Molecular irnrnu~o~o~~, Vol. 29, No. 7/8, pp. lOI3-1023, 1992

0~61-5890/92 $5.00 + 0.00 Pergamon Press Ltd

Printed in Great Britain.

EPITOPE SIZE, SPECIFICITY AND EQUILIBRIUM CONSTANT FOR FOUR MONO~LONAL ANTIBODIES BINDING TO THE 04 POLYSACCHARIDE ANTIGEN OF SALMONELLA SEROGROUP B BACTERIA Srv LIND and ALF A. LINDBERG* Department

of Clinical Bacteriology, Karolinska Institute, Huddinge Hospital, S-14186 Huddinge, Sweden (First received 29 July 1991; accepted in revisedform

3 October 1991)

Abstract-One rat (MAST 83),and three mouse (MAST 107, 108 and 112) monoclonal antibodies (mAbs) directed against Sa~mo~ella serogroup 30 1ipopoIysaccha~de (LPS) were characterized and found to bind to the 0:4 epitope but recognizing different surfaces of the polysaccharide chain. The epitopes were defined from the combined results of: (i) binding specificities in enzyme immuno assay (EIA) against chemically defined LPS and glycoconjugates; (ii) studies of affinity constants in Farr-assay for binding to oligosaccharides purified from LPS, or chemically synthesized; and (iii) knowledge of the conformation of the 0-polysaccharide chain of Salmonella BO bacteria. Two of the antibodies, MAST 83 and 108, bound to the 04 epitope when present in the terminai non-reducing end as well as an intrachain determinant of the 0-polysaccharide, whereas MAST 107 and 112 bound only to the 0:4 epitope when present as an intrachain determinant. The equilibrium constants (K values), determined for binding of the mAbs and a Fab-fragment isolated from one of them to a ‘251-labelled tyramine-derivative of a Salmonella BO dodecasaccharide, were: 4.3 x lo5 (MAST 83), 1.0 x 10’ (MAST 107), 1.3 x 10’ (MAST 107 Fab), 4.5 x lo4 (MAST 108) and 1.9 x 10’ ljmol (MAST 112). The results suggest that each epitope encompasses the 04 specifying D-abequosyl residue together with different numbers of saccharides varying in size from di- to tetrasaccharides from the linear backbone chain. The antibodies also bind to different surfaces of the 0-polysaccharide chain as suggested by its conformation.

LNTRODUCTION

The outer membrane of Salmonella bacteria consists of proteins, phospholipids and lipopolysaccharide (LPS). The LPS molecule is major constituent and is composed of three regions: lipid A, the core oligosaccharide and the 0-antigenic polysaccharide chain built by repeating oligosaccharide units. The major antibody response to Salmonella is directed towards its LPS and particularly the 0-antigenic polysaccharide, whose structure constitutes the basis for classification of ~aim~nelZa bacteria into serogroups (Ka~mann, 1966). Within one serogroup there are different 0-antigenic determinants, one of which is specific for the subgroup: e.g. 0:4 for serogroup B and 0: 9 for D. The structure of the repeating unit of Salmonella serogroup B is known (Fig. 1). Here 0: 4 is the serogroup-specific dete~inant and 0 : 12 a determinant shared with serogroup A, B and D bacteria. The 0:4 epitope as defined by rabbit antibodies, is composed by the u-D-Abep (1 + 3) a-D*Author to whom correspondence should be addressed. abbreviations: EIA, enzyme immunoassay; mAb, monoclonal antibody; LPS, lipopolysa~ha~de; K values, equilibrium constants; A or Abe, o-abequose; M or Man D-mannose; R or Rha, L-rhamnose; G or Gal, D-galactose; Glc, D-glucose; TFA, trifluoroacetic acid; BSA, bovine serum albumin; PBS, phosphate-buffered saline.

Manp structure and the 0: 12 epitope by the trisaccharide a-D-Manp (1 + 4) a -L-Rhap (1 --, 3) a-n-Galp. Besides these major 0-antigenic specificities, Salmonella belonging to serogroup B also may possess some additional antigenic factors. Substitution of a-Abe with an 0-acetyl at C2 gives 0:5 specificity and E-D-Glc linked (1 + 6) or (1 -+ 4) to a-D-Gal constitutes 0: 1 and 0: 12, specificities, respectively. Studies of the interactions between bacterial polysaccharide antigens and antibodies are important for our understanding of the underlying mechanisms for virulence and protection. Use of the Salmoneffa BO polysaccharide as a model system is particularly attractive since its three-dimensional structure is well defined (Bock et al. 1984a). In previous studies, the size of the epitope for a rabbit anti Sa~rnu~el~~ BO antibody was determined to be larger than a tetrasaccharide but smaller, or equal to, an octasaccharide (J&beck et al., 1979). In this paper we report on one rat and three mouse anti Salmonella BO monoclonal antibodies (mAbs) which recognize the 0: 4 epitope. The equilibrium constants for binding of the mAbs to different 0-antigenic oligosaccharide structures were determined using a Farr-assay. The availability of conformational models of the oligosaccharides enabled us to obtain a good picture of the surfaces taking part in the binding.

1013

SIV LIND

1014

LINDBERG

and ALF A.

