Biochem J. (1980) 191, 437-447Printed in Great Britain
Antigenic determinants of a plant proteoglycan, the Gladiolus stylearabinogalactan-protein
P. A. GLEESON*tt and A. E. CLARKE**School ofBotany, University ofMelbourne, Parkville, Vic. 3052, Australia, and
tC.S.I.R.O. Division ofProtein Chemistry, Parkville, Vic. 3052, Australia
(Received 24 March 1980/Accepted 8 May 1980)
Antiserum has been raised to the arabinogalactan-protein of Gladiolus style mucilage.This macromolecule has been characterized and has a structure consistent with a
1 -3-linked f,-galactan backbone with side branches of 1 -.6-linked f,-galactosyl residues,some of which carry terminal a-L-arabinofuranoside residues [Gleeson & Clarke (1979)Biochem. J. 181, 607-6211. The specificity of the antiserum has been investigated byimmunoprecipitation with [3H]arabinogalactan-protein. The 3H label was introducedinto the arabinogalactan-protein by oxidation of the terminal galactose residues withgalactose oxidase, followed by reduction with NaB3H4. The antigenic specificity of theantiserum was shown to be directed towards the carbohydrate component of thearabinogalactan-protein. D-Galactose and L-arabinose were the most effective hapteninhibitors of the antiserum; other monosaccharides, N-acetyl-D-galactosamine, D-
galactono- 1 ,4-lactone, D-glucose, D-mannose, L-rhamnose, L-fucose and D-xylose,were all poor inhibitors. The antiserum showed preference for fl-galactosides over
a-galactosides. Of the haptens examined, the disaccharide 6-O-fl-D-galactopyranosyl-D-galactopyranose was the most potent inhibitor. The antigenic features of thearabinogalactan-protein were investigated by examining the interaction of the antiserumwith chemically and enzymically modified arabinogalactan-protein. Also, the cross-
reactivity of structurally related polysaccharides and glycoproteins with the specificantiserum was assessed by a haemagglutination assay using erythrocytes coupled withspecific antiserum. The results indicate that the dominant antigenic determinants of thearabinogalactan-protein are probably the side branches of 1 -.6-linked #-galactoseresidues bearing the terminal a-L-arabinose residues.
Many interactions between animal cells aremediated by cell-surface macromolecules (Hughes,1979; Frazier & Glaser, 1979). There is increasingevidence that surface macromolecules of plant cellsare also involved in a variety of cellular recognitionreactions (Albersheim & Anderson-Prouty, 1975;Heslop-Harrison, 1978; Clarke & Knox, 1979). Ourinterest is in the male-female recognition reaction inflowering plants, particularly the role of surfacesecretions of the female sexual tissues in the captureand recognition of compatible pollen, and in thenurture of the pollen tubes during their growththrough the style canal to the ovary. We have shownthat the major component of both the receptivestigma surface secretion and the style-canal exudateof Gladiolus is an arabinogalactan-protein (Gleeson& Clarke, 1979; Clarke et al., 1979; Gleeson &
t Present address: The Hospital for Sick Children,Toronto, Canada.
Clarke, 1980a). The style material has been wellcharacterized. It is polydisperse in the mol.wt. range150000-400000. It has a low content of associatedprotein (3%o) and the major monosaccharides of thecarbohydrate component are galactose and arabin-ose in the proportions 6 :1. The carbohydratecomponent was shown to be homogenous. Dataobtained from methylation analysis, mild acidhydrolysis and enzymic hydrolysis are compatiblewith a model based on 1-+3-linked galactanbackbone, branched through C(0)6 to side branchesof 1-.6-linked galactose residues, some of whichcarry terminal a-L-arabinofuranoside residues (Fig.1). Optical-rotation and lectin-binding studies indi-cated that the galactose residues are in the ,B-anomeric configuration (Gleeson & Clarke, 1979).Analysis of the stigma arabinogalactan showed it tobe structurally similar to the style arabinogalactan-protein (Gleeson & Clarke, 1980a).
Antibodies to isolated cell-surface glycocojugates
Fig. 1. A proposed model ofthe Gladiolus style arabinogalactan-protein
of this nature provide specific probes for investi-gating their development and distribution, as well astheir relationship to other glycoconjugates. We haveraised a specific antiserum to this isolated, struc-turally defined style arabinogalactan-protein; herewe report the specificity of the antiserum and a
partial definition of the antigenic determinants of thearabinogalactan-protein.
