Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/unpaywall/characterization-of-sialyloligosaccharide-binding-by-recombinant-YIZS5v00I5
References
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
- Publisher
- Unpaywall
- ISSN
- 0021-9258
- DOI
- 10.1074/jbc.270.13.7523
- Publisher site
- See Article on Publisher Site
Abstract
Vol. 270, No. 13, Issue of March 31, pp. 7523-7532, 1995 THE JOURNAL OF BIOLOGICAL CHEMISTRY Printed in U.S.A. © 1995 by The American Society for Biochemistry and Molecular Biology, Inc. Characterization of Sialyloligosaccharide Binding by Recombinant Soluble and Native Cell-associated CD22 EVIDENCE FOR A MINIMAL STRUCTURAL RECOGNITION MOTIF AND THE POTENTIAL IMPORTANCE OF MULTISITE BINDlNG* (Received for publication, November 30, 1994, and in revised form, January 13, 1995) Leland D. Powell:j:§, Rakesh K. Jainll, Khushi L. Mattall, Subramaniam Sabesan], and Ajit Varki:j: From the :j:Glycobiology Program and UCSD Cancer Center, Department of Medicine, University of California at San Diego, La Jolla, California 92093, 'IIGynecologic Oncology Research, Roswell Park Cancer Institute, Buffalo, New York 14263, and IICentral Science & Engineering, DuPont Co., Wilmington, Delaware 19880 dominantly on resting IgM+IgD+ B cells (1-5). It binds to CD22, a B cell-specific receptor of the immunoglobulin oligosaccharides containing the sequence Siaa2-6GaI131-4Glc/ superfamily, has been demonstrated to bind to oligosac- charides containing a2-6-linked sialic acid (Sia) resi- GlcNAc, and shows no affinity for oligosaccharides containing dues. Previously, we demonstrated that the minimal a2-3-linked Sia residues. By sequence analysis, it is a member structure recognized by this lectin is the trisaccharide of the immunoglobulin superfamily, with an N-terminal V-type Siaa2-6GaI/U-4GlcNAc, as found on N-linked, O-linked, domain followed by six Ig C2-type domains, a membrane span- or glycolipid structures (Powell, L., and Varki, A. (1994) ning region, and a 160-amino acid cytoplasmic tail (6-9). Two J. Biol. Chem. 269, 10628-10636). Here we utilize a solu- isoforms of human CD22 have been identified by cDNA cloning, ble immunoglobulin fusion construct (CD22Rg) to deter- a seven domain CD2213 form and a shorter CD22a form, which mine directly by equilibrium dialysis the stoichiometry lacks the third and fourth domains present in CD2213. The (2:1) and dissociation constant (32 pM) for Neu5Aca2- extent of tissue expression of these two isoforms is at present 6GaI/U-4Glc binding. Inhibition assays performed with unexplored, although most cells examined appear to express over 30 different natural and synthetic sialylated and/or the larger isoform (0). Murine CD2213 shows a 62% sequence sulfated compounds are utilized to define in greater homology to the human form and, likewise, a lectin activity detail specific structural features involved in oligosac- directed toward a2-6-linked sialyloligosaccharides (9, 11). charide-protein binding. Specifically, the critical fea- In vitro assays have demonstrated that CD2213 (hereafter tures required for binding include the exocyclic hy- referred to as CD22) functions in a dual capacity, both as an droxylated side chain of the Sia residue and the a2-6 adhesion molecule and as an activation molecule. Cells induced linkage position to the underlying Gal unit. Surpris- to express CD22 by cDNA transfection acquire the ability to ingly, alterations of the 2-, 3-, and 4-positions of the adhere to a variety of different cell types, including erythro- latter residue have limited effect on the binding. The cytes, lymphocytes (both T and B cells), and a variety of trans- nature of the residue to which the Gal is attached may affect binding. Bi(a2-6)·sialylated biantennary oligosac- formed cell lines (7-9, 12, 13). In certain cases, a higher level of charides are capable of simultaneously interacting with binding has been demonstrated with cell activation, which both lectin sites present on the dimeric CD22Rg fusion seems to correlate with increased expression of l3-galactoside construct, giving a marked improvement in binding a2,6-sialyltransferase (8, 9), the enzyme that synthesizes the over monosialylated compounds. Furthermore, data are Siaa2-6GaI131-4GlcNAc sequence (14, 15). A role in activation presented indicating that full-length native CD22, ex- is indicated by the observations that CD22 defines the subset of pressed on the surface of Chinese hamster ovary cells, is 2+ IgM+ B cells which show increased levels of intracellular Ca structurally and functionally a multimeric protein, dem- in response to stimulation with anti-u, and that anti-CD22 onstrating a higher apparent affinity for multiply aialyl- augments this response (16). Moreover, anti-a stimulation ofB ated compounds over monosialylated compounds. These cells rapidly induces the phosphorylation of cytoplasmic Tyr observations provide a mechanism for strong CD22- residue(s) on CD22, and a small percentage of surface CD22 dependent cell adhesion despite the relatively low K for (~2%) may be found in association with the B cell-sIg complex, protein-sugar binding. including both IgM or IgG of naive or memory B cells (17-19). In addition to playing a role with B cell activation, CD22 participates in T cell activation. CD22 binds to several T cell CD22 is a sialic acid (Sia)! binding glycoprotein found pre- glycoproteins, including CD45, and binding of soluble CD22 to CD45 attenuates the increase in intracellular calcium normally * This work was supported by National Institutes of Health Grants seen in T cells following stimulation with anti-CD3 (8, 12). GM32373 (to A. V.) AI29326 (to K. L. M.) and Clinical Investigator In a series of experiments utilizing both the full-length Award KOI CA01649 (to L. D. P.) and by American Cancer Society CD2213 molecule expressed in COS cells or a truncated three Institutional Grant ACS-IRG93W (to L. D. P.). The costs of publication domain construct fused to the Fc portion of Ig (CD22Rg), we of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in ac- and others have demonstrated that its ability to bind to cells or cordance with 18 U.S.C. Section 1734 solely to indicate this fact. § To whom correspondence should be addressed: Cancer Center 0063, UCSD School of Medicine, La Jolla, CA92093-0063.Tel.: 619-534-2507; lactose; SNA,Sambucus nigra agglutinin; IC , concentration giving 50% s o Fax: 619-534-5792. inhibition of binding; FGP, fibrinogen glycopeptides; RIC, relative 1 The abbreviations used are: Sia, sialic acid type unspecified; AGP, inhibitory concentrations (IC relative to a2-6-sLac); PAGE, poly- s o a,-acid glycoprotein; CD22Rg, CD22-immunoglobulin chimera contain- acrylamide gel electrophoresis; CHO, Chinese hamster ovary; PBS, ing Ig domains 1-3 of human CD22{3; DTSSP, 3,3'-dithiobis(sulfosuc- phosphate-buffered saline; ELISA, enzyme-linked immunosorbent cinimidyl propionate); PNGase F, peptide N-glycosidase F; sLac, sialyl- assay; ONP, o-nitrophenyl. This is an Open Access article under the CC BY license. 7524 Characterization of CD22-0ligosaccharide Binding precipitate glycoproteins from celllysates is dependent on the linked sialic acid residues. Ten grams were dispersed in 250 ml of 0.1 M Tris-Cl, pH 8.0, 1.0 mM MgC12' 1 mM CaCI 0.02% azide, and digested presence of Sia residues on the target cells or molecules (8, 20, 2, at 50°C with 500 mg of predigested Pronase (Calbiochem, San for 24 h 21). A sensitive column assay showed that CD22 has a low but Diego, CAl. At 24 h, fresh Pronase was added and the digestion contin- detectable affinity for sialylated N-linked oligosaccharides, pro- ued for another 24 h. After partial concentration by lyophilization, the viding that the Sia residues are a2-6-linked. In contrast, a2- sample was desalted on Bio-Gel P-2 (Bio-Rad) in 4% pyridine, 2% acetic 3-linked residues are not bound. Using purified ~-galactoside acid in water, and the V (detected by the resorcinol assay) lyophilized. a2,6-sialyltransferase to sialylate a variety of complex oligosac- This material was digested once again with 50 mg of Pronase for 24 h. Thereafter, the pH was adjusted to 11.5 with concentrated NH.OH and charide structures, we further demonstrated that CD22Rg rec- incubated at 37°C for 3-4 h to remove any O-acetylation ofSia residues ognizes only the trisaccharide Siaa2-6Gal~1-4Glc/NAc. Other (known to interfere with binding to CD22Rg, see Ref. 13). The sample structural features in complex N-linked oligosaccharides, in- was