We have observed that antidextran hybridomas with potential N-linked glycosylation sites in VH have higher affinity for polymeric dextran and for isomaltoheptaose than those lacking potential glycosylation sites. In these studies we have used gene transfection and expression techniques to verify that the carbohydrate addition sites in VH were used. The carbohydrate of the VH region was accessible for binding by the lectin Con A. By ELISA analysis it was demonstrated that the aKa of the antibody for dextran was influenced by the presence of carbohydrate in VH, with the aglycosylated antibody having an aKa 15-fold lower than its untreated counterpart. The aKa for antigen of antibodies that contain carbohydrate only in their constant region was unaffected by lack of carbohydrate. Thus, not only the amino acid sequence of the variable region but also its carbohydrate moieties can determine the magnitude of the antigen-antibody interaction.

The fine structure of the combining site of human mAb IgMNOV to poly-alpha(2----8)linked NeuNAc, the epitope of the group B meningococcal and E. coli K1 polysaccharides, has been probed using RIA and ELISA. Inhibition by oligomers ranging from 2 to 12 residues was used to assay binding to IgMNOV by group B meningococcal polysaccharide preparations (GBMP) or by poly(A). The inhibitory properties of the oligomers were almost identical in both assays of the binding of GBMP to horse IgM (H46). This evidence and the finding that both GBMP and poly(A) precipitated IgMNOV equally per unit weight indicated that the epitope of poly(A) must mimic an equivalent epitope on GBMP despite the absence of any apparent common structural features in the two molecules. Unlike most carbohydrate-anticarbohydrate systems in which the site is saturated by oligomers of up to six or seven sugars, all the anti-alpha(2----8)NeuNAc systems above required much larger oligomers. Because these oligomers are larger than the maximum size of an antibody site the epitope must be conformationally controlled, and this has been confirmed by nuclear magnetic resonance spectroscopy. However, despite the above similarities, GBMP and poly(A) were differentiated in that only GBMP bound to H46. Smaller linear molecules obtained by delipidating the GBMP, as well as periodate-oxidized GBMP with its nonreducing end oxidized or linked covalently to BSA, bound to and precipitated IgMNOV and H46. This showed that, despite their differences, terminal nonreducing ends were not involved and that both epitopes were located in the conformationally controlled inner residues of the GBMP. The difference thus must reside in the ability of IgMNOV and H46 to recognize different structural aspects of the same conformationally controlled inner residues. The ELISA data indicate that both IgMNOV and H46 have groove-type sites that bind exclusively to an epitope located on the acidic side of the inner residues. The differences determining the ability of IgMNOV and the failure of H46 to cross-react with poly(A), poly(I), and denatured DNA, may depend on differences in the degree of protonation required by each antibody, and this may be clarified by a study of the effects of pH on the precipitin behavior of IgMNOV and H46.

We have described an IgM antibody from a patient with macroglobulinemia specifically reacting with poly-alpha(2----8)N-acetyl neuraminic acid (NeuNAc) the capsular polysaccharide of two important human pathogens, group B meningococcus and E. coli K1. This antibody has a narrowly defined specificity in its interactions with polysaccharides, being unable to bind poly-alpha(2----9)NeuNAc or alternating poly-alpha(2----8)alpha(2----9)NeuNAc. However, it shows interesting crossreactivity with seemingly unrelated polynucleotides and denatured DNA, supporting the hypothesis that charged groups with a given spacing may determine the specificity of antigen-antibody interactions on otherwise dissimilar molecular structures. Despite the crossreactivity with denatured DNA and polynucleotides, the antibody does not appear to have adverse effects in the patient. The antibody protects newborn rats against E. coli K1 infection, as well as the standard horse antiserum H46, and one would expect it to prove useful in humans as an adjunct to antibiotic therapy in infections with group B meningococcus and E. coli K1. We have attempted to clone the antibody-producing cells from peripheral blood, and have shown that the relevant cells are present and can be cultured.

Six monoclonal IgM from patients with Waldenstrom's macroglobulinemia that react with Klebsiella polysaccharides were tested for their ability to bind to nucleic acid antigens. One of the macroglobulins bound to the polynucleotide poly(G), and one bound to poly(G), poly(I), and single-stranded DNA. The reaction with the polynucleotides was specifically inhibited by the Klebsiella polysaccharide K30. A monoclonal lupus anti-DNA antibody (16/6) was found to react weakly with the Klebsiella polysaccharides K30 and K21. Five of the Waldenstrom macroglobulins shared an idiotypic determinant with the 16/6 anti-DNA antibody. The reaction between the macroglobulins and the antiidiotype serum was specifically inhibited by Klebsiella polysaccharides, an indication that the idiotypic marker was in the antigen-binding site of the macroglobulins. These results indicate the existence of widely dispersed conserved variable region genes that encode idiotypically related immunoglobulins with the capacity to bind to both bacterial polysaccharides and nucleic acids. Such genes can be expressed by patients with either Waldenstrom's macroglobulinemia or systemic lupus erythematosus.

