DB 50 values obtained from EMSA (Tables I  and II), suggesting that it is easier for PY*LKTK to disrupt STAT⅐PY*LKTK complexes than STAT⅐STAT complexes. PY*LKTK Disrupts Stat3 Activation in Vivo-We reasoned that if PY*LKTK disrupts active Stat3 and diminishes levels available to bind DNA in vitro, it should block Stat3 activation in vivo. To test this, we synthesized PY*LKTK and its nonphosphorylated counterpart, PYLKTK, linked to an mts (57) at the C terminus and generated PY*LKTK-mts or PYLKTK-mts. Because it is composed of a hydrophobic amino acid sequence (AAVLLPVLLAAP), studies have shown that mts aids small molecules, including peptides, to go across the plasma membrane (57). We carried out time course studies by introducing for different times PYLKTK-mts or PY*LKTK-mts into Srctransformed fibroblasts that harbor constitutively active Stat3. We prepared nuclear extracts for measuring Stat3 DNA binding activities by EMSA. Results show that on the basis of equal protein, nuclear extracts prepared from cells treated with PY*LKTK-mts peptide have significantly reduced Stat3 DNA binding activities compared with extracts prepared from nontreated cells (Fig. 6A, lanes 5, 10 and 15). Treatment of Srctransformed cells with the non-phosphorylated PYLKTK-mts peptide has no effect on Stat3 activation (Fig. 6A, lanes 4, 9 and  14). Results show significant inhibition of Stat3 activation by PY*LKTK-mts following treatment for 12 h, which continues to be blocked for up to 48 h (Fig. 6). In contrast, no consistent reduction of Stat3 DNA binding activity was observed in nuclear extracts prepared from Src-transformed fibroblasts treated with PY*LKTK-mts for 6 h or less (data not shown).
These results suggest that the presence of PY*LKTK-mts peptide blocks Stat3 activity in intact cells. PY*LKTK might not only disrupt preformed Stat3:Stat3 dimers but may also block de novo Stat3 dimerization and activation by forming heterocomplexes with Stat3 monomers. To test this possibility, we investigated the activation of Stat3 by IL-6 in NIH 3T3 fibroblasts that have been treated with PY*LKTK-mts or its non-phosphorylated counterpart for 1 h prior to stimulation by the cytokine. Consistent with activation of Stat3 by the cytokine, EMSA of nuclear extracts prepared from cells stimulated with IL-6 shows enhanced Stat3 DNA binding activity (Fig. 6B,  lanes 2-5). Furthermore, pretreatment of cells with PY*LKTKmts and not PYLKTK-mts completely blocked Stat3 activation induced by IL-6 (Fig. 6B, lane 6). To identify positively the STAT DNA binding activity in these cells, nuclear extracts were preincubated with anti-Stat1 or anti-Stat3 antibodies. The presence of the anti-Stat1 antibody did not affect the complex, indicating that the protein-DNA complex does not contain Stat1 (Fig. 6A, lanes 2, 7, and 12, and B, lane 3). In contrast, anti-Stat3 antibody blocked and supershifted the complex (Fig. 6A, lanes 3, 8, and 13, and B, lane 4), demonstrating that the protein-DNA complex contains Stat3. These results together show that in vivo, PY*LKTK can block Stat3 activation and thus is expected to inhibit Stat3-mediated signaling.