Experimental
MaterialsFormalin-fixed and heat-killed Staphylococcus
aureus (Cowan I) cells and goat anti-(rabbitglobulin) were purchased from CommonwealthSerum Laboratories, Melbourne, Vic., Australia.NaB3H4 (7Ci/mmol) and iodine-125 were obtainedfrom The Radiochemical Centre, Amersham, U.K.The NaB3H4 was stored frozen in 0.1ml portions(2 mCi) in 0.01 M-NaOH as described by Gahmberg(1978). D-Galactose, L-fucose, L-rhamnose, lactose,citrus pectin (grade II), Ceratonia siliqua (gumlocust bean), gum arabic (from Acacia senegal),porcine thyroglobulin (type II) and bovine sub-maxillary gland mucin (type 1) were obtained fromSigma Chemical Co., St. Louis, MO, U.S.A. Methyla - and ,B-galactopyranosides D-galactono- 1,4-lac-tone and 3-0-fl-D-galactopyranosyl-D-arabinosewere from Pfanstiehl Laboratories, Waukengan, IL,U.S.A. L-Arabinose and D-xylose were from Cal-biochem, San Diego, CA, U.S.A.; D-mannose andNonidet P40 were from BDH, Poole, Dorset, U.K.D-Glucose was from Ajax Chemicals, Melbourne,Vic., Australia. Araban was purchased from Koch-Light Laboratories, Colnbrook, Bucks., U.K.Ovalbumin was from Townsend and Mercer, Mel-
bourne, Vic., Australia. Human transferrin was
obtained from Miles Laboratories, Elkhart, IN,U.S.A. Bio-Gel P100 was purchased from Bio-RadLaboratories, Richmond, CA, U.S.A. Dextran wasfrom Pharmacia Fine Chemicals, Uppsala, Sweden.6-0-,-D-Galactopyranosyl-D-galactopyranose, syn-thesized as described by Baldo et al. (1978a), was agift from Dr. B. Baldo (Roche Research Instituteof Marine Pharmacology, Dee Why, N.S.W.,Australia). The Triticum (wheat) arabinogalactan-peptide, prepared by the method of Fincher & Stone(1974), was a gift from Dr. G. Fincher (Departmentof Biochemistry, La Trobe University, Bundoora,Vic., Australia). The Lolium multiflorum (rye-grass)endosperm arabinogalactan-protein, prepared bythe method of Anderson et al. (1977) was a giftfrom Professor B. A. Stone (Department of Bio-chemistry, La Trobe University, Bundoora, Vic.,Australia). Fetuin was a gift from Dr. C. Ward(Division of Protein Chemistry, C.S.I.R.O., Park-ville, Vic., Australia). Yeast mannan was preparedby the method of Peat et al. (1961). Arabino-galactans from the exudates of Gladiolus stigma andLilium stigma were prepared as previously described(Gleeson & Clarke, 1980a,b). Human erythro-cytes were supplied by the Royal MelbourneHospital, Melbourne, Vic., Australia.
MethodsIsolation of Gladiolus style arabinogalactan-
protein. The arabinogalactan-protein from Gladiolusstyle canal was isolated by affinity chromatographyon tridacnin-Sepharose 4B as previously described(Gleeson et al., 1979) and further chromatographedon a column (20cm x 1 cm) of Bio-Gel P100 in8 M-urea, to remove any non-covalently boundmaterial. The arabinogalactan-protein was recovered
1980
Galp1
6
438
Antigenic determinants of Gladiolus arabinogalactan-protein
in a single symmetrical peak that eluted with the voidvolume.
Preparation of a speciflc antiserum to theGladiolus arabinogalactan-protein. Specific anti-sera to the isolated Gladiolus arabinogalactan-protein were raised in rabbits. Two rabbits (NewZealand White) were inoculated with a relativelyhigh dose of antigen (1 mg) in Freund's completeadjuvant (Difco, Detroit, MI, U.S.A.); the antigenwas administered subcutaneously by multiple injec-tions in each of three sites, behind the neck and oneach flank, and the rabbits were re-inoculated 28days later with a further 1 mg of antigen emulsified inFreund's incomplete adjuvant. The rabbits were bledfrom the ear vein at 7, 10 and 14 days thereafter; thesera from the 7-day bleed gave the highest antibodytitre in both rabbits. The sera were frozen in 400,1portions at -700C. Sera when stored at -700Cshowed virtually no loss of activity over a 1-yearperiod; however, storage at -20 or 4°C resulted inappreciable loss of activity within months.
Immunodiffusion. Immunodiffusion, using themicro-slide apparatus (Gelman Instrument Co., AnnArbor, MI, U.S.A.) was performed in 1% (w/v)agarose (Behring Institut, Marburg, Germany)containing 0.15 M-NaCl and 0.02% NaN3 for 24 h at370C in a humidity chamber; the slides were thenexamined for the presence of immunoprecipitinbands.
Radiolabelling the Gladiolus arabinogalactan-protein. (a) 3H-labelling of terminal galactose resi-dues. The Gladiolus arabinogalactan-protein has ahigh content of terminal galactose residues (29%)(Gleeson & Clarke, 1979) and was labelled byspecific oxidation of these residues with galactoseoxidase, followed by reduction with NaB3H4 (Morell& Ashwell, 1972). The method employed wasessentially that of Morell & Ashwell (1972) andWinand & Kohn (1970). The arabinogalactan-protein (1.2mg) was dissolved in 2ml of buffer(0.02M-sodium phosphate/0.045 M-sodium acetate/0.15M-NaCl/1% toluene, pH7.0) to which 26 unitsof horseradish peroxidase (Calbiochem) and 44 unitsof galactose oxidase (Kabi AB, Stockholm, Sweden)were added. This mixture was incubated at 260C for50h. During the incubation, a precipitate formedthat subsequently redissolved. After incubation thereaction mixture was diluted 5-fold with 0.05 M-sodium phosphate buffer, pH 7.0, containing0.05 M-NaCl. Approx. lOmCi (1.4,umol) of NaB3H4was added and the mixture incubated at room tem-perature for 30min. To ensure complete reduction,unlabelled NaBH4 (3 mg) was added and incubationwas continued for a further 15 min. Excess NaBH4was then destroyed by lowering the pH to 4.6 with4M-acetic acid. The reaction mixture was dialysedfor 24h against distilled water (four changes) andfinally equilibrated with 0.15M-NaCl containing
Vol. 191
0.01 M-CaCl2. The 3H-labelled arabinogalactan-protein was then reisolated by affinity chromato-graphy on a column (1Ocm x 1 cm) of tridacnin-Sepharose 4B as previously described (Gleeson etal., 1979). The isolated [3Hiarabinogalactan-proteinhad a specific activity of approx. 6.7 x 106 d.p.s./mg.A portion of the [3Hiarabinogalactan-protein washydrolysed with 2.5M-trifluoroacetic acid at 100°Cfor 2h and chromatographed on Whatman no. 3paper in ethyl acetate/pyridine/water (8:2:1, byvol.) for 18h. The distribution of the 3H label on thechromatogram was determined by cutting thechromatogram into strips and counting the radio-activity in each strip; 75-80% of the activity co-chromatographed with galactose and a smallpercentage of the label (5%) was found close tothe origin.
