Characterization of novel costimulatory molecules. A protein of 38-42 kDa from B cell surface is concerned with T cell activation and differentiation.

Optimal activation of T cells often requires signals delivered by the ligation of T cell receptor (TcR) and those resulting from costimulatory interaction between certain T cell surface accessory molecules and their respective counter receptors on antigen presenting cells. The molecular events underlying the co-stimulatory activity are still not understood fully. Here we describe a 38-42-kDa (B3) protein, present on the surface of lipopolysaccharide-activated B cells, which can provide costimulation to resting T cells leading to a predominant release of interleukin (IL)-4 and IL-5 and negligible amounts of IL-2 and interferon-. Binding assay and electron microscopic autoradiography data suggest that this molecule binds T cells, and the same can be competed by unlabeled B3. Characterization experiments point out that B3 shows up as a single prominent peak on reverse phase-high performance liquid chromatography, runs as a single spot in reducing two-dimensional gel electrophoresis, and is a phosphoglycoprotein. The Western analysis indicate that it does not cross-react with antibodies directed against murine ICAM-1, LFA-1α, VCAM-1, HSA, and B7 suggesting the novelty of the protein. The internal amino acid sequence of this molecule suggests that it does not belong to a known category of murine B cell surface molecules.

T helper cell activation is accomplished by recognition of antigen-Ia complex, expressed by antigen presenting cell (APC), 1 via a clonally restricted heterodimeric receptor (TcR) (1)(2)(3). The precise mechanism by which APCs activate T cells is quite complex and not fully understood. The current dogma is that at least two signals are required. The first signal is provided by the occupancy of TcR, which is major histocompatibility complex restricted (1), and the second non-major histocompatibility complex restricted signal (costimulatory signal) is delivered by certain molecules present on the surface of APCs (4 -6). The participation of costimulatory signal in T cell activation is of paramount importance as it results in two potential outcomes, activation or clonal anergy (7,8). The two different outcomes of antigen recognition, by T cells, is first explained by the dual signal model of T cell activation by Bretscher and Cohn (9) and updated recently by Jenkins and Schwartz (10). Since then, efforts of numerous researchers have culminated in the identification of various molecules capable of providing costimulatory signal (11). The list of these second signal generating molecules, however, is still far from complete as reports are rapidly appearing in the literature regarding the possible existence of certain hitherto unknown molecules with costimulatory properties .
To identify additional cell surface-associated molecules that provide costimulatory signals to T cells, we have isolated proteins from lipopolysaccharide (LPS)-activated B cell membrane. When reconstituted into lipid bilayer, at least three proteins (B1, B2, and B3) gave differential levels of costimulatory help to primary T cell activation. The present document presents the results obtained with one of the above proteins with a molecular mass range of 38 -42 kDa (B3) (the data of B1 and B2 have been communicated elsewhere) and its relation to primary T cell activation and differentiation.

Animals
Female inbred BALB/C mice, 8 -10 weeks old, were obtained from the National Institute of Nutrition, Hyderabad, India, and from our conventional conditions and were allowed free access to food and water.

Primary T Cells
CD4 ϩ T cells were prepared from mice spleens as follows. Briefly, a single cell suspension of spleens was prepared in balanced salt solution (pH 7.2). Red cells were lysed using hemolytic Gey's solution. Nonadherent cells, collected by allowing cells to adhere to plastic Petri plates (Nunc, Denmark) at 37°C with 7% CO 2 for 2 h, were treated sequentially with a mixture of anti-Mac2 and anti-Mac3 (45 min on ice), 33D1 (45 min on ice), and anti-Ia d (45 min on ice). The cells were then washed and incubated with two rounds of anti-Lyt-2.2 (Cedarlane, Ontario, Canada) with 45 min each on ice followed by treatment with low toxicity baby rabbit complement. CD4 ϩ T cells were enriched by passing through nylon wool column. The cells were collected after five to six washes with prewarmed RPMI, 10% FCS (37°C) and plated on Petri plates, previously coated with goat anti-mouse IgM, for 1 h at 37°C. The non-adherent cells were used as a source of CD4 ϩ T cells and the purity of such cell population routinely exceeded 98% as estimated by FACScan (Beckton Dickinson) in preparations stained with anti-L3T4. * 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.

