Characterization and Binding Specificity of the Monomeric STAT3-SH2 Domain*

Signal transducers and activators of transcription (STATs) are important mediators of cytokine signal transduction. STAT factors are recruited to phosphotyrosine-containing motifs of activated receptor chains via their SH2 domains. The subsequent tyrosine phosphorylation of the STATs leads to their dissociation from the receptor, dimerization, and translocation to the nucleus. Here we describe the expression, purification, and refolding of the STAT3-SH2 domain. Proper folding of the isolated protein was proven by circular dichroism and fluorescence spectroscopy. The STAT3-SH2 domain undergoes a conformational change upon dimerization. Using an enzyme-linked immunosorbent assay we demonstrate that the monomeric domain binds to specific phosphotyrosine peptides. The specificity of binding to phosphotyrosine peptides was assayed with the tyrosine motif encompassing Tyr705 of STAT3 and with all tyrosine motifs present in the cytoplasmic tail of the signal transducer gp130.

Src homology 2 (SH2) 1 domains are highly conserved regions common to a series of cytoplasmic signaling proteins such as the Src family of tyrosine kinases, phospholipase C-␥, the p85 subunit of phosphatidylinositol 3-kinase, and the STAT family of transcription factors (1,2). These noncatalytic domains target the proteins to specific phosphotyrosyl peptide sequences within their binding partners, thereby regulating a wide range of intracellular signaling events. The specificity of this interaction is determined by the amino acid sequence surrounding the phosphotyrosine on the one hand and by the SH2 domain on the other (3).
Activation of transcription factors of the STAT family has been shown to require the transient association of the STATs with cytokine receptors (4,5). STAT factors interact through their SH2 domains with specific phosphotyrosine motifs within the cytoplasmic parts of the activated receptors. In the case of the activation of STAT1 and STAT3 by interleukin-6, four such tyrosine motifs within the interleukin-6 signal-transducing receptor subunit gp130 have been identified (6,7). Two of these motifs (Y 767 RHQ and Y 814 FKQ) give rise to specific STAT3 activation, whereas two others (Y 905 LPQ and Y 915 MPQ) are able to recruit both STAT1 and STAT3 (6). Subsequent to receptor binding, the STAT factors are phosphorylated on a single tyrosine residue by receptor-associated tyrosine kinases of the Janus kinase family (8 -10). This activation of the STAT factors leads to homo-or heterodimerization and translocation to the nucleus, where they bind to enhancers of interleukin-6inducible genes resulting in the activation of transcription of, e.g. acute phase protein genes (11)(12)(13). The dimerization of STAT factors has also been shown to be mediated by the SH2 domains (9). This has been confirmed recently by x-ray structures of the STAT1 and STAT3 dimers bound to DNA (14,15). In this complex the two SH2 domains form a tunnel that is passed by the two phosphotyrosine-containing tail segments.
Previous experiments have shown that the SH2 domain is also the sole determinant of specific STAT factor activation via gp130 and the interferon-␥ receptor (16,17). The mechanism for the binding of STAT monomers to the phosphotyrosinecontaining recruitment sites of the cytoplasmic region of signaltransducing receptor subunits still needs to be elucidated.
Here we describe the expression, refolding, and structural characterization of the STAT3-SH2 domain as well as its specific binding to phosphotyrosine peptides. Furthermore, we demonstrate that this interaction requires a monomeric domain.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-Biotinylated peptides were synthesized as described earlier (18).
Plasmid Construction-Constructions were carried out using standard procedures (19). For construction of pB-STAT3-BS, the cDNA encoding murine STAT3 (kindly supplied by J. Darnell, Jr., Rockefeller University, New York) was provided with BglII and SalI restriction sites at the 5Ј-and 3Ј-ends, respectively, and cloned into a pBluescript vector (Stratagene, Heidelberg, Germany). The sequence encoding the STAT3-SH2 domain (amino acid residues 582-702) was amplified by polymerase chain reaction, and BamHI and AvrII restriction sites were introduced by the 5Ј-and 3Ј-primers, respectively. The BamHI/AvrII DNA fragment was ligated with a modified pRSet5c vector carrying an adaptor consisting of an amino-terminal MRGS(H) 6 -tag and a BamHI and an AvrII restriction site. The resulting vector pRSetS3SH2 coding for the amino-terminally His-tagged STAT3-SH2 domain was verified by DNA sequence analysis.
