Immunology and Cett Biology {1991) 69. 159-165
Monoclonal antibodies which identify carbohydrate-defined MHC Class I epitopes HELEN C. O'NEILL Experimental Haematology Group. John Curiin School oj Medical Research. Australian National University, Canberra, ACT, Australia (Submitted IS March 1991. Accepted for publication 23 April 1991.) Summary Eleven different monoclonal antibodies specific for H-2K- and H-2D-encoded Class I molecules have been screened to determine Class 1 epitopes dependent on both carbohydrate and protein structures. Monoelonal antibodies have been identified which bind to carbohydrate-defined antigens encoded by both the H-2Kand H-2Dgene regions. Sensitivity toglycosidases versus pronase has been used to classify antigens both expressed as cell surface molecules and when prepared as detergent solubilized antigen. Several simple sugars have also been found to act as inhibitors of antibodies which bind to carbohydrate-defined sites. The genelic control of carbohydrate antigen expression by H-2K- and H-2D-linked genes has been verified since a specific antibody does not bind to H-2K''or H-2D''molecules encoded by several mutani strains of mice containing single amino acid substitutions in their protein produet. All of these data are consistent with Class I antigenic structures being encoded in carbohydrate as well as protein moieties.
INTRODUCTION The H-2K and H-2D gene regions ofthe murine major histocompatibility complex (MHC) contain one or more genes encoding a 45 kDa giycoprotein (1). Extensive sequencing of peptides, and also nucleotides. has confirmed a common structure for murine Class I molecules. They each have a hydrophobic amino acid sequence in the C-terminal portion ofthe molecule which anchors them in the cell membrane, and three external domains contribute to an immunoglobulin-like molecule (2). An antigen binding cleft is now known to house peptide fragments of foreign antigens (3). A conserved feature of mouse Class I molecules is the presence of A'-linked oligosaccharide side chains attached to two asparagine residues at amino acid position numbers 86 and 176 in the ai and a2 domains respectively. The H-2D'' molecule is unique in comparison with other known Class I H-2 molecules and has a third oligosaccharide side chain at amino acid position number 256 in the a3 domain (4). The ai and a2 domains represent the polymorphic domains of Class I molecules and many monoclonal antibodies have
been generated to haplotype specific determinants (5-7). These two domains are involved in the formation ofthe antigen binding cleft. Previously we have reported the chemical nature of epitopes identified by several anti-H2K'* antibodies. Epitopes. detected by two antibodies H1OO-3OR3 and 11-4.1, were found to be determined, at least in part, by carbohydrate structures (8). The possibility that antigenic sites can be encoded by carbohydrate as well as protein determinants, is now also indicated for la antigens (9). Here we have extended our analysis and compared the chemical nature of antigens detected with 11 different monoclonal antibodies specific for gene products of both the H-2K and H-2D gene regions.
MATERIALS AND METHODS Animals
All mice were bred in the John Curtin School of Medical Research and used for preparation of spleen-cell suspensions at 6-8 weeks of age. Mutant strains of mice were kindly provided by I. F. C. McKenzie (Melbourne University, Vic, Australia).
Correspondence: H. C. O'Neill, Experimental Haematology Group, John Curtin School of Medical Research, Australian National University. Canberra, ACT 2600, Australia, Abhrevialion.s used in ihi.s paper. FCS, fetal calfserum; MHC, major histocompatibility: NP-40, Nonidet P-40; PBS, phosphate buffered saline; SRBC, sheep red blood eells.
