The Spectral and Thermodynamic Properties of Staphylococcal Enterotoxin A, E, and Variants Suggest That Structural Modifications Are Important to Control Their Function*

The superantigens staphylococcal enterotoxin A and E (SEA and SEE) can activate a large number of T-cells. SEA and SEE have approximately 80% sequence identity but show some differences in their biological function. Here, the two superantigens and analogues were characterized biophysically. SEE was shown to have a substantially higher thermal stability than SEA. Both SEA and SEE were thermally stabilized by 0.1 mm Zn2+ compared with Zn2+-reduced conditions achieved using 1 mmEDTA or specific replacements that affect Zn2+coordination. The higher stability of SEE was only partly caused by the T-cell receptor (TCR) binding regions, whereas regions in the vicinity of the major histocompatibility complex class II binding sites affected the stability to a greater extent. SEE exhibited a biphasic denaturation between pH 5.0–6.5, influenced by residues in the TCR binding regions. Interestingly, enzyme-linked immunosorbent assay, isoelectric focusing, and circular dichroism analysis indicated that conformational changes had occurred in the SEA/E chimerical constructs relative to SEA and SEE. Thus, it is proposed that the Zn2+binding site is very important for the stability and potency of SEA and SEE, whereas residues in the TCR binding site have a substantial influence on the molecular conformation to control specificity and function.

Superantigens (SAgs) 1 such as the staphylococcal enterotoxins (SE) are very potent T-cell-activating proteins known to cause food poisoning or toxic shock (1). SEs bind as unprocessed proteins to MHC class II molecules and activate T-cells displaying certain V␤ regions of T-cell receptor (TCR) (2,3). Because the number of V␤-genes is limited, a much larger fraction of the T-cells is activated by SAgs than by normal antigens (2,4). The strong cytotoxicity induced by these enterotoxins has been explored for cancer therapy by fusing them to tumor-reactive antibodies (5,6).
Nine different SEs have been identified and these are designated SEA-SEE and SEG-SEJ. The sequence identity of these SAgs with SEA ranges from 20% for SEG to 82% for SEE (described in greater detail in Refs. 7 and 8). The binding to MHC class II of both SEA and SEE is known to be Zn 2ϩ -dependent (9). Both SEA and SEE have two MHC class II binding sites, one close to the N terminus with low affinity for MHC class II and one close to the C terminus with moderate affinity. These two sites may cooperate and cross-link MHC class II molecules with a higher affinity (10,11). In addition the ability to cross-link two neighboring MHC class II molecules may stimulate secretion of inflammatory cytokines, such as interleukin-6, interleukin-8 (12), and interleukin-1␤ (13). One of these interactions is stabilized by a Zn 2ϩ ion, coordinated by His187, His225, and Asp227 in SEA (14) and His81 in the MHC class II ␤-chain (15). Replacing any of the residues Asp-227, His-225, and His-187 with Ala dramatically decreases the affinity for Zn 2ϩ and subsequently MHC class II (10). SEA and SEE behave similarly in many functional assays, for example human T-cell proliferation and MHC class II-dependent cytotoxicity assays (16). However, the V␤ specificity (2,16,17) differs slightly between SEA and SEE, and in contrast to SEE, SEA can induce a strong MHC class II-independent T-cell response. This indicates that presentation on MHC class II is more important for optimal activation of T-cells with SEE (16). As a consequence, these intrinsic differences in the TCR binding sites of SEA and SEE indicate that the affinity of SEA for the TCR may be different compared with SEE.
In this study, the differences in stability between SEA and SEE were studied and compared with functional data. To address differences on the molecular level, engineered chimeras as well as variants with specific replacements were investigated. Exchanging parts of the SEA amino acid sequence for the corresponding sequence from SEE or vice versa induces conformational changes in the tertiary structure that could be important for the T-cell-activating properties. Also, the influence of the Zn 2ϩ binding site, which is important for MHC class II binding, on the structural stability of the SAgs was investigated.