neutralized with acetic acid, lyophilized, and desalted as above on cluding branching, fucosylation, and/or other core region sug- Bie-Gel P-2. To remove the brown residue arising from the Pronase, the ars, were not recognized. However, a higher apparent affinity material was passed over a 10-ml column oflipophilic Sephadex (LH-60, Sigma) in pyridine/acetate buffer and then finally through a C SPICE was observed for multisialylated structures, implying that the 1 8 cartridge (Analtech Inc., Newark, NJ), to yield a colorless material. The CD22Rg construct was capable of interacting with adjacent FGP prepared contained a mixture of mono- and bisialylated oligosac- a2-6-Sia residues present on the same molecule. However, this charide chains, as determined by the amount of l3-galactosidase releas- column assay could not exclude higher apparent binding able galactose before and after digestion with Arthrobacter ureafaciens merely due to a higher density of a2-6-Sia residues on a single sialidase (Calbiochem). A sample containing 10-20 nmol of Sia was oligosaccharide. digested with 20 milliunits of jack bean l3-galactosidase (Oxford Glyco- Systems, Rosedale, NY) with or without 10 milliunits of A. ureafaciens To further understand the lectin properties of this molecule, sialidase. Liberated Gal was detected by the Mopper-Gindler assay (27), several assay systems including equilibrium dialysis, ELISA which detects free reducing terminals of all hexoses. Sialic acid, a keto capture, column binding, and cell adhesion, are utilized here to sugar, has a color yield only 17% that of Gal, on a molar basis, permit- examine the binding of CD22Rg to a number of different mono- ting the accurate detection of liberated Gal in the presence of the and bisialylated compounds. Additionally, cross-linking exper- liberated NeuAc. By this approach, the FGP contained approximately iments have been performed to explore the possibility that 40% bisialylated structures, with the remainder being monosialylated cell-surface CD22 might be present in a multimeric form. These structures. This preparation was utilized as the unlabeled FGP in Figs. 2 and 7. To prepare a population enriched in bisialylated structures, experiments, together with additional work in the accompany- ~200 nmol were N-acetylated with acetic anhydride (28) to cap the free ing papers (22, 23) examining the role of CD22 in cell-adhesion amino groups on the amino acid(s) and fractionated by high perform- events, offer new insights into how this Sia-specific lectin may ance liquid chromatography on a Rainin SAX column, equilibrated with function in complex biological systems. water, and developed with a gradient of 0.3 M NaCI, 20 mM Tris-Cl, pH 7.0. Two major and two minor resorcinol-positive peaks were detected, EXPERIMENTAL PROCEDURES and by the approach described above, the second major peak identified Equilibrium Dialysis-Equilibrium dialysis was performed in a mi- to contain approximately 75% bisialylated structures. This enriched crodialyzer chamber consisting of 3/16 inch diameter holes in leucite, bisialylated material was used as an inhibitor for the ELISA assay manufactured locally, utilizing CE M; cut-off 100,000 membranes (25 described in Fig. 4. f-Lmthick; No. 132966, Spectrum Medical Industries, Houston, TX). Use Preparation of f'HJFGP-Radiolabeled bi(a2-6)-sialylated FGP was of thicker membranes, even of M; cut-off 50,000, gave incomplete dial- prepared by treating a portion of the initial FGP preparation (-20 ysis of some oligosaccharide ligands over 24 h. Samples were prepared nmol) with UDP-Gal and bovine galactosyltransferase (Sigma), and utilizing the CD2213 chimera (CD22Rg) constructed from the first three then with CMP-[3HJNeuAc and l3-galactosyl a2,6-sialyltransferase (21). domains of CD2213 fused to murine Ig (8), expressed by stable transfec- This material, when analyzed by binding to the CD22Rg protein A- tion in CHO cells (21), which do not synthesize a2-6-linked Sia residues Sepharose column (21), contained about 60% bisialylated structures (24,25). The CD22Rg chimera purified from CHO cells was >90% pure and 40% monosialylated material. This material was preparatively (by SDS-PAGE utilizing Coomassie Blue staining). Protein concentra- separated into mono- and bisialylated structures on the CD22Rg col- tion was determined by the BCA assay (Pierce), utilizing either bovine umn, to yield a radioactive bisialylated FGP glycopeptide whose elution serum albumin (2 mg/ml stock solution, Pierce) or purified pooled hu- corresponded to the pool III material described previously (21). man IgG (No. 14506, Sigma) as a standard. Pooled IgG was chosen as a Synthetic Sialosides-The syntheses of the sialylated oligosaccha- standard, since CD22Rg is 40% IgG by construction with the remaining rides 2-4, 6-11, and 26-29 (Table I) have been described elsewhere 60% (CD22) being a member of the immunoglobulin superfamily. Thus, (29-31). Included are two diisopropylidene derivatives (2 and 11) which CD22Rg might be expected to have a similar reactivity in this colori- contain a Gal residue with two isopropylidene groups, alkylating metric assay. Both standards were within 95% agreement and thus the the 0 and the ° -0 . molecules. The synthesis of the sialylated, 1-02 value obtained utilizing IgG as a standard was utilized to determine the and/or sulfated oligosaccharides 12-25 (Table I) has been fucosylated, concentration ofCD22Rg solutions. As an additional confirmation of the described elsewhere (32-35). validity of this determination, CD22Rg concentration was determined Sialic Acid Determinations-Sialic acid was quantitated by the ace- by A , utilizing the value of A~8'o"" = 1.35 for pooled human IgG. By this tylacetone assay (36), which quantitates the formaldehyde released 2 8 0 approach, the concentration of CD22Rg was determined to be only 15% after treatment with 1.25 mM NaIO•. This assay is equally sensitive for higher than that determined by the BCA assay. Since the above A both free and glycosidically bound sialic acids, unlike either the resor- 2 8 0 value for pooled IgG depends upon an average extinction coefficient cinol or thiobarbituric acid assays. The same stock tube of Neu5Ac was used as a standard throughout. which is not accurate for a single protein, the BCA values were used throughout. Samples containing CD22Rg (1-1.2 mg/ml, 5-5.1 f-LM pro- Oligosaccharide-CD22Rg Binding Column Assay-A 2.5-mm diame- tein, based on M; = 210,000), 2-10 pmol of [3HJa2-6-sLac, and 1-20 ter column, containing 0.15 ml of protein A-Sepharose containing 25-50 f-Lmol of ll'2-6-sLac (final concentration 0.2-400 J.LM), in 50 J.Ll ofTBS (20 J.Lg of CD22Rg, equilibrated in TBS (20 mM Tris-CI, pH 7.4, 140 mM NaCI, 0.02% sodium azide), was prepared. A 30-f-LI sample containing mM Tris-Cl, pH 7.4, 0.14 MNaCI, 0.02% azide) were dialyzed against 50 3H-bisialylated 950 cpm of FGP (less than 0.1 nmol) was combined with f-Ll of TBS for 14-18 h at 4°C. Experiments done in the absence of 14CJGlc CD22Rg demonstrated complete equilibration of the [3HJcll'2-6sLac 200 cpm of [ and either buffer alone or variable concentrations of across the membrane over this time period. Thereafter, 25-f-Ll aliquots other sialosides, as indicated in figure legends, and the column eluted at were removed from each side of the membrane and counted by liquid 4°C collecting 0.12-ml fractions. On a column of this small size, quan- scintillation spectrometry. From these data, the total, bound, and free titative recovery of the 3H-Iabeled bisialylated FGP is achieved at 4 "C, concentrations of a2-6-sLac were calculated and the data analyzed and warming to ambient temperature as performed previously was not according to Scatehard (26). Slope and intercepts were calculated by a required (21). linear least squares fit program provided by Cricket Graph (Cricket ELISA Assays-An ELISA capture assay was developed to screen the Software, Malvern, PAl. Inclusion of 1 mM a2-6-sLac resulted in total ability of different sialylated oligosaccharides to inhibit the binding of competition of binding of the [3HJa2-6-sLac in this assay. CD22Rg to a sialoglycoprotein. Working at 4 "C, 200 ng of pooled Preparation of FGP-Bovine fibrinogen (Sigma) was utilized as a human IgM (Calbiochern) or 1-2 f-Lg of human a glycoprotein 1-acid (AGP, Sigma) were bound to microtiter wells (Immulon 4, Fisher Sci- source of biantennary oligosaccharides containing one or two a2-6- 7525 Characterization of CD22-0ligosaccharide