A carbohydrate neoepitope that is up-regulated on human mononuclear leucocytes by neuraminidase treatment or by cellular activation

Abstract

The expression of cell-surface antigens can delineate specific leucocyte developmental or functional stages. For example, certain membrane glycoproteins are expressed selectively on leucocyte subsets only after activation. Leucocyte activation can also induce changes in carbohydrate epitopes expressed on surface antigens. In the present studies, we report on a novel monoclonal immunoglobulin M antibody (mAb 13.22) that recognizes a unique carbohydrate epitope expressed on human leucocyte membrane proteins. Characterization of mAb 13.22 specificity by immunoblotting showed that it recognized proteins of MW ∼95 000 and 150 000, including both CD18 and CD11b. The mAb 13.22 epitope was removed by N-glycosidase F but not by endoglycosidase H or fucosidase, demonstrating that it is an N-linked carbohydrate antigen. Interestingly, immunoblot staining was enhanced after neuraminidase treatment, suggesting that the antibody epitope might also be partially masked by sialic acid. In resting leucocytes, the mAb 13.22 antigen was expressed strongly on neutrophils, while dull staining was present on monocytes, and no lymphocyte staining was observed. In marked contrast, treatment of leucocytes with neuraminidase resulted in exposure of a mAb 13.22 neoepitope on a subset of lymphocytes (primarily T lymphocytes and natural killer cells) as well as up-regulated staining more than 18-fold on monocytes. Activation of lymphocytes in culture with phytohaemagglutinin or concanavalin A also unmasked the mAb 13.22 neoepitope on ∼37% of the CD45RO⁺ lymphocytes. Furthermore, analysis of leucocytes collected from the synovial fluid of patients with rheumatoid arthritis showed that ∼18% of the lymphocytes present expressed the mAb 13.22 neoepitope. Taken together, our results suggest that the mAb 13.22 carbohydrate neoepitope could represent a physiologically relevant marker that is up-regulated on leucocyte subsets during the inflammatory response.

Introduction

The ability of circulating leucocytes to exit from the vascular system is crucial to the host defence process, and interaction with endothelial cells is essential for the localized infiltration of neutrophils and monocytes into extravascular inflammatory sites^1–3 or for the homing of lymphocytes to lymphoid tissues.^4,5 These leucocyte trafficking processes are mediated through a variety of cell surface adhesion molecules, including selectins, integrins, mucins and the immunoglobulin superfamily.^2,3,6 The importance of adhesion proteins in host defence is clearly demonstrated, for example, by the genetic disorder known as leucocyte adhesion deficiency (LAD).^7–9 Leucocytes from patients with LAD lack integrin β₂, resulting in severely compromised host defence capability. As a result, LAD patients suffer from severe recurrent bacterial and fungal infections.^7,8,10

The molecular events controlling cell adhesive interactions are not completely understood; however, several regulatory mechanisms have been implicated. One important regulatory mechanism is the modulation of expression of cell surface antigens via protein synthesis, exocytosis of intracellular stores, and shedding membrane adhesion molecules.^4,11,12 For example, neutrophil priming or activation can cause shedding of l-selectin from the cell surface or up-regulation of CD11b/CD18 or both, depending on the stimulus used.^13–16 Activation of lymphocytes also results in alterations in the level of expression of cell-surface ligands, including transferrin receptors,¹⁷ Ia antigens,¹⁸ insulin receptors,¹⁹ thymus leukaemia antigen²⁰ and interleukin-2 receptors.^21,22 These changes in antigen expression may help to control leucocyte adhesion and responsiveness during the inflammatory response.