EXPERIMENTAL PROCEDURES
Materials-Horseradish peroxidase-labeled goat anti-human IgG (␥specific), Sigma Fast™ OPD peroxidase substrate tablets, and L-methionine sulfoximine were purchased from Sigma. Purified IgG pool was from Pharmacia & Upjohn AB (Stockholm, Sweden).
Dry milk (fat-free) was from Semper AB (Stockholm, Sweden) and horseradish peroxidase-labeled rabbit anti-SEA antibodies were from Toxin Technology (Sarasota, FL). Na 2 51 CrO 4 was from Amersham Pharmacia Biotech. The restriction enzymes BglII and HindIII were from Roche Molecular Biochemicals and Life Technologies, Inc., respectively, and T4 ligase was from Roche Molecular Biochemicals.
RPMI 1640 cell growth medium and HEPES were from Bio Whittaker (Verviers, Belgium) and Hank's balanced salt solution without phenol red was from Life Technologies, Inc. R10 fetal bovine calf serum * 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.
¶ To whom correspondence should be addressed. Thermal Denaturations-The thermal denaturations were carried out using a HP 8453 UV diode-array spectrophotometer with a heating cell (Hewlett-Packard, Waldbronn, Germany). The spectrum was monitored between 190 and 1100 nm. Before measurement the cuvette was cleansed with 6 M guanidine-HCl and Millipore water. SAg solution was added to the cuvette and diluted with 20 mM phosphate buffer to give a final SAg concentration of 0.1 mg/ml. The temperature was raised in steps of 0.5°C, and at every temperature the sample solution was left to equilibrate before the absorbance measurement. The typical temperature interval was 40 -75°C. The denaturation was visualized by plotting the absorbance difference A 286 :A 264 (18), and the denaturation temperatures were defined as the EC 50 value.
Expression and Purification of SEA/E Chimeras-The different superantigens were expressed constitutively in the Escherichia coli strain UL 635 (xyl-7, ara-14, T4 R , ⌬ompT) using a vector (10,19) with a staphylococci protein A promoter, a signal peptide, and a kanamycin resistance gene. In the Fab-SAg vector (5, 6), expression was controlled by a lacUV5 promoter.
The SAgs (Table I) were expressed using either shaker flasks or fermenters (Belach Bioteknik, Sweden), whereas the C215Fab-SAgs (Table I) were produced in fermenters. When using shaker flasks, bacteria containing the respective production plasmid (Table I) were cultivated for approximately 18 h in 500 ml of 2ϫ YT medium containing 25 g/ml kanamycin and 12 g/ml isopropyl ␤-D-thiogalactoside at 25°C. The cells, containing the SAgs, were then harvested by centrifugation at 5000 ϫ g for 30 min.
Production of the Fab-SAgs in fermenters was carried out essentially as described in Ref. 6. A similar procedure was used for the SAg expression, although no isopropyl ␤-D-thiogalactoside was added. The fermentations were discontinued after 45 h. Metabolic activity was interrupted by cooling the cultivation to 15°C, and 1 mM EDTA was added. SAgs were recovered from the growth medium after fermenter cultivations and from the periplasm after shaker flask cultivations.
The periplasmic extraction was carried out by resuspending the cell pellet in 30 ml of 0.3 M Tris-HCl, 1 mM EDTA, 20% sucrose, pH 7.5, and incubating the suspension for 30 min on ice. After centrifugation for 15 min at 9300 ϫ g the pellet was resuspended in 30 ml of 5 mM MgCl 2 . The cell debris was removed by centrifugation for 15 min at 6300 ϫ g; the supernatant containing the SEA/E chimera was filtered to remove remaining debris, and 3 ml of 10ϫ PBS (0.14 M NaCl, 3 mM KCl, 10 mM phosphate buffer, pH 7.4, and 0.05% Tween 20) were added to the SEA/E solution, which was stored at Ϫ20°C until purification.
The growth medium or periplasmic extract was diluted to approximately 2 millisiemens/cm, adjusted to pH 5.0, and applied on a HiLoad 26/10 ® SP-Sepharose ® column (Amersham Pharmacia Biotech). The sample was eluted with a linear gradient from 20 mM to 500 mM ammonium acetate, pH 6.0, with 0.02% Tween 80 over 30 min and a flow rate of 5 ml/min. To wash the column between samples. 30 ml of 1 M NaOH was used.