Binding entific Co., Tustin, CAl in PBS (for IgM) or 50 mM Tris-CI, pH 10 (for 0.10 AGP), for 18-36 h. After blocking with PBS, 1% bovine serum albumin (3-4 h), 2-3 tJ-g of CD22Rg in 100 tJ-I of TBS, with or without the indicated sugar inhibitor, were added and incubated for 18 h. The plates 0.08 were rinsed twice with TBS, incubated for 2 h with 100 tJ-Vwell of horseradish peroxidase-conjugated goat anti-murine IgG sera (Bio-Rad; c:: <, 1/200 dilution in TBS, 0.1% Nonidet P-40), and then rinsed 4 times, 5 CI) 0.06 min/rinse, with TBSlNonidet P-40. The wells were developed with 0- phenylenediamine, and A read with a MicroTek plate reader. Assays <, s 6 0 '0 were performed in duplicate throughout. For each set of experiments, :;, 0.04 a2-6-sLac was included as a positive reference compound. As the IC so .c for a2-6s-Lac ranged from 30 to 120 tJ-M between assays, the relative inhibitory concentration (RIC) of a given compound was calculated as 0.02 the ratio of IC values for a test compound to a2-6-sLac. so Creation of CD22-CHO Cells-Stably transfected cell lines were es- tablished as described (21), utilizing a plasmid coding for the full-length 0.00 transmembrane form of CD22f3 (8). Colonies (lifted by EDTA) were 2 3 screened for CD22 expression using phycoerythrin-conjugated Leu-14 monoclonal antibody (Becton Dickinson, Oxnard, CAl; and flow cytom- bound/R etry utilizing a FACscan® instrument (Becton Dickinson Irnrnunocy- tometry Systems, Mountain View, CAl. FIG. 1. Equilibrium dialysis binding of a2-6-sLac to CD22Rg. Plate Adherence Assay-Daudi cells, a human B cell lymphoma line, CD22Rg (5 tJ-M protein concentration based on a M, of 210,000) was were labeled for 18 h with [3Hlthymidine (DuPont NEN, Wilmington, incubated with increasing amounts of unlabeled a2-6-sLac in the pres- DE), washed 3 times in PBS, 0.5% bovine serum albumin, and adjusted ence of a fixed amount (2-4 pmo!) of [3H12-6-sLac, in a total volume of to 5 x 10s/ml in PBS, 0.5% bovine serum albumin, 2 mM MgC12' 50 tJ-1. Dialysis was against 50 tJ-I of buffer, across a M; cut-off 100,000 dialysis membrane, for 18 h at 4°C. Aliquots were taken from either Confluent wells of a 48-well tray containing either CHO or CD22-CHO side of the membrane and counted, the concentration of bound and free cells were rinsed 3 times with PBS. Samples (100 tJ-!) containing a2- a2-6-sLac determined, and the data plotted according to Scatchard. 6s-Lac, FGP, or no inhibitor, were added, incubated at 4°C on an orbital The data were fit to a single line by a linear least squares program. The shaker (l00 rpm) for 30 min, and rinsed four times with ice-cold TBS. slope corresponds to a K of 32 liM, and the x intercept indicates a Bound cells were lifted with 0.4 ml of TBSlNonidet P-40 and quanti- stoichiometry of 2.2. R" total concentration of receptor in (tJ-M). tated by liquid scintillation counting. Each inhibitor concentration was studied in triplicate. Cross-linking Studies-CD22-CHO cells were metabolically labeled with the presence of a single sialic acid-binding site located in with [3SS1EXPRE3sS3sS (DuPont NEN) for 12 h in methionine-free one of the first three N-tenninal domains of CD2213. This stoi- media, then chased with complete media for 4 h. Working at 4°C, cells were lifted with PBS, 2 mM EDTA, washed in PBS, and then cross- chiometry is based on protein concentration detennined by the linked with 0-3 mM DTSSP (Pierce). After quenching excess cross- BCA protein assay, using human IgG as a standard. Determi- linker with 10 mM Tris-Cl in PBS, cells were lysed (21) in the presence nation of protein concentration by A was in close agreement 2 8 0 of aprotinin and phenylmethylsulfonyl fluoride, lysates were pre ad- (see "Experimental Procedures"). The apparent binding affinity sorbed with protein A-Sepharose (3 h) and then adsorbed with anti- for a2- 6-sLac is 32 /LM. CD22 (To15, Dako Corp., Carpenteria, CAl and protein A-Sepharose for Previously, we utilized a column elution assay to determine 12 h. Antigen-resin complexes were washed sequentially with TBS, Nonidet P-40, 2 mM EDTA; TBS, Nonidet P-40, 2 mM EDTA, 1.0 MNaCI; CD22Rg-oligosaccharide binding (21, 37). By this approach, TBS, 1% NonidetP-40, 0.1% SDS, 0.1% chelate; and finally with 50 mM sialylated oligosaccharides with two or more a2-6-linked Sia Tris-Cl, 0.1% Nonidet P-40. Bound proteins were eluted with SDS- residues were found to be retained longer as compared to PAGE sample buffer lacking reducing agents, and one-half of each monosialylated structures. This observation suggested sample was reduced (1% dithiothreitol), Reduced and nonreduced sam- tighter binding of bisialylated structures with the two possi- ples were boiled for 2 min and analyzed by SDS-PAGE (6%), followed by fluorography. ble binding sites on the dimeric CD22Rg chimera. To more Similarly prepared radioactive lysates were subjected to irnmunopre- quantitatively study interactions with a bisialylated bianten- cipitation with CD22Rg in the absence or presence of a2-6-sLac, uti- nary oligosaccharide, Pronase glycopeptides were generated lizing the same washing procedure. Total amount of precipitable radio- from bovine fibrinogen, which contains exclusively bianten- activity was determined, and the samples analyzed by SDS-PAGE/ nary N-linked structures containing only a2-6-linked Sia fluorography. residues (38). On several batches of commercial fibrinogen RESULTS examined, sialylation was incomplete, and large quantities of Equilibrium Dialysis Analysis ofCD22-a2-6-sLac Binding- pure bisialylated material could not be generated. However, Previous work established that CD22 functions as a sialic acid- as described under "Experimental Procedures," a small quan- specific lectin (8, 21, 37). These studies were performed both tity of radioactive bisialylated material could be generated with the native transmembrane CD2213 molecule, transiently using glycosyltransferases and ion exchange chromatogra- expressed in COS cells, as well as with a fusion protein termed phy. This material was employed in a "single point" dialysis CD22Rg. The latter was fonned by the fusion of the first three experiment utilizing radiotracer amounts of either [3H]a2- N-terminal domains of CD2213 with the hinge and two C-ter- 6-sLac or 3H-bisialylated FGP. Under conditions of large minal domains of human IgG (8), forming a protein with an excess of lectin over ligand, the free receptor concentration apparent M; of 210,000 and 105,000 under nonreducing and closely approximates the total receptor concentration (R,). reducing conditions by SDS-PAGE, respectively (data not Thus, provided the stoichiometry is known or can be esti- shown). Thus, the native molecule is a dimer, akin to native mated, the K can be calculated directly from the ratio immunoglobulins, and some of its binding properties may be of bound to free ligand and the R,. By this approach, a bisi- due to the proximity of two identical lectin sites, located in the alylated glycopeptide, prepared from fibrinogen, exhibited a first three domains. To further examine the sialic acid-lectin 17-fold higher apparent affinity to CD22Rg than a2-6-sLac. properties of CD22, we directly examined the binding of the This ratio would correspond to an apparent K of 1-2 /LM. chimera to [3H]a2-6-sLac by equilibrium dialysis, performed The enhanced binding affinity of the bisialylated FGP over at 4°C. These results (see Fig. 1, which includes data pooled that of a2-6-sLac was further demonstrated by examining from three experiments) demonstrate the presence of two bind- their ability to inhibit the binding of a radiolabeled oligosac- ing sites (n = 2.2) per dimeric 210,000 M; chimera, consistent charide to immobilized CD22Rg. The CD22Rg columns initially 7526 Characterization of CD22-0ligosaccharide Binding a2-6-Sia groups, then the FGP must have an intrinsically --Glc higher affinity for CD22Rg than with a2-6-sLac. These results E are consistent with those of the single point dialysis experi- Inhibitor at: a. ment. The most likely mechanism for this increased affinity --O-O~M ____ 10~M would be its ability to simultaneously interact with more than 'S; one sialic acid binding site. Alternatively, other segments of the ~30~M --I:r- 90 ~M A ra N-linked glycopeptides may interact with CD22Rg (e.g. Man residues, chitobiose core sugars). However, in prior investiga- :s ra .. tions we have found no evidence to suggest that other struc- tural features on an N-linked oligosaccharide are recognized by CD22Rg (37). Sialyloligosaccharide Binding Specificity of CD22-To fur- ther define structural features recognized by CD22, a number of chemically synthesized glycosides were screened as inhibi- tors in an ELISA assay using either IgM or AGP coated onto microtiter wells. While the accompanying papers (22, 23) show that interactions of these proteins with CD22 is strictly Sia- dependent, IgM was found to be a more reliable reagent for 'tl Gl these assays as (a) IgM intrinsically has a much higher appar- .. :l 40 ent affinity for CD22Rg (23); (b) a good signal with IgM re- Gi ~ quired only 100-200 ng/well, as compared with 1 JLg/well of 'S; AGP; (c) significant lot-to-lot variability in AGP was found :;:; ra (which paralleled total Sia content); and (d) immunoglobulins adhere to polystyrene surfaces more efficiently than highly :s ra .. glycosylated proteins such as AGP. Wells coated with IgM were incubated with 20 JLg/ml CD22Rg in the absence or presence of Gl a potential inhibitor at various concentrations, and the bound u 80 .. chimera detected with horseradish peroxidase-conjugated goat Gl a. anti-mouse immunoglobulin antisera. The structures of the compounds examined, and the numbering scheme employed in Figs. 3-5, are listed in Table I. By this assay, the IC of 5 0 a2-6-sLac ranged between 30 and 120 JLM ie.g, Fig. 3, A and D), a result consistent with its K determined from equilibrium dialysis (Fig. 1). The variability most likely reflects intrinsic limitations of the assay and differences in reagents employed, 0 as these experiments were conducted over 6 months time. In 0 10 20 each series of experiments, a2-6-sLac was included as a refer- fraction number ence compound, and the RIC of these different compounds to a2-6-sLac are summarized in Table I. FIG. 2. Comparison of a2-6-sLac and FGP as inhibitors of These experiments indicate that CD22Rg is capable of bind- [sH]FGP binding to CD22Rg. A 30-1L1 sample containing bisialylated 14ClGlc [3HlFGP (less than 0.1 nmol), combined with [ and buffer alone ing a broad range of a2-6-sialosides and, moreover, that struc- (0), 10 ILM (.), 30 ILM (0), or 90 ILM (6) of unlabeled FGP (a mixture of tural features away from the Sia residue may significantly mono and bisialylated structures), was applied to CD22Rg-PAS and 14ClGlc influence CD22Rg binding. In confirmation of prior studies, eluted at 4 °C (Panel A). The elution profile of [ for only one of Q2-3-linked Sia residues are not recognized (5, Fig. 3). Several the four runs is also shown (e). From these data, a running sum on a percentage basis of the eluted radioactivity was calculated, and is molecules containing a 6-thio derivative of Gal (2, 3, 7, 8, 9, and presented in Panel B. A similar series of experiments was done with 0, 11) were examined. These compounds were originally devel- 10, 30, and 90 ILM a2-6-sLac, and these data are presented only as a oped as (potentially) nonhydrolyzable inhibitors for different running sum in Panel C (symbols as in Panel A). bacterial and viral sialidases (31). All these compounds bound to CD22Rg although, for some, with a significantly poorer utilized to demonstrate oligosaccharide binding contained affinity as indicated by a higher RIC. Compound 8 had a RIC 20200 JLg of protein, and the elution of most multisialylated 3-fold higher than its non-thio derivative (6), while 9 had a compounds required warming the column to 22-24 °C (21, 37). lower RIC than its non-thio derivative (10). However, this A new column was constructed which contained -25-50 JLg of chemical modification did not produce dramatic changes in the protein on 0.15 ml of protein A-Sepharose. With a column of apparent binding affinity of these compounds for CD22Rg. this small scale, the [3H]FGP, applied in a total volume of 30 JLl, 14C]Glc Several C sialosides were similarly examined. eluted significantly slower than the nonbinding [ 6-methyl-Gal This chemical modification limits the rotamer conformations marker (Fig. 2), and warming of the column was not required possible with the Sia-Gal disaccharide. In aqueous solution, for successful elution, When this same 30-JLl sample, contain- rotamer orientations (tg and gt) are commonly found, ing 10-90 JLM unlabeled FGP (with concentrations based on Sia two formed by a 120 rotation around the C bond in Gal. In one groups, not peptide), is applied to the column, the radioactivity 5-C 6 elutes significantly earlier (Fig. 2A). These column profiles can (gt), the two saccharide residues are bent back over themselves, and in the other (tg), the two residues are in an extended be presented with greater clarity by calculating a running sum, conformation (31). The (6S)- or (6R)-C group sterically on a percentage basis, of the eluted radioactivity (Fig. 2B). In 6-methyl contrast to this result, when 10-90 JLM Q2-6-sLac is included, limits this rotation and shifts the equilibrium in favor of one or significantly less inhibition of the binding of the [3H]FGP is the other of these two rotamers. These compounds have been observed (Fig. 2C). As a greater level of inhibition is seen with useful in determining the orientation of the Sia-Gal disaccha- FGP than a2-6-sLac for identical concentrations of competing ride preferred by different sialidases, most of which show a 7527 Characterization of CD22-0ligosaccharide Binding TABLE I Summary of relative inhibitory concentrations (RIC) of sialylated and / or sulfated compounds as determined by ELISA competition assay From the ELISA competition experiments presented in Figs. 3 and 4, the RIC (ratio of observed IC to that of a2-6sLac in the same assay) so for each compound was calculated. RIC No. Compound" ___ 1 (2,6 SL) -0- 2 1.0 NeuAca2-6Gal{31--4Glc -3 NeuAca2-6(6-thio )Gal diisopropylidene 1.0 -0- 4 2.3 3. NeuAca2-6(6-thio)Gal ----.- 5 (2,3 SL) 4. NeuAca2-6Gal f3-TMS 2.8 NIc 5. NeuAca2-3 Galf31--4Glc 4.3 6. NeuAca2-6(6R)(6-Me)Gal {3-TMS NeuAca2-6(6S)(6-Me)(6-thio)Gal 10.0 8. NeuAca2-6(6R)(6-Me)(6-thio)Gal f3-TMS 9. NeuAca2-6( 6S)( 6-Me)(6-thio)Gal f3-TMS 13 NeuAca2-6(6S)(6-Me)Gal f3-TMS 20 11. NeuAca2-6(6R)(6-Me)(6-thio) Gal diisopropylidene NeuAca2-6Galf31-3GlcNAc f3-0Bn 2.7 -0- 4.3 13. NeuAca2-6Galf31-3GaINAc a-OBn -0- 8 14. 6-0-S0 a-ONP 6.7 3Galf31-3(NeuAca2-6)GaINAc Galf31-3(NeuAca2-6)GlcNAc f3-0Bn 13 ----.- 16. Galf31-3(NeuAca2-6)GaINAc a-OBn 17. NeuAca2-6GalNac a-ONP 0.33 --- Cl Gal{31-3(NeuAca2-6)GlcNAc {3-0NP 0.33 0.33 C 19. 6-0-S0 f3-0NP 3Galf31-3(NeuAca2-6)GlcNAc :a NIc 20. 6-0-S0 f3-0NP 3Galf31-3GlcNAc NIc 21. Gal{31-3(6-0-S0 f3-0Me :c 3)GlcNAc NF 22. 6-0-S0 a-ONP 3Galf31-3GalNAc "i6 NIc Galf31-3( 6-0-S0 3)GaINAca-O-allyl NIc 24. 6-0-S0 3Galf31--4Glc 'x 25. 6-0-S0 NF CU 3(Fucal-2)Galf31--4Glc -0- 26. NeuAca2-6Galf31--4GlcNAcf31--4 \ -0- Gal-f3-0R,d 0.5 - ----.- 14 NeuAca2-6Galf31--4GlcNAcf31-2/ Gl 27. NeuAca2-6Galf31--4GlcNAcf31-6\ .. Gl Gal-f3-0R, 1.0 Q. --- NeuAca2-6Galf31--4GlcNAcf31-3/ 28. NeuAca2-6Galf31--4GlcNAc{31-6\ Gal-f3-0R, 1.3 N euAca2-6Gal{31--4GlcNAcf31--4/ 29. NeuAca2-6Gal{31--4GlcNAc{31-6\ Gal-f3-0R, 1.3 NeuAca2-6Galf31--4GlcNAcf31-2/ 30. Pooled AGP oligosaccharides" 1.3 -0- 1 Sialylated fibrinogen glycopeptides" 2.7 ----.- 15 a The numbering of the compounds corresponds to the numbering -l::r- 17 scheme used in Figs. 3 and 4. --- 18 -0- 19 b TMS, trimethylsilane; Bn, benzyl. c No inhibition detected at highest concentration employed, as indi- cated in respective figure. d R1 is -(CH • 2)sCOOCH3 e Total PNGase F released oligosaccharides from AGP. f A mixture of mono(a2-6)sialylated (25%) and bi(a2-6)sialylated (75%) glycopeptides. -0- 20 marked preference for tg over the gt (31). For CD22Rg binding, _ 21 the C derivatives all showed poorer RICs relative to 6-methyl -0- 22 non-methyl derivatives (Fig. 3B and Table I). However, the --- 23 ----.- 24 (6S)-C modification appeared to be more detrimental 6-methyl --+-- 25 to sialoside binding (6 versus 10), indicating a preference for 4 3 2 1 0 the tg rotamer. This preference is lost in the methyl-thio de- 10- 10. 10. 10. 