Preparation of Resting B Lymphocytes
Resting B cells from mice spleens were prepared as follows. Briefly, a single cell suspension of spleens was prepared in balanced salt solution (pH 7.2). The red blood cells were removed by treatment with Gey's solution. After removing the macrophages by allowing them to adhere twice to plastic surface (1 h each at 37°C and 7% CO 2 ), the cells were treated twice (45 min each on ice) with a mixture of anti-Mac2, anti-Mac3, and 33D1 and a mixture of anti-Thy1, anti-L3T4, and anti-CD8 followed by labeling (30 min at 37°C) with low toxicity baby rabbit complement. The cells obtained were then loaded on a discontinuous Percoll gradient (100, 70, and 50%) and centrifuged at 1600 ϫ g for 30 min at 4°C. The cells collected from 100 -70% interface layer were considered as resting B cells.

Preparation of LPS-activated B Lymphocytes
Activated B cells were prepared as follows. Briefly, a single cell suspension of mice spleens was prepared in balanced salt solution (pH 7.2). The red blood cell were depleted by treatment with hemeolytic Gey's solution. The cells were then plated on plastic Petri plates (Nunc, Denmark) for 2 h at 37°C and 7% CO 2 . The non-adherent cells were treated sequentially on ice for 45 min each with a mixture of anti-Mac2 and anti-Mac3, and a mixture containing anti-Thy1, anti-L3T4, and anti-CD8 antibodies followed by complement mediated killing. The cells were then incubated at a concentration of 4 ϫ 10 7 /10 ml/pertiplate with 10 g/ml LPS (from Salmonella typhosa; Sigma) for 72 h at 37°C and 7% CO 2 . The purity of such cells was over 98% as analyzed by FACscan (Beckton Dickinson).

Isolation of LPS-activated B Cell Surface Proteins
The LPS-activated B cells were harvested and washed three times with PBS (pH. 7.2) and frozen overnight at Ϫ70°C. The cells were thawed and homogenized in the presence of 0.25 M sucrose, 20 mM Tris-HCl (pH 7.4), and 1 mM EDTA along with a protease inhibitor mixture (leupeptin 10 g/ml, aprotinin 10 g/ml, iodoacetamide 10 mM, antipain 10 g/ml, pepstatin 10 g/ml, chymostatin 10 g/ml, and phenylmethylsulfonyl fluoride (1 mM). The nuclear fraction was removed by centrifugation for 10 min at 700 ϫ g at 4°C. The supernatant (S1) was collected and kept aside. The pellet was rehomogenized and spun as above. The supernatant (S2) was collected, and the pellet was discarded. S1 and S2 were mixed and subjected to centrifugation at 1,10,000 ϫ g for 2 h at 4°C. The supernatant was discarded, and the pellet was solubilized in 1% Triton X-100, 20% glycerol, and 20 mM Tris-HCl (pH 7.5) and protease inhibitor mixture (composition as mentioned above) and agitated overnight at 4°C followed by centrifugation at 100,000 ϫ g for 1 h at 4°C. The proteins, from the supernatant, were separated using 10% preparative polyacrylamide gel electrophoresis according to Laemmli (15). After electrophoresis, the protein bands were located by staining a strip of the gel with Coomassie Blue, and the appropriate unstained regions were crushed and eluted with 1% SDS, 100 mM NH 4 HCO 3 ,50 mM Tris-HCl, 0.1 mM EDTA, and 0.15 M NaCl (pH 8.0) at 37°C for 48 h. After filtration and centrifugation to remove polyacrylamide particles, the solution was dialyzed against 0.1% SDS, 10 mM NH 4 HCO 3 , 10 mM Tris-HCl, 0.01 mM EDTA, 50 mM NaCl (pH 8.0) for 24 h at 4°C.

SDS Removal, Estimation, and Partial Protein Renaturation
To the protein solution was added Lubrol Px (a neutral detergent) to effect a final 10% of the detergent concentration, and it was kept at 37°C for 6 h to enable a micelle mixture of SDS-Lubrol Px. This was followed by dialysis against 10 mM Tris-HCl (pH 8.0), 1% Lubrol Px, 0.01 mM EDTA, 50 mM NaCl at 4°C for 96 h. The dialysis buffer was replaced by a fresh one after every 8 h. The extent of SDS removal, from the protein in the dialyzing bag, was estimated with a basic fuschin method (16).