Expression, Purification, and Refolding of the Recombinant MRGS(H) 6 -tagged STAT3-SH2 Domain-Escherichia coli strain BL21(DE3)pLysS transformed with the pRSetS3SH2 plasmid was grown at 37°C in LB medium containing chloramphenicol (50 g/ml) and ampicillin (100 g/ml) to an A 595 of 0.6 -0.7. Cells were induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside for 3 h at 37°C and subsequently harvested by centifugation. The bacterial pellet was resuspended in lysis buffer (26 mM Tris-HCl, pH 7.5, 10 mM EDTA, 1 mM * This work was supported by the Deutsche Forschungsgemeinschaft (Bonn) and the Fonds der Chemischen Industrie (Frankfurt/Main). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Immunoblot Analysis-The harvested cell fragments containing the inclusion bodies were resolved by SDS-PAGE and transferred to an Immobilon polyvinylidene difluoride membrane (Millipore, Eschborn, Germany) using a semidry electroblotting apparatus. STAT3-SH2 detection was performed using a monoclonal MRGS His antibody (QIAGEN, Hilden, Germany). A polyclonal goat anti-mouse horseradish peroxidase-conjugated secondary antibody (DAKO, Hamburg, Germany) was used to visualize the immunoreactive bands by Western blot techniques.
Peptide Binding Assay-Peptide binding of the isolated pure STAT3-SH2 domain was performed by the means of an ELISA. A 96-well ELISA Maxisorb plate (NUNC, Roskilde, Denmark) was coated with 2.5 g/ml streptavidin (100 l/well; 16 h at room temperature). Unoccupied binding sites were blocked with 2% bovine serum albumin in phosphate-buffered saline (10 mM Na 2 HPO 4 /KH 2 PO 4 , pH 7.4, 200 mM NaCl, 2.5 mM KCl) (200 l/well; 2 h at room temperature). After four washings with phosphate-buffered saline and 0.02%Tween (200 l/well), the biotinylated peptides were immobilized by incubating the streptavidin surface with 100 l/well of a 500 ng/ml peptide solution in phosphatebuffered saline for 1 h at room temperature. The wells were washed four times with 200 l/well phosphate-buffered saline and 0.02% Tween, and the surface was equilibrated with buffer C or D. STAT3-SH2 solutions were incubated for 1 h at room temperature, and unbound protein was removed by washing four times with the appropriate phosphate buffer. The unoccupied phosphotyrosine residues were detected by incubation with PY20 phosphotyrosine antibody (Transduction Laboratories, Lexington, KY; 1:2,000, 100 l/well) for 45 min at room temperature. Bound PY20 was visualized using a polyclonal goat anti-mouse horseradish peroxidase-conjugated antibody (DAKO; 1:2,000, 100 l/well, 45 min at room temperature). Staining reagent was 0.1 mg/ml 3,3Ј,5,5Ј-tetramethylbenzidine in 0.1 M acetate buffer, pH 5.5, containing 0.003 vol % H 2 O 2 . The reaction was stopped with 2 M sulfuric acid. Inhibition of the PY20/phosphopeptide interaction by STAT3-SH2 was determined by calculating the decrease in absorbance with increasing amounts of STAT3-SH2 relative to a SH2-free sample.
Size Exclusion Chromatography-Size exclusion chromatography was performed on a Bio-Silect TM SEC 125-5 column (Bio-Rad). The column was equilibrated with the refolding buffer C (pH 7.5) or D (pH 5.5), respectively, loaded with 500 l of STAT3-SH2 (70 -80 g/ml), and run at a constant flow rate of 0.7 ml/min. The collected 0.7-ml fractions were resolved on a 15% polyacrylamide gel by SDS-PAGE and visualized by silver staining.
Circular Dichroism Spectroscopy-CD measurements were carried out on an AVIV (Lakewood, NJ) 62DS CD spectrometer, equipped with a temperature control unit, and a Jasco J-600 spectropolarimeter, both calibrated with a 0.1% aqueous solution of D-10-camphorsulfonic acid according to Chen and Yang (20). The spectral band width was 1.5 nm. The time constant ranged between 1 and 4 s and the cell path length between 0.1 and 10 mm.
Fluorescence Spectroscopy-Steady-state fluorescence spectra were recorded on a Spex Fluorolog 211 photon-counting spectrofluorometer (Spex Industries, NY) with a band width of 2.7 nm (excitation monochromator) and 2.2 nm (emission monochromator). The excitation wavelength was 295 nm. The spectra are corrected for changes in lamp intensity and for spectral sensitivity of the emission-monochromator/ photomultiplier system. All fluorescence measurements were carried out at 20°C.