10 10 1 rivatives (8, 9, and 11), which are poorer inhibitors of CD22Rg- mM inhibitor concentration IgM binding. FIG. 3. Structural parameters influencing the interaction of A large number of different Sia 0<2-6-Hex(NAc) sialosides sialosides with CD22Rg. The relative affinities of several different (with Hex(NAc) being GlcNAc, GalNAc, or Gal) are recognized sialylated oligosaccharides for CD22Rg was inferred by determining their ability to inhibit the binding ofCD22Rg to immobilized IgM in an by CD22Rg (Fig. 3, C and D, and Table I). A preference for ELISA assay. Binding was performed in the absence or presence of the GlcNAc over GalNAc is suggested by the different RICs of indicated inhibitor at 4 °C for 12-15 h, and then bound CD22Rg deter- compounds 12 versus 13 and 15 versus 16, although a differ- mined as described under "Experimental Procedures." The different ence in the linkage (0< versus f3) of the blocking groups may also compounds tested are identified by number corresponding to the listing in Table I. The data presented are representative experiments, and explain the differences seen. CD22Rg binding was significantly each data point represents duplicates :!: S.D. For each series of exper- influenced by the structure of the group attached to the Hex- iments performed on a given day, a2-6-sLac was included as a refer- (NAc) residue. For example, the RIC of 18, which has a ONP ence compound. The data presented in Panels A, B, and C are from one group attached to a GlcNAc residue, is 40-fold better than that series of experiments, and the inhibition profile of a2-6-sLac is shown of 15, which contains a benzyl alcohol group instead. Addition- only in Panel A for simplicity. 7528 Characterization of CD22-0ligosaccharide Binding 120 -r-----------------, "C oS! :::l Cii Cl 's c: '5 80 c: III :c c;; 'S ~ ___ Glc 60 -0- 1 'x ____ 26 ell --0-- 26 ---I:r-- 27 'E 40 --.- 27 Gl --+-- 28 ---+- 28 ... --0-- 29 ______ 29 Gl --tJ- no Inhibitor c. 20 ---t:r- 30 ____ 31 0 10 20 o+-~~_~~....-~~....,--~~ ........~~......; fraction number 10 -4 10 -3 10 1 FIG. 5. Demonstration of relative binding ofbisialylated struc- tures to CD22Rg. Compounds 26-29 were screened for their ability to mM inhibit [3H]FGP-CD22Rg binding in the column assay as described in FIG. 4. Inhibition of CD22Rg-IgM binding by bisialylated com- Fig. 2, using a single concentration of 30 JLM (based on Sia groups). The pounds. Several synthetic or naturally occurring bisialylated com- elution profiles are presented as a running sum. The elution profiles of pounds were examined in the ELISA assay. The different compounds [3H]FGP in the absence of any inhibitor (no inhibitor) and of one typical 14C]Glc (Gle) are also shown. tested are identified by number corresponding to the listing in Table I. profile of [ ally, Glc is favored over trimethylsilane (compare 1 and 4), and tri-, and tetraantennary structures. Surprisingly, the RIC of all we previously demonstrated that Glc was favored over glucitol of these compounds, relative to 0/2-6-sLac, was approximately (37). Other pairs of compounds differing only in their reducing 1. The RIC for the FGP preparation, however, consistently was in group were not available. Several structures based on the 0/2- the range of 2.5 to 3, whereas one of the synthetic bisialylated 6-sialylated Gal{:l1-3GalNAc/GlcNAc trisaccharide (12 and 13), compounds (26) was 2-fold better. None of the bisialylated com- which is not found in nature, also were reasonable inhibitors of pounds exhibited RICs in the range of 0.03, as would be predicted CD22Rg-IgM binding. Moreover, significant modifications of from the equilibrium K values measured above. Moreover, the the Hex(NAc) residue are permissible, including 3-0-galacto- marked difference between 0/2-6-sLac and FGP seen in the sylation (compounds 15 and 16) as well as capping ° with CD22Rg-oligosaccharide column assay (Fig. 2) was not reflected 1-°4 alkyl groups, as found in the diisopropylidene derivatives. in this ELISA assay. This apparent discrepancy probably reflects Taken together, these results indicate that Sia0/2-6Hex(NAc) the intrinsic differences between these two assays, particularly sialosides are recognized by CD22Rg. These observations are for measuring binding events characterized by relatively fast on significant as they indicate that a wide range of 0/2-6-sialy- and off rates (unlike antibody-antigen binding, which is typically lated structures, found on both N- and O-linked structures and characterized by slow on and off rates). In the ELISA assay, glycolipids, may be potential ligands for CD22Rg. washing steps (lacking the inhibitor) are performed multiple Previously, we demonstrated that glycopeptides from bovine after the binding of the CD22Rg to the sialylated glycoprotein submaxillary mucin, a rich source of Sia0/2-6GalNAcO/-(pep- (which is done in the presence of the inhibitor). In contrast, in tide) residues, was not recognized with high affinity by CD22 both the equilibrium dialysis experiments and in the column even after de-O-acetylation of sialic acids (37). When this same assay, the soluble oligosaccharide (as either a ligand or an inhib- bovine submaxillary mucin preparation was screened by the itor) is present throughout the binding process. Thus, these as- column assay as described in Fig. 2, a level of inhibition of says are fundamentally different, which may explain the appar- [3H]FGP binding corresponding to 30 J.LM 0/2-6-sLac was seen ently discrepant results in the inhibitory potency of the with 900 J.LM bovine submaxillary mucin (based on Sia concen- bisialylated compounds. tration; data not shown), confirming our earlier result (37). Given this limitation with the ELISA assay, compounds Thus, the low RIC of compound 17 (Table I), which is compa- 26-29 were examined with the column retention assay, using each at a concentration of 30 J.LM. These results indicate meas- rable to the mucin structure must be explained either by a positive effect of the ONP aglycone, or by a negative effect of urable differences in the ability of these compounds to compete with the binding of the [3H]FGP sample to CD22Rg, with 26 the polypeptide in the case of bovine submaxillary mucin. being the most potent inhibitor (Fig. 5). Although it would Several modifications were without effect on sialoside bind- ing. Sulfation did not affect the binding of sialylated com- appear to be more potent than FGP at 30 J.LM, the FGP prepa- ration contained a mixture of mono- and bisialylated struc- pounds, and sulfate did not substitute for sialic acid (Fig. 3E). tures. While these four sialosides are structurally very similar, The sulfated derivative of 16 did exhibit a 4.5-fold improve- ment in its RIC (16 versus 14, Table n, yet these compounds they are all branching isomers and, consequently, the terminal 0/2-6-Sia residues will be oriented differently (29, 30). Molec- also differ in their reducing terminus blocking group. As with ular modeling predicts that the Cz-C distance between the two 17,18, and 19, the ONP group improves the compound's RIC Sia molecules is 17-19 Afor 26 and 29, and 9-10 Afor 27 and considerably. Recognition of Bisialylated Structures by CD22Rg-Several 28 (29), thus, these measurements do not correlate with the bisialylated compounds were examined (Fig. 4; 26-31, Table I). different RICs seen for these four compounds (Table I). Other structural features, such as the relative orientation of the Included here were several synthetic bisialylated structures Sia0/2-6Gal units, must be involved. Earlier we presented ev- (29-31) as well as FGP glycopeptides and N-linked oligosac- charides released from AGP. The FGP preparation utilized idence indicating that CD22Rg was capable of discriminating between two different bi(Sia0/2-6)-tetraantennary isomers, al- here contained approximately 75% bisialylated structures (see though we were not able to directly prove which antennae were "Experimental Procedures"). The AGP mixture utilized here contained a mixture of 0/2-6- and 0/2-3-linked residues in an sialylated (37). The observations with these four synthetic hi- sialylated compounds confirms our earlier hypothesis that approximate ratio of 1.2 to 1 (37) on a mixed population of bi-, 7529 Chara cterization of CD22 -0ligosa ccharide B inding 1 2 0~------ - .-- ---- ---' 1 0 0 Cl 200'- :a c: :.a "E CIl (J ... CIl a. 4 0 --.