The combining sites of 12 mouse hybridoma antibodies to dextran B1355S have been characterized by quantitative precipitin assay. All antibodies preferentially bind the immunizing antigen B1355S and two other class I dextrans, B1498S and B1501S, but show substantial differences in the extents to which they cross react with class I dextrans, suggesting their clustering into five groups. Three myeloma proteins, CAL20 TEPC1035, J558, and MOPC104E, which bind dextran B1355S, each fall into a different group. There appears to be a substantial, but imperfect, correlation of DH region structure and individual idiotypic determinants with dextran binding patterns. Proteins with RY DH segments and IdI (J558) idiotypes are in groups 1 or 3, and proteins with YD DH segments and IdI (MOPC104E) idiotypes are exclusively in group 5. However, identical patterns of precipitin curves accompany very different sequences in CDR3. Antibodies of group 1, which react only with class II dextrans, differ the most in primary sequence, a finding suggesting that subsites responsible for cross reactivity with class I dextrans may be blocked and that this may be effected by side chains of different amino acids. This finding delineates a new aspect of the relationship of variability in amino acid sequence to antibody complementarity.

Isomaltose oligosaccharides varying in size from two sugars, isomaltose (IM2), to seven sugars, isomaltohepatose (IM7), were coupled to stearylamine by reductive amination with sodium cyanoborohydride. Each compound was purified by column chromatography to yield a series of glycolipids containing oligosaccharides differing in length. Stearyl-isomaltotriose to stearyl-IM7 could be incorporated into liposomes and could render them agglutinable by specific antibodies to alpha 1 leads to 6 dextran and could be lysed if complement was also added, whereas those containing stearyl-IM2 were not agglutinated or lysed, indicating that stearyl-IM2 may not be protruding from the liposome surface sufficiently to react with the antibody. Stearyl-isomaltosyl oligosaccharides by themselves or incorporated into liposomes were equally antigenic when emulsified in complete Freund's adjuvant. They elicited pauciclonal responses, and the antibodies were alpha 1 leads to 6 specific, cross-reacted with dextran, and gave semirestricted isoelectric focusing patterns.

The immunochemical specificity of the combining sites of murine myeloma protein CAL20 TEPC1035 was studied by quantitative precipitin and precipitin inhibition assays. Myeloma protein CAL20 TEPC1035 precipitated with only three dextrans, B1355S4, B1498S, and B1501S, with high proportions of alpha(1 leads to 3) linkages, but not with any other dextrans, glycogen, and pullulan. Inhibition tests with various sugars show that the combining site of myeloma protein CAL20 TEPC1035 is most complementary to panose, a trisaccharide DGlc alpha(1 leads to 6)DGlc alpha(1 leads to 4)DGlc. Panose was 3.3 times more potent than a tetrasaccharide DGlc alpha(1 leads to 6)DGlc alpha(1 leads to 4)DGlc alpha(1 leads to 4)DGlc and 8, 23, 42, > 42 times more active than maltose, nigerose, isomaltose, and kojibiose, respectively. These findings were paralleled by their binding properties as determined by affinity electrophoresis. The association constants (Ka) of these three dextrans to myeloma protein CAL20 TEPC1035 ranged from 3.8 X 10(3) ml/g to 5.02 X 10(3) ml/g. The association constant of inhibitor (Kia) of panose was 8.19 X 10(3) M-1. Myeloma protein CAL20 TEPC1035 is an antidextran with specificity different from those of other murine myeloma antidextrans and from human antidextrans reported previously and its combining site size is at least as large as a trisaccharide. The binding constant of methyl alpha-D-glucoside (7.2 X 10(2)) was 73% of that of panose and comparable to that of myeloma protein W3129 (9.4 X 10(2)) with a cavity-type site and 600 times lower (1.6 X 10(0)) for QUPC52 with a groove type site, indicating that the terminal nonreducing residue is held in a cavity. Inhibition data with various alpha(1 leads to 4)-linked oligosaccharides also indicate that the internal portions of these inhibitors may react directly with a portion of the combining site. These findings suggest that myeloma antidextran CAL20 TEPC1035 has a partial cavity-type combining site in which the terminal nonreducing dGlc alpha(1 leads to 6) moiety is held in a cavity with the other two sugars forming a groove. However, oligosaccharides with one or more alternating [leads to 3DGlc alpha(1 leads to 6)DGlc alpha(1 leads to 3)DGlcl leads to] units with and without terminal nonreducing DGlc alpha(1 leads to 6) or DGLc alpha(1 leads to 3) side chains remain to be tested to determine whether structures known to be present in the three dextrans which precipitate CAL20 TEPC1035 may not prove to be more active than panose.