The SAg-containing fraction was diluted to approximately 2 millisiemens/cm, adjusted to pH 5.0, and applied on a 1-ml Resource-S column. The sample was eluted with an ammonium acetate buffer gradient from 20 mM to 500 mM, pH 5.0, with 0.02% Tween 80 over 30 min and a flow rate of 1 ml/min. To wash the column between samples, 2 ml of 1 M NaOH was used. The purified SAg was concentrated to at least 0.5 mg/ml using ultrafiltration (Centriprep 10, Amicon Inc., Beverly, MA).
The Fab-SAgs were purified using protein G affinity chromatography and ion-exchange chromatography (6). Typical yields for the SAg chimeras were 5 mg/l in shaker flasks and 50 mg/l or more in fermenters. The yield for the Fab-SAgs in fermenters was typically 50 -400 mg/l.
Analytical Procedures-SDS-polyacrylamide gel electrophoresis was carried out using precast 10 -15% polyacrylamide gradient gels in the Phast system (Amersham Pharmacia Biotech) and isoelectric focusing with gels ranging from pH 3 to 9. For staining, Coomassie Blue was used. The purity of the samples was visually estimated from the SDSpolyacrylamide gel electrophoresis. The protein concentration of the product solutions was determined with a UV-spectrophotometer (Hewlett-Packard) at 280 nm. Mass spectrometry was carried out on a Lasermat 2000 laser desorption/ionization mass spectrometer (Finnigan Mat ltd, Hemel Hempstead, United Kingdom) with a matrix con-sisting of 10 mg/ml sinapinic acid (20) in 30% acetonitrile with 0.1% trifluoroacetic acid.
CD Measurements-Both the far UV and near UV CD spectra were recorded in a 20 mM NaH 2 PO 4 buffer, pH 6.0, using a Jasco J720 (Japan Spectroscopic Co. Ltd., Hachioji City, Japan). The path length of the cuvettes used for far and near UV measurements were 0.1 and 1.0 cm, respectively. The concentration of the protein solutions used for near UV CD was approximately 0.8 mg/ml and for far UV CD it was approximately 0.2 mg/ml (21). Corrections were made in the CD spectra for concentration differences between samples.
Enzyme-linked Immunosorbent Assay Analysis of C215Fab-SAgs-Flat bottomed high binding EIA/RIA 96-well plates (Corning Costar Co., Cambridge, MA) were coated overnight at 4°C using 100 l of 1 g/ml goat anti-mouse -chain in 50 mM NaHCO 3 , pH 9.6. Residual binding sites were blocked using 3% fat-free dry milk in PBS-Tween, 200 l/well for 1 h. To each well 100 l of the respective C215Fab-SAg solution, 0.5 g/ml in PBS-Tween, was added and incubated for 1 h. As a negative control, 100 l of 0.5 g/ml C215Fab was used. The final steps were carried out using either human or rabbit antibodies. Using human antibodies, 100 l of purified IgG pool at different concentrations in 3% fat free dry milk in PBS-Tween was added to the wells and incubated for 2 h, and 100 l of 1.66 g/ml horseradish peroxidaselabeled goat anti human IgG (␥-specific) were added and incubated for 1 h. Alternatively horseradish peroxidase-labeled rabbit anti-SEA at different concentrations, diluted in blocking solution, was added and incubated for 2 h. The plates were developed with the Sigma Fast™ OPD peroxidase substrate tablet as recommended by the supplier. Between each step of the assay the wells were washed four times with PBS-Tween.
T-Cell Proliferation-The growth medium used was RPMI 1640 with 10% fetal calf serum, 50 M 2-mercaptoethanol, and 0.1 mg/ml gentamycin sulfate. Spleen cells from C57B1/6 mice were obtained in house as a suspension in growth medium. The cells, 2 ϫ 10 5 /well as determined by counting in Bü rker chambers using trypan blue viability staining, were incubated in 96-well flat bottomed plates (Nalgene Nunc International, Denmark). The SAgs used were analyzed in triplets in the concentration interval 0.001-100 pM. After incubation for 3-4 days the cells were pulsed with 10 l of Scintillation Proximity-based Binding Assay-Chinese hamster ovary cells transfected with human CD80 and HLA-DR4 were cultured in RPMI 1640, supplemented with 10% (v/v) fetal bovine serum, 0.1 mg/ml gentamycin sulfate, and 1 mM L-methionine sulfoximine.