OAc

OAc

i

I

2

2

a-D-Abep

a-E-G@

:. 3

1 416

a-D-Manp(l-+4)

a-L-Rhap(l+3)

a-D-Galp(l[-tZ)

a-D-Glcp

a-D-Abep 1 3 a-D-Manp(l-14)

1 416 a-L-Rhap(l-13)

a-D-Galp(l-+]n

Fig. 1. The structure of Salmonella serogroup B oligosaccharide. EXPERIMENTAL Bacterial

strains

Salmonella typhimurium SH 4809 (O-antigens 4, 5 and 12) S. typhimurium SL 3622 (O-antigens 1, 4, 5 and 12) S. typhimurium SH 4305 (O-antigens 4, 5 and 12,) S. essen (O-antigens 4 and 12), S. enteritidis SH I262 (Oantigens 9 and 12), S. typhi 253 Ty (O-antigens 9 and 12,), S. thompson IS 40 (O-antigens 6 and 7) S. Ra) and S. typhimurium TV 119 (chemotype typhimurium SL 53 13 (O-antigens 1,4, 5,9 and 12) were from the strain collection of the Department of Clinical Bacteriology, Karolinska Institute, Huddinge Hospital, Huddinge, Sweden. Preparation

Bacteria

of LPS

were cultivated

as described

earlier (Jansson

et al., 198 1). LPS was extracted from lyophilized bacteria by phenol-water (Westphal et al., 1952) and the water

phase collected and lyophilized. The LPS preparation was treated with RNase in 0.1 M acetate buffer, pH 5.0. For preparation of LPS from the rough strain S. typhimurium TV 119, the phenol-chloroform-petroleum ether method was used (Galanos et al., 1969). Preparation

of 0-antigenic

oligosaccharides

The procedures for preparation and titration of phage P22 have been described (Svenson and Lindberg, 1978). Oligosaccharides were prepared by enzymatic degradation with phage P22 glycanases (Svenson and Lindberg, 1978) of alkaline treated (0.15 M NaOH at 100°C for 2 hr) LPS from S. typhimurium SH 4809 and polysaccharide from S. typhimurium SH 4305. For preparation of hexasaccharide devoid of D-abequose, SH 4809 octasaccharide was hydrolysed in 0.5 M TFA for 24 hr at room temp, neutralized and purified on a column of Bio-Gel P-2 (Bio-rad). The structures of the oligosaccharide preparations were verified with sugar analysis by gas liquid chromatography (Sawardeker et al., 1965) and by methylation analysis (Jansson et al., 1976). In addition, the SH 4809 hexasaccharide was analysed by ‘H NMR in deuterium oxide solution at 40°C with JEOL GX-400 (400 MHz) spectrometer. Sodium trimethylsilyl propanoate-d, (TSP; SO,OO) was used as internal reference. S. typhimurium

0 -antigenic

saccharides

and glycoconju -

gates

The structures of the compounds used are shown in Fig. 2A and B. Synthesis of a-D-Abe (1 + 3) cr-D-Manp

E-D-Abe (1 -3) U-D(I + aminophenylethylamine, Manp (1 -4) a-L-Rhap-(1 -3) r-D-Galp (1 +p-trifl1984) and uoroacetamidophenyl) (Classon et ul., a-D-Galp (1 -+ 2) [E-D-Abe (1 + 3)] CI-D-Manp (1 + 4) sr-L-Rhap (1 + 8)-methoxycarbonyloctyl (Bock and Meldal, 1984) have been described elsewhere. Coupling of oligosaccharides to BSA was done as described previously (Svenson and Lindberg, 1979). The substitution degree of the saccharide-BSA glycoconjugates was in the range of 7-25 mol of saccharide per mol BSA. EZA

EIA was performed as described earlier (Carlin and Lindberg, 1983). The antibodies were added to LPS(10 pgg/ml) or glycoconjugate(5 pg/ml) coated microtiter plates (A/S Nunc, Roskilde, Denmark). The conjugates used were alkaline phosphatase conjugated goat anti mouse polyvalent Ig, diluted l/l000 (Sigma) and goat anti rat IgG, diluted l/l000 (Sigma). “‘Z-labelling

of ligand

From the EIA results, the SH 4809 [G(A)MR], dodecasaccharide was selected as the ligand to be labelled. The dodecasaccharide was subjected to reductive amination with a lOO-fold excess of tyramine (Sigma) in the presence of sodium cyanoborohydride. The product was chromatographed on a column of Bio-Gel P-4 (Bio-Rad) in 0.1 M pyridinium acetate buffer, pH 5.6, followed by purification on a column of Bio-Gel P-2 in water, containing 0.05% (w/v) trichlorobutanol as preservative. The dodecasaccharide-tyramine conjugate was labelled with “‘1 by the chloramine T method (Hunter, 1978). Ten micrograms of conjugate was mixed with 2 mCi of NaI (Amersham). The reaction was stopped by addition of sodium disulphite and the labelled product was separated from free iodine on a Sephadex G- 10 column (Pharmacia LKB Biotechnology Inc.) in phosphate-buffered saline (PBS). The sp. act. of the conjugate was 40-100 Ci/mmol. Farr-assay

The assay was performed essentially as described earlier (Carlin et al., 1987 a). Forty microliters of unlabelled saccharide or saccharide-tyramine conjugate in PBS, containing 0.02% NaN, (w/v), was incubated with 40 pl of a mixture of 60-90,000 cpm of ‘251-labelled dodecasaccharide-tyramine, 65,000 cpm of “NaCl in PBS and 10% heat inactivated foetal bovine serum, and

Mapping of Salmonella 0 : 4 specific monoclonal SH

4809

1015

antibodies dodecasaccharide-tyramine a-g-AbeQ 1

(a-g-GalQ(l42)

Synthetic Synthetic

G(A)MR

3 a-g-ManQ(l+4)

AMRG

_L-Rha),&yramine

tetrasaccharide

tetrasaccharide-BSA a-;-Abeg

a-g-Abep

a-g-GalQ(l+2)

1

:

1

3 a-L-ManQ(ld4)

a-g-ManQ(l+4)

a-c-RhaQ(1 --f mo-BSA

a-_L-RhaQ(l+3) a-E-GalQ(1 +pf

SH 4809 tetraSynthetic

AMRG

octa-

and dodecasaccharide

tetrasaccharide-BSA a-g-AbeQ

a-g-AbeQ

3

(tx-D-GalQ(ld2)

a-_D-ManQ(l+4) a-l-RhaQ(l+3)