Two human IgM myeloma proteins, IgMWEA and IgMMAY, were found to react with agar and Klebsiella polysaccharides that contain pyruvylated D-galactose (DGal). Quantitative precipitin data and precipitin inhibition studies with methyl alpha- and beta-glycosides of 4,6-pyruvylated-D-galactose showed their combining sites to be different, although each was directed against the pyruvylated-D-Gal, one reacting most specifically with Klebsiella polysaccharides with terminal nonreducing beta-linked 2,4 pyruvylated-D-Gal, whereas the other reacted equally well with Klebsiella polysaccharides that contain 3,4 beta-linked and 4,6 alpha-linked terminal nonreducing pyruvylated-DGal. Inhibition studies showed that both sites are directed toward one of the two space isomers of 3,4- or 4,6-pyruvylated DGal, the form in which the methyl group of the pyruvate is equatorial, or endo, and its carboxyl group axial, or exo, to the plane of the acetal ring. Coprecipitation studies showed the combining site of IgMWEA to be located on an (Fab')2 fragment and not on the (Fc)5mu fragment. The monoclonal peak in the serum of IgMMAY was specifically precipitated by Klebsiella polysaccharide. Myeloma proteins with specificities of this type may occur with reasonable frequency in humans and may be a consequence of clonal expansion from inapparent infection, carrier states, or disease produced by various Klebsiella organisms.

Amino acid sequences of rabbit light chains show considerable evidence of independent assortment of framework (FR) and complementarity-determining (CDR) segments. This suggests that they are coded for by independent genetic units (minigaenes) and that individual light chains are assembled somatically by recombining these units. Identical FR sets with multiple members generally comprise chains with different specificities, whereas identical CDR sets tend to have chains of a single specificity. A J segment, which, by analogy with mouse light chains, is made up of the last two residues of CDR3 plus all of FR4, contained 18 different sets and could contribute to diversity generated by CDR3. The longest segment, FR3, had a very large number of sets. Evidence is presented showing that the number of sets could be substantially reduced by permitting FR3 to be formed by two independently assorting segments comprising residues 57-68 and 69-88.

Two sets of apparently conflicting data on the genes coding for the variable region are being accumulated. One suggests that the sets of nucleotides coding for the framework segments of immunoglobulin light and heavy (VL and VH) chains assort independently and are therefore germ-line minigenes which, together with sets of nucleotides coding for the complementarity-determining regions (CDR) or segments assemble to form complete variable (V)-region genes (15, 16, 33). The other, based on the findings with clones from 12-d-old embryo and adult mouse coding for V-regions, infer that the first three frameworks and the three complementarity-determining segments are already assembled as germ-line V-genes (17-21). It is now generally accepted that the J segment, which in the one instance sequenced (21) is made up of nucleotides coding for framework (FR)4 plus two residues of CDR3, is a minigene. An examination of sequences of human, mouse, and rabbit V-regions, assuming the latter hypothesis, indicates that individual framework sets would have to be present in many copies. The FR2 segment found in one human, 20 mice, and 13 rabbits would have to be present in at least 10/14 copies in the NZB, and 5/6 in the BALB/c mouse, and 12/13 in the rabbit. The X-ray crystallographic data show this region to be a loop, projecting out from the V-domain, capable of accommodating many substiutions and 12 and 8 alternative sequences for this FR2 segment have been found in mouse and rabbit VK chains with substitutions possible at 13 of the 15 positions. These alternative sequences occur much less frequently than the preserved FR2 segment. Thus, there is no basis in the protein structure to account for evolutionary stability of this FR2 segment if it occurs in so many copies in germ-line genes coding for residues 1-96, but its stability is easily explained if it were coded for by a separate germ-line minigene present as a single copy; the alternative forms could then have arisen by duplication and mutation of this minigene. Somatic assembly of the minigene segments for the three framework and three complementarity-determining segments during differentiation would account completely for our assortment data from which FR4 was inferred to be a minigene.