(b) Iodination. Attempts to iodinate the arabino-galactan-protein with 1251I by the lactoperoxidasemethod (Marchalonis, 1969) were unsuccessful;tyrosine is present in the protein moiety of thearabinogalactan-protein (Gleeson & Clarke, 1979)but these residues are presumably inaccessible toiodination by this method.
Quantitative immunoprecipitation studies. Quanti-titative immunoprecipitation experiments wereperformed with the 3H-labelled arabinogalactan-protein. Two different methods of immunoprecipi-tation were compared. (a) Immunoprecipitationwith Staphylococcus aureus (Cowan 1; protein A-producing strain) cells. The formalin-fixed and heat-killed S. aureus cells were handled as described byKessler (1975). The cells were stored as supplied asa 10% suspension at -200C. Before use the cellswere centrifuged at 9000g for 30s in a Microfuge(Beckman Instrument Co.) and incubated in 0.5%Nonidet P40 in 50mM-Tris, 5mM-EDTA, 150mM-NaCl and 0.02% NaN3, pH 7.4 (Tris/EDTA/NaClbuffer) for 15min at room temperature. The cellswere then washed twice in Tris/EDTA/NaCl buffercontaining 0.05% Nonidet P40, and finally sus-pended in this 0.05% Nonidet P40/buffer solution asa 10% (v/v) suspension.The [3Hiarabinogalactan-protein in phosphate-
buffered saline (5mM-phosphate buffer/0.85%NaCl, pH 7.4) was added to the specific antiserum orto non-immune serum in a total volume of 50,1 andincubated for 1 h at 370C in 400,1 Microfuge tubes.A portion of the 10% (v/v) S. aureus suspensionwas then added, the amount used being sufficient forcomplete precipitation of the immunoglobulin frac-tion (lOO,l of cells for undiluted serum; 50,1 cellsfor all dilutions of serum). After 15 min at roomtemperature the cells were spun at 9000g for 20 s ina Microfuge (Beckman) and washed twice in 200,1of Tris/EDTA/NaCl buffer containing 0.05% Non-idet P40. The cells were resuspended using aSuper-Mixer (Lab-Line Instruments). The washed
439
P. A. Gleeson and A. E. Clarke
cells were finally resuspended in 200,1 of 0.5 M-NaOH. Portions (50,u1) were removed, neutralized,liquid scintillant (PCS; Amersham/Searle, ArlingtonHeights, IL, U.S.A.) (5 ml) added and the radio-activity measured in a liquid-scintillation counter(Isocap 300; Nuclear-Chicago, Des Plaines, IL,U.S.A.).
The titration of specific -antiserum with [3HI-arabinogalactan-protein (2500d.p.s., approx. 0.3 ,g)is shown in Fig. 2. Equivalence was obtained byusing a dilution of antiserum of 1 in 4, and 45% ofthe total radioactivity was precipitated. The extentof non-specific precipitation, determined by usingnon-immune serum in the same assay, was con-sistently below 0.5% of the total radioactivity.
(b) Indirect antibody precipitation. The PHI-arabinogalactan-protein in phosphate-buffered salinewas added to specific antiserum or non-immuneserum in a total volume of 75Al and incubated for1 h at 370C. Goat anti-(rabbit globulin) (25 p1) wasthen added at a concentration pre-determined to giveoptimal precipitation. The mixture was incubated fora further 1 h at 370 C, centrifuged at 9000g for1 min, and reincubated at 4°C overnight. Theimmunoprecipitate was collected by centrifugationat 9000g for min, and washed three times with200,ul portions of phosphate-buffered saline. Theprecipitate was then redissolved and the radioactivitymeasured as described above. Titration of specificantiserum by this method showed a maximum of25% of the total radioactivity precipitated atequivalence. Non-specific precipitation ranged
. 70
5: 60
o 50esa
0 501. i 40- 0
°0 30o 0
01> E 20CXS. _
G 10
1/250 1/8 1/21/20 1/4 1
Antiserum dilutionFig. 2. Titration of the specific antiserum with PHi-
arabinogalactan-proteinA portion (25 pl) of [PHlarabinogalactan-protein(2500d.p.s. or about 0.3,ug) in phosphate-bufferedsaline, pH 7.0, was added to a range of dilutions ofthe specific antiserum in a total volume of 50,l andincubated for lh at 370C. Staphylococcus aureuscell suspension (10%, v/v) was added and themixture incubated at room temperature for 15min.The cells were washed and the radioactivity deter-mined as described under 'Methods'. Each deter-mination was carried out in duplicate.
between 1 and 5% of the total radioactivity. Thusimmunoprecipitation by this method gave lowerspecific precipitation and higher non-specific pre-cipitation than immunoprecipitation using S. aureuscells. Method (a) was used for subsequent experi-ments in the present study. The relative efficiency ofimmunoprecipitation with S. aureus cells has beennoted previously (Kessler, 1975).
Inhibition studies were performed by preincu-bation of the antiserum with inhibitor in phos-phate-buffered saline, pH7.0, at 370C for 30min.The final dilution of the antiserum in this incubationmixture was 1 in 4. Portions of the antiserum-inhibitor mixture (25,p) were placed in 400,u1Microfuge tubes together with 25,u1 of [3Hiarabino-galactan-protein (2500d.p.s. or approx. 0.3,ug) andincubated at 370C for 1 h. S. aureus cells (50,u1 of a10% suspension) were added and the incubationcontinued at room temperature for 15 min. The cellswere washed and the radioactivity measured asdescribed above. Control assays in the absence ofinhibitors were performed at the same time. Allinhibition assays were performed in duplicate, andcontrol assays in triplicate.
Enzymic and chemical modifications of theGladiolus arabinogalactan-protein. (a) Enzymichydrolysis. The arabinogalactan-protein was treatedwith a-L-arabinofuranosidase as previously de-scribed (Gleeson & Clarke, 1979). Analysis of thetreated material showed that all arabinose residueswere removed (Gleeson & Clarke, 1979); thisarabinose-depleted material is referred to as the'galactan-protein'.
