Proc. Nati. Acad. Sci. USA Vol. 87, pp. 5031-5035, July 1990 Immunology
T-cell antigen CD28 mediates adhesion with B cells by interacting with activation antigen B7/BB-1 PETER S. LINSLEY*, EDWARD A. CLARKt, *Oncogen, 3005 First Avenue, Seattle, WA 98121; and
AND
JEFFREY A. LEDBETTER*
tDepartment of Microbiology, University of Washington, Seattle, WA 98195
Communicated by Seymour J. Klebanoff, March 30, 1990
intercellular adhesion mediated by major histocompatibility complex (MHC) class I (13) and class II (14) molecules with the CD8 and CD4 accessory molecules, respectively. We have expressed the CD28 antigen to high levels in Chinese hamster ovary (CHO) cells and have used these transfected cells to develop a CD28-mediated cell adhesion assay. By using this assay as a screening method, we have demonstrated an interaction between CD28 and a natural ligand expressed on activated B lymphocytes, the B7/BB-1 antigen.
Studies using monoclonal antibodies (mAbs) ABSTRACT have implicated the homodimeric glycoprotein CD28 as an important regulator of human T-cell activation, in part by posttranscriptional control of cytokine mRNA levels. Although the CD28 antigen has functional and structural characteristics of a receptor, a natural ligand for this molecule has not been identified. Here we show that the CD28 antigen, expressed in Chinese hamster ovary (CHO) cells, mediated specific intercellular adhesion with human lymphoblastoid and leukemic B-cell lines and with activated primary murine B cells. CD28mediated adhesion was not, dependant upon divalent cations. Several mAbs were identified that inhibited CD28-mediated adhesion, including mAb BB-1 against the B-cell activation antigen B7/BB-1 and some mAbs against major histocompatibility complex class I antigens. B7/BB-1 expression correlated closely with CD28-mediated adhesion, but class I expression did not. Transfected COS cells expressing the B7/BB-1 antigen adhered to CD28+ CHO cells; this adhesion was blocked by mAbs to CD28 and B7/BB-1. The specific recognition by CD28 of the B-cell activation antigen B7/BB-1 represents a heterophilic interaction between members of the immunoglobulin superfamily that may serve to regulate T-cell cytokine levels at sites of B-cell activation.
MATERIALS AND METHODS
The generation of a T-lymphocyte immune response is a complex process involving cell-cell interactions (1) and production of soluble immune mediators (cytokines or lymphokines) (2). This response is regulated by several T-cell surface molecules, including the T-cell receptor complex (3) and other "accessory" surface molecules (1). One such accessory molecule is the CD28 antigen, a homodimeric glycoprotein of the immunoglobulin superfamily (4) found on most mature human T cells (5). Current evidence suggests that this molecule functions in an alternative T-cell activation pathway distinct from that initiated by the T-cell receptor complex (for review, see ref. 6). Monoclonal antibodies (mAbs) to CD28 can augment T-cell responses initiated by various polyclonal stimuli (5-7). These stimulatory effects may result from mAb-induced cytokine production (8) as a consequence of increased mRNA stabilization (9). Anti-CD28 mAbs can also have inhibitory effects; i.e., they can block autologous mixed lymphocyte reactions (10) and activation of antigen-specific T-cell clones (11). The in vivo function of CD28 is not known, although its structure (4) suggests that like other members of the immunoglobulin superfamily (12) it might function as a receptor. CD28 could conceivably function as a cytokine receptor, although this seems unlikely since it shares no homology with other lymphokine or cytokine receptors (4). Alternatively, CD28 might be a receptor that mediates cell-cell contact. In this paper, we describe experiments designed to test this possibility. For this purpose, we have taken an approach based on experiments used to demonstrate