H,C. O'NEILL
160
Antibodies All mouse monoclonal antibodies used in this study have been listed in Table 1 together with a reference to their origin and Ig class. Most were obtained from donor labs as hybridoma culture supernatant, but some were prepared as mouse ascitic fluid. The H141-31. H141-30. B22-249. HIOO-27R9 and H1OO-3OR3 antibodies were purchased as ascitic fluid from Biotest Folex (Birmingham. UK) and the 11-4.1 antibody, an affinity purified antibody, was purchased from Becton Dickinson (Mountain View, CA, USA). Preparation of cells Spleen-cell suspensions of high viability were prepared for serological titrations by dissociating tissue on a wire mesh. Cells were prepared in Eagle's minimal essential medium (F15, Grand Island Biological Co.. Grand Island. NY, USA), containing 5% fetal calfserum (FCS). Cell preparations were depleted of red and dead cells by centrifugation through 14% Isopaque/Fieoil (8). For serological assays, cells were cleared of surface Ig by a capping procedure using sheep IgG specific for mouse Ig. In some experiments Ig-capped spleen cells were treated with glycosidases. Briefly, this involved a 30 min incubation of cells at a concentration of 5 X 10^/mLat 37''C with 100 |ig/mL mixed glycosidases (Mills Laboratories Inc.. Kankakce. IL. USA), and neur-
aminidase at 10 U/mL (Vibrio cholera. B grade. Calbiochem, CA. USA). Control cells were subjected to the same treatment, but in the absence of enzyme. The exact procedure has been described previously. Preparation of cell lysates Cells at a concentration of 2 X lO^/mL were solubilized in 0.5% v/v Nonidet P-40 (NP-40) in phosphate buffered saline (PBS) containing I.5mmol/L MgCb and lO'^mol/L phenylmethylsulphonyl fluoride. This method, and the method for removing detergent from lysates using XAD-8 resin beads, has been described elsewhere (10). For some experiments, lysates were treated with enzymes. This involved either a24 h incubation at 37°C in the presence of 0.1% NaNi with 1 mg/mL pronase (Calbiochem B Grade. San Diego. CA, USA), or 200ug/mL mixed glycosidases together with 10 U/mL neuraminidase. The exact conditions for enzyme treatment of lysates have been reported previously (8). Enzyme activity was halted by the maintenance of lysates at 4*'C. In rosette inhibition studies, serial two-fold dilutions of enzyme treated and untreated spleen cell lysates were absorbed overnight with a constant amount of antibody. Residual antibody activity was measured in the rosetting assay using Ig-capped spleen cells. This method has been described in detail previously (10).
Table 1. Sensitivity of antibody-defined Class I epitopes to giycosidase treatment.
Antibody B22-249 HI41-3I HI4I-3O 28-14-8S 20-8-4 27-II.13S H1OO-27R9 HIOO-3OR3 3-83P 11-4.1 16-3-22S
Spleen cell targets C57BL/6J C57BL/6J C57BL/6J B10.A(4R) B10,A(4R) B10.A(4R) BIO. A BIO. A BIO. A BIO. A BIO.A(2R)
,^^tibodytitre(% maximum cell binding)' GlycosidaseClass I determinant Untreated cells treated cells^" Db
D*> DO
D*Db
D" K'' Kk Kk Kk
80 000 (88) 2 560 (70) 80 000 64(86) 800(76) 64 000(86) 480 000(77) 16 000(73) 5 120(93) 8000(56) 1 280(86)
2 500(88) 2 560(66) < 10 000** 16** 75** 3 000 •* 480 000(80) 2 000 ** 5 120(93) 500'* 1 280(80)
'Antibody binding was determined by lhe rosetting assay and the litre represents the reciprocal of the antibody dilution giving 50% reduction in maximum cell binding. Spleen cell targets were chosen to give antibody binding specific for the listed Class 1 determinant, "Reductions in antibody titrc of > four-fold, ^Some cells were treated with a mixed giycosidase preparation stjpplemented with neuraminidase prior to use in the rosetting assay.
MONOCLONAL ANTtBODtES AND MHC CLASS I EPITOPES
Serological assays The rosetting assay was commonly used to detect the binding of antibody to Ig-capped spleen cells. Sheep anti-mouse Ig coupled sheep red blood cells (SRBC) were used to detect antibody bound to cells, by the formation of rosettes. The exact method has been described in detail previously (8). Sugar inhibition studies A battery of >30 mono-, di- and oligosaccharides were screened for their capacity to inhibit the binding of monoclonal antibodies to Igcapped spleen cells using the rosetting assay. This procedure has been described previously (8.11). Briefly, lO^iL of each sugar (used at 20 mg/mL in PBS) was added to 10 (iL antibody at a given minimal saturating concentration, and incubated for 60 minat 4°C.Ten microiitres of Ig-capped spleen cells at a concentration of 1.5 X 10^/mL was then added and the rosetting assay performed as described above.