In addition to changes in the level of expression of leucocyte cell-surface antigens, recent studies have shown that activation of leucocytes also results in changes in carbohydrate epitopes expressed on these antigens, possibly as a mechanism for regulating negative charge density on the cell surface [reviewed in ref. 6]. A number of studies have shown that modulation of sialic acid content of protein and lipid antigens on the surface of T cells occurs with changes in functional capacities of these cells,^23,24 and this response appears to be due, in part, to the increased expression of endogenous T-cell neuraminidase activity.^25,26 T cells activated in vitro with concanavalin A (Con A) are hyposialylated compared to resting cells,²⁵ and activated antigen-responsive lymphocytes that aggregate in vivo in lymph node germinal centres are also hyposialylated [reviewed in ref. 6]. Hyposialylation of lymphocyte cell-surface antigens has also been shown to occur in processes other than activation. For example, hyposialylation of lymphocyte proteins has been observed in human immunodeficiency virus (HIV) infection^27,28 and in the Wiskott–Aldrich immunodeficiency syndrome.²⁹ Furthermore, recent studies by Stamatos and co-workers³⁰ showed that desialylation of peripheral blood mononuclear cells actually helps to promote growth of HIV-1 in vitro. Thus, it is clear that leucocyte cell-surface protein carbohydrate decorations also play key roles in the immune function of these cells.

The identification of leucocyte surface markers which correlate with various states of leucocyte activation has been essential in dissecting the components of the immune response.^31–33 Although the majority of these markers are protein antigens, a number of studies have indicated that changes in carbohydrate expression can also be linked to functional changes in leucocytes (see above). In the present studies, we characterize a monoclonal immunoglobulin M (IgM) antibody (mAb 13.22), which recognizes a unique carbohydrate epitope expressed on human leucocyte membrane proteins. In resting cells, the mAb 13.22 antigen is strongly expressed on neutrophils, but only at very low levels on monocytes and not at all on lymphocytes. In contrast, treatment of leucocytes with neuraminidase resulted in exposure of a mAb 13.22 neoepitope on a subset of lymphocytes as well as up-regulated staining more than 20-fold on monocytes. Multi-colour flow cytometric analysis demonstrated that the mAb 13.22 neoepitope was expressed on distinct subsets of T lymphocytes and natural killer (NK) cells, but not on B lymphocytes. We also show that activation of lymphocytes in culture with mitogens resulted in exposure of the mAb 13.22 neoepitope on a subset of CD45RO⁺ lymphocytes. Furthermore, we found that this neoepitope was expressed on a subpopulation of the lymphocytes that were present in synovial fluid from patients with rheumatoid arthritis. Overall, our results suggest that the mAb 13.22 carbohydrate neoepitope could represent a physiologically relevant marker that is exposed on leucocyte subsets during the inflammatory response.

Materials and methods

Reagents

Monoclonal antibodies used in these studies included mAb 13.22,³⁴ anti-CD11a [MCA1149 from Serotech Ltd, Oxford, UK; MM12A from VMRD, Pullman, WA; TS2/4 from the American Type Culture Collection (ATCC, Manassas, VA; HB-244)], anti-CD11b (44a),³⁵ anti-CD11c (RDI-CBL146 from Research Diagnostics, Inc., Flanders, NJ; SHCL-3 from Becton Dickinson Biosciences, San Jose, CA), anti-CD18 [RDI-CBL514 from Research Diagnostics, Inc., Flanders, NJ; IB4 from ATCC (HB-10164); TS1/18 from ATCC (HB-203)], antil-selectin (DREG 56),³⁶ and anti-CD16 (57.1.9) (M. T. Quinn, K. L. Jutila, and M. A. Jutila, unpublished results). Polyclonal antibodies used in these studies included anti-CD11b (R7928a)³⁴ and anti-CD18 (R7928c), which was prepared as described previously³⁷ by immunizing rabbits with a synthetic peptide having the sequence of the C-terminus of CD18 (NH₃-REYRRFEKEKLKSQWNNDNPL-COOH) coupled to keyhole limpet haemocyanin with glutaraldehyde. Fluorescein isothiocyanate (FITC)-conjugated anti-CD4 (Leu-3a/3b) and anti-CD45RA (Leu-18); phycoerythrin (PE)-conjugated anti-CD8 (Leu-2a), anti-CD11b (Leu-15), anti-CD62L (Leu-8) and anti-CD16 (Leu-11c); biotinylated anti-CD3 (Leu-4) and anti-T-cell receptor (TCR) γ/δ; and allophycocyanin (APC)-conjugated anti-CD19 (SJ25C1) and anti-CD45RO (Leu-45RO) were all from Becton Dickinson Biosciences. FITC-conjugated anti-CD2 (T11) and anti-CD19 (B4) were from Coulter (Miami, FL). Affinity-purified goat anti-mouse IgM (μ-chain-specific), alkaline phosphatase-conjugated goat anti-mouse IgG (H + L), alkaline phosphatase-conjugated goat anti-mouse IgM (μ-chain-specific), FITC-conjugated goat anti-mouse IgG (H + L), and PE-conjugated goat anti-mouse IgM (μ-chain-specific) were all from Jackson ImmunoResearch, Inc. (West Grove, PA). Streptavidin Cy-Chrome was from Pharmingen (San Diego, CA). Goat anti-rabbit IgG (H + L) was from BioRad Laboratories (Hercules, CA).