mAbs. mAb 9.3 (anti-CD28, ref. 15) was purified from ascites fluid before use. A number of mAbs to B-cell associated antigens were tested for their abilities to inhibit CD28mediated adhesion. mAbs 60.3 (CD18); iF5 (CD20); G29-5 (CD21); G28-7, HD39, and HD6 (CD22); HD50 (CD23); KB61 (CD32); G28-1 (CD37); G28-10 (CD39); G28-5 (CD40); HERMES1 (CD44); 9.4 (CD45); LB-2 (CD54); and 72F3 (CD71) have been described and characterized (16-18). These mAbs were purified before use. &TA4-1 (ref. 19, antiIgD); 2C3 (ref. 20, anti-IgM); Nambl, H1DE, p10.1, and W6/32 (refs. 20 and 21, anti-human class I); and HBiOa (ref. 20, anti-MHC class II) were also purified before use. mAbs B43 (CD19); BL-40 (CD72); AD2, 1E9.28.1, and 7G2.2.11 (CD73); EBU-141, LN1 (CDw75); CRIS-1 (CD76); 424/4A11 and 424/3D9 (CD77); Leu 21, Ba, 1588, LO-panB-1, FN1, and FN4 (CDw78); and M9, G28-10, HuLymlO, 2-7, F2B2.6, 121, L26, HD77, NU-B1, BLAST-1, BB-1, anti-BL7, antiHC2, and L23 were used as coded samples provided to participants in the Fourth International Conference on Human Leukocyte Differentiation Antigens (September 1989, Vienna) (22). Most of these were used in ascites fluid form. mAbs BB-1 and LB-1 (23) were also purified from ascites fluid before use. Anti-integrin receptor mAbs, P3E3, P4C2, and P4G9 (24), were used as hybridoma culture supernatants. Plasmids and Transfections. cDNA clones encoding the T-cell antigens CD4, CD5, and CD28 (4) in the expression vector TH3M (4) were kindly provided by Sandro Aruffo and Brian Seed (Massachusetts General Hospital, Boston). An expressible cDNA clone encoding the B7/BB-1 antigen (25) was provided by Gordon Freeman (Dana-Farber Cancer Institute, Boston). Dihydrofolate reductase-deficient CHO cells were cotransfected as described (26) with a mixture of plasmids, irH3M-CD28 (4) and pSV2dhfr (27). COS cells were transfected with B7/BB-1, CD4, or CD5 cDNAs in the presence of DEAE-dextran using a protocol supplied by B. Seed and A. Aruffo. CD28-Mediated Adhesion Assay. Cells to be tested for adhesion were labeled with 51Cr, washed, and preincubated in complete RPMI medium [RPMI 1640 medium containing 10% (vol/vol) fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 ,g/ml)] containing 10 mM EDTA
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: mAb, monoclonal antibody; MHC, major histocompatibility complex; FITC, fluorescein isothiocyanate; LPS, lipopolysaccharide. 5031
5032
Proc. Natl. Acad. Sci. USA 87 (1990)
Immunology: Linsley et al.
unless otherwise indicated. mAbs to be tested for adhesion inhibition were then added to 10 tig/ml, and cells were incubated for -1 hr at 230C. In some experiments, a mouse mAb having irrelevant specificity was added to the labeling and preincubation reaction mixtures to saturate Fc receptors. Labeled cells (1-10 x 106 cells per well in 0.2 ml of complete RPMI medium containing EDTA and mAbs, where indicated) were then added to the CHO monolayers. When adhesion inhibition by mAb 9.3 was measured, the mAb or its F(ab')2 were added to the CHO cells 1 hr prior to addition of labeled cells. Adhesion was initiated by centrifugation in a plate carrier and continued at 370C for 1 hr. Monolayers were then washed five times with ice-cold complete RPMI medium and solubilized by addition of 0.5 M NaOH, and radioactivity was measured in a y counter. Numbers of bound cells were calculated by dividing total bound radioactivity (cpm) by the specific activity (cpm per cell) of labeled cells. When COS cells were used, results are expressed as cpm bound since their viability at the end of the experiment was generally 70% by mAb 9.3) was clearly observed with T51, 5E1, Raji, and Jijoye cells. Daudi cells also showed specific adhesion, although to a lesser extent. Other cell lines did not show specific CD28-mediated adhesion, although some (e.g., Namalwa) showed relatively high nonspecific adhesion. Primary mouse splenic B cells did not show CD28-mediated adhesion but acquired the ability to adhere after activation with LPS. In other experiments, six additional lymphoblastoid lines showed CD28-mediated adhesion, but the U937 cell line, cryopreserved human tonsil B cells, and phytohemagglutinin-stimulated T cells did not show adhesion. These experiments indicate that a CD28 ligand is found on the cell surface of activated B cells of human or mouse origin. CD28-Mediated Adhesion Is Specifically Blocked by BB-1, a mAb to a B-Cell Activation Antigen. In initial attempts to define B-cell molecules involved in CD28-mediated adhesion, we measured adhesion by lymphoblastoid cell lines with mutations in other known cellular adhesion molecules. The 616 cells (MHC class II deficient, ref. 28) bound to CD28' CHO cells equally well or better than parental T51 cells. Likewise, a CD18-deficient cell line derived from a patient Er~I T51 1 A2
Raji
Jijoye No Mb _~~~~
_
Daudi
=3 MAbi.