The affinity was estimated by calculating half-maximal binding (B max 1 ⁄2) from the saturation curve. The C215Fab moiety was used to attach the C215Fab-SEs to the SPA beads coated with anti-mouse antibodies. Fifty microliters of C215Fab-SE (at different concentrations) and 50 l of anti-mouse SPA-polyvinyl toluene beads (40 mg/ml in assay buffer without bovine serum albumin) were mixed for 1 min in a microtiter plate (OptiPlate, Packard, Greve, Denmark), covered with a plastic film, and incubated for 30 min to 2 h at 4°C (depending on the time needed for the preparation of cells). The 3 H-labeled cells were added to the microtiter plate with preincubated C215Fab-SE-SPA beads at a density of 1.5 ϫ 10 4 cells/50 l/well. The plate was sealed by a plastic film, mixed for 1 min on an orbital shaker platform, and incubated in the dark for 8 h at room temperature. The radioactivity in each well was measured during 3 min/well using a ␤-top counter (Packard). As a negative control, C215Fab was used.

RESULTS
Denaturations of SEA and SEE-To investigate the stability of SEA and SEE and understand how it depends on variables such as pH, salt concentration, and various additives, thermal denaturations were carried out using a UV spectrophotometer with a heating cell and a diode array detector. The sample consisted of 0.1 mg/ml SAg dissolved in 20 mM phosphate buffer of varying pH. At denaturation the UV absorbance spectra are changed, because buried residues become exposed. The denaturation was monitored by measuring the difference in absorbance at 286 and 264 nm, because drastic changes in the absorbance difference of these two wavelengths occur upon denaturation (18). After denaturation, the SAgs usually precipitated.
The melting points for SEE were generally 5-15°C higher compared with SEA. Notably, a biphasic denaturation was observed for SEE between pH 5.0 and pH 6.5, but at a higher pH the denaturation was monophasic (Fig. 1). A possible biphasic denaturation was observed for SEA at pH 5.0. At pH 4.0 -5.0 for SEA and 4.0 for SEE, the melting was less distinct suggesting unfolding in a more complex way. At pH 4.0, both SEA and SEE remained soluble after denaturation while precipitating at the other pHs.
Denaturations were also carried out with SEA and SEE at pH 4.0 -9.0 with the addition of 150 mM NaCl, which substantially reduced the stability of both SEA and SEE at lower pH by approximately 6°C, while increasing the thermal stability of SEE at higher pH with 2-5°C (data not shown).
To see whether SEE could obtain its original structure from the partially unfolded state, the temperature gradient was stopped there, and the sample slowly cooled. However, during this procedure most material precipitated.
Denaturations of SEA and SEE in guanidine-HCl monitored by far and near UV CD measurements showed that the tertiary structure unfolds with midpoints at 1.52 M for SEA and 1.96 M for SEE (Fig. 2). The secondary structure for SEA and SEE is lost at 5.10 and 4.14 M, respectively (Fig. 2). Thus, at pH 6.0 the tertiary structure of SEE is more resistant than SEA to chemical denaturants, whereas the opposite is true for the secondary structure.
In conclusion SEE has a substantially more stable structure than SEA, and the melting points differ up to 15°C. At a lower pH the denaturation of SEA and SEE was less distinct.