PY*LKTK Disrupts Stat3-mediated Gene Regulation in Vivo-We wanted to determine whether by suppression of Stat3 activation, PY*LKTK could attenuate Stat3 transcriptional activity in vivo. For this purpose, we generated NIH 3T3/v-Src fibroblasts that stably express a Stat3-dependent firefly luciferase reporter, pLucTKS3 (23,49), based on the C-reactive protein gene promoter (NIH 3T3/v-Src/pLucTKS3 cells). Because Stat3 is constitutively activated in v-Src-transformed fibroblasts (47), NIH 3T3/v-Src/pLucTKS3 cells exhibit constitutive firefly luciferase activity (49) that correlates with constitutive Stat3 activation. Into this background, we stably expressed a different luciferase reporter, pRLSRE, and so generated double stable transfectants, NIH 3T3/v-Src/pLucTKS3/ pRLSRE cells, that concurrently express both reporters. The reporter, pRLSRE, contains the SRE of the c-fos promoter that is driving a Renilla luciferase gene. We have shown previously that v-Src activates the SRE-based promoter independent of Stat3 (23) and thus pRLSRE serves as Stat3-negative control reporter. Src-transformed fibroblasts that stably express pRLSRE have constitutive induction of Renilla luciferase reporter activity (data not shown). On the other hand, the single stable transfectant, NIH 3T3/v-Src/pRLSRE, was also generated, and equal proportions of this and of NIH 3T3/v-Src/ pLucTKS3 were pooled together to make a mixed population of NIH 3T3/v-Src/pLucTKS3 and NIH 3T3/v-Src/pRLSRE. The firefly and Renilla luciferases utilize different substrates and a Values are the means and S.D. of at least 5 independent assays. DB 50 , concentration of peptide at which DNA binding activity is reduced by 50%.

Inhibition of Stat3 with Phosphotyrosyl Peptides
thus can be assayed independently in the same cell lysates.
To examine changes in Stat3 transcriptional activity with peptides, these indicator cell lines were treated with or without PYLKTK-mts or PY*LKTK-mts for specified times, and cytosolic extracts were prepared for luciferase assays. Results show that treatment of indicator cells with PY*LKTK-mts peptide FIG. 5. Interaction of STAT proteins with bead-coupled PY*LKTK. Bead-coupled PY*LKTK was incubated with STAT-containing lysates prepared from Sf-9 cells infected with recombinant baculovirus expression vectors encoding Stat1 or Stat3 for 30 min, and then further incubated with or without increasing amounts of PY*LKTK for 20 min. After removing unbound proteins by extensive washing of beads, SDS sample buffer was added to samples of STAT proteins bound to bead-coupled PY*LKTK and subjected to SDS-polyacrylamide gel electrophoresis and Western blotting with anti-STAT antibodies. A, equal amounts of Sf-9 lysate containing Stat3 were added to bead-coupled PY*LKTK or PYLKTK to which was subsequently added 0.1-0.8 mM PY*LKTK or PYLKTK. B, equal amounts of Sf-9 lysate containing Stat1 were added to bead-coupled PY*LKTK or PYLKTK to which was subsequently added 0.1-0.8 mM PY*LKTK or PYLKTK. Upper panels represent STAT protein bound to bead-coupled peptide, and lower panels represent total STAT protein bound to bead-coupled peptide. significantly suppresses induction of the Stat3-dependent pLucTKS3 luciferase reporter (Fig. 7, B and C). Consistent with the DNA binding studies (Fig. 6), this inhibitory effect was significant by 12 h and persisted for 48 h after phosphopeptide treatment, with minimal effects detected at 6 h duration of treatment or less (data not shown). In contrast, treatment of FIG. 6. In vivo effect of PY*LKTK-mts on Stat3 DNA binding activity. A, NIH 3T3/v-Src or their counterparts stably expressing the Stat3-dependent (pLucTKS3) or -independent (pRLSRE) luciferase reporters were treated with or without PY*LKTK-mts or PYLKTK-mts for the indicated times. Both nuclear and cytosolic extracts were prepared from the same cultures and used for EMSA using radiolabeled hSIE probe (A) and luciferase or ␤-galactosidase activity measurements (Fig. 7), respectively. B, nuclear extracts prepared from NIH 3T3 cells that were stimulated with IL-6 for 30 min following treatment with peptides for 1 h were subjected to EMSA using radiolabeled hSIE probe. Antibodies against Stat1 (A, lanes 2, 7, and 12 and B, lane 3) or Stat3 (A, lanes 3, 8, and 13 and B, lane 4) were added to nuclear extracts to identify the STAT proteins bound to probe. Positions of activated Stat3-DNA complexes are indicated, as well as the band corresponding to complexes supershifted by anti-Stat3 antibody. The bottom band (A) is free hSIE probe and contains no bound protein. MTS, membrane translocating sequence.