-- a2 -6s Lac fibrinoge n 45~ glyco pe ptide TD~ --- 3 4 5 2 1 2 0 1 0- 10- [ in hib ito r 1mM .F IG. 6. Inhibition of CD22Rg binding to Daudi cell g lycopro- 3 5 t ems b y a 2- 6-s L ac . Equal al iquots of [ S 1Me t -la be led Daud i ce ll FIG. 7. Inhibition of Daudi c ell binding to C D 22 e x p ressi ng l y ~ at e wer e pr ecleared with pr ot ein A-Sep h arose an d t he n pr ecipit a t ed CUO cells by sialylated oligosaccharid e s. Ra diolabel ed Daudi cells with CD22Rg a n d additiona l pr ot ein A-Se pharose in t h e pr esen ce of wer e a llow ed to bin d to con flu ent cu lt ures of C D22-CH O cell s at 4 °C buffer al on e (l a ne 2 ), or 30 JL~I (l a ne 3), 90 JLM (l a ne 4 ), or 1 mxt (lan e 5 ) a nd boun d cells qu an t it a t ed by sci nt ill a t ion countin g. During th e initi al a 2- 6:s La c. Materi a l . non sp ecific all y a ds orbed to an eq u a l volume of bi nding s te p, eit he r a 2- 6-s La c (.) or FGP (mixt u re of mono - a nd protem A-Sepharose in the ab senc e of CD22 Rg is show n in lane 1. After bis ial yl a ted s t r u ct u r es; e ) wa s in clud ed at th e indicated concen t ra t ions . boiling th e be ads with samp le buffer, eq u a l a liq u ots of t h e sol ubilized Each point represents t he a ver a ge of triplicate assa ys. r adi oa ctivity wer e exa m ine d by SDS-PAGE a n a lys is . Ot her a liq u ot s wer e us ed to de t e r min e the amount loa ded ( cp m) of 800 (lan e 1), 12,6 20 (la ne 2 ), 6,860 (lan e 3 ), 3 ,620 (lan e 4 ), and 2 ,500 (lan e 5 ); a nd to show probe t his question, CHO cells were sta bly transfected with th at a ll of this ra dioactivity is trich loroacet ic a cid-pr ecipit a b le (da t a n ot CD 22f3 con tai n ing pl a smid , to create a CD22-expressi ng s u b- s ho wn). TD, tracking dy e . lin e . Wh en grown as a m on ola ye r , thes e cells are ca pable of binding Daudi cells , wh ile th e pa r en t al CHO cell s do not (22) . CD22 Rg bindi ng is influ en ced by t he r el a t ive positionin g of Bindin g can be pa r ti all y inh ibi t ed by in clu si on of a 2- 6-s La c or multipl e Sia gro u ps on a si ngle oligosaccha r ide. F GP . H owe ver, sign ificantly di ffe r en t in h ib ition curves a re a2-6-sLac Inhib it ion of Precip itat ion of Dau di Gly coprotein s see n with t hese two differen t comp ounds . With FGP, approxi- by CD22Rg-Previous s t u dies de m onstr ated t ha t C D22Rg im - m a t el y 50% inh ib iti on is obse rve d at 300 MM, whil e les s than munoprecipitates a spectru m of a2-6-sialyla te d glycopro te i ns 50% inhibi t ion was see n at 1 m Ma2-6-sLac (F ig. 7 and accom- 35SJM from differ en t T a n d B cell popul a ti on s (8 , 20 , 2 1). An [ et- panying pap er (22 »). Th a t t he IC va lues for t he se com pounds s o labele d lys a t e , pr ep a r ed from Daudi cells, was precipi t ated do no t ma t ch that seen in t he ELISA assay is not surprising, with CD22 Rg in t he a bsence or pres enc e of increa sing a mo u nts give n t he mul t ival en t n a ture of cell- cell a dhesion . However, the of a2-6 -s Lac . Th e ove ra ll pa t tern of precipi t a t ed proteins see n differenc e in inhibi ti on seen wi th t hese two com pou n ds is s t r ik - is s imi lar to that publish ed pre viou sl y a n d is ex pe cte d to in- in g a n d reproducibl e . I n a dditiona l ex pe ri me nts (n ot shown ) 1 clu de both glyco prote ins an d pr ot eins associa te d wi t h glyco- m M2- 6-sL a c consiste n tly ga ve onl y pa r ti al inh ibit ion ofC D22 - pr ot ein s whic h bin d to C D22 Rg (8 , 2 1). Th e r esults (F ig . 6 ) CHO cell bi n ding , a n d F GP at t he sa me conce ntration always in dicate d that a2-6-s Lac inhib it ed CD22 Rg-glycopro tei n bind- gave n early com plete inhibi ti on. Th e extent of i n h ibition se e n ing i n a range predi ct ed by its K , with - 50% inhibi t ion see n a t with 1 m M 2-6-sL a c was so me what vari abl e , and m a y reflect 30 MM, furthe r i n d icating that C D2 2Rg is func t ion in g in th ese variables s u ch as t he den si t y of CD22 on t he C HO cell surface pr ecip it a ti on assays so le ly by r ecogni t ion of a 2- 6-li n ke d s ia lic (e ve n t h oug h th e cell s we re stab ly tra ns fecte d, t he le vel s of ac id resi dues . Of particul ar n ot e , t he binding of a ll precipi t a t ed ression wer e n ot a lways con si s t en t afte r exte n de d cu lt ure ). ex p proteins a p pea re d to b e equa lly se nsitive to inhibition by a 2- We therefor e did n ot do an in d ep th stu dy on 2-6-s Lac i n hibi - 6-s Lac, with t he exce ptio n of t he low m olecul ar weight prot eins tion a n d FGP i n h i bi tio n versus CD22 de n s i ty . Rega rdl ess , th e whic h com igrate wit h prot eins that n on sp ecific all y a ds or b t o abili t y of F GP to produc e h igh er le vel s of inh ibit ion at t h e s a me pr ot ei n A-Se pharose (F ig. 6, lanes 1 a n d 5 ). It is known t ha t tration (ba sed on Sia content) suggests that C D22 mo l- conc en many of t he n on precipi t a t ed Daudi cell gly copr ote ins al so con - ecu les on t he surface of t he CH O-C D22 cell are in close e n ou gh tain a 2- 6-lin k ed sialic acid res id ues .f Th e segregat ion of dif- pro ximi t y to a llo w a s i ng le F GP m olecul e to interact with more fere n t glyco proteins in t o eit her CD2 2Rg binding or nonbinding th an on e lectin binding site on two CD22 chai ns . fractio ns mu s t t he re fore be b a s ed on othe r fa ctors, s u ch a s Cross-link ing S t ud ies of Cell-surface CD22 -Given this r e- den s it y of a2- 6-sialic aci d r esidues (wh ich would con trol th e s u lt, t he quartenary structu re of cell-s urface ex pressed CD22 ab ility of t he chime r a to form a precipi tin reaction ) or steric wa s exa mi ne d by cr oss- li n k ing t he m et a bolicall y lab eled pro - h indr a n ce of a 2-6-sia lic aci d r esidues by t he prot ein s t r u ct u re t ein wi t h a clea vabl e cr oss-linking reagent DT SSP. Low lev els (a s was observed wit h t ra ns ferrin (3 7)). of DTSSP r esulted i n a p a r t ial re ductio n i n intens ity of th e Inh i bition ofCD22-CH O Cell B inding to Daudi Cells- Giv en 140 -kD a CD 22f3 band con comi t a ntly with t h e appearance of a t hat a ll of t hese ex pe r ime nts we re performed wi th th e biv al ent hi gh M; band wh ich barely ente re d a 6% S DS -polyacrylamide CD22Rg chi me ra , t he qu es ti on arises whethe r n ati ve C D22 ge l (F ig . 8 ). Cleavage of t he cr oss-linke r re su lte d in t he loss of ex p resse d as a integr al m emb rane prot ein , is capabl e of distin- th is hi gh M ; band . N o a dditiona l prot ein bands are seen fol- gu is hing be t ween m on o- a n d mul tiply s ia lyla te d oligosaccha- low in g r educ t ion , eve n on ge ls whic h are expose d for a consid- ri de structures. N o inform a ti on is curren tly a vailabl e on t he erably lon ge r tim e , indica t in g t hat the h igh M r band does not quarte nary str uctu re of n a t i ve CD22 , a n d ea r ly s t u d ies indi- con t ain additi on al m etab olicall y la beled (bu t low abundance or cated t hat it is n ot fou nd as a di sulfide-linked dimer (4 ). To mol ecul arly h eterogen ou s ) prot ei n s . Moreover, the di s a p pea r - a nce of t he hi gh m olecul a r m a s s ba nd (followi ng re duction ) is asso cia te d wit h an i nc rease i n t he inte ns ity of t he 140 -k Da 2 L. D. Powell a n d A. Varki, u n pu blish ed data . 7530 Characte riz atio n of CD22-0l igosaccharide Bind ing 3 mM 1 mM 0 mM DTSSP 2 -ME + + + 200'- AcHN OH F IG. 9 . Proposed m o d e l of C D22·sialoside bi n di ng. T he di sac cha - 2 3 4 5 6 ride Si a a2 -6Ga l, in t h e tg rot am er con for ma t ion, is pres en t ed interact- FI G. 8 . Cross -li n k i n g of C D22 on C D 22 e x p ressi n g C H O c e lls. ing with t h e bin ding pock et of CD2 2. Th is mode l is desi gn ed to incor - Cells met ab olica lly lab eled with [' S1Me t wer e cros s-linked wit h t h e porate t he dat a pr esented in Tab le I, a nd under "Discus s ion." R , : Gal, thi o-cleav able age n t , DTSSP, a t th e indica t ed conce nt ra t ion , and im- GlcNAc, Ga INAc, trimethylsilane, ben zyl, ONP, a lly l, isopropy lid en e or mun opr ecipitated with mAb To15, direct ed aga ins t hum an CD22 . After H; R : isopropylid ene , OH , or NAc; R ,,: Gal , isopropyliden e , or OH ; R.,: purifi cation , one -ha lf of eac h sa m ple was reduced with 2-mercaptoeth- isopropy lid en e or OH ; and R : methyl or OH . a no l (2-M El , as indic ated , a nd both reduced a n d n on r educed sa mples were a na lyzed on a 6% SDS -PAGE . Th e m olecul ar weight s ta n da r ds are s ive studies on antibo dy-hapten binding (r eviewed in Ref. 26 ) ind ica te d on th e left . TD, t rack in g dye. wou ld argue that th is enhanced affinity is due to th e simulta- neous binding of the two Sia a2-6GalJ31-4GlcNAc moieti es t o CD22J3 band , demonstrating th at thi s high molecular mass C D22Rg. Unambiguous proof of t h is point, of course , would band must contain at least some CD22 molecules . The bands req uire dir ect equ ilibri u m binding st u dies wi th mono- and migrating at the dye front adsorb to protein A-Sepharose in the bivalent forms of CD22 with mono- and