Functionally active CD11b/CD18 was purified to homogeneity by LM2/1 immunoaffinity chromatography as described previously.^38,39

Preparation of leucocyte membrane detergent extracts

Neutrophils and mononuclear cells were purified from peripheral blood using dextran sedimentation and gradient separation on Histopaque 1077 as described previously.⁴⁰ To purify lymphocytes further, the mononuclear cell layer was collected, washed, resuspended at 5 × 10⁶ cells/ml in Dulbecco's modified Eagle's minimal essential medium containing 0·08% bovine serum albumin, and incubated in T-175 tissue culture flasks (20 ml/flask) for 30 min at 37°. The non-adherent lymphocytes were then collected by gentle aspiration.

To obtain membrane detergent extracts, neutrophil and lymphocyte preparations were resuspended in membrane buffer (100 mm KCl, 10 mm NaCl, 10 mm HEPES, pH 7·4), treated with 3 mm diisopropylfluorophosphate for 15 min on ice, and homogenized by N₂ cavitation for 15 min as described previously.⁴¹ The cavitate was collected, centrifuged at 1000 g for 5 min at 4° to remove unbroken cells and nuclei, and the supernatant was centrifuged at 100 000 g for 30 min at 4°. The membrane pellet was washed once with membrane buffer containing 2 mm MgCl₂ and protease inhibitors (1 mm phenylmethylsulphonyl fluoride, 10 µg/ml chymostatin, 1 µg/ml aprotinin, and 1 µg/ml pepstatin A) and once with the same buffer containing 1 m NaCl. The resulting pellet was resuspended at 2·5 × 10⁸ cell equivalents/ml and extracted for 1 hr on ice with 1% octyl-β-d-glucopyranoside in membrane buffer with inhibitors. The sample was then centrifuged at 100 000 g for 30 min at 4°, and the supernatant was collected (membrane detergent extract).

Lymphocyte cultures

Human peripheral blood mononuclear cells, obtained by separation on sterile Histopaque 1077, were resuspended in sterile RPMI-1640 medium supplemented with 10% fetal calf serum (Hyclone, Logan, UT) and divided into aliquots into T-75 tissue culture flasks (20 ml containing 5 × 10⁶−7 × 10⁶ cells/flask). Control cells, untreated cells, or cells treated with 5 µg/ml phytohaemagglutinin (PHA) (data not shown) or 5 µg/ml Con A were incubated at 37°/10% CO₂. At 7 days, cells were harvested gently to avoid collecting any adherent cells and analysed by flow cytometry as described below.

Preparation of leucocytes from synovial fluid

Patients who fulfilled the American College of Rheumatology criteria for rheumatoid arthritis were studied. Synovial fluid was collected into anticoagulant-containing (5 mm ethylenediaminetetraacetic acid or acid-citrate dextrose) tubes during therapeutic aspiration from knees of patients and analysed within 2 hr of collection. The fluid was diluted 2 : 1 with Hanks' balanced salt solution (HBSS), centrifuged at 200 g to pellet the cells, and the cells were resuspended in HBSS and stained for one- and two-colour flow cytometric analyses, as described below.

Flow cytometry

Single-colour flow cytometric analysis was performed as described previously.⁴² For two-colour flow cytometric analysis, 1 × 10⁶ cells in 100 µl Dulbecco's phosphate-buffered saline (DPBS) + 2% normal goat serum were incubated for 30 min on ice together with 100 µl DPBS+2% goat serum containing two different primary antibodies (directly conjugated with FITC, PE or APC) diluted at 50 µg/ml. The cells were then washed again with DPBS+2% goat serum, resuspended in 500 µl of the same buffer, and analysed on a FACSCalibur flow cytometer (Becton Dickinson).