Synthetic

3 a-q-ManQ(l+4)

SH

AM disaccharide-BSA

4809

hexasaccharide

(tx-o-GalQ(1-12) a-g-ManQ(l+4)

a-g-Abep 1

SH 4305 3 a-D-ManQ(l+ _

deca-

papea-BSA

and

k-Rha)n,,

pentadecasaccharide

OAc

SH 4809 tetra- octa- and dodecasaccharide-BSA

(a-g-GalQ(l+2)

_L-Rha)n=1,2,3

a-g-GalQ(l-+pf-BSA

a-g-AbeQ

tr-+lCQ

: 3 a-E-ManQ(l+4)

z 4 (tx-D-GalQ(l+2)

I-Rha)n,t

,2,3 pf-BSA

I 2 a-g-Abeg 1 3 a-g-ManQ(l-14)

k-Rha)n=z,z

Fig. 2(A). Structures of glycoconjugates containing Salmonella serogroup B structural elements. mo, %methylcarbonyloctyl; pf, p-trifluoroacetamidophenyl; papea, p-aminophenylethylamine; BSA, bovine serum albumin; A, D-abequose; M, D-mannose; R, L-rhamnose; G, D-galactose. (B) Structures of saccharide inhibitors containing Salmonella serogroup B structural elements. pf; p-trifluoroacetamidophenyl; tyramine; 4-hydroxyphenethylamine; A, D-abequose; M, D-mannose; R, L-rhamnose; G, D-galactose.

2Opl of purified monoclonal antibody diluted in PBS, in a micro Eppendorf tube. **NaCl was used as a volume marker to eliminate washing steps. After incubation overnight, 100 p 1 of saturated ammonium sulphate in PBS was added. Tubes were mixed extensively and incubated in an ice bath for 30 min. To separate bound and free antigen, tubes were centrifuged at 20,OOOg for 5 min at 4°C. Approximately 180~1 of supernatant was withdrawn using a suction device, and the tubes were counted for 10 min in a LKB-Wallac 1260 multigamma, instrument. The procedure for counting of the samples has been described in detail (Carlin et al., 1987 a). Twenty different concns of unlabelled dodecasaccharide-tyramine (from 0.45 x lo-’ to 1.0 x 10F4 M) were incubated with a constant amount of labelled dodecasaccharide-tyramine (4-l 0 x 10m9M) containing **NaCl. In this type of assay, the K value for binding of the monoclonal antibody to dodecasaccharide-tyramine was determined. The

antibody concn was set to be within one order of magnitude of l/K (Pinchard, 1978). To estimate K values for binding of monoclonal antibody to saccharide inhibitors, 17 concns in two-fold dilution steps from 1 x lop3 to 1 x lop8 of saccharides were incubated with a constant amount of labelled dodecasaccharidetyramine. Duplicate tubes were used for each concn of antigen, and negative control ascites was included in each experiment. Calculations

Data were processed in the computer program EBDA (McPherson, 1983) to get initial estimates of K,, to get initial estimates of K,, K,, and R as described (Carlin et al., 1987 a). K values were calculated in a general ligand receptor binding program, LIGAND (Munson and Rodbard, 1980), which accepted a file created in EBDA.

1016

Production

Srv

and characterization

of monoclonal

LIND

and ALF A.

antibodies

Salmonella BO specific monoclonal antibodies MAST 107, MAST 108 and MAST 112 were generated by immunizing BALB/c mice i.p. with 5 x 10’ bacteria/mouse of whole heat-killed S. typhimurium SH4809 (0:4, 5 and 12) or SL 3622 (0: 1, 4, 5 and 12) on day 0. 12 and 25. On day 28, spleen lymphocytes (approx. 6 x 10’ cells) were fused with Sp 2/O myeloma cells (approx. 5 x 10’ cells). Cell culture media and fusion procedure were as described earlier (Lind et al., 1985). MAST 83 was derived by immunizing a LOUjC rat in the foot pads with 2 x 10” whole heat-killed SH 4809 bacteria. Nine days later, approx. 1 x 10’ cells from the lymph nodes were fused with an equal amount of cells from the rat myeloma cell line Y3 Ag 1.2.3. Culture supernatants were tested in EIA lo-12 days after the fusion against LPS from the strain used for immunization. Positive clones were tested against LPS from S. typhimurium SL 3622, SH 4305, TV 119, S. rsscn, S. enteritidis SH 1262, S. typhi 253 Ty and S. thompson IS 40 before cloning at least twice by limiting dilution. Cloned hybridomas were expanded either in serum free culture medium (Lind et al., 1985) or as ascites fluid in pristane primed BALBic mice or LOU/C rats. Antibody class and subclass of the monowere determined as described clonal antibodies previously (Lind et a/., 1985). Pur$cation

qf monoclonal

antibodies

Monoclonal antibodies MAST 107 and MAST 112 (all IgG3) were purified from serum free culture fluid or from ascites. Antibodies in serum free culture fluid were concentrated in an Amicon ultrafiltration cell (Amicon) on a PM-30 filter. Antibodies in ascites were precipitated with saturated ammonium sulphate and dialysed against PBS. The concentrated antibodies were chromatographed on a column of Protein A Sepharose 4B (Pharmacia LKB Biotechnology Inc.) in 1.5 M glycine, 3 M sodium chloride, pH 8.9, and eluted with 0.1 M citric acid, pH 4.0. MAST 83 (IgG2a) was purified on a column of Protein G Sepharose 4FF (Pharmacia LKB Biotechnology Inc.) in 0.2 M sodium phosphate, pH 7.0, and eluted with 1.0 M glycine-HCl, pH 2.7. MAST 108 (IgM) was purified from ascites as above and chromatographed on a column of Sephacryl S-300 Superfine (Pharmacia LKB Biotechnology Inc.) in 0.1 M Tris-HCl, 0.5 M sodium chloride, pH 8.0. Antibody concn was determined either with the Lowry method or by measuring A,,,. Preparation

of Fab fragments

Fab fragments were prepared from MAST 107 (IgG3) by papain digestion, where 6 mg/ml of papain (Sigma) was activated with 0.3 M 2-mercaptoethanol for 1 hr at room temp and purified on a column of Sephadex G-25 (Pharmacia LKB Biotechnology Inc.) in 10 mM sodium acetate buffer, pH 5.6. The antibodies were cleaved with the activated papain (antibody: papain 200 : 1) in 3 mM EDTA and I mM dithiothreitol at room temp overnight.