Reconstitution of Proteins into Liposomes
Preparation of Liposomes-These were prepared by a reverse phase evaporation method using DL-␣-phosphatidylcholine, dipalmitoyl, Sigma) in 1:1 chloroform/methanol ratio. The lipid film was left in vacuum for 2 h at room temperature. The film was then hydrated in 10 mM Tris-HCl (pH 8.0), 60 mM NaCl and sonicated in a bath type sonicator (Bransonic, model B 2200 E4, Danbury, CT) to clarity at 4°C for 30 min. The liposomes were sequentially sized through 0.4 and 0.2 m polycarbonate membranes before use.
Protein-Liposome Coupling-A 1 to 500 ratio of SDS-depleted and partially renatured protein sample to liposomes was placed in a dialysis bag and dialyzed for 96 h at 4°C against 10 mM Tris (pH 8.0), 0.01 mM EDTA, and 10 mM NaCl with at least 12 changes of dialysis buffer. The contents of the dialyzing bag were spun down for 2 h at 4°C at 178,000 ϫ g. The supernatant was discarded, and the pellet was dissolved in physiological saline and later passed through a Sephadex G-50 minicolumn according to Fry et al. (17) to remove unliposomized protein.
Density Gradient Centrifugation-2 ml of the above sample was layered on top of a discontinuous gradient of 5-40% sucrose (w/v) in 10 mM Tris-HCl (6.8), 0.15 M NaCl, and 0.1 mM EDTA. The sample was centrifuged overnight at 98,000 ϫ g in a Beckman SW28 rotor at 4°C. 2-ml samples were collected from the 5 and 10% interface, washed in 5 volumes of physiological saline by centrifuging at 178,000 ϫ g for 2 h at 4°C. The pellet was dissolved in physiological saline and stored sterile for not more than 3 months at Ϫ20°C. The protein content, coupled to liposomes, was estimated after lysing with 1% SDS by the BCA method (18). The lipid phosphorous was estimated as per the method of Ames and Dubin (19).
Lymphokine Bioassay-CD4 ϩ T cells were cultured in 24-well plates at a density of 0.25 ϫ 10 6 /well with different stimuli. The culture supernatants were collected after 22 h for lymphokines assay.
Interleukin-2 and -4 Assay-Interleukin-2 and interleukin-4, in culture supernatants, were determined by the induction of HT-2 proliferation as described by Fernandez-Bortan et al. (2). Briefly, HT-2 cells (1 ϫ 10 4 /well) were cultured in 96-well flat bottomed plates, containing RPMI 10% FCS and various concentrations of culture supernatants from control and experimental wells. Since HT-2 cells are responsive to both these lymphokines, therefore, while measuring IL-2 level, the activity of IL-4 was neutralized using anti-murine IL-4 antibody (600 ng/ml). Likewise, for IL-4 assay, the activity of IL-2 was inhibited by adding anti-IL-2 and anti-IL-2 receptor antibodies. The cultures were incubated for 24 h at 37°C/7% CO 2 followed by pulsing with 1 Ci [ 3 H]thymidine during the last 6 h of culture. The cells were harvested, and the incorporated radioactivity was measured by liquid scintillation counting. The lymphokines activity, in terms of units, was derived by extrapolating the standard curve values obtained by using recombinant interleukin-2 and -4 (Genzyme).
Interferon-␥ Assay-Interferon-␥ was assayed by its ability to inhibit the proliferation of WEHI-279 cells. Cells were cultured in 96-well flat bottomed plates with different concentrations of culture supernatants from control as well as experimental wells. The cultures were pulsed with [ 3 H]thymidine and the incorporated radioactivity was measured as above. Recombinant IFN-␥ served as standard to extrapolate the lymphokine activity in terms of units.