I6t
sensitivity of spleen cell antigens to a mixed giycosidase preparation supplemented with neuraminidase has been found to consistently reflect the results of other tests designed to determine the chemical nature of Class I epitopes. These other tests have included the following: (i) an assessment of comparative sensitivity of a detergent lysate of cells to pronase v.v giycosidase treatment using enzyme-treated lysates to inhibit antibody binding in the rosetting assay; and (ii) a test ofthe capacity of simple sugars to inhibit antibody binding. Data in Table 1 summarize the binding endpoints of 11 antibodies which bind specifically to H-2Dt'- or H-2K'^ encoded gene products expressed on spleen cells. Binding on untreated and mixed giycosidase treated cells using the rosetting assay has been compared. Seven of these antibodies specific for both H-2K''- and H-2Db-encoded molecules bind to treated cells, giving a reduction in binding endpoint of between four-fold and 32-fold. These antibodies appear to be specific for carbohydrate-defined epitopes.
Removal of sialic acid, which is one of two sugars usually found at the terminus of branched oligosaccharide side chains, can also be used to define the protein vs carbohydrate nature of RESULTS antigens detected by monoclonal antibodies. The 11-4.1 antibody has been shown to bind to a Giycosidase sensitivity of antigens recognized by neuraminidase sensitive carbohydrate site (8) monoclonal antibodies but the 3 anti-Df^ antibodies shown in Fig. 1 bind to sites which are not neuraminidase sensitive, The sensitivity of Class I antigens to giycosidase and are probably not present at the terminus of treatment has been used as a simple gauge ofthe oligosaccharide side chains. carbohydrate nature of Class I epitopes. The
Antibody dilution
Fig. I. The binding of anti-D^J antibodies to normal and glycosidase-treated B10. A(4R) spleen cells. The rosetting assay was employed to measure the bitiding ofthe 28-14-8S. 20-8-4 and 27-11 -1 3S antibodies to cells untreated (•). neuraminidase treated (A) or treated with a mixture of glycosidases supplemented with neuraminidase (•).
162
H,C. O'NEILL
Further attempts to define the sugar specificity of these monoclonal antibodies have included tests for the inhibitory capacity of simple sugars on antibody binding to cells. More than 30 mono-, di- and oligosaccharides were screened for rosette forming cells. A summary of sugars which inhibit binding of five different antibodies is shown in Table 2. Since different sugars were found to be inhibitory for different antibodies it is possible that they are mimicking different carbohydrate linkages representing unique epitopes for each ofthe different monoclonal antibodies. The giycosidase sensitivity of six of these antibodies was also confirmed in the lysate inhibition assay. Detergent lysates of spleen cells were either pronase-treated. mixed glycosidasetreated or left untreated. A minimal saturating concentration of each antibody was absorbed to two-foid dilutions of these lysates. and then free antibody left unbound was assayed in the rosetting assay. For all antibodies tested here, there was a clear division of epitope sensitivity to either pronase or glycosidases (Table 3). This result was consistent with results shown in Table 2 describing giycosidase sensitivity of cell surface antigens. The predicted carbohydrate or protein nature of antibody-specific Class I H-2K and H-2D epitopes has been summarized in Table I. Untreated lysates were ineffective inhibitors of 28-14-8S. 20-8-4 or 27-11-13S binding. This result has been shown to hold true for lysates prepared from several strains of mice (data not shown), and we predict that some conformational change must occur following enzyme treatment of solubilized antigen to expose epitopes specific for these antibodies. Attempts to genetically map carbohydratedefined H-2K/D antigens Several monoclonal antibodies specific for carbohydrate-defined epitopes appear to be
Table 3. Inhibitory activity of enzyme-treated lysate for antibody binding.