(b) Mild acid hydrolysis. The arabinogalactan-protein was hydrolysed in 12.5 mM-oxalic acid at100°C for 5 h and the hydrolysate fractionated byprecipitation with 80% (v/v) ethanol (Gleeson &Clarke, 1979). Analysis of the fractions showed thatall the arabinose had been removed, as well as 25%of the galactose residues, which were released bothas the free monosaccharide and as galactosyloligosaccharides (Gleeson & Clarke, 1979).Methylation analysis of the galactan remaining afterhydrolysis (ethanol-insoluble fraction) indicated thatthe molecules had been degraded both at the sidechains and at the backbone (Gleeson & Clarke,1979).
(c) Periodate oxidation. The arabinogalactan-protein (5mg) was dissolved in sodium periodate(0.02M; 5ml) and incubated at 40C in the dark.Samples (100,ul) were removed at intervals (1, 4, 7,24, 48, 72h), diluted to 25 ml with distilled water andthe A223 measured. Oxidation was complete after 7 h.During oxidation of the arabinogalactan-protein,0.92mol of periodate/mol of saccharide residueswere consumed. The theoretical periodate uptake,based on methylation data (Gleeson & Clarke,1979) is 0.83mol of periodate/mol of saccharide
1980
440
Antigenic determinants of Gladiolus arabinogalactan-protein
residues. To monitor the efficiency of the period-ate-oxidation procedure, a sample of maltose wasoxidized at the same time under the same con-ditions; the observed periodate uptake equalled thetheoretical value. The oxidized arabinogalactan-protein was dialysed exhaustively against phos-phate-buffered saline, pH7.0, and stored at 40C.Under these conditions the oxidized arabino-galactan-protein remained in solution, but afterfreeze-drying or freezing the solution at -200C, itbecame totally insoluble.Removal of sialic acid from glycoproteins. Sialic
acid was removed from glycoproteins by hydrolysisin 0.05 M-H2SO4 at 800C for 1 h (Kieda et al., 1978).The desialylated glycoproteins were dialysed ex-haustively against distilled water and freeze-dried.
Haemagglutination assay using specific antiserumcoupled to erythrocytes. The globulin fraction of theanti-(arabinogalactan-protein) serum was precipi-tated with 40%-satd. (NH4)2SO4, washed once with40% (NH4)2SO4, dissolved in distilled water anddialysed exhaustively against 0.9% NaCl. Theglobulin fraction (1 mg) was coupled to humanerythrocytes (125,u1 of packed cells in 2 ml of 0.9%NaCl) by using 0.1% (w/v) chromic chloride(diluted from a 'matured' 1% (w/v) stock solution]as described by Parish & McKenzie (1978). Variousamounts of the 0.1% (w/v) chromic chloride weretested, and the amount that gave optimal coupling ofthe globulin fraction to the erythrocytes was used.The washed coupled erythrocytes were diluted to a2% (v/v) suspension and stored at 40C, in thepresence of 0.5% bovine serum albumin as astabilizer, for periods up to one week. Before use theywere washed once and resuspended to 2% (v/v) inphosphate-buffered saline, pH 7.4.
The cross-reactivity of a number of polysac-charides and glycoproteins with the anti-(arabino-galactan-protein) serum was assessed by a haemag-glutination assay using the specific-antiserum-coupled erythrocytes. Doubling dilutions of thesamples (25,u1) were made in a microtitre tray(Disposable Products, Adelaide, S. Australia,Australia) and an equal volume of the couplederythrocytes was added. The trays were incubated atroom temperature for 1 h and the lowest dilution atwhich agglutination was detected was taken as theend point. As controls to the specificity of thehaemagglutination, assays were performed in asimilar manner by using untreated erythrocytes aswell as erythrocytes coupled with non-immuneserum.
Results
The immune sera of both rabbits gave two bandswith the Gladiolus arabinogalactan-protein inimmunodiffusion tests; however, in both cases the
Vol. 191
sera had to be concentrated 2-fold for theseprecipitin bands to be observed. The followingstudies on the specificity of the antiserum toGladiolus arabinogalactan-protein were carried outwith serum obtained from a single rabbit.
Inhibition of the [3Hlantigen-antibody binding withsaccharide inhibitors
The specificity of the anti-(arabinogalactan-pro-tein) serum was determined initially by studying theability of various mono- and di-saccharides to inhibitbinding of specific antiserum to [3Hlarabino-galactan-protein. The hapten-inhibition curves of the[3Hlantigen-antibody binding by mono- and di-saccharides are shown in Fig. 3. The haptenconcentrations required for 50% inhibition of thebinding between [3Hlarabinogalactan-protein andantiserum are shown in Table 1. Of the haptensexamined, the disaccharide 6-0-fJ-D-galactopyrano-syl-D-galactopyranose was the most potent inhibi-tor. It was approx. 80 times more effective on amolar basis than the next most powerful inhibitor,methyl ,6-D-galactopyranoside, and 280 times moreeffective than D-galactose. The antiserum showedpreference for the galactosides in the ,B-configura-tion, as the ,J-galactosides examined, methyl 1)-D-galactopyranoside, 3-0-fI-D-galactopyranosyl-D-arabinose and 4-O-fl-D-galactopyranosyl-D-gluco-pyranose (lactose) were all considerably betterinhibitors (about 10 times) than the only galactosidein the a-configuration tested, methyl a-D-galacto-pyranoside. The monosaccharide L-arabinose wasequally as effective an inhibitor as D-galactose.However, D-arabinose in the disaccharide 3-0-f,-D-galactopyranosyl-D-arabinose did not enhancethe inhibitory capacity of the galactoside.