Interleukin-5 Assay-The activity of IL-5 was assayed by the induction of proliferation of mouse splenic resting B cells using dextran sulfate as a comitogen as described by Swain (21). Resting B cells were obtained from splenocyte suspension as mentioned above. After washing, 1 ϫ 10 5 cells/well were added into individual wells of 96-well flat bottomed microtiter plate. 100-l aliquots of a range of dilutions of the culture supernatants under test were added in triplicate wells along with 50 g/ml dextran sulfate. Murine recombinant IL-5 added in different dilutions to obtain a standard curve whose specificity was cross-checked with anti-IL-5 (500 ng/ml). The cultures were pulsed with 1 Ci of [ 3 H]thymidine after 72 h and harvested 16 h later. The incorporated radioactivity was determined by liquid scintillation counting. IL-5 activity, expressed in terms of units/ml, was obtained from the standard values.
Northern Blotting of mRNA-CD4 ϩ T cells (5 ϫ 10 6 /ml) were cultured in 24-well plate (Costar, MA) for 8 h at 37°C, 7% CO 2 in the presence of previously immobilized anti-CD3 (10 g/ml), PMA (10 ng/ ml), and B3 (0.01 g/ml) in separate sets of experiments. Thereafter, the cells were harvested and washed repeatedly in cold PBS (pH 7.2) and stored at Ϫ70°C in pellets until RNA extraction was performed.
Gel Electrophoretic and Lectin Gel Binding Analysis-SDS-PAGE was performed in 0.5-mm thick slab gels containing 4% acrylamide in stacking gel and a 10% acrylamide in separating gel, with a buffer system according to Laemmli (15). The lane containing B3 was transferred on to Immobilon-P (Millipore, CA), washed, and visualized by staining with 0.2% aqueous Ponceau S, destained, and its ability to bind 125 I-ConA was tested in conjunction with autoradiography on x-ray film.
Phosphorylation Procedure-The assay mixture (final volume, 0.2 ml) contained 50 mM MOPS/KOH (pH 6.0), [␥-32 P]ATP (0.2 mCi/ml), 0.3 M MgCl 2 , and approximately 75 g protein of LPS-activated B cell membrane lysate. The reaction carried out at room temperature was started by addition of the proteins and stopped by adding 50 l of 50% trichloroacetic acid, 0.3 M unlabeled ATP, and 0.3 M MgCl 2 . All the following operations were carried out at 4°C. The proteins were washed four times. The pellet, extracted by 1 ml of diethylether, was further processed and electrophoresed according to Amory et al. (23) After electrophoresis, the gel was stained with Coomassie Blue, destained, and B3 was located, excised, dried, and exposed to x-ray film.
Iodination and Competitive Binding Assay-125 I-Labeling of B3 was performed using the IODO-bead method with PD-10 column (Pharmacia, Sweden). Iodinated samples were trichloroacetic acid-precipitated and an equivalent fraction containing 2 ng of protein was allowed to bind the CD4 ϩ T cells preactivated with anti-CD3 (10 g/ml) or PMA (10 ng/ml) and/or both as the case may be. The reaction was carried out for 30 min at 37°C in RPMI containing 10% FCS, 0.2% sodium azide, 20 mM HEPES (pH 7.0). For competition binding, a range of (5-100 times the protein concentration) non-radiolabeled dilutions of B3 was used. The reaction mixture was incubated at 4°C for 2 h on an orbital shaker and mixed with a vortex mixer at 15-min intervals. After extensive washing, the cell pellets were subjected to gamma counting (Beckmann).

Electron Microscopy
Negative Staining of Liposomes-A 20-l droplet of aqueous suspension of liposomes was placed on a fresh piece of parafilm, and a sample droplet was picked up by touching a carbon-coated grid to it. After allowing the excess liquid to drain off, the grid was gently dipped in 1% phosphotungstic acid (pH 7.0) and dried by blotting on a filter paper. Grids were observed in a transmission electron microscope (JEOL 1200 EXII, Japan) and representative fields were photographed.