HL60
T51 CEM Jurkat
0
10
i
I--q
20
HSB2
30
5(0
40
with leukocyte-adhesion deficiency (Gambaro cells, ref. 29) also adhered specifically to CD28. Thus, MHC class II and CD18 molecules do not mediate adhesion to CD28. We then tested a panel of mAbs to B-cell surface antigens for their ability to inhibit CD28-mediated adhesion of T51 cells. For these experiments, we tested a total of 57 mAbs reactive with T51 cells. Initial screening experiments were carried out in the absence of EDTA, and any mAbs that blocked adhesion were subsequently retested in the presence and absence of EDTA. Of these mAbs, only those directed against MHC class I molecules (Nambl, H1DE, P10.1, and W6/32) and one to an unclustered B-cell antigen [BB-1, originally described as a B-cell activation marker (23)] were consistently able to block CD28-mediated adhesion by >30%. The dose dependence of adhesion inhibition by the anticlass I mAb H1DE, by BB-1, and by 9.3 were compared in the presence of EDTA in the experiment shown in Fig. 4. mAb 9.3 was most effective at blocking, but mAb BB-1 was able to block =60% of adhesion at concentrations
T-cell antigen CD28 mediates adhesion with B cells by interacting with activation antigen B7/BB-1 PETER S. LINSLEY*, EDWARD A. CLARKt, *Oncogen, 3005 First Avenue, Seattle, WA 98121; and
AND
JEFFREY A. LEDBETTER*
tDepartment of Microbiology, University of Washington, Seattle, WA 98195
Communicated by Seymour J. Klebanoff, March 30, 1990
intercellular adhesion mediated by major histocompatibility complex (MHC) class I (13) and class II (14) molecules with the CD8 and CD4 accessory molecules, respectively. We have expressed the CD28 antigen to high levels in Chinese hamster ovary (CHO) cells and have used these transfected cells to develop a CD28-mediated cell adhesion assay. By using this assay as a screening method, we have demonstrated an interaction between CD28 and a natural ligand expressed on activated B lymphocytes, the B7/BB-1 antigen.
Studies using monoclonal antibodies (mAbs) ABSTRACT have implicated the homodimeric glycoprotein CD28 as an important regulator of human T-cell activation, in part by posttranscriptional control of cytokine mRNA levels. Although the CD28 antigen has functional and structural characteristics of a receptor, a natural ligand for this molecule has not been identified. Here we show that the CD28 antigen, expressed in Chinese hamster ovary (CHO) cells, mediated specific intercellular adhesion with human lymphoblastoid and leukemic B-cell lines and with activated primary murine B cells. CD28mediated adhesion was not, dependant upon divalent cations. Several mAbs were identified that inhibited CD28-mediated adhesion, including mAb BB-1 against the B-cell activation antigen B7/BB-1 and some mAbs against major histocompatibility complex class I antigens. B7/BB-1 expression correlated closely with CD28-mediated adhesion, but class I expression did not. Transfected COS cells expressing the B7/BB-1 antigen adhered to CD28+ CHO cells; this adhesion was blocked by mAbs to CD28 and B7/BB-1. The specific recognition by CD28 of the B-cell activation antigen B7/BB-1 represents a heterophilic interaction between members of the immunoglobulin superfamily that may serve to regulate T-cell cytokine levels at sites of B-cell activation.