The inhibition curves of the effective inhibitors allshowed a maximum of 70-80% inhibition at thehighest concentrations tested. At a concentration of
Table 1. Comparison ofsaccharides as hapten inhibitorsof the 13Hlarabinogalactan-protein-specific antiserum
Fig. 3. Saccharide inhibition ofthe [3H]arabinogalactan-protein-specific antiserum bindingInhibition of binding of [3Hlarabinogalactan-protein (2500d.p.s. or about 0.3,ug) to specific antiserum (dilution 1 in4). The immunoprecipitation assays were performed as described under 'Methods'. The inhibitor concentration is thatin the incubation mixture before addition of Staphylococcus aureus cells. All inhibition assays were performed induplicate and analyses varying by more than + 10% were rejected and the assay repeated. Triplicate assays in theabsence of inhibitors were performed at the same time and the average of these assays was used in calculatingthe percentage inhibition. Symbols: *, 6-O-6-D-galactopyranosyl-D-galactopyranose; 0, methyl f,-D-galacto-pyranoside; l, 3-0-,8-D-galactopyranosyl-D-arabinose; A, 4-0-,6-D-galactopyranosyl-D-glucopyranose (lactose);0, D-galactose; o, L-arabinose; *, methyl a -D-galactopyranoside; *, L-rhamnose;-&, N-acetyl-D-galactosamine; 0,D-xylose; o), D-glucose; *, L-fucose; 0, D-mannose; 4 D-galactono-1,4-lactone.
100mM, both D-galactose and L-arabinose showed70% inhibition; there was only a marginal increasein inhibition when both monosaccharides, each at100mm, were tested together.A range of other related monosaccharides, i.e. N-
glucose, D-mannose, L-rhamnose, L-fucose andD-xylose, were all poor inhibitors.
Inhibition of the [3Hlantigen-antibody binding withenzymically and chemically modified arabino-galactan-protein and arabinoxylan
The antigenic features of the arabinogalactan-protein were investigated by examining the con-
sequence of modifying the arabinogalactan-proteinboth enzymically and chemically. This was assessedby assaying the ability of the modified arabino-galactan-protein to inhibit binding between ['HI-arabinogalactan-protein and specific antiserum. Theresults are shown in Fig. 4. On a weight basis, thegalactan-protein gave an inhibition curve similar tothat given by the native arabinogalactan-protein, andboth gave 100% inhibition at 250,ug/ml. The oxalicacid-treated arabinogalactan-protein also gave an
inhibition curve similar to that produced by thenative material; however, at low concentrations, theoxalic acid-treated material was a slightly moreeffective inhibitor than either the native arabino-galactan-protein or the galactan-protein, and at highconcentrations it was a slightly less effective inhibi-tor, giving a maximum inhibition of 90% at250,ug/ml. Periodate oxidation destroyed virtuallyall the antigenic determinants, as the oxidizedmaterial was only a weak inhibitor (Fig. 4). Thepolysaccharide arabinoxylan was not an effectiveinhibitor of the antigen-antibody binding.
Antigenic cross-reactivity of polysaccharides andglycoproteins with anti-(arabinogalactan-protein)serum
The cross-reactivity of a number of polysac-charides and glycoproteins with the anti-(arabino-galactan-protein) serum was assessed by using ahaemagglutination assay with the specific antiserumcoupled to erythrocytes. The relevant structuralfeatures of the polysaccharides and glycoproteinsexamined are given in Table 2, together with theresults, which are expressed as the lowest con-
1980
00.-
.0
I0
442
Antigenic determinants of Gladiolus arabinogalactan-protein
100 J
80
40
26
0 10 100 1000
[Inhibitor] (,ug/ml)Fig. 4. Inhibition of the [3H]arabinogalactan-protein-specific antiserum binding with enzymically and chemi-cally modified arabinogalactan-protein and arabinoxylan
Inhibition of binding of [3Hlarabinogalactan-pro-tein (2500d.p.s. or about 0.3,ug) to specific anti-serum (dilution 1 to 4). The immunoprecipitationassays were performed as described under'Methods'. The inhibitor concentration is that in theincubation before addition of Staphylococcus aureuscells. All assays were performed in duplicate andthose varying by more than + 10% were rejected andthe assay repeated. Triplicate assays in the absence ofinhibitors were performed at the same time and theaverage of these assays were used in calculating thepercentage inhibition. Symbols: 0, unlabelledarabinogalactan-protein; El, galactan-protein; 0,oxalic acid-treated arabinogalactan-protein; A,periodate-oxidized arabinogalactan-protein; U,arabinoxylan.
centration giving detectable haemagglutination.Haemagglutination with the native Gladiolus stylearabinogalactan-protein was detectable at 31,ug/ml.The galactan-protein and the oxalic acid-treatedarabinogalactan-protein showed no detectable haem-agglutination at 500,ug/ml and 1000lg/ml respect-ively. The antiserum cross-reacted strongly withthe closely related arabinogalactan from Gladiolusstigma surface exudate and to a lesser extent witharabinogalactans from Triticum and from Loliummultiflorum (rye-grass) endosperm cell-culturemedium. The stigma surface exudate of Liliumlongiflorum also contains arabinogalactans (Aspinall& Rosell, 1978). Two groups of arabinogalactanshave been separated from this exudate by affinitychromatography on tridacnin-Sepharose 4B. Boththe bound and unbound fractions contain arabino-galactans (Gleeson & Clarke, 1980b) that cross-reacted strongly with the specific antiserum. Thewheat arabinoxylan cross-reacted strongly, haem-agglutination being detected at a concentration of15,ug/ml. A pectin preparation containing a mixtureof cell-wall components cross-reacted strongly, asdid the galactomannan from Ceratonia siliqua seeds.Mannan, dextran, araban and gum arabic showedessentially no cross-reactivity. A number of glyco-proteins were also examined; all except for oval-bumin were desialylated to expose terminal galactose
Table 2. Cross reactivity ofpolysaccharides and glycoproteins with antiserum to Gladiolus arabinogalactan-proteinThe antiserum to Gladiolus arabinogalactan-protein was coupled to human erythrocytes as described under'Methods'. Cross-reactivity of the polysaccharides and glycoproteins was assessed by a haemagglutination assayusing erythrocytes coupled with specific antiserum. Control assays were performed by using untreated erythrocytesas well as erythrocytes coupled with non-immune serum.