LINDBERG

The fragments were alkylated with 3 mM iodoacetamide, dialysed against 10 mM Tris-HCl buffer, pH 7.5, and purified on a column of DEAE-Bio-Gel A-l (Bio-Rad). Sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) was run on the fragments to follow the digestion of the antibodies and to evaluate the purity of the fragments. Conformational antigens

calculations

of Salmonella

serogroup

B

Plots were generated using the Insight/Discover package (Biosym Technologies Inc. San Diego) and plotted with Postscript-code on an Apple LaserWriter. Dihedral angles in the glycosidic linkages and in the acyclic side chains were taken from Bock et al. (19846). RESULTS Preparation

of 0 -antigenic

oligosaccharides

Structural analysis of oligosaccharides with two or three repeating units, isolated from S. typhimurium SH 4305 polysaccharide by enzymatic degradation with phage P22 glycanase, showed that approximately 85% of the molecules were substituted with a-D-Glc bound (1 -4) to r-D-Gal, constituting the 0: 12, specificity. These SH 4305 oligosaccharides retain their 0 : 5 specificity, the 0-acetyl group (Fig. 1) linked to C-2 of abequose. By contrast, oligosaccharides prepared from alkaline-treated SH 4809 LPS lack 0: 5 specificity, because alkaline treatment of LPS removes the 0-acetyl group. ‘H NMR analysis of abequose-lacking (GMR), hexasaccharide prepared from SH 4809 [G(A)MR], octasaccharide by TFA hydrolysis, showed that about 5% of the abequose was still present. Production and characterization qf Salmonella B specljic monoclonal antibodies

serogroup

Fusions of spleen cells from BALB/c mice immunized with whole heat-killed S. typhimurium SH 4809 or SL 3622 with Sp2/0 myeloma cells, yielded 28 positive clones after initial screening in EIA against homologous LPS. When the clones were tested against 0-antigenic oligosaccharide-BSA conjugates in EIA, only three of them were found to bind to the conjugates. These mAbs, MAST 107, MAST 108 and MAST 112 were selected for further testing. MAST 83 was generated by immunizing a LOU/C rat with whole heat/killed S. typhimurium SH 4809 and fusing cells from the lymph nodes with Y3 Ag 1.2.3. rat myeloma cells. MAST 83 was selected from five SH 4809 LPS and S. essen LPS binding clones. The antibody classes and subclasses were IgG3 for MAST 107 and MAST 112, IgG2a for MAST 83 and IgM for MAST 108. All light chains were rc. In order to determine the binding specificities in detail, the four mAbs were tested against a panel of LPSs with different 0-antigenic structures and against oligosaccharide-BSA conjugates (Table 1). All mAbs bound to LPS with the O-antigen 4 determinant. In addition, MAST 112 and MAST 83 bound to the following glycoconjugates: SH 4809 (0:4, 5, 12)

Mapping of Salmonella 0 : 4 specific monoclonal

1017

antibodies

Table 1. Binding activity in EIA of anti-Salmonella BO monoclonal antibodies. AdO5at 20 min for 0.1 pg of antibody binding to 1 pg of antigen A

Antibody Subclass Immunogen

MAST 83 MAST 107 %AST 108 MAST 112 IgG3, K IgG3, K IgGZa, K IgM K SH 4809 SL 3622 SL 3622 SL 3622

Antigen SH 4809 LPS (0:4, 5, 12) SL 3622 LPS (0: 1,4,5,12) S. Essen LPS (0 : 4, 12) SH 4305 LPS (0:4, 5, 12,) SH 1262 LPS (0:9, 12) 253 Ty LPS (0:9,12,) SL 5313 LPS 0: 1, 4, 5, 9, 12) IS 40 LPS (0 : 6,7) TV 119 LPS (Chemotype Ra) Synthetic tetra G(A)MR-BSA Synthetic tetra AMRG-BSA SH 4809 G(A)MR tetra-BSA SH 4809 [G(A)MR], octa-BSA SH 4809 [G(A)MR],dodeca-BSA AM-BSA

2.5 2.7 1.0 0.36 < 0.050 < 0.050 0.25 < 0.050 < 0.050 3.0 2.2 3.1 3.0 3.2 0.75

A = a-D-abequose,

R = cr-L-rhamnose, G = E-D-galactose.