indicator cells with the non-phosphorylated control peptide, PYLKTK-mts, has no effect on induction of the Stat3-dependent luciferase reporter (Fig. 7, B and C). For control experiments, treatment of cells with PY*LKTK-mts or PYLKTK-mts did not inhibit induction of the Stat3-independent pRLSRE luciferase or ␤-galactosidase reporter activities (Fig. 7, A and  C). Thus, PY*LKTK-mts has no significant effect on transcriptional events that are independent of Stat3, suggesting that in our studies PY*LKTK-mts does not exhibit undue nonspecific effects. These findings provide evidence that suppression of Stat3 activation by PY*LKTK, either by disruption of preformed dimers of Stat3 and/or blockade of de novo Stat3 dimerization, prevents Stat3-mediated gene expression in vivo. This effect is specific and requires Tyr(P), consistent with specific Tyr(P)-SH2 interactions as the likely basis for obstructing Stat3 activation.
Inhibition of v-Src Transformation by PY*LKTK-mts-Previous studies have demonstrated that constitutive Stat3 activation is required for transformation by the Src oncoprotein (22,23). Because PY*LKTK-mts blocks Stat3 activity in vivo, we investigated whether this suppression of Stat3 signaling has biological significance. We tested the effect of PY*LKTKmts and PYLKTK-mts on the growth of NIH 3T3/v-Src cells in soft agar suspension as a measure of v-Src transformation. Treatment of Src-transformed fibroblasts in agar with PY*LKTK-mts significantly suppresses anchorage-independent growth of cells (Fig. 8A), suggesting that the Stat3 phosphopeptide inhibits v-Src transformation. Together with the results from the DNA binding activity and transcriptional reporter assays, this finding indicates that PY*LKTK blocks Src transformation by suppressing constitutive Stat3 signaling. This is in agreement with previous studies (22,23) using dominant-negative forms of Stat3 protein and further demonstrates a requirement for Stat3 signaling in Src transformation. In contrast, treatment of NIH 3T3/v-Src cells in agar with the non-phosphorylated counterpart, PYLKTK-mts, has no influence on growth, suggesting that Tyr(P) is required for the effect of this peptide on Src transformation. In other control studies, neither PY*LKTK-mts nor its non-phosphorylated counterpart has any effect on the growth in soft agar of Rastransformed cells (Fig. 8B) that do not require Stat3 signaling (22,23,49). This result also strongly argues that under the conditions of this study, PY*LKTK does not exert undue nonspecific effects on transformation or cell growth. Together, these studies provide compelling evidence that PY*LKTK selectively blocks Stat3 signaling and its biological consequences. DISCUSSION The critical role of Stat3 in certain types of human tumor cells and the potential benefits of blocking its signaling have validated Stat3 as a target for drug discovery (19,25), and thus intensified the drive for inhibitors of Stat3 signaling. Because dimerization is a key event in the activation of STATs, its interference is predicted to compromise STAT signaling and the resulting biological functions. Our current study utilized a peptide approach to disrupt Stat3 dimerization, both for proofof-concept testing and also to provide leads for rational design of Stat3 inhibitors. We demonstrate that the Stat3 phosphopeptide, PY*LKTK, significantly inhibits Stat3 activity both in vitro and in vivo. Because the sequence PY*LKTK is the same as that in the SH2-binding region of Stat3, we propose the following model for its interaction with Stat3 in vitro (Fig.  9). By associating with the SH2 domain of Stat3, PY*LKTK disrupts the Tyr(P)-SH2 interactions that stabilize active Stat3:Stat3 dimers and causes their dissociation into two Stat3 monomers, forming inactive Stat3⅐PY*LKTK heterocomplexes. Depending on the levels of PY*LKTK, there may be residual FIG. 7. In vivo effect of PY*LKTK-mts on Stat3-mediated transcriptional activity. Luciferase and ␤-galactosidase activities of cytosolic extracts prepared from stable transfectants expressing Stat3-dependent (pLucTKS3), Stat3-independent (pRLSRE), and ␤-galactosidase (an internal control) reporters. A, Renilla luciferase activities of extracts prepared from peptide-treated and non-treated cells expressing pRLSRE. B, firefly luciferase activities of extracts prepared from peptide-treated and non-treated cells expressing pLucTKS3. C, luciferase and ␤-galactosidase activities of cytosolic extracts prepared from peptide-treated and nontreated cells. MTS, membrane translocating sequence.