SourcePolysaccharides
(a) Arabinogalactan-protein Gladiolus stylecanal
(b) Arabinogalactan-protein Gladiolus styletreated with a-L- canalarabinofuranosidase
(c) Arabinogalactan-protein Gladiolus styletreated with oxalic acid canal
1 - 3-linked f-galactan back-bone with side chains of1-+6-linked f-galactoseresidues substituted withRhaf, Araf, Galp, GlcpA,Arap
1 -4-linked f-xylan backbonewith single a-L-arabinose(f) residues as side branches
Contains a branched polymerof L-arabinose
1 -4-linked f-mannan withsingle 1-+6-linkedgalactose residues as sidebranches
Mixture of soluble cell-wallcomponents
1-+6-linked (x-mannan withside branches of 1 -.2- and1-+3-linked mannose residues
1-6-linked a-glucan withside branches of 1-3-linked a-glucose residues
Terminal saccharide sequence(Man),-GlcNAc ...
f-Gal-(1+3)-a-GalNAc. . .
4-Gal-(1 4)-#-GlcNac ...
#-Gal-GlcNAc
(Man),-GlcNAc...4-Gal-(l-I4)-f-GlcNAc...
GalNAc. . . (major)
Gal. . . (minor)
Lowestconcentration
giving detectablehaemagglutination
(pg/ml)500
ReferenceFincher et al.
(1974);Neukom &Markwalder(1975)
Aspinall &Rosell (1978)
78 Gleeson &Clarke(1980b)
155 Gleeson &Clarke(1980b)
8000 Aspinall (1969)
15 Perlin (1951);Neukomet al. (1967)
8000 Hough &Powell(1960)
250 Smith &Montgomery(1959)
125
8000 Ballou (1974)
8000 Walker (1978)
>4000 Spiro (1973)>8000 Spiro (1973)
Spiro &Bhoyroo(1974)
>8000 Jamieson et al.(1971)
>4000 Fukuda &Egami(1971a,b)
>8000 Gottschalk &Graham(1959)
Bertolini &Pigman(1970)
1980
444
Antigenic determinants of Gladiolus arabinogalactan-protein
residues. However, none of the glycoproteinsshowed any haemagglutination at the highest con-centrations used.
Discussion
Many cell-surface components of animal cells,yeast cells and other micro-organisms are antigenic.In some cases the antigenic components are thecarbohydrate moieties of surface glycoconjugates,e.g. human blood-group-specific antigens (Watkins,1978) and microbial surface antigens (Jann &Westphal, 1975).
Increasingly, immunological approaches are beingused to investigate the nature and function of plantsurface components (Knox, 1980), but, in mostcases, little chemical characterization of the anti-genic components has been undertaken. We haveraised a specific antiserum to the Gladiolus stylearabinogalactan-protein and have carried outimmunochemical studies to describe the nature ofthe antigenic determinants. The antiserum is directedtowards the carbohydrate component of the arabino-galactan-protein. This is indicated by the destruc-tion of antigenic determinants by periodate oxida-tion (Fig. 4). Further, hapten-inhibition studiesshowed that the specificity of the antiserum wasdirected to the monosaccharide components D-galactose and L-arabinose. This was confirmed bythe ability of the antiserum to bind the poly-saccharides arabinoxylan and galactomannan,which have a high proportion of accessible terminalarabinose and galactose residues respectively.Although hapten-inhibition studies indicated a pre-ference for galactosides in the fl-configuration,a-galactosides were also effectively bound in boththe hapten inhibition and the haemagglutinationassays. This is in keeping with the f-configuration ofgalactose residues in the arabinogalactan-protein(Gleeson & Clarke, 1979).None of the monosaccharides tested gave 100%
inhibition of the antigen-antibody binding. Thisprobably reflects the relatively low affinity ofmonosaccharides for the antibody-binding sitescompared with that of the antigenic determinantsof the arabinogalactan-protein, suggesting that thedeterminants include a number of saccharide resi-dues. The binding sites of other anti-glycosyl anti-bodies are known to extend over a number ofsaccharide residues, for example a hexasaccharide inthe dextran-anti-dextran interaction (Kabat, 1966)and a tetrasaccharide in the antigenic determinantsof human blood-group antigens A and B (Lloyd etal., 1966). Other carbohydrate-binding proteins, e.g.lectins and glycosidases, are also known, in anumber of cases, to accommodate an extendedregion of the saccharide chain in their binding sites(Goldstein & Hayes, 1978; Blake et al., 1967).
Vol. 191
Inhibition studies of the anti-(arabinogalactan-pro-tein) serum with disaccharides showed that 6-O-fl-D-galactopyranosyl-D-galactopyranose was afar more potent inhibitor than was the mono-saccharide D-galactose, also suggesting that thebinding site can accommodate at least two galactoseresidues.
Further characterization of the antigenic deter-minants by hapten inhibition is limited by theavailability of model oligosaccharides. Therefore analternative approach was adopted; the arabino-galactan-protein was chemically and enzymicallymodified and the effect of these modifications on theantigenic determinants was assessed.