bival ent s ia lyla te d ab sence of a n t i-CD22 and thu s repres ent nonspecific contami- oligosaccharides . nants . Thus, a portion of cell-surface CD22 m ust be present on Ev en if a single mul ti sialylated oligosaccharid e does sim u l- t he su r fa ce in a multim eric structure. taneously interact with the two lectin binding s ites of CD22 , DISCUSSI ON t he latter is in fact an artifici ally cre ated chimeric protein . It is Previ ousl y, we established that th e B cell sialic acid-binding mor e important to know if a s imila r ph enomenon can occur pr ot ein , CD22 , specifically recogniz ed th e tri saccharide Siaa2- with native full -length CD22J3 which is initially sy n t hes ized a s 6GalJ31-4GlcNAc as commonly found on N-linked oligosaccha - a monomer. Ind eed , two lin es of inv estigation (in h ibit ion of rides , but al so on some O-linked oligosacchari des and glycoli p- bi nding by oligosacc hari des and cell s ur fa ce cross-linking ) sug - ids (2 1, 37 ). These studies were all accomplishe d wi th a gest that the n a t ive molecu le expr ess ed on t he surface of CHO colu mn -bin ding assay, employing the bivalent im m u nog lobu lin ce lls, associates to form non coval en t oligomers. T hi s multirn- fus ion constr uct CD22 Rg bou n d to protei n A-Se pha rose, in eri c associatio n offers a mec hani sm of pr odu cin g a hig her ap- whic h t he elutio n of a p pro priately sialylated stru ct u res was pa r en t binding a ffinity t han wou ld be obse rve d wit h sing le signi fica n tly reta r de d beyond th a t of a nonbinding mon os a c- ch a in CD22. chari de . Stru ct u res with mor e a2 -6-Sia resi dues we re se pa - Colle ct ive ly, t he r esul t s obtai ne d wit h t he differ en t mono- rate d fro m t hose with less Sia resi d ues. Th is observatio n sug - sia lylated com pou n ds exami ne d in Ta ble I in dicate t hat t h e geste d t he poss ibi lity of mul t iva len t interactio n bet ween a minimum structu re req uired for CD22 Rg bindin g is Siaa2-6- m ultiply sialylated oligos a cchari de and the bivalent CD22 Rg H ex(NAc), where Hex(NAc) is Ga l, Ga INAc, or GlcNAc. An construct. Alternatively , the bet t er retention cou ld be mer ely exa mination of t he di fferen t RI Cs of t he com pou n ds listed in due to a statistical effect, i.e. increased pro ba bility of interac- T abl e I, togethe r with our earlier studies offers con sid erable tions with the colu m n du ri n g the elution. The results descri bed insight into t he regions of t he sialoside which are recogniz ed by herein demonstrate that CD22 Rg chimera h a s two bin di ng t he pr ot ei n . F ig. 9 shows Neu5Aca2-6Ga lJ3 1-R in th e t g rota- sit e s for a 2- 6-s La c, each with a K of - 30 P.M. Alt ho ug h eq ui- mer (t h e gt rota mer is formed by a 120 °C rotation around th e librium dialysis coul d not be performed on a bi(Sia a 2-6 )-bi- Ga l Cs-C bond ), an d we pr opose a model for CD22- si alo sid e antennary oligosaccharide, a dditional studies indica t ed t hat bi nd in g ba sed on t hese data and ea r lier studies 03 , 21 , 39 ). this compound bound to CD22 Rg with ap proximately 17-fold Th e C -C si de chain ofSia is es sential for binding, a s either 7-CS lower K . Additionally, a comparison of the in h ibitory pot enc y its re moval or its 9-0 -acetylation abolis he s binding ( 13 , 20 , 21 ). of seve r a l bisialylated oligosacchari des agai nst a 2- 6-s Lac in T he N substituent on Sia (glycolyl vers us acetyl) ma y be in retention assay furt he r s upports t he lik elih ood that ding sit e , a s the m urin e form of CD22 th e column close pr oximi t y to t he bi n simultaneous interaction is occurring. In t hese a ssays, t he shows a pa r ti al pr efer en ce for Neu5Gc over Neu5Ac (a pr efer- bisialylated compounds are clearly more pot en t in h ibito rs t ha n e nce not see n wit h h u ma n CD22 ) (1). The Gal 0 s-C face of a2-6-sLac whe n com pared on an equi mola r basis r el a ti ve to t h is di sa cch ari de mu s t a lso be invo lved in CD22 binding a s a2 -6-Sia gro ups , in dicating t hat t hey h a ve a higher a ffin ity t hese posit ion s are t he k ey structural fea tu r es which d ist in - tha n a 2- 6-s La c a lone . Th e wor k presented here a n d pr eviou sly gui sh a 2- 6-s La c fro m a 2-3 -s Lac. Th e genera lly un fa vor a ble (2 1) h a s indicate d t hat n o ot h er structural fe atures in a mul - effect of C methyla t ion of Ga l lik ewise in dicates that t hi s tiantennary oligosacchari de are r ecogn ized by CD22Rg. Th u s, r egion of t he di sacch a r id e faces t he bindin g s ite . In contrast, t he enha nce d a ffin ity of th e bis ial yl a t ed oligosa cch a r ides is t h e ot h er hyd r oxyl gro u ps of th e Ga l r esidue are not in volved in most lik ely du e to t he bindi n g of CD22 Rg to t he t wo Sia a2 - CD22 bin din g , as (a) a 2-6-sialylated Ga l, Ga INAc, a n d GlcNAc 6GalJ31-4GlcNAc structures terminati ng t he a nte n n ae . Ex t en- a r e a ll r ecogn ized; ( b ) ca pping some or a ll of t he hy droxyl Characterization of CD22-0ligosaccharide Binding 7531 groups with alkyl groups (as in the case of the di-O-isopropyli- multiple washing steps in between, is very different from a dene derivative) or with an additional Gal residue was without solution phase equilibrium or column assay. This discrepancy indicates that ELISA capture assays may not always be the effect on binding; (c) the configuration at Gal C is not signifi- cant (as both a2-6-sialylated GalNAc and GlcNAc are recog- best assay for describing interactions which may potentially be multivalent. nized). The group at the reducing terminus of Gal (R in Fig. 9) The binding affinity for a monovalent ligand seen here, ~30 must also face against the protein (as opposed to being oriented fLM, is in line with that observed for other lectin-oligosaccharide out into the solvent) as different blocking groups at this posi- interactions (45). For a single receptor-ligand interaction, this tion (trimethylsilane, benzyl, ONP, H, Glc, Glc-OH, or peptide) affinity is comparatively low and would suggest that in nature, markedly influence binding. Thus, we would predict that the CD22's function as an adhesion molecule is dependent upon CD22 binding site would be an area sufficiently large to accom- multiple interactions. Conversely, it is possible that higher modate these multiple points of contact. affinity interactions will be found with naturally occurring The binding specificity of CD22 is somewhat comparable to ligands. The CD22-containing oligomers we have reported here that of the lectin SNA. As studied by precipitation assays, this may be part of a mechanism to create a receptor structure lectin is specific for the sequence Siaa2-6GallGaINAc, with a capable of distinguishing between different a2-6-sialyloligo- K (by equilibrium dialysis) of 2.5 fLM (25, 40). As with CD22, saccharides. Such a mechanism(s) could be very important in the C side chain of Sia is required for binding (40). 7-C8-C 9 the rolets) CD22 plays in cell-cell adhesion and signaling. Unlike CD22, however, SNA recognizes additional structural The cross-linking studies presented here indicate that native determinants in an oligosaccharide structure, as evidenced by CD22 may also utilize multivalency to achieve functional in- the 100-fold greater inhibitory potency of Siaa2-6Gal{H- teractions. No dimers were seen in these studies, indicating 4GlcNAc{3l-3Gal{3l-4Glc over a2-6-sLac (40). Moreover, lac- that at least a portion of cell-surface CD22 exists in a multim- tose at concentrations of 0.1 M are capable of inhibiting SNA- eric state. Also, no association was seen with any other meta- oligosaccharide binding (25, 40), an effect which is not found bolically-labeled proteins, implying that these complexes are with CD22Rg (37). Mucins containing the Siaa2-6GalNAc se- homomultimers. In support of this, the intensity of the CD22 quence bind well to SNA (judged by a precipitin assay) (40), monomer band decreased in the presence of the cross-linker to while preparations of such mucins bind very poorly to CD22 an extent qualitatively consistent with the appearance of the (37). While 9-0-acetylated a2-6-Sia groups are not recognized multimer bands. Furthermore, the anti-CD22 antibody ad- by CD22, it is not known if SNA binding can be affected by this 35S1Met less than 0.2% of the total macromolecular [ sorbed substitutions. 