For three- and four-colour flow cytometric analysis, 1 × 10⁶ cells in 100 µl DPBS+2% normal goat serum were incubated for 30 min on ice together with 100 µl DPBS+2% goat serum containing one unconjugated primary (diluted at 50 µg/ml). After washing with 3–4 ml of DPBS+2% goat serum, the cells were incubated for 30 min in the dark on ice with 100 µl secondary antibody (anti-mouse FITC-IgG or anti-mouse PE-IgM diluted at 1 : 250). The cells were then washed again with DPBS+2% goat serum, incubated with 100 µl 10% normal mouse serum for 20 min on ice (to block available anti-mouse immunoglobulin binding sites on the second-stage reagent), and then incubated for 30 min in the dark on ice with additional primary antibodies directly conjugated with a fluorochrome (e.g. APC, FITC, or PE) or biotin. Avidin CyChrome was used to reveal biotin-labelled antibodies. The cells were then washed, resuspended in 500 µl DPBS+2% goat serum, and analysed on a FACSCalibur flow cytometer calibrated for two-, three-, and four-colour analysis with Calibright beads (Becton Dickinson). Negative controls included (1) cells alone, (2) second-stage alone, (3) single colour stains, and (4) the differences in the staining patterns of the individual directly conjugated mAbs. A minimum of 10 000 cells was analysed for each sample. Data are shown in single colour histograms or in two-dimensional plots. Marker placement for determination of per cent positive cells and for statistical comparisons was established by placing the marker outside the upper limit of background staining.

Glycosidase treatment of neutrophil membrane detergent extracts

For endoglycosidase H treatment, 25 µl neutrophil membrane detergent extract containing 20 µg protein were mixed 1 : 1 with 0·1 m β-mercaptoethanol and 0·1% sodium dodecyl sulphate (SDS) and heated to 100° for 5 min. After cooling, 15 µl 0·5 m sodium citrate pH 5·5, 75 µl H₂O, 1 : 100 dilution of protease inhibitor cocktail (Sigma), and 5 mU endoglycosidase H were added and incubated overnight at 37°. For α-l-fucosidase treatment, 25 µl neutrophil membrane detergent extract containing 20 µg protein were mixed 1 : 1 with 0·1 m β-mercaptoethanol and 0·1% SDS and heated to 100° for 5 min. After cooling, 20 µl 0·1 m sodium acetate pH 5·5, 1 : 100 dilution of protease inhibitor cocktail and 5 µg α-l-fucosidase were added and incubated overnight at 37°. For neuraminidase treatment, 25 µl neutrophil membrane detergent extract containing 20 µg protein were mixed 1 : 1 with 0·1 m β-mercaptoethanol and 0·1% SDS and heated to 100° for 5 min. After cooling, 20 µl 0·1 m sodium acetate pH 5·5, 1 : 100 dilution of protease inhibitor cocktail, and 25 mU neuraminidase were added and incubated overnight at 37°. For N-glycosidase F treatment, 25 µl neutrophil membrane detergent extract containing 75–100 µg of protein were mixed 1 : 1 with 0·1 m β-mercaptoethanol and 0·5% SDS and heated to 100° for 5 min. After cooling, 25 µl 0·5 m Tris pH 8·0, 5 µl 20% octyl-β-d-glucopyranoside, 1 : 100 dilution of protease inhibitor cocktail, 10 µl of 0·1 m 1-10 phenanthroline, and 2 mU N-glycosidase F were added and incubated at 37° overnight. For controls on all glycosidase treatments, the enzyme was replaced with buffer, and the samples were processed identically. After treatment, all samples were heated for 5 min at 100° to inactivate the enzyme, mixed with SDS–polyacrylamide gel electrophoresis (SDS–PAGE) sample buffer, and analysed by SDS–PAGE and Western blotting as described below.

Immunoprecipitation

For immunoprecipitation with IgG isotype antibodies, neutrophil membrane detergent extract (100 µl) was incubated with 5–10 µg antibody for 18 hr at 4°. The samples were then incubated for 3 hr at room temperature with 100 µl of a 50 : 50 slurry of Protein G–Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ) in membrane buffer containing 1% octyl-β-d-glucopyranoside and 1 : 100 Sigma protease inhibitor cocktail. After seven washes with membrane buffer containing 1% octyl-β-d-glucopyranoside and protease inhibitors, the Sepharose beads were boiled for 5 min in 120 µl of SDS–PAGE sample buffer and analysed by SDS–PAGE and Western blotting as described below.