M = a-D-mannose,

G(A)MR tetra-, [G(A)MRh octa- and [G(A)MR], dodecasaccharide-BSA and to the synthetic G(A)MR (0:4) tetrasaccharide-BSA which all represent intra O-chain structures. MAST 83 also bound to the synthetic AMRG (0 : 4) tetrasaccharide-BSA which represents the repeating unit in the terminal non-reducing end of the polysaccharide chain. MAST 107 bound to synthetic G(A)MR (0:4) tetrasaccharide-BSA and to SH 4809 [G(A)MR], dodecasaccharide-BSA, but not to SH 4809 G(A)MR tetra-or [G(A)MR], octasaccharide-BSA. MAST 108 bound to SH 4809 [G(A)MRlz octa- and [G(A)MR], dodecasaccharide-BSA and to synthetic AMRG- and G(A)MR-BSA, although weakly to the latter. MAST 83 was the only mAb binding to the disaccharide conjugate AM-BSA. None of the mAbs bound to LPS possessing O-antigens 9, 12, 9, 12, or 6, 7 or to the LPS from the rough mutant S. typhimurium TV 119 (lacking only the 0-antigenic polysaccharide chain). Farr -assay

The mAbs were tested subsequently in a Farr-assay, using as hapten the dodecasaccharide [G(A)MR], isolated from SH 4809 0 : 4, 12 LPS labelled with tyramine. A Fab-fragment prepared from MAST 107 was also included in the test. The K values for the binding of four mAbs and the MAST 107 Fab-fragment to the dodecasaccharide-tyramine ranged from 4 x lo4 to 4.8 x 10sl/mol (Table 2). MAST 83 had the highest K value followed by MAST 112, MAST 107 and its Fab-fragment, and MAST 108. Farr-inhibition with various 0 -antigenic saccharides

Farr-inhibition experiments were then performed by displacing the binding of mAbs to dodecasacchaMIMM 29,7.8-N

3.2 2.9 2.6 1.1 < 0.050 < 0.050 2.6 < 0.050 < 0.050 1.1 < 0.050 < 0.050 < 0.050 2.1 < 0.050

2.6 2.7 2.4 2.3 < 0.050 < 0.050 2.6 < 0.050 < 0.050 0.12 2.3 < 0.050 1.9 1.9 < 0.050

3.0 2.9 3.0 2.4 < 0.050 < 0.050 2.8 < 0.050 < 0.050 2.7 < 0.050 0.36 2.6 2.3 0.090

ride-tyramine with different oligosaccharides (Fig. 2B). For MAST 83 the SH 4809 [G(A)MR], dodecasaccharide gave a K value that was 14 times higher than the K value for dodecasaccharide-tyramine. SH 4809

octasaccharide and SH 4305 pentaKWWNs decasaccharide bound nine times stronger and SH 4305 decasaccharide three times stronger than the dodecasaccharide-tyramine. Synthetic AMRG tetrasaccharide was three times less efficient in binding and SH 4809 (GMR)2 hexasaccharide seven times less efficient than the dodecasaccharide-tyramine. The best inhibitor of the MAST 107 dodeca-tyramine binding was SH 4809 [G(A)MR], dodecasaccharide, followed by SH 4809 [G(A)MRh octasaccharide that was four times lower, SH 4305 pentadecasaccharide that was 50 times lower and synthetic AMRG tetrasaccharide and SH 4809 G(A)MR tetrasaccharide which were 125 and 240 times lower, respectively. SH 4809 (GMR), hexasaccharide did not inhibit at all. The binding profile for the Fab-fragment from MAST 107 was similar to that of the original antibody. Binding of MAST 108 to dodecasaccharide-tyramine was efficiently inhibited by SH 4809 [G(A)MR], octa- and [G(A)MR], dodecasaccharide. Inhibition with SH 4305 decasaccharide was three times less efficient and SH 4305 pentadecasaccharide six times less efficient. Synthetic AMRG tetrasaccharide, SH 4809 G(A)MR tetrasaccharide and SH 4809 (GMR), hexasaccharide did not inhibit at all. The K value for MAST 112 binding to SH 4809 [G(A)MR], dodecasaccharide was twice that of the K value for binding to dodecasaccharide-tyramine, which in turn was equal to the K value for SH 4809 [G(A)MR], octasaccharide. The efficiency of the other inhibitors were: SH 4809 G(A)MR tetrasaccharide (three times lower), SH 4305 decasaccharide (23 times lower),

1018

SIV LIND and ALF A. LINDBERG

Mapping of Salmonella 0:4 specific monoclonal antibodies synthetic AMRG tetrasaccharide (27 times lower), SH 4305 pentadecasaccharide (77 times lower) and SH 4809 (GMR), hexasaccharide (91 times lower). DISCUSSION

We showed in previous studies (Bock et al., 1984a) that the preferred conformation of Salmonella serogroup B O-antigen polysaccharide in solution has a helical character with approximately three repeating units per turn. The conformation of the individual tetrasaccharide repeating units was calculated semiempirically using the Hard-Sphere-Exo-Anomeric (HSEA) approach (Bock et al., 1984a) with a correlation with NMR spectra on natural and synthetic saccharides (Bock et al., 19846). The D-Abe residues protrude from the backbone oligosaccharide chain forming the immunodominant sugar of the OCantigen. It has, however, been established for rabbit antibodies that the epitope of Salmonella B O-antigen comprises not only the immunodominant sugar but at least one tetrasaccharide repeating unit (J&-beck et al., 1979). Since the conformation and the topography of an antigen is important in antibody recognition, the availability of three-dimensional structures of the oligosaccharides used in this study, facilitates the investigation of the molecular requirements for the binding of the mouse and rat mAbs. The results obtained in EIA and in Farr-assay showed that the four mAbs, directed against the Salmonella serogroup 0:4 epitope had different specificities. The epitopes recognized by the antibodies differ in the following respects: (i) size; (ii) terminal end or intrachain epitope; (iii) involvement in the epitope of a-D-Glc linked 1 44 to a-D-Gal; and (iv) requirements of an intact reducing end. MAST 83 binds to structures ranging in size from a di- to a dodecasaccharide, with increasing inhibitory activity in the Farr-assay with increasing size of the oligosaccharide (Table 2). The antibody recognizes its antigen both terminally AMRG and intrachain [G(A)MR],,Z,3. This result was also confirmed in the EIA (Table 1), showing equal binding activity to the two synthetic AMRG and G(A)MR tetrasaccharides. In addition, MAST 83 is the only mAb binding to a-D-Abe (1 + 3)) a-D-Man 1 -+ papea-BSA in the EIA, which indicates that the epitope is small. This suggests that the a-D-Gal residue does not play an important role in the binding. Accordingly, substitution of D-Gal with a 1,4linked a-d% had no measurable effect on the antibody binding. From the results obtained in the EIA and Farr-inhibition studies we suggest that the epitope recognized by MAST 83 is the trisaccharide AMR (Fig. 3A) and that the increased affinity to larger saccharides is due to conformational changes in the epitope. The covalent linkage of tyramine to the r_-rhamnosyl residue in the reducing end of the dodecasaccharide affected the binding of MAST 83 since the non-modified octa- and dodecasaccharide were IO-fold better as inhibitors in the Farr-assay (Table 2). We are presently unable to explain this preference of MAST 83 for a non-substituted O-l in