Zn 2ϩ Dependence of the Thermal Stability of SEA and SEE-To investigate the importance of Zn 2ϩ binding for thermal stability of SEA and SEE, denaturations were made with buffers containing 1 mM EDTA to bind Zn 2ϩ ions or buffers containing 0.1 mM ZnCl 2 to saturate the SAgs with Zn 2ϩ . Compared with conditions with no Zn 2ϩ , accomplished by the addition of 1 mM EDTA, the increase in melting points for SEA with Zn 2ϩ addition was up to 6°C (Fig. 1). The stabilizing effect of Zn 2ϩ for SEA was larger at pH 6.0 compared with pH 7.0, but for SEE (Fig. 1) it was almost independent of pH at pH 5.0, 6.0, and 7.0. The melting points of SEA with the addition of 0, 0.01, and 0.1 mM ZnCl 2 increased almost linearly from 59.5 to 61.4°C but no detectable increase in melting point occurred with the addition of 1.0 mM ZnCl 2 compared with 0.1 mM (data not shown) indicating Zn 2ϩ saturation of the SAgs at 0.1 m ZnCl 2 . In contrast 0.1 mM MgCl 2 or 0.1 mM CaCl 2 at pH 6.0 stabilized neither the SEA nor the SEE structure (data not shown).
With the Zn 2ϩ addition, SEE exhibited a biphasic denaturation at pH 5.0 and 6.0, as did SEA at pH 5.0 (Fig. 1). Interestingly, the denaturation curve of SEA at pH 5.0 was much less distinct than at pH 6.0 and 7.0, and no precipitate was formed. Unlike SEA, the EDTA-treated SEE showed a substantially less distinct denaturation behavior at pH 7.0 (Fig. 1).
Denaturations were also made on the mutants SEA D227A ,  Table I). The residues Asp-227 and His-187 (10) are essential for coordination of Zn 2ϩ in the MHC class II ␤-chain binding site. The residue Phe-47 is located in the MHC class II ␣-chain binding site, which is unlikely to be metal ion-dependent. Similarly Ser1 may be involved in the Zn 2ϩ coordination (22) according to one crystal structure.
The replacement of the Zn 2ϩ binding residues strongly affected the thermal stability of the SAgs. The melting point of SEA D227A was 1.8°C lower than SEA at pH 6.0 and 6.3°C at  (Table I) were generally lower than those for wild type SAgs, although the difference was larger at pH 7.0 compared with pH 6.0.  regions a, f, a and a ϩ h (Fig. 5), respectively, from SEA, whereas SEA/E-bdeg is SEA with the regions b ϩ d ϩ e ϩ g (Fig. 5) from SEE. In the molecular mass column the theoretical masses determined from the amino acid sequence (Fig. 5) are displayed. The EC 50 values in the T-cell proliferation assay are displayed as related to the EC 50 values for SEA. The thermal denaturations were carried out in buffers containing 0.1 mM ZnCl 2 . When calculating the charge difference (dCharge) between the SAgs the amino acids Lys and Arg were regarded as positive, whereas Glu and Asp were regarded as negative. The other amino acids were regarded as neutral. n/a, not analyzed. n/a n/a n/a SEA H187A 62.0 57.5 7.3 0 n/a n/a n/a SEE/A-a D227A 69.0 69.3 6.9 0 n/a n/a n/a SEE D227A 70.1 67.4 n/a n/a n/a n/a n/a

FIG. 2. Guanidine denaturations of SEA and SEE.
Denaturation of SEA (q) and SEE (E) using guanidine followed by CD at 280 nm (left) and 220 nm (right). Background signal subtracted. pH 7.0 (Fig. 3). For SEA H187A the melting points at pH 6.0 and 7.0 were lowered 0.5 and 3.9°C, respectively. The stability decrease for SEE D227A relative to SEE was 2°C at pH 6.0 and 4°C at pH 7.0 (Fig. 3). SEE/A-a D227A had melting points 5.5 and 6°C lower than SEE/A-a at pH 6.0 and 7.0, respectively (Fig. 3). SEA D227A, F47A was slightly more stable than SEA D227A but still less stable than SEA (Fig. 3). Des(1-5)SEA denaturated at approximately 56°C at pH 7.0, a thermal stability decrease relative to SEA of 5.5°C (Fig. 3). Interestingly the D227A/H187A mutants with substantially lower affinity than SEAwt for Zn 2ϩ (10) were also to some extent stabilized by the addition of Zn 2ϩ (data not shown).