Periodate oxidation of the arabinogalactan-pro-tein causes extensive degradation of side branchescontaining 1_-6-linked galactose residues, terminalgalactose and arabinose residues, but nooxidation of the 1-.3-linked galactan backbonewould be expected. The loss of antigenic deter-minants after periodate oxidation indicates that theantigenic determinants are located on the sidebranches. This location of determinants at the sidebranches is a common feature of polysaccharideantigens (Jann & Westphal, 1975); e.g. the anti-genic determinants of a number of yeast mannansare side branches containing 1_-.3- and 1-.2-linkedmannose residues (Ballou, 1970). Apparently, in thearabinogalactan-protein, both arabinose andgalactose are included in the antigenic determinantsas both monosaccharides inhibited the antigen-antibody binding. The question of the number of setsof specific antibodies has not been resolved. How-ever, it is likely that a single set of specific antibodiescapable of binding to determinants containing bothD-galactose and L-arabinose is involved, as a mixtureof these monosaccharides inhibited the antigen-antibody binding to a similar extent as did theindividual monosaccharides. There is some indirectevidence that the galactose residues may be moreimportant than the arabinose residues in deter-mining the extent of binding by the specific anti-serum; thus the antiserum can bind to at least twogalactosyl residues; it also has a high affinity forboth the arabinose-free galactan-protein and oxalicacid-treated arabinogalactan-.protein. In contrast,arabinoxylan was a poor inhibitor of the binding(Fig. 4). Oxalic acid hydrolysis of the arabino-galactan-protein not only removes all the arabinosebut also fragments the molecule at both the sidebranches and the backbone, to give an in-creased proportion of 1-.6-linked galactose residues(Gleeson & Clarke, 1979). This high proportion of1-_6-linked galactose residues could account for itsefficiency as an inhibitor.The differences in effectiveness of various poly-
saccharides in the two assays reflects the funda-mental differences in the assays; the inhibition
445
446 P. A. Gleeson and A. E. Clarke
assays measure relative affinities of the determi-nants for the antiserum; on the other hand thehaemagglutination assay, using antibody-coatederythrocytes, measures the extent of cross-linkingbetween the cells and may be dependent on thenumber and distribution of the determinants, as wellas their affinity. The cross-reactivity of relatedarabinogalactans confirms the specificity of the anti-serum. The arabinogalactan from Gladiolus stigmais structurally most similar to the native antigenand this is reflected in its strong cross-reactivity withthe antiserum. The arabinogalactan-protein fromLolium multiflorum endosperm cells cross-reacts toa lower extent, possibly owing to its more highlybranched nature, and the low molecular weight ofthe wheat arabinogalactan-peptide (22000) is con-sistent with its poorer cross-reactivity. Althoughgum arabic shares some structural features with theGladiolus style arabinogalactan-protein, there aremajor differences in the terminal sequences, con-sistent with its poor cross-reactivity. The araban didnot cross-react; this polysaccharide preparationcontains a number of monosaccharide components,but the proportion and arrangement of the arabin-ose residues is not known. Arabinoxylan, which hada low affinity for the antiserum, cross-reactedstrongly in the haemagglutination assay, presum-ably because the xylan chains are extensivelysubstituted with terminal arabinose residues. Neitherthe oxalic acid-treated Gladiolus arabinogalactan-protein nor the galactan-protein showed any detect-able haemagglutination. The fragments of thearabinogalactan-protein remaining after oxalic acidhydrolysis may have been too small to alloweffective cross-linking of the erythrocytes. Thenon-reactivity of the galactan-protein was, however,unexpected, but may be related to its alteredsolubility properties (Gleeson & Clarke, 1979).None of the glycoproteins examined showed any
detectable haemagglutination; all except ovalbuminhad terminal non-reducing galactose residues, andthe non-reactivity probably reflects a low content ofthese residues in the glycoproteins. In contrast,galactomannan, which has high content of terminala -galactose residues, cross-reacted strongly.A number of mouse myeloma immunoglobulin A
proteins have been described that also bind specifi-cally to polysaccharides containing 1--6-linked,B-D-galactosides (Glaudemans, 1975). Oligosacchar-ides containing these linkages were more potentinhibitors of the myeloma proteins than was themonosaccharide, with the 1-+6-linked fl-D-galacto-tetraose being the most effective inhibitor tested(Jolley et al., 1974; Glaudemans, 1975). Theimmunoglobulin A protein of one of these, the J539myeloma, has been shown to bind to the Gladiolusarabinogalactan-protein (Gleeson & Clarke, 1979)as well as to a number of other arabino-3,6-
galactans (Baldo et al., 1978b). Thus the specificityof this myeloma protein is similar to the anti-(arabinogalactan-protein) serum described here,except that the anti-(arabinogalactan-protein) serumalso binds terminal a-L-arabinofuranose, residues,whereas the myeloma proteins tested were not inhi-bited by a-L-arabinofuranoside (Potter et al., 1972).
In summary, the antiserum to Gladiolus arabino-galactan-protein is directed to side chains of thecarbohydrate component; both monosaccharides,D-galactose and L-arabinose, contribute to theantigenic determinants.
Carbohydrate-specific antibodies, together withlectins, are extremely useful tools in investigations ofboth the form and function of glycoconjugates. Theantiserum described here provides an additionalspecific reagent for such investigations.
We gratefully acknowledge gifts of material from: Dr.B. Baldo, Roche Research Institute of Marine Pharmac-ology, Dee Why, N.S.W., Australia; Dr. G. Fincher,Department of Biochemistry. La Trobe University, Bun-doora, Vic., Australia; Professor B. A. Stone, Depart-ment of Biochemistry, LaTrobe University, Bundoora,Vic., Australia; Dr. C. Ward, Division of ProteinChemistry, C.S.I.R.O., Parkville, Vic., Australia; Profes-sor H. Neukom, Swiss Federal Institute of Technology,Zurich, Switzerland. We would like to thank ProfessorB. A. Stone, Department of Biochemistry, La Trobe Uni-versity, Bundoora, Vic., Australia, Professor S. J. Leach,Department of Biochemistry, University of Melbourne,Parkville, Vic., Australia, and Professor R. B. Knox,School of Botany, University of Melbourne, Parkville,Vic., Australia, for critical review of the manuscript. Thiswork was supported in part by the Australian ResearchGrants Committee.