1019

the reducing end. The use of the acid-treated, and therefore abequos deficient, (GMR), hexasaccharide proved the requirement for D-Abe in the epitope. The low K value observed with (GMR), is readily explained by the residual D-abequosyl residues detected by ‘H NMR spectroscopy, amounting to 5% substitution of mannose. MAST 108, the only IgM mAb in this group, binds in the EIA to the synthetic tetrasaccharide AMRG, which has an intact D-galactosyl residue in the reducing end, but it does not bind to the other synthetic tetrasaccharide G(A)MR. None of these tetrasaccharides inhibit in the Farr-assay, which might be due to the multivalency of the IgM antibody, which requires higher concns of the tetrasaccharides for inhibition. Like MAST 83, MAST 108 binds both terminal and intrachain, but it has a larger epitope, probably a tetra-AMRG or a pentasaccharide AMRGM (Fig. 3B). MAST 108 binds to structures having the 0: 12, specificity, although the interaction is four or five times weaker than with the octa- or dodecasaccharides, which indicates that the a-Dglucosyl substituent forming the 12, specificity interferes only slightly with the binding (Fig. 3C). MAST 107 is an intrachain-specific antibody with a plausible epitope size of a pentasaccharide RG(A)MR with a need for the terminal r_-rhamnosyl residue to be either free or glycosidically linked (Fig. 4A). This conclusion was drawn from the Farr-inhibition experiments resulting in the highest K value for the [G(A)MR13 dodecasaccharide, the only saccharide exhibiting this epitope. Also the EIA studies show that the antibody is intrachain specific. However, no binding to [G(A)MRlz octasaccharide-BSA was observed in the EIA. This is in accordance with the suggested epitope, since RG(A)MR is missing in the octasaccharide glycoconjugate, which has the reducing L-rhamnosyl in a linear form, because of the reductive amination step in the work-up procedure whereas it is present in ring form in the native octasaccharide [G(A)MR-h used as inhibitor in the Farrassay. However, MAST 83 binds to synthetic tetrasaccharide G(A)MR in the EIA and would therefore be expected also to bind to the [G(A)MR12 octasaccharide. We believe that the reason for this contradiction is the higher substitution degree of the tetrasaccharide-BSA glycoconjugate (25 mol saccharide/BSA for the tetrasaccharide compared to 7 mol saccharide/BSA for the octasaccharide). In contrast to MAST 83 and MAST 108 this antibody binds with a very low constant to the SH 4305 pentadecasaccharide in the Farrinhibition assay. Our interpretation is that a-l,4 linked a-D-Glc forming the 0: 12, specificity interferes with the binding by blocking the interaction (Fig. 4C). There is, however, a binding in the EIA to the corresponding LPS. The substitution degree of a-D-Gal is ca 85%, which means that 15% of the repeating units do not have the 12, specificity which most likely is sufficient to get a binding in the sensitive EIA. The Fab-fragment from MAST 107 has K values for binding to the octa- and dodecasaccharides from SH 4809 which are the same as for the antibody

1020

SIV

LIND and ALF A. LINDBERG 1

5 Glc

1 Abe

GC

Abe

1 4

1 3

:. 4

3

Gal 1+2 Man 144

:.

Rha l-+3 Gal l-+2 Man I+4

Rha

C

Fig. 3. Stereoplots

showing the conformation

in ball-and-stick

model of two repeating units of

SaImonella typhimurium SH 4809 O-antigen (A and B) and two repeating units of Salmonella typhimurium SH 4305 O-antigen (C), obtained after degradation with phage P22 glycanases. The

proposed epitopes for MAST 83 and 108 antibody recognition are shown. The view is from the “front” side with the rhamnose in the reducing end to the right. The arrows show the binding site for the o-glucosyl residue. (Table 2). This most likely excludes the possibility of MAST 107 having an attachment of both combining sites when binding to the octa- or dodecasaccharide antigens. Such a hypothesis was proposed, but not proven, in studies of anti S. Jlexneri Y monoclonal antibodies binding to S. jexneri O-chain saccharides (Carlin et al., 1987a). The size of the antigenic determinant recognized by MAST 112 is a tetrasaccharide G(A)MR (Fig. 4B). The antibody is intrachain specific, since it does not bind to the AMRG tetrasaccharide in the EIA. Presence of the cr-D-Glc substituent of SH 4305 oligosaccharides decreases the binding significantly (Fig. 4C). For the two mAbs (MAST 107 and 112) that do not tolerate

cr-D-Glc linked tcl C-4 of D-Gal, a low affinity constant is observtd in the Farr-inhibition assay for binding to SH 4305 deca- and pentadecasaccharides. Likewise, there is a binding to the corresponding LPS in the EIA. This can be explained by the incomplete substitution of D-Gal which is sufficient to get a detectable value in the Farr-inhibition assay and a binding in the EIA. Deca- and pentadecasaccharides prepared from SH 4305 LPS retain the 0: 5 specificity, the 0-acetyl linked to C-D-Abe. We surmise that this substituent has no effect on the antibody binding, because all the mAbs bind to S. essen LPS (0 :4, 12) which lacks the 0 : 5 specificity, in the EIA.