Stat3:Stat3 dimers, which are measured as a low level of DNA binding activity after equilibrium is reached. A key element of this model is the critical role of Tyr(P)-SH2 interactions, as all of the inhibitory effects of the PYLKTK-derived peptides require the presence of Tyr(P). Our current findings are consistent with previous reports (44,58,59) that characterized the interactions of Tyr(P) with the SH2 domains of STATs using phosphopeptides and together provide evidence for the mechanistic importance of Tyr(P)-SH2 binding in formation of active STAT dimers (44, 60 -62).
Our studies show that PY*LKTK disrupts the DNA binding activity of not only Stat3 but also Stat1 complexes. This observation is consistent with the finding that the Stat1 SH2 domainbinding peptide, GY*IKTE, disrupts Stat1 (3, 44) and reflects the closer relationship between these two STATs compared with other STAT family members. For example, Stat1 and Stat3 form heterodimers in vivo and share at least 50% amino acid sequence homology in the SH2 domain-binding region (compare GY*IKTE for Stat1 with PY*LKTK for Stat3). Moreover, both STATs bind to the same DNA-response elements in certain promoters (54,56) and functionally appear to have opposing but balancing effects (63). In contrast, no substantial interaction with Stat5 or disruption of its activity by PY*LKTK is observed, strongly arguing against an indiscriminate mode of action for the phosphopeptide and suggesting that Stat3 and Stat5 may share very limited similarities. Except for one study (64), there are no other reports of Stat3:Stat5 heterodimer formation in vivo. Moreover, comparison of amino acid sequences of the SH2 domain-binding regions of Stat3 and Stat5 reveal only about 30% identity (compare PY*LKTK for Stat3 with GY*VKPQ for Stat5), consistent with the finding that the Stat3-derived phosphopeptide does not interact with Stat5.
By alanine scanning, the minimum peptide sequence sufficient to confer the interaction with Stat3 was reduced to a tripeptide, of structure XY*L (where X represents a group that engages in amide bond formation with Tyr(P), as in AY*L or PY*L). Whereas Ala, Pro, or a simple acetyl group is sufficient for inhibitory activity of XY*L, it remains to be determined whether other amino acids at the YϪ1 position would do the same. Similar to PY*LKTK, these tripeptides, AY*L and PY*L, disrupt Stat3, and to a lesser extent Stat1 dimers, with no effect on Stat5 dimers. From the standpoint of peptide-protein interactions, this observation indicates that the tripeptides can adopt a conformation that is similar to that of PY*LKTK in the Tyr(P) region. Thus, the three key requirements for the activity of PY*LKTK are the Tyr(P), the YϪ1 position, and the Yϩ1 Leu residue. The importance of the Y-1 position is intriguing, as it is believed that the binding specificity of Tyr(P) peptides to SH2 domains is determined by residues C-terminal to the Tyr(P) (65). The crucial amino acid residue in the Tyr(P)-binding pocket of SH2 domains is the positively charged Arg (60), suggesting a strong ionic interaction with the negatively charged Tyr(P). Based on this, it is likely that the absence of an amide derivative at the Y-1 position leaves the free N terminus (NH 3 ϩ ), which may interact unfavorably with positively charged residues in the SH2 domain or disrupt the bound conformation of the tripeptide. In support of this, we observed that while Y*LK lacks effect on Stat3 DNA binding (Table I), acetylation of the Tyr, as in Ac-Y*LK, or the formation of a peptide bond at the YϪ1 position, such as in AY*LK, makes these peptides strong disrupters of Stat3 activation.