MATERIALS AND METHODS
The generation of a T-lymphocyte immune response is a complex process involving cell-cell interactions (1) and production of soluble immune mediators (cytokines or lymphokines) (2). This response is regulated by several T-cell surface molecules, including the T-cell receptor complex (3) and other "accessory" surface molecules (1). One such accessory molecule is the CD28 antigen, a homodimeric glycoprotein of the immunoglobulin superfamily (4) found on most mature human T cells (5). Current evidence suggests that this molecule functions in an alternative T-cell activation pathway distinct from that initiated by the T-cell receptor complex (for review, see ref. 6). Monoclonal antibodies (mAbs) to CD28 can augment T-cell responses initiated by various polyclonal stimuli (5-7). These stimulatory effects may result from mAb-induced cytokine production (8) as a consequence of increased mRNA stabilization (9). Anti-CD28 mAbs can also have inhibitory effects; i.e., they can block autologous mixed lymphocyte reactions (10) and activation of antigen-specific T-cell clones (11). The in vivo function of CD28 is not known, although its structure (4) suggests that like other members of the immunoglobulin superfamily (12) it might function as a receptor. CD28 could conceivably function as a cytokine receptor, although this seems unlikely since it shares no homology with other lymphokine or cytokine receptors (4). Alternatively, CD28 might be a receptor that mediates cell-cell contact. In this paper, we describe experiments designed to test this possibility. For this purpose, we have taken an approach based on experiments used to demonstrate
mAbs. mAb 9.3 (anti-CD28, ref. 15) was purified from ascites fluid before use. A number of mAbs to B-cell associated antigens were tested for their abilities to inhibit CD28mediated adhesion. mAbs 60.3 (CD18); iF5 (CD20); G29-5 (CD21); G28-7, HD39, and HD6 (CD22); HD50 (CD23); KB61 (CD32); G28-1 (CD37); G28-10 (CD39); G28-5 (CD40); HERMES1 (CD44); 9.4 (CD45); LB-2 (CD54); and 72F3 (CD71) have been described and characterized (16-18). These mAbs were purified before use. &TA4-1 (ref. 19, antiIgD); 2C3 (ref. 20, anti-IgM); Nambl, H1DE, p10.1, and W6/32 (refs. 20 and 21, anti-human class I); and HBiOa (ref. 20, anti-MHC class II) were also purified before use. mAbs B43 (CD19); BL-40 (CD72); AD2, 1E9.28.1, and 7G2.2.11 (CD73); EBU-141, LN1 (CDw75); CRIS-1 (CD76); 424/4A11 and 424/3D9 (CD77); Leu 21, Ba, 1588, LO-panB-1, FN1, and FN4 (CDw78); and M9, G28-10, HuLymlO, 2-7, F2B2.6, 121, L26, HD77, NU-B1, BLAST-1, BB-1, anti-BL7, antiHC2, and L23 were used as coded samples provided to participants in the Fourth International Conference on Human Leukocyte Differentiation Antigens (September 1989, Vienna) (22). Most of these were used in ascites fluid form. mAbs BB-1 and LB-1 (23) were also purified from ascites fluid before use. Anti-integrin receptor mAbs, P3E3, P4C2, and P4G9 (24), were used as hybridoma culture supernatants. Plasmids and Transfections. cDNA clones encoding the T-cell antigens CD4, CD5, and CD28 (4) in the expression vector TH3M (4) were kindly provided by Sandro Aruffo and Brian Seed (Massachusetts General Hospital, Boston). An expressible cDNA clone encoding the B7/BB-1 antigen (25) was provided by Gordon Freeman (Dana-Farber Cancer Institute, Boston). Dihydrofolate reductase-deficient CHO cells were cotransfected as described (26) with a mixture of plasmids, irH3M-CD28 (4) and pSV2dhfr (27). COS cells were transfected with B7/BB-1, CD4, or CD5 cDNAs in the presence of DEAE-dextran using a protocol supplied by B. Seed and A. Aruffo. CD28-Mediated Adhesion Assay. Cells to be tested for adhesion were labeled with 51Cr, washed, and preincubated in complete RPMI medium [RPMI 1640 medium containing 10% (vol/vol) fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 ,g/ml)] containing 10 mM EDTA
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: mAb, monoclonal antibody; MHC, major histocompatibility complex; FITC, fluorescein isothiocyanate; LPS, lipopolysaccharide. 5031
5032
Proc. Natl. Acad. Sci. USA 87 (1990)
Immunology: Linsley et al.