Electron Microscopic Autoradiography-The B3⅐liposome complex as well as goat anti-mouse IgM (Sigma) were 125 I-labeled by IODO-bead method essentially as described in the previous section. The radiolabeled samples were incubated with CD4 ϩ T cells (previously activated at 37°C for 30 min with 10 g/ml anti-CD3) for 30 min at 37°C. After incubation, the cells were washed thrice with PBS to remove unbound material by centrifuging for 10 min at 500 ϫ g at 4°C. To the pellet was added an equal volume of low melting agarose (Sigma) and allowed to gel, and the latter was cut into 1-mm 3 pieces. Trapped cells were fixed in 1% paraformaldehyde, 1% glutaraldehyde in PBS at 4°C for 1 h. After washing with PBS, the cells were post-fixed in 1% osmium tetraoxide for 90 min at 4°C in the dark. After washing with PBS and passing through graded acetone series, the samples were embedded in epoxy resin (Bio-Rad). Ultrathin sections, cut on Ultracut S (Richert-Jung, Austria), were coated in dark with photographic emulsion (Ilford nuclear L4 emulsion) and incubated in the dark for 2-3 weeks in a dessicator at 4°C. The autoradiographs were developed and stained with aqueous uranyl acetate and lead citrate and observed in a transmission electron microsope (JEOL 1200 EXII, Japan).
Western Blotting-Western immunoblots were made from SDS-PAGE (24). Samples were electrophoresed for 90 min with constant amperage of 14 mA in a minigel apparatus (Atto, Japan). The proteins were transferred to Immobilon-P polyvinylidine difluoride membranes (Millipore, MA) using a Bio-Rad electrophoretic transfer unit with 10 mM CAPS buffer (pH 11.0). After washing in PBS, pH. 7.2, for 15 min, the membranes were blocked with 3% nonfat dry milk and 0.2% Tween-20 in PBS for 2 h at 37°C with gentle agitation. After washing in PBS, the membranes were incubated overnight at 4°C with appropriate antibody in PBS-Tween, washed, and incubated with anti-rat horseradish peroxidase-conjugated IgG for 2 h at room temperature on an orbital shaker. Bound antibody was detected with metal ion enhanced diaminobenzidine.
Reverse Phase-HPLC-B3 was located, isolated, and eluted as mentioned earlier in this section. The protein sample was then diluted with 0.5% trifluoroacetic acid and loaded on to a microbore HPLC column (C8) (Aquapore RP 300, Brownlee columns, ABI) and monitored.
Protein Sequencing-B3 isolated from activated B cell membranes, as described above, was purified by 10% SDS-PAGE (15). After the electrophoresis, B3 was blotted on to "ProtBlot" (Applied Biosystems) in CAPS buffer (pH 11.0) followed by staining in 0.2% Ponceau S, 1% acetic acid and washed with PBS. The protein was then digested with trypsin as its N terminus was found to be blocked and a sample equivalent of 2 pmol, from one of the digested fractions, was subjected to internal sequencing up to 15 residues on Applied Biosystem (model 492 A) Procise Sequencer (at The Protein Sequencing Facility, Worcester Foundation for Experimental Biology, Shrewsbury, MA). The repetitive yields of the first 15 amino acids sequenced were found to be at least 90%.
Two-dimensional Electrophoresis-The reagents for this purpose were prepared essentially according to Amory et al. (23) and performed as advocated by Penin et al. (25).

Identification and Partial Characterization of B3-
The membranes of LPS-activated murine splenic B cells, when subjected to SDS-PAGE analysis, revealed about 15 major protein bands when stained with Coomassie Brilliant Blue (Fig. 1a). B3 was localized, crushed, and eluted as described under "Experimental Procedures." The accuracy with which B3 was isolated was checked by rerunning the protein on SDS-PAGE. When stained with Coomassie Brilliant Blue, it demonstrated a single band both under non-reducing (Fig. 1b), reducing (Fig. 1c), and as a single spot in two-dimensional gel electrophoretic (Fig.  1e) conditions. When B3 was subjected to lectin gel binding (Fig. 1f) and phosphorylation (Fig. 1g) assays it was noticed that B3 binds 125 I-ConA and incorporates radiolabeled phosphate indicating the possibility that it is a phosphoglycoprotein. The results of our attempts to localize B3 on the surface of resting B cells demonstrated that it is hardly detectable even when probed by silver stain of SDS-gel possibly indicating that its high expression is induced when pretreated with LPS (Fig. 1d).