Fluorescence lifetimes and anisotropy decay were measured in the single photon-counting mode with an Edinburgh Instruments Ltd. (U. K.) spectrometer, model 199. The full width at half maximum of the lamp pulse from the hydrogen flashlamp was 1.4 ns. The excitation wavelength was 295 nm and the band width 8 nm. The emitted light was passed through a combination of a UV-transmitting black glass and a cutoff glass filter to create a band pass (WG320, DUG11, Schott, Mainz, Germany). At least 80,000 counts were accumulated in the peak channel of the total fluorescence intensity, I(t). The lamp pulse was recorded with a suspension of Ludox (NEN Life Science Products) at 345 nm. Data handling and the iterative nonlinear least squares fit of the decays were accomplished with a program supplied by Edinburgh Instruments Ltd. Intensity decays (I(t)) were fit to the multiexponential model using Fluorescence anisotropy decays were analyzed by an exponential fit.
The parameters of r(t) are as follows. r is the anisotropy; , the rotational correlation time; r 0 and r ϱ are the limiting anisotropies, r(t)(t 3 0) ϭ r 0 and r(t)(t 3 ϱ) ϭ r ϱ . The quality of the fits was gathered from plots of weighted residuals and from the statistical parameter 2 (21).
Protein Concentrations-Protein concentrations were calculated Proteins of the inclusion bodies were resolved by SDS-PAGE, transferred to an Immobilon membrane, and detected with a monoclonal MRGS His antibody (lane 6). The apparent molecular mass is in good agreement with the calculated mass of 15 kDa. B, reverse phase HPLCpurified STAT3-SH2 domain was refolded by dialysis. For purity determination, the proteins were resolved by SDS-PAGE. Panel 1 shows silver staining of the refolded STAT3-SH2 domain. For quantification, the polyacrylamide gel was stained with Sypro TM Orange reagent and analyzed using a Storm 840 scanner. The quantitative analysis (panel 2) proved the protein to be more than 99% pure (peak a) with the sole detectable impurity being a disulfide bonded STAT3-SH2 dimer (peak b).
from absorption spectra in the range of 240 -320 nm using the method of Waxman et al. (22).

RESULTS
Expression and Purification of Recombinant STAT3-SH2-To obtain sufficient amounts of protein, the amino-terminally His-tagged STAT3-SH2 domain was expressed in E. coli. The recombinant protein was found entirely in inclusion bodies (Fig. 1A). Repeated sonication and centrifugation yielded inclusion bodies containing about 90% STAT3-SH2 protein. 1 liter of medium contained 40 -50 mg of inclusion body proteins. After solubilization of the inclusion bodies in GdnHCl the proteins were separated on a reverse phase HPLC column, and STAT3-SH2-containing peak fractions were lyophilized. The STAT3-SH2 protein proved to be at least 99% pure (Fig.  1B). This procedure yielded 10 -15 mg of pure STAT3-SH2/liter of culture.
Refolding and CD Spectroscopy-The purified protein was dissolved in 6 M GdnHCl, 1 mM EDTA, and 100 mM dithiothreitol and dialyzed for refolding against buffers of pH 7.5, 5.5, and 4.5, respectively. Subsequently the protein samples were characterized by CD spectroscopy. Fig. 2 shows the far UV and near UV CD spectra of the STAT3-SH2 domain at pH 7.5 (solid line) and pH 4.5 (dashed line). Although the far UV CD spectra are remarkably different at the two pH values, they look similar to spectra of other SH2 domains (23,24). Even more pronounced differences were detected between the near UV CD spectra at the two different pH values (Fig. 2B). For instance, at pH 4.5 a distinct band appeared at 292 nm which can be assigned to a tryptophan residue. This effect can be attributed to a local change rather than to a change of the overall fold of the protein.
The reversibility of this structural change was determined by changing the pH of the solution from pH 4.5 to 7.5 and vice versa (data not shown).
To determine the thermal stability of the folded proteins at the different pH values we recorded a series of CD spectra with increasing temperature. At pH 7.5 a melting curve with a midpoint around 43°C was obtained (Fig. 3). At pH 4.5, however, the protein precipitated with increasing temperature (data not shown).
Unfolding by GdnHCl-Because the thermal stability of the protein could only be estimated at pH 7.5, we monitored the GdnHCl-induced unfolding of the protein by fluorescence spec- Fluorescence Spectroscopy-Steady-state fluorescence, fluorescence lifetime, and anisotropy decay measurements were performed at pH 7.5, 5.5, and 4.5. The steady-state fluorescence data of the SH2 domain upon excitation at 295 nm are compiled in Table I. With a pH decrease from 7.5 to 4.5, the maximum of the emission band shifted from 336 to 330. This shift was accompanied by a decrease of the full width at half maximum from 56 to 51 nm. The shift of the emission maximum and the decrease of the full width at half maximum are indicative of an increasingly hydrophobic environment of the sole tryptophan present in the SH2 domain.