In conclusion, the addition of Zn 2ϩ to SEA and SEE increased the stability compared with Zn 2ϩ -free conditions. The substitution H187A in SEA and D227A in SEA, SEE, and SEE/A-a had a clear destabilizing effect, indicating that these residues are very important for the stability of the SAgs. The destabilizing effect was more pronounced at pH 7.0 than at pH 6.0. The substitution F47A in SEA stabilized the structure at both pH 6.0 and 7.0. Because of the substantial influence of Zn 2ϩ on the stability of the SAgs, all denaturations were henceforth carried out in the presence of 0.1 mM ZnCl 2 .
Characterization of the SEA/E Chimeras-SEA and SEE have a sequence identity of approximately 80%. The main differences are found in eight different regions, four close to the two MHC class II binding sites and four in the vicinity of the TCR binding site (16) (Fig. 4) (Table I) were prepared to study the impact of replacement of these regions. Region f in SEE contains two unique histidine residues, which might influence the biphasic denaturation behavior of SEE (Fig. 1), and regions a and h are in the vicinity of region f. In the chimera SEA/E-bdeg the influence of the whole TCR binding site on the denaturation behavior of SEA and SEE was investigated.

. The chimeras SEE/A-a, SEE/A-f, SEE/A-h, SEE/A-ah, and SEA/E-bdeg
The purification of the SAgs was carried out in two ionexchange chromatography steps, whereas the Fab-SAgs were recovered using a protein G affinity chromatography step followed by ion-exchange chromatography (6). The purity of the products was determined to be at least 90% (data not shown) using SDS-polyacrylamide gel electrophoresis and isoelectric focusing. The identities were confirmed using matrix-assisted laser desorption/ionization-mass spectroscopy (Table I).
Notably, the isoelectric points differed substantially between the chimeras and the wild type SAgs (Table I). SEE/A-ah and SEE/A-h, with a positive charge difference of one or two relative to SEE, had isoelectric points at 8.0 or 8.2 compared with 7.4 for SEE, which suggests some conformational differences between the SAgs. Therefore CD spectra were acquired for SEA, SEE, SEE/E-f, and SEE/A-a. The far-UV CD spectra, 200 -250 nm, derived primarily from the secondary structure of the proteins (23), were indistinguishable for all SAgs (Fig. 5). The near UV CD spectra, primarily arising from the tertiary structure in the vicinity of the aromatic amino acid residues Tyr and Trp (23), showed larger differences (Fig. 5) indicating subtle conformational differences in the tertiary structure between the SAgs. Further evidence for structural differences was obtained by enzyme-linked immunosorbent assay analysis of SEA, SEE, and the chimeras, which showed that the binding of antibodies to SEE was substantially less than for SEA (data not shown). Interestingly the chimera SEE/A-f, although being sequentially more SEA-like than SEE, produced a lower response against rabbit anti-SEA than SEE. Similarly, SEE/A-a produced a lower or equal response than SEE against a human IgG pool. To

study whether regions that contain the TCR binding site influence the unfolding properties of SEA and SEE, thermal denaturations of SEE/A-a, SEE/A-f, SEE/A-h, SEE/A-ah, and
SEA/E-bdeg were studied at pH 6.0 -7.0 ( Table I). The regions a, f, and h contain putative TCR-binding residues (16). SEA/ E-bdeg has the two MHC class II binding regions from SEE and the TCR binding region from SEA. To get an excess of Zn 2ϩ , avoiding differences in this important parameter, 0.1 mM ZnCl 2 was added. The results of the thermal denaturation of these molecules are shown in Fig. 6 and Table I. The melting points for the chimeras were all lower than for SEE, with the exception of SEE/A-a at pH 6.0 and 7.0 and SEE/A-h at pH 6.0 but substantially higher compared with SEA (Table I). Notably SEE/A-a and SEE/A-h, at pH 6.0, had higher melting points than SEE (Table I), whereas SEE/A-ah had a lower melting point, indicating that a combination of these two stabilizing regions reversed this effect. Interestingly the chimeras where region h had been replaced, SEE/A-h and SEE/A-ah, were significantly stabilized by the lower pH contrasting the other chimeras.