Mapping of Salmonella 0:4 specific monoclonal G:c

Abe 1

GC

antibodies

1021

1

Abe 1

z

1

z

4

3

4

3

Gal 142 4

Man 1+4 Rha l-+3 Gal 1+2 Man 1+4 Rha 2 3 4 2 3

C

Fig. 4. Stereoplots showing the conformation in ball-and-stick model of two repeating units of Salmonella typhimurium SH 4809 O-antigen (A and B) and two repeating units of Salmonella typhimurium SH 4305 O-antigen (C), obtained after degradation with phage P22 glycanases. The proposed epitopes for MAST 107 and 112 antibody recognition are shown. The view is from the “back” side, obtained by turning the “front” view of the saccharide 180” around the y-axis, with the rhamnose in the reducing end to the left. The arrows show the linking site for the r>-gtucosyl residue giving 0 : 12, specificity.

In the Fart--inhibition assay, only decasaccharide and pentadecasaccharide structures prepared from SH 4305 with the 0: 4, 5, 12, specificity were tested. The corresponding saccharides were prepared from SL 3622, with the 0: 1, 4, 5, 12 specificity, i.e. with a-D-Glc linked I ,6 to a-D-Gal. Unfortunately, the substitution degree of a-D-Glc was by methylation analysis found to be only 20% which made it unsuitable as inhibitor. In conclusion, these four mAbs bind to different antigenic determinants. Two of the mAbs, MAST 83 and

MAST 108, bind to the terminal end but can also recognize their epitopes in the saccharide chains. The a-D-Gal residue is not important for binding and consequently a-l ,4 linkage of a D-Glc to a -D-Gal does not have any influence on the interaction. The epitope for MAST 108 is in the size of a pentasaccharide and for MAST 83 probably the size of a trisaccharide. These two mAbs do, however, differ in the respect that MAST 83 binds also to di- and tetrasa~haride structures and that MAST 83 prefers an intact reducing end of the terminal a+Rha.

1022

SW LIND and ALF A. LINDBERG

The other two mAbs, MAST 107 and MAST 112, are intrachain specific. MAST 107 and MAST 112 have combining sites, probably in the size of a tetra- to pentasaccharide. None of these two MAbs accept the a-D-Glc substituent linked to O-4 of D-Gal. Earlier conformational studies of the O-repeating units of Sahnonelia BO (Bock ef a(., 1984a) showed the presence of a hydrophobic surface stretching from the 3,6dideoxyfunctions of D-Abe to the 6-deoxy group of the L-Rha which may be of importance in the antibody recognition. It was also shown (Norberg et al., 1985) that the C-2 to C-4 region of the D-Abe is important since configurational changes in C-4 or C-2 and C-4, change the 0:4 antigenic specificity of D-Abe to 0:9 (D-TYV) or 0:2 (D-Par). Furthermore, the O-l region of r_-Rha was shown to be important. The conformational figures (Figs 3C and 4C) also show that the cc-D-Glc substituent linked to O-4 of @-D-Gal disrupts the hydrophobic area with its hydroxymethyl group (Bock et al., 1984a). This explains why two of our mAbs (MAST 107 and 112) had a significantly decreased binding when the cc-~-Glc substituent was introduced. Obviously, there is also another surface available for binding since MAST 83 and 108 both are independent of the presence of the I-D-Glc substituent. These two surfaces must be present in both immunogens (SH 4809 and SH 4305) used for generation of the mAbs, because the two binding specificities are found regardless of the immunogen used. Our studies have established the specificities for four mAbs recognizing the 0:4 epitope in the O-polysaccharide chain of Salmonella serogroup B bacteria. Although they appeared similar in the EIA binding studies, the Farr-inhibition assays, using different saccharides and glycoconjugates, defined their individual unique specificities. For an absolute epitope mapping more synthetic derivatives are required. For a definitive knowledge of the events involved in the antibody-antigen interaction direct measurements are required such as the crystallography studies of a Fab in the presence of a hapten, being pioneered by Rose et al. (1990). Such studies are of a great interest because of the important role of the abequosyl residues in a terminal non-reducing end in the repeating units of Salmonella bacteria. Thus Sahonella serogroup B bacteria are more virulent than Salmoneilae of other serogroups (Valtonen et al., 1975). One reason may be that the abequosyl containing O-polysac~haride chains were the least potent in triggering C3 activation in the complement system (Grossman et al., 1990). Antibodies directed against the 0:4 epitope were also superior to antibodies directed against other epitopes in passive protection as shown by Carlin et al. (19876). The unravelling of the molecular events involved in the host-parasite interactions require the availability of precisely mapped and thereby defined antibodies. Acknowledgements-This work was supported by the Swedish Medical Research Council (grant No. 16 x 656) and the Swedish Board for Technical Development {grant No. 870201 1F). We acknowledge Dr P.-E. Jansson at the Department of Organic Chemistry, Arrhenius Laboratory of Stockholm

University, for making the stereoplots of the oligosaccharides and Christina Jlderberg for making the rat monoclonal antibody.

REFERENCES Bock K. and Meldal M. (1984) Synthesis of tetrasaccharides related to the O-specific determinants of Salmonella serogroup A, B and D,, Acta Chem. &and. B38, 255-266. Bock K., Meldat M., Bundle D. R., Iversen T., Garegg P. J., Norbcrg T., Lindberg A. A. and Svenson S. B. (1984~) The conformation of SalmonelIo O-antigenic polysaccharide chains of serogroups A, B and D, predicted by semiempirical, hard-sphere (HSEA) calculations. Carbohydr. Res. 130, 23-34.