unless otherwise indicated. mAbs to be tested for adhesion inhibition were then added to 10 tig/ml, and cells were incubated for -1 hr at 230C. In some experiments, a mouse mAb having irrelevant specificity was added to the labeling and preincubation reaction mixtures to saturate Fc receptors. Labeled cells (1-10 x 106 cells per well in 0.2 ml of complete RPMI medium containing EDTA and mAbs, where indicated) were then added to the CHO monolayers. When adhesion inhibition by mAb 9.3 was measured, the mAb or its F(ab')2 were added to the CHO cells 1 hr prior to addition of labeled cells. Adhesion was initiated by centrifugation in a plate carrier and continued at 370C for 1 hr. Monolayers were then washed five times with ice-cold complete RPMI medium and solubilized by addition of 0.5 M NaOH, and radioactivity was measured in a y counter. Numbers of bound cells were calculated by dividing total bound radioactivity (cpm) by the specific activity (cpm per cell) of labeled cells. When COS cells were used, results are expressed as cpm bound since their viability at the end of the experiment was generally 70% by mAb 9.3) was clearly observed with T51, 5E1, Raji, and Jijoye cells. Daudi cells also showed specific adhesion, although to a lesser extent. Other cell lines did not show specific CD28-mediated adhesion, although some (e.g., Namalwa) showed relatively high nonspecific adhesion. Primary mouse splenic B cells did not show CD28-mediated adhesion but acquired the ability to adhere after activation with LPS. In other experiments, six additional lymphoblastoid lines showed CD28-mediated adhesion, but the U937 cell line, cryopreserved human tonsil B cells, and phytohemagglutinin-stimulated T cells did not show adhesion. These experiments indicate that a CD28 ligand is found on the cell surface of activated B cells of human or mouse origin. CD28-Mediated Adhesion Is Specifically Blocked by BB-1, a mAb to a B-Cell Activation Antigen. In initial attempts to define B-cell molecules involved in CD28-mediated adhesion, we measured adhesion by lymphoblastoid cell lines with mutations in other known cellular adhesion molecules. The 616 cells (MHC class II deficient, ref. 28) bound to CD28' CHO cells equally well or better than parental T51 cells. Likewise, a CD18-deficient cell line derived from a patient Er~I T51 1 A2
Raji
Jijoye No Mb _~~~~
_
Daudi
=3 MAbi.
HL60
T51 CEM Jurkat
0
10
i
I--q
20
HSB2
30
5(0
40
with leukocyte-adhesion deficiency (Gambaro cells, ref. 29) also adhered specifically to CD28. Thus, MHC class II and CD18 molecules do not mediate adhesion to CD28. We then tested a panel of mAbs to B-cell surface antigens for their ability to inhibit CD28-mediated adhesion of T51 cells. For these experiments, we tested a total of 57 mAbs reactive with T51 cells. Initial screening experiments were carried out in the absence of EDTA, and any mAbs that blocked adhesion were subsequently retested in the presence and absence of EDTA. Of these mAbs, only those directed against MHC class I molecules (Nambl, H1DE, P10.1, and W6/32) and one to an unclustered B-cell antigen [BB-1, originally described as a B-cell activation marker (23)] were consistently able to block CD28-mediated adhesion by >30%. The dose dependence of adhesion inhibition by the anticlass I mAb H1DE, by BB-1, and by 9.3 were compared in the presence of EDTA in the experiment shown in Fig. 4. mAb 9.3 was most effective at blocking, but mAb BB-1 was able to block =60% of adhesion at concentrations