References
Albersheim, P. & Anderson-Prouty, A. J. (1975) Annu.Rev. Plant Physiol. 26, 31-52
Anderson, R. L., Clarke, A. E., Jermyn, M. A., Knox,R. B. & Stone, B. A. (1977) Aust. J. Plant. Physiol. 4,143-158
Aspinall, G. 0. (1969) Adv. Carbohydr. Chem. Biochem.24, 333-379
Aspinall, G. 0. & Rosell, K. G. (1978) Phytochemistry17, 919-921
Baldo, B. A., Sawyer, W. M., Stick, R. V. & Uhlenbruck,G. (1978a) Biochem. J. 175,467-477
Baldo, B. A., Neukom, H., Stone, B. A. & Uhlenbruck,G. (1978b) Aust. J. Biol. Sci. 31, 149-160
Ballou, C. E. (1970) J. Biol. Chem. 245, 1197-1203Ballou, C. E. (1974)Adv. Enzymol. 40, 239-270Bertolini, M. & Pigman, W. (1970) Carbohydr. Res. 14,
53-63Blake, C. C. F., Mair, G. A., North, A. C. T., Phillips,
D. C. & Sarma, V. R. (1967) Proc. R. Soc. LondonSer. B 167, 365-388
1980
Antigenic determinants of Gladiolus arabinogalactan-protein 447
Clarke, A. E. & Knox, R. B. (1979) Dev. Comp.Immunol. 3,571-589
Clarke, A. E., Gleeson, P., Harrison, S. & Knox, R. B.(1979) Proc. Natl. Acad. Sci. U.S.A. 76, 3358-3362
Fincher, G. B., Sawyer, W. M. & Stone, B. A. (1974)Biochem. J. 139,535-545
Fincher, G. B. & Stone, B. A. (1974) Aust. J. Biol. Sci.27, 117-132
Frazier, W. & Glaser, L. (1979) Annu. Rev. Biochem. 48,491-523
Fukuda, M. & Egami, F. (197 la) Biochem. J. 123,407-414
Fukuda, M. & Egami, F. (1971b) Biochem. J. 123,415-420
Gahmberg, C. G. (1978) Methods Enzymol. 50, 204-206Glaudemans, C. P. J. (1975) Adv. Carbohydr. Chem.
Biochem. 31, 313-346Gleeson, P. A. & Clarke, A. E. (1979) Biochem. J. 181,
607-621Gleeson, P. A. & Clarke, A. E. (1980a) Carbohydr. Res.
in the pressGleeson, P. A. & Clarke, A. E. (1980b) Phytochemistry in
the pressGleeson, P. A., Jermyn, M. A. & Clarke, A. E. (1979)
Anal. Biochem. 92, 41-45Goldstein, I. J. & Hayes, C. D. (1978) Adv. Carbohydr.
Chem. Biochem. 35, 128-340Gottschalk, A. & Graham, E. R. B. (1959) Biochem.
Biophys. Acta 34, 380-391Heslop-Harrison, J. (1978) Cellular Recognition Systems
in Plants (Studies in Biology no. 100), E. Arnold,London
Hough, L. & Powell, D. B. (1960) J. Chem. Soc. 16Hughes, R. C. (1979) in Glycoconjugate Research
(Gregory, J. D. & Jeanloz, R. W., eds.), vol. 2, pp.985-1005, Academic Press, London and New York
Jamieson, G. A., Jett, M. & DeBernardo, S. L. (1971) J.Biol. Chem. 246, 3686-3693
Jann, K. & Westphal, O. (1975) in The Antigens (Sela,M., ed.), vol. 3, pp. 1-125, Academic Press, Londonand New York
Jolley, M. E., Glaudemans, C. P. J., Rudikoff, S. &Potter, M. (1974) Biochemistry 13, 3179-3183
Kabat, E. A. (1966)J. Immunol. 97, 1-11Kessler, S. W. (1975) J. Immunol. 115, 1617-1624Kieda, C. M. T., Bowles, D. J., Ravid, A. & Sharon, N.
(1978) FEBS Lett. 94, 391-396Knox, R. B. (1980) in The Antibody as a Tool (Warr, G.& Marchalonis, J. J., eds.), John Wiley and Sons, NewYork, in the press
Lloyd, K. O., Kabat, E. A., Layug, E. J. & Gruezo, F.(1966) Biochemistry 5, 1489-1501
Marchalonis, J. J. (1969) Biochem. J. 113, 299-302Morell, A. G. & Ashwell, G. (1972) Methods Enzymol.
28,205-208Neukom, H. & Markwalder, H. (1975) Carbohydr. Res.
39, 387-389Neukom, H., Providoli, L., Gremli, H. & Mui, P. A.
(1967) Cereal Chem. 44, 238-244Parish, C. R. & McKenzie, I. F. C. (1978) J. ImmunoL
Methods 20, 173-183Peat, S., Whelan, W. J. & Edwards, T. E. (196 1)J. Chem.
Soc. 29Perlin, A. S. (1951) Cereal Chem. 28,382-393Potter, M., Mushinski, E. B. & Glaudemans, C. J. P.
(1972)J. Immunol. 108, 295-300Smith, F. & Montgomery, R. (1959) ACS Monogr. 141,
pp. 324-338, Reinhold, New YorkSpiro, R. G. (1973) Adv. Protein Chem. 27, 349-467Spiro, R. G. & Bhoyroo, V. D. (1974) J. Biol. Chem. 249,
5704-5717 t
Walker, G. J. (1978) Int. Rev. Biochem. 16, 75-126Watkins, W. M. (1978) Proc. R. Soc. London Ser. B
202, 3 1-53Winand, R. J. & Kohn, L. D. (1970) J. Biol. Chem. 245,