For immunoprecipitation with mAb 13.22 (IgM isotype), neutrophil membrane detergent extract (100 µl) was incubated with 16 µg mAb 13.22 for 18 hr at 4°, followed by the addition of 6 µg goat anti-mouse IgM secondary antibody (Jackson ImmunoResearch) and a 1-hr incubation at room temperature. The samples were then incubated with Protein G–Sepharose and processed exactly as described above.

Electrophoresis and Western blotting

SDS–PAGE using 7–18% polyacrylamide gradient gels and Western blotting were performed as described previously,⁴³ except that the blots were blocked with Blotto-Tween (5% dry milk+0·2% Tween-20) overnight at 4°. Transfers were blotted with primary antibody for 3 hr at 25°, followed by the appropriate alkaline phosphatase-conjugated secondary antibody for 1 hr at 25°, and developed using a BioRad alkaline phosphatase development kit.

Results

Previously, mAbs were generated against human neutrophil membrane proteins by immunizing mice with heparin Ultrogel/lectin affinity purified neutrophil membrane proteins.^34,44 Interestingly, one of the IgM isotype antibodies produced in this fusion, mAb 13.22, appeared to recognize both CD11b and CD18 on Western blots.³⁴ This unique staining profile suggested a common epitope on both subunits of CD11b/CD18 and led us to perform further studies to characterize the specificity of this antibody.

To further characterize the epitope(s) recognized by mAb 13.22, we analysed its reactivity by Western blot analysis of neutrophil and lymphocyte membrane extracts. As shown in Fig. 1(a), mAb 13.22 recognized proteins of ∼95 000 and 150 000 MW in neutrophil membrane extracts, while little or no staining was observed in detergent extracts of lymphocyte membranes. Since this staining pattern on Western blots was similar to that of the integrin β₂,⁴⁵ we considered whether mAb 13.22 might be recognizing one or more of the forms of integrin β₂. Consistent with this hypothesis, mAb 13.22 specifically stained both CD11b and CD18 in samples of affinity-purified neutrophil CD11b/CD18 (Fig. 1b), while there was no staining in samples of affinity-purified CD11b/CD18 precleared of CD11b/CD18 by immunoprecipitation with anti-CD11b mAb 44a (Fig. 1c). The effectiveness of preclearing was confirmed by lack of CD11b staining in the precleared sample (Fig. 1c). In neutrophil membrane detergent extracts immunoprecipitated with an anti-CD11b mAb 44a, mAb 13.22 also specifically stained both CD11b and CD18 (Fig. 1d). Finally, mAb 13.22 was able to directly immunoprecipitate CD11b/CD18 from neutrophil membrane detergent extracts, as determined by staining mAb 13.22 immunoprecipitates with antibodies specific for CD11b (Fig. 1e) and CD18 (not shown). Thus, our results demonstrate that mAb 13.22 does indeed recognize both CD11b and CD18 on human neutrophils. Furthermore, immunoprecipitation of leucocyte detergent lysates with a panel of antibodies specific for CD11a (mAbs MCA1149, MM12A, and TS2/4) and CD11c (mAbs RDI-CBL146 and SHCL-3) followed by immunoblotting with mAb 13.22 demonstrated that mAb 13.22 specifically recognized CD11b/CD18; no staining was observed in immunoprecipitated CD11a or CD11c, even though analysis of the immunoprecipitated proteins by SDS–PAGE and silver staining showed that we had indeed immunoprecipitated these integrin β₂ α subunits (n = 5; data not shown).

Figure 1.