B3 Binds to T Cell Surface-B3, before checking for its costimulatory ability, was reconstituted into lipid bilayers. Prior to reconstituting the protein, it was ensured that a fairly homogeneous preparation of liposomes was obtained. Fig. 2a reveals the negative staining of liposomes indicating a near homogeneous preparation of lipid vesicles. In order to demonstrate the protein reconstitution in lipid vesicles, the aid of electron microscopic (EM) autoradiography was employed. Since B3 is a membrane protein, presumably with a hydrophobic stretch, it was expected to be inserted into the lipid bilayer. In order to illustrate the liposome-protein coupling, the recon-stituted protein was iodinated and processed for EM autoradiography. Fig. 2b (arrows) shows the predominant presence of iodinated protein on the vesicle surface suggesting by that each of the liposome has a fairly even distribution of B3. Such 125 I-B3-bearing liposomes when incubated with anti-CD3 activated T cells, and it was observed that B3 was distributed uniformly all along their surface (arrows, Fig. 2c).
That the binding of B3 to T cell surface was specific was demonstrated in experiments with nonspecific control like murine anti-IgM. The data obtained clearly shows that there was no binding of liposome-coupled 125 I-anti-IgM on the T cell surface presumably because T cells do not possess receptors for IgM (Fig. 3a). On the other hand, when liposomized 125 I-anti-IgM was incubated with A20 cells, as a positive control, the radiolabeled antibody was seen to be evenly distributed on the plasma membrane (PM, arrows, Fig. 3b) and also localized in the interior of the cell (arrows, Fig. 3c).
To further confirm the specificity of binding of B3 to T cells, competitive binding assay was performed. When unreconstituted 125 I-B3 was incubated with T cells in the absence of anti-CD3, very less binding (1,250 Ϯ 252 cpm) was noticed. However, the number of receptors for B3 appeared to be upregulated when the T cells were preactivated with anti-CD3 (10,509 Ϯ 2,562 cpm). Further, the binding specificity of B3 to T cells was tested by competing 125 I-B3 with unreconstituted and unlabeled B3 to bind the anti-CD3-activated T cells. The data clearly show that the binding capacity of 125 I-B3 is diminished by unlabeled B3 (Fig. 4, a-g).
Antibodies against Murine LFA-1␣, ICAM-1, HSA-1, B7, and VCAM-1 Do Not Cross-react with B3-It is known that a majority of costimulatory molecules identified thus far are adhesive in nature. In our efforts to rule out the possibility of B3 being a known costimulatory molecule, a Western analysis was performed. The data obtained are depicted in Fig. 5 which demonstrates that B3 did not cross-react with any of the antibodies against murine LFA-1-␣, ICAM-1, HSA, B7, and VCAM-1. LPS-activated B cell membrane lysate was run in appropriate lanes as a positive control for these adhesive molecules.
B3 Costimulates Primary T Cells to Proliferate- Fig. 6a shows that the addition of B3 to the cultures led to an increase in [ 3 H]TdR incorporation of T cells in a dose-dependent manner. A maximum statistically significant (p Ͻ 0.05) proliferation (39,426 Ϯ 4,214 cpm) was noticed at a B3 concentration of 1 g/ml as against the basal value (955 Ϯ 268 cpm) obtained with cells plus anti-CD3 only. When the concentration of B3 was increased, no further amplification in T cell proliferation was observed. Thus, in all the subsequent experiments, only the half-maximal concentration (0.01 g/ml) of B3, as determined by dose-response pattern, was used. Further, when the cultures were stimulated with controls like liposomes, SDS, B3 alone, gel eluate, and LPS, no statistically significant (p Ͻ 0.05) proliferation (Ͻ2,000 cpm) of T cells was noticed (Fig. 6b) pointing out thereby that the activity elicited by B3 was indeed specific. On the other hand, when PMA was added to anti-CD3-activated T cells, as a positive control, a maximum incorporation of [ 3 H]TdR (10,871 Ϯ 2,827 cpm) was observed.