Antibody B22-249 H14I-31 28-14-8S 20-8-4 27-il-13S HIOO-27R9 3-83P 11-4.1
Untreated
512 256
Monoclonal antibodies which identify carbohydrate-defined MHC Class I epitopes HELEN C. O'NEILL Experimental Haematology Group. John Curiin School oj Medical Research. Australian National University, Canberra, ACT, Australia (Submitted IS March 1991. Accepted for publication 23 April 1991.) Summary Eleven different monoclonal antibodies specific for H-2K- and H-2D-encoded Class I molecules have been screened to determine Class 1 epitopes dependent on both carbohydrate and protein structures. Monoelonal antibodies have been identified which bind to carbohydrate-defined antigens encoded by both the H-2Kand H-2Dgene regions. Sensitivity toglycosidases versus pronase has been used to classify antigens both expressed as cell surface molecules and when prepared as detergent solubilized antigen. Several simple sugars have also been found to act as inhibitors of antibodies which bind to carbohydrate-defined sites. The genelic control of carbohydrate antigen expression by H-2K- and H-2D-linked genes has been verified since a specific antibody does not bind to H-2K''or H-2D''molecules encoded by several mutani strains of mice containing single amino acid substitutions in their protein produet. All of these data are consistent with Class I antigenic structures being encoded in carbohydrate as well as protein moieties.
INTRODUCTION The H-2K and H-2D gene regions ofthe murine major histocompatibility complex (MHC) contain one or more genes encoding a 45 kDa giycoprotein (1). Extensive sequencing of peptides, and also nucleotides. has confirmed a common structure for murine Class I molecules. They each have a hydrophobic amino acid sequence in the C-terminal portion ofthe molecule which anchors them in the cell membrane, and three external domains contribute to an immunoglobulin-like molecule (2). An antigen binding cleft is now known to house peptide fragments of foreign antigens (3). A conserved feature of mouse Class I molecules is the presence of A'-linked oligosaccharide side chains attached to two asparagine residues at amino acid position numbers 86 and 176 in the ai and a2 domains respectively. The H-2D'' molecule is unique in comparison with other known Class I H-2 molecules and has a third oligosaccharide side chain at amino acid position number 256 in the a3 domain (4). The ai and a2 domains represent the polymorphic domains of Class I molecules and many monoclonal antibodies have
been generated to haplotype specific determinants (5-7). These two domains are involved in the formation ofthe antigen binding cleft. Previously we have reported the chemical nature of epitopes identified by several anti-H2K'* antibodies. Epitopes. detected by two antibodies H1OO-3OR3 and 11-4.1, were found to be determined, at least in part, by carbohydrate structures (8). The possibility that antigenic sites can be encoded by carbohydrate as well as protein determinants, is now also indicated for la antigens (9). Here we have extended our analysis and compared the chemical nature of antigens detected with 11 different monoclonal antibodies specific for gene products of both the H-2K and H-2D gene regions.
MATERIALS AND METHODS Animals
All mice were bred in the John Curtin School of Medical Research and used for preparation of spleen-cell suspensions at 6-8 weeks of age. Mutant strains of mice were kindly provided by I. F. C. McKenzie (Melbourne University, Vic, Australia).
Correspondence: H. C. O'Neill, Experimental Haematology Group, John Curtin School of Medical Research, Australian National University. Canberra, ACT 2600, Australia, Abhrevialion.s used in ihi.s paper. FCS, fetal calfserum; MHC, major histocompatibility: NP-40, Nonidet P-40; PBS, phosphate buffered saline; SRBC, sheep red blood eells.