The mAb 13.22 recognizes both subunits of CD11b/CD18. (a) Membrane detergent extracts prepared from purified human lymphocytes (L) and neutrophils (N) were analysed by SDS–PAGE followed by silver staining, as well as by Western blotting with mAb 13.22. Representative of three independent experiments. (b) Affinity-purified CD11b/CD18 (5 µg) was analysed by SDS–PAGE followed by silver staining, as well as by Western blotting with mAb 13.22. Representative of three independent experiments. (c) Affinity-purified CD11b/CD18 (5 µg) was analysed by immunoprecipitating with mAb 44a (anti-CD11b) followed by Protein G–Sepharose. The supernatant (lane 1 in both blots) and immunoprecipitate (lane 2 in both blots) were analysed by SDS–PAGE and Western blotting with mAb 13.22 and an anti-peptide antibody recognizing CD11b (R7928a), as indicated. Representative of three independent experiments. (d) Membrane detergent extract prepared from purified human neutrophils (lane 1) was immunoprecipitated with Protein G–Sepharose alone (lane 2) or with mAb 44a (anti-CD11b) followed by Protein G–Sepharose (lane 3). The samples were then analysed by SDS–PAGE and Western blotting with mAb 13.22 and antibodies R7928a and R7928c, as indicated. Representative of four independent experiments. (e) Membrane detergent extract prepared from purified human neutrophils (starting material shown in lane 1) was immunoprecipitated with Protein G–Sepharose alone (lane 2) and with mAb 13.22 followed by goat anti-mouse IgM and then Protein G–Sepharose (lane 4). As a background control, mAb 13.22 alone was incubated with goat anti-mouse IgM followed by Protein G–Sepharose (lane 3). All samples were then analysed by SDS–PAGE and Western blotting with mAb 13.22 and antibody R7928a, as indicated. Representative of four independent experiments. In all panels, prestained molecular weight standards (STD) are indicated, and the upper and lower arrowheads indicate the locations of CD11b and CD18, respectively.

As stated above, mAb 13.22 was shown to be an IgM isotype. Since IgM antibodies commonly recognize carbohydrate epitopes on proteins,⁴⁶ we investigated whether mAb 13.22 staining was altered by glycosidase treatment of the samples before Western blot analysis. As shown in Fig. 2, treatment of neutrophil membrane detergent extracts with endoglycosidase H, which cleaves high-mannose-type and hybrid oligosaccharide chains between the two N-acetylglucosamine (GlcNAc) residues in the core region of the oligosaccharide chain, did not affect staining of CD11b/CD18 by mAb 13.22, even though the size of CD11b was reduced by ∼10 000–15 000. In contrast, treatment with N-glycosidase F, which cleaves the bond between the asparagine residue of the protein and the GlcNAc residue, resulted in a complete loss of staining by mAb 13.22 (Fig. 2), clearly demonstrating that the epitope recognized by mAb 13.22 was carbohydrate. In samples treated with neuraminidase, which removes terminal sialic acid groups, mAb 13.22 was still able to stain CD11b/CD18 (Fig. 2); however, the staining intensity observed in neuraminidase-treated samples was consistently enhanced by ∼20–30% over the staining intensity in untreated samples (determined by densitometric analysis of the blots). This difference in staining intensity suggested that removal of sialic acid might make the antigenic epitope even more accessible to staining by mAb 13.22. Finally, treatment with α-l-fucosidase, which cleaves terminal fucose groups, had no effect on mAb 13.22 staining and also did not alter the size of CD11b (Fig. 2). Note that inclusion of a mixture of protease inhibitors was essential in the fucosidase-treated samples, as a contaminating protease activity was present in our commercial fucosidase preparations, and even in the presence of protease inhibitors some proteolysis of the antigen was observed (Fig. 2). In any case, our glycosidase experiments show that mAb 13.22 recognizes a carbohydrate epitope that can be removed by N-glycosidase F and may be partially masked by sialic acid.

Figure 2.

Analysis of the sensitivity of the mAb 13.22 antigen to glycosidase treatment. Membrane detergent extract prepared from purified human neutrophils (corresponding starting material shown in lanes 1, 3, 5, 7) was treated with endoglycosidase H (lane 2), α-l-fucosidase (lane 4), neuraminidase (lane 6), and N-glycosidase F (lane 8). The untreated and treated samples were then analysed by SDS–PAGE (15 cm gels) and Western blotting with mAb 13.22 and antibody R7928a, as indicated. In both panels, prestained molecular weight standards (STD) are indicated. Representative of five independent experiments.