The melting points of SEE/A-a and SEE/A-f increased from pH 6.0 to 7.0 with approximately 1°C for SEE/A-a and 2°C for SEE/A-f (Table I). For the chimeras SEE/A-h and SEE/A-ah the melting points decreased from pH 6.0 to 7.0 with approximately 11°C for SEE/A-h and 3°C for SEE/A-ah. The melting point for SEA/E-bdeg was approximately 68 -69°C at both pH 6.0 and 7.0. The differences in melting points between pH 7.0 and 6.0 could be caused by residues, which are important for the structural stability of the protein in the present conformation, becoming protonated between pH 7.0 and 6.0 and thereby acquiring a charge, which might change the stability of the structure. Biphasic denaturations were observed for SEA/Ebdeg at pH 6.0 and by SEE/A-h at pH 7.0. The denaturation behavior of SEE/A-h at pH 6.0 seems to be more complex with the SAg not precipitating at denaturation (Fig. 6). SEE/A-a, -f, and -ah showed no biphasic denaturation behavior.
Interestingly, SEA/E-bdeg composed of SEA with the two MHC class II binding sites from SEE was substantially more stable than SEA, indicating that several different regions contribute to the greater stability of SEE. Of the chimeras, SEE/ A-h and SEA/E-bdeg showed biphasic denaturations.
In conclusion, most likely there are conformational differ- ences between the SEA/E chimeras indicated by isoelectric focusing and near UV CD spectra as well as the low antibody binding properties of SEE/A-a and SEE/A-f. The thermal stability of all the chimeras was higher than for SEA with the highest melting points obtained for SEE/A-a and SEE/A-h, but a combination of these two chimeras, SEE/A-ah, was substantially less stable.
Binding of SE Fusion Proteins to MHC Class II-expressing Cells-The interaction between C215Fab-SEs and human MHC class II (HLA-DR4) presented on cells was studied using a binding assay based on the SPA technology. To detect the binding of SEs to HLA-DR4, the C215Fab-SEs was attached to SPA beads coated with anti-mouse Ig antibodies via the Fab moiety, whereas the cells expressing HLA-DR4 were labeled by the incorporation of [ 3 H]leucine. The concentration yielding half-maximal binding (B max 1 ⁄2) was used as an estimate of the binding affinity. B max 1 ⁄2 concentrations for the different C215Fab-SEs analyzed (Fig. 7) were found to be approximately in the same range (2 ϫ 10 Ϫ8 M) (data not shown). This indicates a similar affinity for HLA-DR4, for SEA, SEE, and the chimeras. A striking difference, however, in the saturation amplitude was noticed for the C215Fab-SEs (Fig. 7). The highest amplitude was seen for  (Table I) as determined by EC 50 values. Using human T-cells (16), SEE/A-f was less potent, whereas the proliferating activity for SEA was similar to the other SAgs. DISCUSSION Several studies have speculated in the importance of conformational changes for superantigen function (24,25). In this study we have investigated the structural properties of several SEA and SEE variants. The results clearly support the previous hypotheses. Although the amino acid sequences of SEA and SEE are very similar, there are differences in biological function. The V␤-specificities for SEA and SEE differ (2, 17) as do their affinities to different MHC class II alleles (9), and SEA may also have a different affinity for the TCR than SEE (16). Interestingly, in many cases chimerical molecules of SEA and SEE acquire properties that are unique and not the predicted combinations between SEA and SEE (Figs. 5-7). Replacing Phe-47 with Ala reduces the number of V␤s that can be activated (25). This could be caused by the inability of this, more structurally stable variant (Fig. 3) to undergo a necessary conformational change.
These findings can be explained by the subtle differences in the tertiary structure between the SAgs (Fig. 5). The MHC class II binding regions in SEE are mostly responsible for this stabilization, whereas the TCR binding sites seem more flexible. Notably, the tertiary structure of SEE was more stable than that of SEA, although the opposite was true concerning the secondary structure (Fig. 2). Thus, it is likely that some side chains primarily contribute to tertiary interactions, e.g. by burying hydrophobic surface area in the core of the protein whereas others mainly stabilize secondary structure, and that SEE and SEA differ in both types of side chains. Interestingly, both guanidine-induced and thermal (data not shown) denaturations of SEA and SEE monitored using CD showed that the tertiary structure was unfolded before the secondary structure. Tertiary structure is generally lost before secondary structure when the two levels of structure do not disappear concomi-tantly, as in e.g. the molten globule, and indicates that the transition observed when performing UV monitored thermal denaturations is mainly an unfolding of tertiary structure.