H,C. O'NEILL
160
Antibodies All mouse monoclonal antibodies used in this study have been listed in Table 1 together with a reference to their origin and Ig class. Most were obtained from donor labs as hybridoma culture supernatant, but some were prepared as mouse ascitic fluid. The H141-31. H141-30. B22-249. HIOO-27R9 and H1OO-3OR3 antibodies were purchased as ascitic fluid from Biotest Folex (Birmingham. UK) and the 11-4.1 antibody, an affinity purified antibody, was purchased from Becton Dickinson (Mountain View, CA, USA). Preparation of cells Spleen-cell suspensions of high viability were prepared for serological titrations by dissociating tissue on a wire mesh. Cells were prepared in Eagle's minimal essential medium (F15, Grand Island Biological Co.. Grand Island. NY, USA), containing 5% fetal calfserum (FCS). Cell preparations were depleted of red and dead cells by centrifugation through 14% Isopaque/Fieoil (8). For serological assays, cells were cleared of surface Ig by a capping procedure using sheep IgG specific for mouse Ig. In some experiments Ig-capped spleen cells were treated with glycosidases. Briefly, this involved a 30 min incubation of cells at a concentration of 5 X 10^/mLat 37''C with 100 |ig/mL mixed glycosidases (Mills Laboratories Inc.. Kankakce. IL. USA), and neur-
aminidase at 10 U/mL (Vibrio cholera. B grade. Calbiochem, CA. USA). Control cells were subjected to the same treatment, but in the absence of enzyme. The exact procedure has been described previously. Preparation of cell lysates Cells at a concentration of 2 X lO^/mL were solubilized in 0.5% v/v Nonidet P-40 (NP-40) in phosphate buffered saline (PBS) containing I.5mmol/L MgCb and lO'^mol/L phenylmethylsulphonyl fluoride. This method, and the method for removing detergent from lysates using XAD-8 resin beads, has been described elsewhere (10). For some experiments, lysates were treated with enzymes. This involved either a24 h incubation at 37°C in the presence of 0.1% NaNi with 1 mg/mL pronase (Calbiochem B Grade. San Diego. CA, USA), or 200ug/mL mixed glycosidases together with 10 U/mL neuraminidase. The exact conditions for enzyme treatment of lysates have been reported previously (8). Enzyme activity was halted by the maintenance of lysates at 4*'C. In rosette inhibition studies, serial two-fold dilutions of enzyme treated and untreated spleen cell lysates were absorbed overnight with a constant amount of antibody. Residual antibody activity was measured in the rosetting assay using Ig-capped spleen cells. This method has been described in detail previously (10).
Table 1. Sensitivity of antibody-defined Class I epitopes to giycosidase treatment.
Antibody B22-249 HI41-3I HI4I-3O 28-14-8S 20-8-4 27-II.13S H1OO-27R9 HIOO-3OR3 3-83P 11-4.1 16-3-22S
Spleen cell targets C57BL/6J C57BL/6J C57BL/6J B10.A(4R) B10,A(4R) B10.A(4R) BIO. A BIO. A BIO. A BIO. A BIO.A(2R)
,^^tibodytitre(% maximum cell binding)' GlycosidaseClass I determinant Untreated cells treated cells^" Db
D*> DO
D*Db
D" K'' Kk Kk Kk
80 000 (88) 2 560 (70) 80 000 64(86) 800(76) 64 000(86) 480 000(77) 16 000(73) 5 120(93) 8000(56) 1 280(86)
2 500(88) 2 560(66) < 10 000** 16** 75** 3 000 •* 480 000(80) 2 000 ** 5 120(93) 500'* 1 280(80)
'Antibody binding was determined by lhe rosetting assay and the litre represents the reciprocal of the antibody dilution giving 50% reduction in maximum cell binding. Spleen cell targets were chosen to give antibody binding specific for the listed Class 1 determinant, "Reductions in antibody titrc of > four-fold, ^Some cells were treated with a mixed giycosidase preparation stjpplemented with neuraminidase prior to use in the rosetting assay.
MONOCLONAL ANTtBODtES AND MHC CLASS I EPITOPES
Serological assays The rosetting assay was commonly used to detect the binding of antibody to Ig-capped spleen cells. Sheep anti-mouse Ig coupled sheep red blood cells (SRBC) were used to detect antibody bound to cells, by the formation of rosettes. The exact method has been described in detail previously (8). Sugar inhibition studies A battery of >30 mono-, di- and oligosaccharides were screened for their capacity to inhibit the binding of monoclonal antibodies to Igcapped spleen cells using the rosetting assay. This procedure has been described previously (8.11). Briefly, lO^iL of each sugar (used at 20 mg/mL in PBS) was added to 10 (iL antibody at a given minimal saturating concentration, and incubated for 60 minat 4°C.Ten microiitres of Ig-capped spleen cells at a concentration of 1.5 X 10^/mL was then added and the rosetting assay performed as described above.