The results of the fluorescence lifetime and anisotropy decay measurements are compiled in Tables II and III. The decays of the tryptophan fluorescence can be fitted by a sum of three exponentials with fractional intensities B i and corresponding lifetimes i and lead to calculated mean lifetimes (ϽϾ) of 4.7, 4.1, and 3.7 ns for the pH values of 7.5, 5.5, and 4.5, respectively (Table II). Such a behavior is in good agreement with the blue shift of the emission maximum in Table I ( 3. Thermal stability of the refolded STAT3-SH2 domain at pH 7.5. Thermal stability was determined by CD spectroscopy. The graph shows the ellipticity ⌰ MRW (MRW, mean residue weight) at 215 nm as a function of temperature. different pH values were fitted with one exponential and led to rotational correlation times of ⌽ ϭ 12.4, 6.4, and 6.1 ns at pH values of 7.5, 5.5, and 4.5, respectively (Table III). Rotational correlation times ⌽ can be used to calculate the molecular mass from the equation M r ϭ f ϫ ⌽ (f ϭ 2.6 kDa/ns) for spherical particles on the basis of the Stokes-Einstein relationship (26). The expected rotational correlation time ⌽ for a monomeric SH2 domain is therefore about 5.7 ns assuming a spherical shape. The measured rotational correlation times at acidic pH values are in good agreement with the overall tumbling rate expected for a monomer. The twice as high value found at neutral pH indicates the existence of a dimeric SH2 domain. The reversibility of the monomer/dimer transition was determined by changing the pH from 4.5 to 7.5 and vice versa by dialysis (data not shown).
Size Exclusion Chromatography-Additional evidence for the dimerization was provided by size exclusion chromatography experiments (Fig. 5) using a calibrated column.  (Table IV). As the STAT3-SH2 domain turned out to undergo a pH-dependent dimerization, we investigated the specificity of the interaction with the various phosphopeptide motifs under neutral (dimer) and acidic (monomer) conditions. For the interaction of the peptides with the monomeric SH2 domain, we performed the assay at pH 5.5 to maintain the stability of streptavidin. Fig. 6 shows a schematic representation of the ELISA used. After incubation of the biotinylated phosphopeptides with the streptavidin-coated surface, the immobilized phosphotyrosine residues were detected with the phosphotyrosine antibody PY20 (Fig. 6A). Incubation with increasing amounts of STAT3-SH2 led to a decrease in absorbance because PY20 was unable to recognize the phosphopeptides bound to the SH2 domain (Fig. 6B). The relative decrease in absorbance with increasing  Biotin-␤A-PGSAAPpYLKTKFI pY X Biotin-␤A-QPVRSHVpYSVTGVH amounts of STAT3-SH2 compared with an SH2-free sample was used to determine the inhibition of the PY20/phosphopeptide interaction by STAT3-SH2 which indicates SH2/peptide binding. Fig. 7A shows the inhibition of the PY20/phosphopeptide interaction at pH 5.5. The STAT3-SH2 domain interacts specifically with the four-membrane distal phosphotyrosine motifs of gp130 (pY 767 , pY 814 , pY 905 , and pY 915 ) whereas the peptides pY 683 and pY X as well as the pY 759 motif, known to bind to the SH2 domain of SHP-2, show only weak binding. In addition, the isolated STAT3-SH2 domain binds to the motif pY 705 of STAT3. Interestingly, the binding of the SH2 domain to pY 767 is significantly impaired by a Q/E exchange at the pYϩ3 position (pY Q770E ). This emphasizes the importance of the pYϩ3 position for specific recognition of receptor motifs by STAT3-SH2. Fig. 7B shows the results of the same experiment under neutral buffer conditions (pH 7.5) where the SH2 domain exists as a dimer. Whereas the monomeric SH2 domain is able to distinguish between the different phosphopeptides, the dimeric molecule is not. The phosphopeptides show only low affinity to the dimer. DISCUSSION The SH2 domains of the STAT factors are involved in receptor recognition as well as in their dimerization. Dimerization is induced by tyrosine phosphorylation and is a prerequisite for DNA binding. To investigate the interaction of the SH2 domains with the respective phosphotyrosine motifs on a molecular level we expressed the SH2 domain of STAT3 in E. coli. After purification the protein was refolded. CD spectra of the refolded protein correspond to those observed with other SH2 domains (23,24). Spectral differences were observed at neutral and acidic pH. The changes seen in the far UV are a reflection of limited rearrangements of the overall structure (Fig. 2). The near UV spectrum, for instance, shows a distinct band at 292 nm at pH 4.5 which is not detectable at pH 7.5, indicating a loss in conformational mobility for the side chain of the sole tryptophan. This result correlates well with the corresponding fluorescence spectra where a blue shift of the emission maximum and a decrease of the full width at half maximum are observed with decreasing pH, reflecting a transition of the sole tryptophan from a hydrophilic to a more hydrophobic environment.