The regions a, f, and h are important for the conformation of SEE, as suggested by isoelectric focusing, CD, antibody recognition, and MHC class II binding assays ( Fig. 5 and 7, Table I). We therefore propose that the TCR binding site in SEs can adopt different conformations that are controlled by subtle differences in this region. This is supported by observed differences in V␤-specificity between SEA and SEE (2,17) and the SEA/E chimeras (16). Thus, several of the residues in the TCR binding site have important functions in controlling the structure, such as the ability for structural modification upon receptor binding and affinity and specificity. Interestingly, these regions may also influence the stochiometry in binding to MHC class II (Fig. 7), perhaps by having structures that facilitate or hinder binding via the low affinity binding sites.
At certain pH values a biphasic denaturation occurs indicating a partial unfolding in the SAgs. SEE/A-f and SEE/A-ah showed no biphasic denaturation, whereas SEE/A-a, SEE/A-h, and SEA/E-bdeg did. Because the biphasic melting is only observed at pH 5.0 and 6.0 but not at pH 7.0, it was hypothesized that it involved His residues getting ionized. Notably SEE/A-f, with two His residues replaced, lacks a biphasic melting suggesting a role for His-161 or His-164 in this local destabilization. Thus, at least one of these two His residues may form parts of the flexible elements that control the structure of the TCR binding site. This hypothesis is further supported by the similar T-cell-proliferating properties of SEE/A-f and SEA. However, there is a biphasic denaturation with a larger absorbance difference in SEA/E-bdeg, which also lacks the two His-161 and His-164, indicating more complete unfolding or that another region is denatured. Therefore, biphasic melting occurs with certain combinations of residues that destabilize the structure.
Zn 2ϩ is important for the MHC class II affinity of both SEA and SEE (9), as well as the reduction in monokine release triggered by SEA and SEE, but Zn 2ϩ does not influence the V␤-specific T-cell stimulation (26). Loss of Zn 2ϩ coordination significantly lowers the potency of SEA (27). From the data presented here, it is clear that Zn 2ϩ also stabilizes the structures of SEA and SEE, especially at physiological pH. Coordination of metal ions often leads to stabilized structures and can increase melting points with approximately 10°C including both functional Mg 2ϩ (28) and engineered Zn 2ϩ binding sites (29,30). The importance of the Zn 2ϩ binding region for the thermal stability of SEA and SEE was further shown by the dramatic decrease in melting point at pH 7.0 for the SEA Zn 2ϩ binding site mutants, such as SEA D227A . Although having a reduced Zn 2ϩ binding, the variants were all stabilized by the addition of a high concentration of Zn 2ϩ indicating that they bind Zn 2ϩ through the known site but with a significantly reduced affinity. Alternatively, an unknown Zn 2ϩ binding site could exist (24). Because SEA D227A was less stable than EDTAtreated SEA, this replacement destabilizes or affects the structure of this MHC class II binding site. This suggests that the very low affinity of this variant for MHC class II is caused by two independent mechanisms, removal of the coordinated Zn 2ϩ ion and structural disturbance. In contrast SEA H187A is more stable than SEA D227A and has a much higher MHC class II affinity and activity (10,11), indicating that here the structure is less affected. In the SEA crystal structure Asp-227 is less exposed compared with His-187 (22).
In conclusion residues in the Zn 2ϩ binding site are very important for the stability and potency of SEA and SEE, whereas residues in the TCR binding site have a substantial influence on the molecular conformation, which may control specificity and function. Our findings will further guide us to understand how bacterial superantigens have evolved and how their potent T-cell stimulatory capacity is maintained on the molecular level. However, these finding will also help us to design superantigens that could have clinical benefits, such as cancer therapy (5,31).