I6t
sensitivity of spleen cell antigens to a mixed giycosidase preparation supplemented with neuraminidase has been found to consistently reflect the results of other tests designed to determine the chemical nature of Class I epitopes. These other tests have included the following: (i) an assessment of comparative sensitivity of a detergent lysate of cells to pronase v.v giycosidase treatment using enzyme-treated lysates to inhibit antibody binding in the rosetting assay; and (ii) a test ofthe capacity of simple sugars to inhibit antibody binding. Data in Table 1 summarize the binding endpoints of 11 antibodies which bind specifically to H-2Dt'- or H-2K'^ encoded gene products expressed on spleen cells. Binding on untreated and mixed giycosidase treated cells using the rosetting assay has been compared. Seven of these antibodies specific for both H-2K''- and H-2Db-encoded molecules bind to treated cells, giving a reduction in binding endpoint of between four-fold and 32-fold. These antibodies appear to be specific for carbohydrate-defined epitopes.
Removal of sialic acid, which is one of two sugars usually found at the terminus of branched oligosaccharide side chains, can also be used to define the protein vs carbohydrate nature of RESULTS antigens detected by monoclonal antibodies. The 11-4.1 antibody has been shown to bind to a Giycosidase sensitivity of antigens recognized by neuraminidase sensitive carbohydrate site (8) monoclonal antibodies but the 3 anti-Df^ antibodies shown in Fig. 1 bind to sites which are not neuraminidase sensitive, The sensitivity of Class I antigens to giycosidase and are probably not present at the terminus of treatment has been used as a simple gauge ofthe oligosaccharide side chains. carbohydrate nature of Class I epitopes. The
Antibody dilution
Fig. I. The binding of anti-D^J antibodies to normal and glycosidase-treated B10. A(4R) spleen cells. The rosetting assay was employed to measure the bitiding ofthe 28-14-8S. 20-8-4 and 27-11 -1 3S antibodies to cells untreated (•). neuraminidase treated (A) or treated with a mixture of glycosidases supplemented with neuraminidase (•).
162
H,C. O'NEILL
Further attempts to define the sugar specificity of these monoclonal antibodies have included tests for the inhibitory capacity of simple sugars on antibody binding to cells. More than 30 mono-, di- and oligosaccharides were screened for rosette forming cells. A summary of sugars which inhibit binding of five different antibodies is shown in Table 2. Since different sugars were found to be inhibitory for different antibodies it is possible that they are mimicking different carbohydrate linkages representing unique epitopes for each ofthe different monoclonal antibodies. The giycosidase sensitivity of six of these antibodies was also confirmed in the lysate inhibition assay. Detergent lysates of spleen cells were either pronase-treated. mixed glycosidasetreated or left untreated. A minimal saturating concentration of each antibody was absorbed to two-foid dilutions of these lysates. and then free antibody left unbound was assayed in the rosetting assay. For all antibodies tested here, there was a clear division of epitope sensitivity to either pronase or glycosidases (Table 3). This result was consistent with results shown in Table 2 describing giycosidase sensitivity of cell surface antigens. The predicted carbohydrate or protein nature of antibody-specific Class I H-2K and H-2D epitopes has been summarized in Table I. Untreated lysates were ineffective inhibitors of 28-14-8S. 20-8-4 or 27-11-13S binding. This result has been shown to hold true for lysates prepared from several strains of mice (data not shown), and we predict that some conformational change must occur following enzyme treatment of solubilized antigen to expose epitopes specific for these antibodies. Attempts to genetically map carbohydratedefined H-2K/D antigens Several monoclonal antibodies specific for carbohydrate-defined epitopes appear to be
Table 3. Inhibitory activity of enzyme-treated lysate for antibody binding.
Antibody B22-249 H14I-31 28-14-8S 20-8-4 27-il-13S HIOO-27R9 3-83P 11-4.1
Untreated
512 256