Competitive binding assays using 3H-labeled blood group A substance and insolubilized Dolichos biflorus lectin or human anti-A were carried out, measuring competition by blood group A1 and A2 glycoproteins, and by unabsorbed anti-A sera, and with these sera absorbed with the A1 and A2 glycoproteins. With Dolichos lectin specific for (formula: see text) A1 substances had about 11 times as many determinants as did A2 substances, but the slopes of the lines in the competitive binding assays were the same. With insolubilized anti-A, A2 substances gave lines of lower slopes. Although individual A1 populations varied in the amounts giving 50% inhibition in the assays, as did A2 substances, the slopes of the lines for the A1 substances were the same and always higher than the slopes of the lines for the A2 substances. Competitive binding assays with unabsorbed anti-A sera and with these sera absorbed with insoluble polyleucyl A1 and A2 substances showed that partial absorption of polyleucyl A1 substances left antibodies of lower slope in the supernate, whereas absorption with polyleucyl A2 substance left antibodies (anti-A1) having the same or an even higher slope than the unabsorbed sera. The findings indicate that human A1 and A2 glycoproteins differ in their determinants, and that A2 specificity is determined by the type 2 chain in which the A trisaccharide (formula: see text) is linked beta 1 leads to 4 to DGlcNAc, whereas the A1 specificity is determined by the type 1 chain in which this trisaccharide is linked beta 1 leads to 3 to DGlcNAc; most of the determinants in the glycoproteins have a second LFuc linked alpha 1 leads to 3 and alpha 1 leads to 4 to the DGlcNAc of the type 2 and type 1 chains, respectively.

The purified lectins from Lotus tetragonolobus and Dolichos biflorus were coupled to Sepharose 2B to make insoluble adsorbents for purification and fractionation of blood group A and H active glycoproteins. With both adsorbents, hog gastric mucin A + H blood substance (HGM), purified by phenol-ethanol precipitation, yielded fractions showing only A, only H, or AH activities. The AH fraction was obtained when the adsorbent column was overloaded with HGM and its A and H specificities seem to be carried on the same molecules since they were not separable by chromatography on either column. However A and H specificities of blood group substance from the stomach of a presumably heterozygous individual hog were both on the same molecules as they too could not be fractionated on either column. Analytical properties of the isolated fractions were generally similar to those of the unfractionated material, the purfied A substances had a higher galactosamine/fucose ratio than did the H substances. Although the original A + H showed very little specific optical rotation, the separated A and H substances rotated positively and negatively, respectively. The lectin-Sepharose adsorbents have also proven useful in isolating A or H substances directly from the crude commercial hog gastric mucin. Blood group A2 substance from a human ovarian cyst yielded two fractions on the Lotus-Sepharose column; the effluent did not interact with the Lotus lectin but precipitated the Ulex and Dolichos lectins and anti-A, and appears to contain type 1 H determinants. The other fraction reacted with Lotus and Ulex lectin as well as with Dolichos and anti-A.

Binding constants of the dextran-reactive BALB/c mouse IgA myeloma proteins W3129 and QUPC 52 have been determined for each member of the isomaltose series of oligosaccharides and for methyl alphaDglucoside. Protein W3129 has maximum complementarity for isomaltopentaose (IM5) deltaf degrees = 7,180 cal/mol) with 55-60% of the total binding energy directed against methylalphaDglucoside. Protein QUPC 52 gives maximum binding with isomaltohexaose (IM6) (deltaF degrees = -5,340 cal/mol) and has about 70% of its total binding energy for isomaltotriose (IM3), but at most only 5% for isomaltose (IM2) or methyl alphaDglucoside. Protein W3129 precipitates with branched dextrans high in alpha (1 yields 6) linkages and reacts with but does not precipitate a synthetic alpha (1 yields 6)-linked linear dextran. Protein QUPC 52 precipitates both branched and linear dextrans. Thus, the immunodominant group for protein W3129 is mimicked by methyl alphaDglucoside and this protein reacts exclusively at the terminal nonreducing ends of alpha (1 yields 6)-linked dextran chains. Protein QUPC 52 has an immunodominant group which is expressed by IM3 but not smaller oligosaccharides and this protein can react at nonterminal locations along alpha (1 yields 6)-linked dextran chains.Precipitation of linear dextran seems to be a valid although not quantitative assay for antidextrans with nonterminal specificity. Quantitative precipitin reactions with branched and linear dextrans suggest that alpha (1 yields 6)-specific human antidextrans are mixtures of molecules having terminal and nonterminal specificities and that the fraction of each type can vary among individuals. Rabbit antisera against IM3 or IM6 coupled to bovine serum albumin also appear to contain antibodies with nonterminal specificity for dextran chains although a large fraction has terminal specificity. Low molecular weight clinical dextran N-150N (congruent to 60,000) reacted more like linear dextran than like its parent native-branched dextran B512. This is thought to result from an abundance of nonterminal determinants in clinical dextran N-150N but a very small number of functional terminal determinants per molecule. An appreciation of terminal and nonterminal specificities and of the different immunodominant structures in isomaltosyl chains has proven to be of a great value in understanding the immunochemical reactions of dextrans. Moreover, certain previous findings with fructosan-reactive mouse myeloma proteins and human antilevans (55, 84) also suggest terminal and nonterminal specificities for levan chains.