The fluorescence anisotropy decay measurements enabled us to assign these spectral differences to a pH-dependent dimerization of the recombinant SH2 domain which exists as a monomer under acidic conditions and as a dimer at neutral pH. The less exposed tryptophan in the monomeric state correlates with a higher stability of the monomer compared with the dimer as revealed by GdnHCl-induced denaturation. Thus, the small structural changes induced by dimerization are accompanied by a destabilization of the molecule.
Taken together, the fluorescence and CD measurements show that dimerization leads to a conformational change in the SH2 domain involving tryptophan 623. In the x-ray structure of the (STAT1) 2 -DNA complex the corresponding tryptophan is located at the surface of the molecule and is accessible to water. The higher B factors of this amino acid residue in the crystal structure are a further indication of its enhanced flexibility in the dimer. For other SH2 domains such as those of Src or Lck kinase it has been shown that the BG loop is attached to the body of the molecule (27,28), whereas it is completely detached in the (STAT3) 2 -DNA and (STAT1) 2 -DNA complexes (14,15). Because the BG loop is also involved in the dimer interface we raise the idea that in the monomeric state this loop resembles the situation seen in other SH2 domains. This would bury the tryptophan (Trp 623 ) within the structure, a fact that we indeed observe for the monomeric SH2 domain. The two conformational states might reflect different modes of SH2/phosphotyrosine peptide interactions in STAT receptor binding and STAT dimer formation.
Recently, heterodimeric complexes of STAT1 with STAT2 or STAT3 prior to cytokine stimulation have been described (29). The ability of the STAT3-SH2 domain to form dimers may reflect such an interaction between unphosphorylated STAT molecules. On the other hand, it cannot be ruled out that the observed dimerization is a property of the isolated domain and that its formation is prevented within the entire  Table IV was determined by monitoring the inhibition of the PY20/peptide interaction with increasing concentrations of STAT3-SH2 at pH 5.5 (A) and pH 7.5 (B).

protein.
To study the interaction of the STAT3-SH2 domain with different phosphotyrosine peptides we established an ELISA based on the competition of a phosphotyrosine monoclonal antibody with the recombinant protein in binding to phosphotyrosine peptide motifs. Whereas the monomeric STAT3-SH2 domain was able to bind to specific phosphotyrosine peptides no such interaction could be observed with the dimeric protein.
We detected a specific interaction between the monomeric STAT3-SH2 domain and the four distal phosphotyrosine motifs present in the cytoplasmic part of the signal transducer gp130. In contrast, the phosphotyrosine peptides corresponding to the two membrane-proximal tyrosine residues did not show a specific interaction (Fig. 7). These results are in good agreement with the previous observation that in transiently transfected COS cells STAT3 is activated only through the four distal tyrosine motifs of gp130 (6,7). Interestingly, we found a Q770E substitution in the Y 767 motif of gp130 to lead to a loss in STAT3-SH2 binding, corroborating the finding that a glutamine residue at the Yϩ3 position is important for STAT3 activation (Fig. 7). Thus, our STAT3-SH2/phosphopeptide interaction studies fully confirmed the results obtained with native STAT3 in COS cells. Furthermore, the phosphotyrosine peptide of STAT3 itself (pY 705 ) showed specific binding to the recombinant STAT3-SH2 domain. A comparison of the affinities would require a common binding mechanism. As deduced from the x-ray structure of the (STAT1) 2 -DNA and (STAT3) 2 -DNA complexes the mode of SH2/peptide binding therein shows fundamental differences to SH2/peptide interactions known so far (2). An expected higher affinity to the phosphotyrosine peptide of STAT3 itself (pY 705 ) compared with receptor motifs was not observed. This is presumably because the complex interaction seen in the x-ray structures cannot be reconstituted in the ELISA.
Thus far, structure/function studies of recombinant STAT-SH2 domains were hampered by the fact that they did not show specific binding to phosphotyrosine peptides. We found that at physiological pH the recombinant STAT3-SH2 domain is forming dimers that do not bind to phosphopeptides. The observation that under acidic conditions, the STAT3-SH2 domain exists as a monomer that specifically binds to phosphotyrosine motifs will enable us to elucidate how STAT factors interact with their receptors.