Induction of Secretion of IL-4 and IL-5 by B3-After confirming that B3 was able to enhance the proliferation of primary T helper cells, we next determined whether or not this protein could elicit the secretion of any lymphokine. To verify this, T cells were cultured with anti-CD3 and/or B3, and the supernatants collected after 22 h were tested for IL-2, IL-4, IL-5, and IFN-␥ levels. The data show that only the cultures stimulated with anti-CD3 and B3 produced significant levels of IL-4 and IL-5 and a very poor level of IL-2 and IFN-␥ (Fig. 7). As seen in the proliferative studies, cultures containing T cells and B3 did not bring about any significant secretion of the above lymphokines. The trend obtained with bioassay of lymphokines was further confirmed by Northern analysis, the details of which are highlighted in Fig. 7 (see inset).
Reverse Phase-HPLC and Protein Sequence Analysis-B3, when subjected to reverse phase-HPLC, showed a single prominent peak (at 18.21 min) in the chromatogram (Fig. 8a) demonstrating thereby that this protein is homogeneous in its present form of isolation coinciding with our two-dimensional data. This protein was subjected to internal amino acid sequence upon tryptic digestion (because of its blocked N terminus) up to 15 residues, and the details are depicted in Fig. 8b. DISCUSSION Optimal T cell activation depends not only on the occupancy of TcR, but also on accessory molecules, provided by the APCs, that function in cell-cell adhesion and/or signal transduction (26). During the recent past, several new costimulatory molecules have been identified (12)(13)(14), and the list still appears to be incomplete as the evidence for existence of additional cell surface proteins involved in signal transduction is being raised by several workers (27,28). It may be mentioned here that on the basis of their ability to secrete specific lymphokines, T helper cells have been divided into Th1 and Th2. Th1 cells produce IL-2 and IFN-␥, lymphotoxin, etc., and primarily participate in stimulting the cell-mediated immunity (29,30), while Th2 cells secrete IL-4, IL-5, IL-6, etc., and are involved mainly in the induction of humoral immunity (29,(31)(32)(33)(34). It has been postulated that these two T helper subsets are not only functionally different but also need qualitatively and quantitatively distinct requirements for costimulation (35). However, to date and to the best of our knowledge, no costimulatory molecules are reported which exclusively activate Th2 cells.
In the present study, we describe the biochemical and functional analysis of a novel LPS-activated murine splenic B lymphocyte cell surface-associated costimulatory molecule, provi- Proteins were run on SDS-PAGE (10%) and transferred onto Immobilon-P membrane and stained in 0.2% aqueous Ponceau S to ensure that all lanes contained the transferred protein. After washing and blocking, the membrane was probed with antibodies against ICAM-1, LFA-1␣, HSA, B7, and VCAM-1. As a positive control, LPS-activated B cell membrane sample was run in appropriate lanes, and the bound antibody was detected by diaminobenzidine.
sionally termed B3, that chiefly activates Th2-like cells. Our data suggest that B3 molecule is specifically involved in the costimulation of resting T helper cells upon cross-linking TcR⅐CD3 complex with anti-CD3 monoclonal antibody resulting in predominant secretion of IL-4 and IL-5 and very poor levels of IL-2 and IFN-␥. Our rationale for choosing B cell surface molecules to provide costimulatory signal to T cells lies in the observation that B lymphocytes are major APCs for the clonal expansion of normal murine CD4 ϩ T cells (36). Further, our selection of LPS-activated B cells for identifying the costimulatory molecules is based on the premise that resting B cells are poor APCs (37) and do not constitute costimulatory activity (38,39); only upon treatment with either LPS or IL-1 or immunoglobulin or IFN-␥, or cross-linking surface major histocompatibility complex class-II molecules or neuraminidase etc. (36, 40 -42) do the B cells acquire enhanced ability to stimulate T cells. Moreover, cytokine secretion is induced from naive T cells only when activated B cells are used as APCs (43). It may also be stressed here that the molecule, described in the present study, is bearly detectable on the membranes of resting B cells even when loaded five times the concentration of LPSactivated B cell membrane lysate probed with silver strain.