6-Thio derivatives are not recognized by SNA, material, and in the presence of the cross-linker, the total nor does it show preferential binding to any of the synthetic amount of counts/min adsorbed to To15/protein A-Sepharose disialosides examined here (26-29) (29, 30). was -75% of that in its absence. If CD22 were being cross- The CD22Rg chimera was constructed to be bivalent by linked to a heterogeneous array of different proteins (and thus inclusion of the hinge region of IgG. The ability of a bi(Siaa2- not visible as a discrete band after reduction) then the amount 6)-biantennary oligosaccharide to simultaneously interact with of counts/min would be expected to increase in the presence of both binding sites is indicated by the higher levels of inhibition the cross-linking agent. Of course, we cannot rule out the of the bisialylated compounds over a2-6-sLac, both in the cell presence of non-labeled proteins in these multimers (either binding assay and the column assay. The different levels of long-lived or poor in MetJCys residues). Regardless, since CRO inhibition achieved by synthetic bisialylated compounds 26-29 cells (a fibroblastic cell line) are capable of supporting the indicate that the construct shows a preference for certain formation of these multimers, we can be certain that no B branching orientations. The ability of multiantennary oligosac- lymphocyte-specific proteins are required. It is possible that charides to simultaneously interact with multiple binding sites the CD22 molecules self-associate in the membrane following on multisubunit lectins has been demonstrated with several biosynthesis. In this regard, CD22 might be similar to the lectins (29, 41-44). One well studied example is the hepatic cation-dependent Man-6-P receptor, which although it has a GallGalNAc lectin. This protein is a trimer, with each monomer stoichiometry of 1:1 for the cognate oligosaccharide (46), is containing two GallGalNAc binding sites. An oligosaccharide able to generate high-affinity binding via the reversible for- with a single Gal residue on its nonreducing terminus binds mation of dimers and multirners (47-49). with a K of 1 mM, and branched structures containing two and three terminal Gal residues bind with affinities approximately Acknowledgments-We acknowledge the technical assistance of 3 6 Jeannette Moyer and Jennifer CalJow, and thank Graham Long for his 10 and 10 tighter, respectively (41, 43, 45). Likewise, the careful review of the manuscript. affinity of the cation-independent Man-6-phosphate receptor, which contains two binding sites, for ligands containing two REFERENCES Man-6-P groups is 300-fold higher than for ligands containing 1. Darken, B., Moldenhauer, G., Pezzutto, A., Schwartz, R., Feller, A., Kiesel, S., and Nadler, L. M. (1986) J. lmmunol. 136,4470-4479 just one (44). From basic thermodynamic principles, the appar- 2. Law, C.-L., Sidorenko, S. P .• and Clark, E. A. {l994) Immunol. Today 15, ent K for dimeric receptor-ligand binding should be the prod- 442-449 uct of each individual binding interaction; the extent that this 3. Pezzutto, A., Dorken, B.• Moldenhauer, G., and Clark, E. A (1987) J. Immunol. 138,98-103 theoretical limit is not met is reflective of strain or steric 4. Boue, D. R., and Lebien, T. W. (1988) J. Immunol. 140, 192-199 factors which limit the simultaneous independent interaction 5. Bone, D. R., and Lebien, T. W. (1988) Blood 71, 1480-1486 6. Stamenkovic, 1., and Seed, B. (1990) Nature 345, 74-77 of the two receptor-ligand pairs (26). The differences in appar- 7. Wilson, G. L., Fox, C. H., Fauci, A. S., and Kehrl, J. H. (1991)J. Exp. Med. 173, ent binding affinity of the different bisialylated compounds 137-146 (26-29; Table I) is indicative that the two Sia binding sites on 8. Stamenkovic, 1., Sgroi, D., Aruffo, A, Sy, M. S., and Anderson, T. (1991) Cell 66, 1133-1144 CD22Rg are not optimally oriented for all bisialylated struc- 9. Torres, R. M., Law, C. L., Santos-Argumedo, L., Kirkham, P. A, Grabstein, K., tures, i.e. that different levels of intramolecular strain are Parkhouse, R. M., and Clark, E. A. (1992) J. Immunol. 149,2641-2649 10. Engel, P., Nojima, Y., Rothstein, D., Zhou, L. J., Wilson, G. L., Kehrl, J. H., and induced in these compounds when bound to CD22Rg. Tedder, T. F. (1993) J. Immunol. 150,4719-4732 The inability of the ELISA assay to detect the bivalent na- 11. KeIrn, S., Schauer, R., Manuguerra, J-C., Gross, H.J., and Crocker, P. R. (1994) Glycocon] J 11, 120-126 ture of oligosaccharide recognition is surprising and without 12. Aruffo, A, Kanner, S. B., Sgroi, D., Ledbetter, J. A., and Stamenkovic, I. (1992) obvious explanation. However, the ELISA capture assay, in- Proc. Natl. Acad. Sci. U. S. A. 89, 10242-10246 volving a primary binding and then a secondary reagent, with 13. Sjoberg, E. R., Powell, L. D., Klein, A., and Varki, A (1994) J. Cell Bioi. 126, 7532 Characterization of CD22-0ligosaccharide Binding 549-562 Chern. Soc. 116,1616-1634 14. Paulson, J. C., Rearick, J. 1., and Hill, R. L. (1977) J. Bioi. Chem. 252, 32. Chandrasekaran, E. V., Jain, R K, Larsen, R. D., Wlasichuk, K., and Matta, 2363-2371 K. L. (1995) Biochemistry, in pre ss 15. Weinstem, J., de Souza-e-Silva, D., and Paulson, J. C. (1982) J. Bioi. Chern. 33. Jain, R. K. and Matta, K. L. (1994) XVlIth International Carbohydrate 257, 13835-13844 Symposium, July 17-22, 1994, Ottawa, Ontario, C2.6, 443, National Re- 16. Pezzutto, A., Rabinovitch, P. S., Dorken, B., Moldenhauer, G., and Clark, E. A. search Council of Canada, Ottawa, Ontario, Canada (1988) J. Immunol. 140,1791-1795 34. Jain, R. K., Piskorz, C. F., and Matta, K L. (1995) Carbohydrate Res., in pres. 17. Schultz, P. G., Campbell, M.-A., Pischer, W. H., and Sefton, B. M. (1992) 35. Vig, R., Jain, R K, and Matta, K L. (1994) Carbohydrate Res., in press Science 258, 1001-1004 36. Shukla, A K, and Schauer, R. (1982) Hoppe-Seyler's Z. Physiol. Chem. 363, 18. Peaker, C. J., and Neuberger, M. S. (1993) Eur. J. [mmunol. 23, 1358-1363 255-262 19. Leprince, C., Draves, K E., Geahlen, R. L., Ledbetter, J. A., and Clark, E. A 37. Powell, L. D., and Varki, A. (1994) J. Bioi. Chern. 269,10628-10636 (1993) Proc. Nat!. Acad. Sci. U. S. A. 90, 3236-3240 38. Debeire, P., Montreuil, J., Moczar, E., van Halbeek, H., and Vliegenthart, J. F. 20. Sgroi, D., Varki, A., Braesch-Andersan, S., and Stamenkovic, 1. (1993) J. Bioi. (1985) Ear. J. Biochem. 151,607-611 Chern. 268,7011-7018 39. Blinkovsky, A. M., and Dordick, J. S. (1993) Tetrahedron: Asymmetry 4,1221- 21. Powell, L. D., Sgroi, D., Sjoberg, E. R, Stamenkovic, 1., and Varki.. A. (1993) J. Bioi. Chern. 268,7019-7027 40. Shibuya, N., Goldstein, I. J., Broekasrt, W. F., Nsimba-Lubaki, M., Peeters, B., 22. Hanasaki, K., Varki, A, and Powell, L. D. (1995) J. Bioi. Chern. 270,7533- and Peumans, W. J. (1987) J. Bioi. Chern, 262, 1596-1601 41. Baenziger, J. U., and Maynard, Y. (1980) J. Bioi. Chem. 255,4607-4613 23. Hanasaki, K., Powell, L. D., and Varki, A. (1995) J. Bioi. Chern. 270,7543- 42. Baenziger, J. D., and Fiete, D. (1979) J. Bioi. Chem. 254,9795-9799 43. Lee, Y. C., Townsend, R. R., Hardy, M. R., Lonngren, J., Arnarp, J., 24. Lee, E. D., Roth, J., and Paulson, J. C. (1989) J. Bioi. Chern. 264, 13848-13855 Haraldss on, M., and Lonn, H. (1983) J. Bioi. Chern. 258, 199-202 25. Shibuya, N., Tazaki, K., Song, Z., Tarr, G. E., Goldstein, 1. J., and Peumans, 44. Tong, P. Y., Gregory, W., and Kornfeld, S. (1989) J. Bioi. Chem. 264, 7962- W. J. (1989) J. Biochem. (Tokyo) 106, 1098-1103 26. Berzofsky, J. A., Ep.tein, S. L., and Berkower, I. J. (1989) in Fundamental 45. Lee, Y. C., and Lee, R. T. (1994) Neoglycoconjugates: Preparation and Immunology (Paul, W. E., ed) pp. 315-353, Raven Press, New York Applico.tions, Academic Pre as, San Diego.. CA 27. Mapper, K, and Gindler, E. M. (1973) Anal. Biochem. 56,440-442 46. Tong, P. Y., and Kornfeld, S. (1989) J. Bioi. Chern. 264, 7970-7975 28. Wang, W. C., and Cummings, R. D. (1988) J. Bio!' Chem. 263,4576-4585 47. Dahms, N. M., and Kornfeld, S. (1989) J. Bioi. Chem. 264, 11458-11467 29. Sabesan, S., Duus, J. 0., Neira, S., Domaille, P., Keirn, S., Paulson, J. C., and 48. Hille, A., Waheed, A., and von Figura, K. (1989) J. Bioi. Chern. 264, 13460- Bock, K. (1992) J. Am. Chern. Soc. 114,8363-8374 30. Gupta, D., Sabssan, 8., and Brewer, C. F. (1993) Eur. J. Biochem. 216, 13467 789-797 49. Waheed, A., Hille, A., .Iunghans, D., and von Figura, K (1990) Biochemistry 31. Sabesan, S., Neira, S., Davidson, F., Duus, J. 0., and Bock, K. (1994) J. Am. 29, 2449-2455
Journal
Journal of Biological Chemistry – Unpaywall
Published: Mar 1, 1995