The requirement for a Leu residue at the Yϩ1 position for PY*LKTK activity suggests its importance for the interaction between Stat3 and the phosphopeptide, consistent with observations that residues C-terminal to the Tyr(P) residue are critical for Tyr(P)-SH2 interactions (36,58,62,65). On the other hand, our observation is in contrast to previous reports that three to five amino acid residues C-terminal to Y* are essential for interactions involving Tyr(P) and SH2 domains (58,62). Our current study shows that amino acid residues at the Yϩ2 and C-terminal positions are not essential. They may, however, have contributions to the overall interaction of PY*LKTK with Stat3. In addition, we cannot exclude a possible role for these residues in other protein-protein interactions involving Stat3. For example, studies have found that amino acid residues C-terminal to the Tyr are key determinants in Stat3 phosphorylation during IL-6 signaling (61).
We demonstrate that the Stat3 phosphopeptide, PY*LKTK, has significant in vivo activity against Stat3 signaling, in agreement with in vitro binding assays. In particular, the phosphopeptide effectively inhibits signaling induced by v-Src and cytokines such as IL-6 that induce Stat3 activation. We theorize that the Stat3 phosphopeptide interacts with and disrupts pre-existing Stat3:Stat3 dimers in Src-transformed cells in the manner observed in vitro. This leads to non-functional Stat3 complexes and ultimately a reduction in Stat3 signaling. PY*LKTK could also inhibit Stat3 signaling in vivo by blocking its de novo activation. From the in vitro protein binding studies, it appears that PY*LKTK could associate with non-phosphorylated cytoplasmic Stat3 monomers and form Stat3⅐PY*LKTK heterocomplexes that would be incapable of binding to Tyr(P)containing docking sites of receptors for subsequent de novo phosphorylation and activation. Although our data do not distinguish among these and other possible mechanisms by which PY*LKTK may inhibit Stat3 signaling in intact cells, inhibition of IL-6-mediated Stat3 activation demonstrates that PY*LKTK can block de novo Stat3 activation when present prior to stimulation. We also provide evidence that by suppressing Stat3 signaling, PY*LKTK significantly and selectively suppresses Stat3 biological activity, as measured by inhibition of Src but not Ras transformation, consistent with previous studies (22,23) that demonstrate a requirement for constitutive Stat3 activity for cell transformation by Src and not Ras.
Our current findings demonstrate that the Stat3-derived phosphopeptide can block Stat3 activation in vitro and in vivo by interaction with Stat3. We show that direct targeting of Stat3 by a designed peptide can suppress Stat3 signaling and consequent biological function, providing proof-of-principle that inhibitors derived from part of the structure of Stat3 can block its activity and biological effects. Present studies do not address membrane permeability and stability of the Tyr(P) peptides, which are two key challenges for future drug discovery. By the use of a membrane translocation sequence of hydrophobic amino acids (57) linked to the C terminus of PY*LKTK, we overcame one of these problems in tissue culture. Instability to peptidases and phosphatases can be reduced by designing appropriate peptidomimetics. Inhibitors of Stat3 signaling may offer greater selectivity for this pathway and fewer nonspecific side effects than those that block signaling upstream of Stat3, such as tyrosine kinase inhibitors. Our present study sets the stage for the design of Stat3 blockers based on key structural features of Stat3 protein. Mutational analyses of STATs have revealed additional critical residues that are obligatory for their DNA binding and transcriptional activities as well as for cooperative interactions with coactivators (63). These insights can be exploited for direct targeting of Stat3 for further mechanistic studies and therapeutic intervention.