Bock K., Meldal M., Bundle D. R., iversen T., Mario Pinto B., Garegg P. J., Kvanstrom J., Norberg T., Lindberg A. A. and Svenson S. B. (19846) The conformation of SalmoneNu 0-antigenic oligosaccharides of serogroups A, B and D, inferred from ‘H- and 13C-nuclear magnetic resonance spectroscopy. Carhohydr. Res. 130, 35-53. Carlin N. I. A., Bundle D. R. and Lindberg A. A. (1987a) Characterization of five Shigeifn @exneri variant Y-specific monoclonal antibodies using defined saccharides and glycoconjugate antigens. J. Imm~n. 138, 441994427. Carlin N. I. A. and Lindberg A. A. (1983) Monoclonal antibodies specific for 0-antigenic polysaccharides of Shigella Jiexneri: Clones binding to iI, II: 3, 4 and 7, 8 epitopes. J. cl/n. Microbial. 18, 118331189. Carlin N. I. A., Svenson S. B. and Lindberg A. A. (1987b) Role of monoclonal O-antigen antibody epitope specificity and isotype in protection against experimental mouse typhoid. Microbial Pathogen. 2, 171-I 83. Classon B., Garegg P. J. and Norberg T. (1984) Synthesis of the 2’-acetate and some deoxy analogues of p-trifluoracetamidophenyl 3-O-(3,6-dideoxy-~-D-xylo-hexopy~dnosyl)2 -D-mannopyranoside, Acta Chem. &and. 838, 195-20 I. Galanos C., Liideritz 0. and Westphal 0. (1969) A new method for extraction of R lipopolysaccharides~ Eur. J. Biochem. 9, 245-249.

Grossman N., Svenson S. B., Leive L. and Lindberg A. A. (1990) Sabnonella 0 antigen-specific oligosaccharide--actyl conjugates activate complement via the alternative pathway at different rates depending on the structure of the Oantigen. Moiec. Immun. 27, 859-865. Hase S. and Rietchel E. Th. (1976) Methylation analysis of glucosaminitol and glucosaminyl-glucosaminitol disaccharides. Eur. J. Biochem. 63, 93-99. Hunter W. M. (1378) Radioimmunoassay. In ~andbouk oj E~~perimenta~immunology, 3rd Edn (Edited by Weir D. M.). pp. 14.1-14.40. Blackwell Scientific, Oxford. Jansson P. E., Kenne L., Liedgren H., Lindberg B. and Liinngren J. (1976) A practical guide to methylation analysis of carbohydrates. Clrem. Commun. (Unio Stockh.) 8. Jansson P. E., Lindberg A. A., Lindberg B. and Wollin R. (198 1) Structural studies on the hexose region of the core in lipopolysaccharides from Enterobacteriaceae. Eur. J. Biochem. 115, 571-577. J&beck H. J. A., Svenson S. B. and Lindberg A. A. (1979) Immunochemistry of Saimonella O-antigens: Specificity or rabbit antibodies against the O-antigen 4 determinant elicited by whole bacteria and O-antigen 4 specific saccharide-protein conjugates. J. rrnm~. 123, 1376-138 1. Kauffmann F. (1966) The Bacteriology of Enterobacter~a~eae. Munsgaard, Copenhagen.

Mapping of Salmonella 0:4 Lind S. M., Carlin N. I. A. and Lindberg A. A. (1985) Production and characterization of KDO-specific monoclonal antibodies recognizing lipopolysaccharides from heptoseless mutants of Salmonella. FEMS Microbial. Left. 28, 45-49.

McPherson G. A. (1983) A practical computer-based approach to the analysis of radioligand binding experiments. Comp. Prog. Biomed.

17, 107-114.

Munson P. J. and Rodbard D. (1980) LIGAND: a versatile computerized approach for characterization of ligandbinding system. Analyt. Biochem. 107, 220-239. Norberg T., Svenson S. B., Bock K. and Meldal M (1985) Immunochemistry of Salmonella O-antigens: studies of Salmonella BO antigen epitopes by enzyme-linked immunosorbent inhibition assay. FEMS Microbial. Left. 28, 171-176.

Pinchard R. N. (1978) Equilibrium dialysis and preparation of hapten conjugates. In Handbook of Experimental Immunology, 3rd Edn (Edited by Weir D. M.), p. 17.1. Blackwell Scientific, Oxford. Rose D. R., Cygler M., To R. J., Przybylska M., Sinnott B. and Bundle D. R. (1990) Preliminary crys-

specific monoclonal

antibodies

1023

ta1 structure analysis of an Fab specific for a Salmonella 0-polysaccharide antigen J. molec. Biol. 215, 489-492.

Sawardeker J. S., Sloneker J. H. and Jeanes A. R. (1965) Quantitative determination of monosaccharides as their alditol acetates by gas liquid chromatography. Analyt. Chem. 37, 1602-l 604.

Svenson S. B. and Lindberg A. A. (1978) Immunochemistry of Salmonella O-antigens: Preparation of an octasaccharidebovine serum albumin immunogen representative of Salmonella serogroup B O-antigen and characterization of the antibody response. J. Immun. 120, 1750-1757. Svenson S. B. and Lindberg A. A. (1979) Coupling of acid labile Salmonella specific oligosaccharides to macromolecular carriers. J. Immun. Meth. 25, 323-335. Valtonen M. V., Plosila M., Valtonen V. V. and Make12 P. H. (1975) Effect of the quality of the lipopolysaccharide on mouse virulence of Salmonella enteritidis. Infect. Immun. 12, 828-832. Westphal O., Liideritz 0. and Bister F. (1952) ijber die extraction von bacterien mit Phenol/Wasser. Z. Naturforsch. 7, 148-155.