In biochemical experiments, we have characterized B3 as a single molecule with an approximate molecular mass of 38 -42 kDa when analyzed by SDS-PAGE. Upon reducing, this molecule was recovered as a single sharp band. The reverse phase-HPLC approach to purify this protein clearly showed a solitary and distinct peak (at 18.21 min) in the chromatogram. This information, in conjunction with the SDS-PAGE analysis, conclusively proves that this protein is homogeneous in its present form of isolation. In addition to these aproaches, the twodimensional profile of B3 always consistently yielded a single pattern (in about 12 repetitions) which also reiterates the absence of any other contaminants sticking either specifically or nonspecifically to the said protein. As assessed by its ability to bind 125 I-ConA, B3 appears to be glycosylated a fact which was further substantiated by partial digestion of this protein by N-glycosidase F that resulted in two distinct fragments of 22 and 18 kDa (data not shown). The phosphorylation assay, on the other hand, revealed that this molecule is capable of incorporating radiolabeled phosphate. These results indicate that B3 is a phosphoglycoprotein.
For an effective signal transduction, a costimulatory molecule is expected to bind its counter ligand on the target cell. Studies were undertaken to explore this possibility, and the results obtained indicate that B3 molecule binds to T cells and this binding can be diminished by competing with unlabeled B3. This fact is further strengthened by the results obtained with electron microscopic autoradiographic studies which show that 125 I-labeled reconstituted B3 molecule when incubated with anti-CD3-activated T cells, this protein is found associ- ated with the surface of the T cell. It is interesting to note that the receptors for this molecule, on T cells, appear to be significantly up-regulated when prior activated with anti-CD3. In the light of these observations, it is safer to assume that T cells bear ligands for B3 molecule.
The internal amino acid sequence was obtained by tryptic digestion (as its N-terminal was found to be blocked) of the protein blotted onto the polyvinylidene difluoride membrane. This yielded a single sequence, and at present we only have a partial sequence of 15 amino acids. A data base (non-redundant amalgamation data base of SwissProt, PDB, SP update, PIR, GP update) search of this yielded somewhat surprising results, as it showed a high homology (95%) with pyruvate kinase and is not a part of any known surface proteins. The pyruvate kinase is a well known cytosolic enzyme (49). In a rare event of misidentifying pyruvate kinase as the present protein, is not possible due to the following reasons: 1) this protein is isolated using the well known procedures for isolation of membranes and was washed extensively to remove any nonspecific contaminants; 2) the present protein is found to be heavily glycosylated from the data shown as under "Results," and it is very uncertain for cytosolic protein be heavily glycosylated; and 3) the molecular mass of pyruvate kinase is about 56 -60 kDa (50) while the protein described in the present study has an approximate molecular mass of 38 -42 kDa. These points strongly disfavor the argument that B3 is pyruvate kinase. The fact that B3 showed high homology with pyruvate kinase (PK), prompted us to investigate as to whether B3 possesses any PK-like activity. When B3 was allowed to react with phosphoenol pyruvate, substrate for pyruvate kinase, it did not show any enzymatic activity thereby further demonstrating that B3 and PK are unrelated entities (data not shown).
The only known costimulatory molecule closest to B3, in terms of molecular weight, is B7-2. Like B7 (now called CD80), B7-2 (a 34-kDa protein) is a counterreceptor for CD28 and CTLA-4 T cell surface molecules and induces the predominant secretion of IL-2 (51). In contrast, the molecule described in the present study activates CD4 ϩ Th cells to secrete IL-4 and IL-5 but a very little IL-2 and IFN-␥. These observations lend support to the view that B3 molecule is not a counter ligand for either CD28 or CTLA-4. Also, when anti-CD3-activated T cells incubated with anti-CD28 and were allowed to interact with 125 I-labeled B3, there was no substantial change in the binding capacity of labeled B3 to T cells. Further on, our experiments to identify the receptor for B3 on T cells, using the homobifunctional cross-linker disuccinimidyl suberate, clearly demonstrated B3 binds a 60-kDa protein on T cell surface (data not shown). It may be mentioned here that B7 (CD80) binds a 44-kDa glycoprotein (CD28) on T cell surface.
Thus, all the generated evidence favors the conclusion that B3 is a novel costimulatory molecule that activates Th2-like cells. Using a similar approach, we have recently demonstrated the presence of a 150-kDa protein (M150) from the membranes of thioglycollate-elicited murine peritoneal macrophages that selectively activate Th1 type of cells leading to the secretion of significant levels of IL-2 and IFN-␥ but negligible amounts of IL-4 (52).