One interesting feature of CD11b/CD18 staining by mAb 13.22 on Western blots was that it stained a much broader band for CD11b and CD18 than did the antipeptide antibodies, suggesting that CD11b and CD18 might represent a subset of leucocyte membrane glycoproteins that express the mAb 13.22 carbohydrate epitope. To evaluate this possibility, neutrophil membrane detergent extracts were first precleared of CD11b/CD18 by immunoprecipitation with mAb 44a, separated on high-resolution, large-format (15 cm long) SDS–PAGE gels, transferred to nitrocellulose, and then immunoblotted. As shown in Fig. 3, preclearing of CD11b/CD18, as determined by the absence of CD11b staining in precleared samples (lane 5), did not remove all of the mAb 13.22 staining and diffuse staining was still present in both the ∼95 000 and 150 000 MW regions (lane 2). As expected, CD11b and CD18 were stained by mAb 13.22 (lane 3) and CD11b was stained by anti-CD11b antibody R31789 (lane 6) in the immunoprecipitated material. Thus, mAb 13.22 appears to recognize a carbohydrate expressed on a group of surface proteins, of which CD11b and CD18 represent a major subset. Studies are currently in progress to determine the identity of other proteins expressing this epitope.

Figure 3.

Analysis of the specificity of mAb 13.22. Membrane detergent extract prepared from purified human neutrophils was precleared of CD11b/CD18 by immunoprecipitation with anti-CD11b mAb 44a followed by Protein G–Sepharose. The starting material (lanes 1 and 4), supernatant (lane 2 and 5), and immunoprecipitate (lanes 3 and 6) were then analysed by SDS–PAGE and Western blotting with mAb 13.22 and antibody R7928a (anti-CD11b), as indicated. Representative of three independent experiments.

Flow cytometric analysis of mixed leucocyte preparations with mAb 13.22 showed that it stained essentially all neutrophils very brightly (mean fluorescence intensity of 428 versus 3 for background control), while very dull staining was observed on resting monocytes (mean fluorescence intensity of 17 versus 5 for background control) (Fig. 4). In contrast, no staining above background control was observed on resting lymphocytes (mean fluorescence intensity of 3 versus 2 for background control) (Fig. 4). These results are consistent with the Western blots shown in Fig. 1, where mAb 13.22 stained CD11b/CD18 in neutrophil membrane extracts but not in lymphocyte membrane extracts.

Figure 4.

Expression of the mAb 13.22 antigen on neuraminidase-treated leucocytes. Human peripheral blood leucocytes were isolated and treated for 1 hr with 100 mU/ml neuraminidase (bold line). Control, untreated cells were incubated under the same conditions without neuraminidase (solid line). The cells were then stained with mAb 13.22 followed by a PE-conjugated second-stage antibody. As a control for background staining, cells were also incubated with the PE-conjugated second-stage antibody alone (dotted line). Expression of the mAb 13.22 antigen on the indicated leucocyte subsets, which were identified by their distinctive forward and side light scatter profiles, is shown in the representative histograms. Representative of at least five independent experiments.

As mentioned above, treatment of neutrophil membrane detergent extracts with neuraminidase resulted in an enhancement of mAb 13.22 staining intensity on Western blots (see Fig. 2), suggesting that the mAb 13.22 epitope might be partially masked by sialic acid. Therefore, we evaluated what effect neuraminidase treatment of leucocytes had on mAb 13.22 staining by flow cytometric analysis. As shown in Fig. 4, neuraminidase treatment of whole leucocyte populations resulted in enhanced staining of neutrophils when compared to untreated cells (mean fluorescence intensity of 2084 versus 428 for untreated cells). This up-regulation of neutrophil CD11b/CD18 expression, as determined by mAb 13.22 staining, is probably due to mild activation of the cells during the incubation at 37°, resulting in secretory and specific granule mobilization.^47–49 In support of this conclusion, we found that CD18 expression was up-regulated (mean fluorescence intensity of 867 versus 69 for cells maintained at 4°) and l-selectin expression was down-regulated (mean fluorescence intensity of 35 versus 152 for cells maintained at 4°) on neutrophils incubated at 37° without neuraminidase. Furthermore, the level of CD18 and l-selectin staining on neuraminidase-treated neutrophils was not significantly different than that observed on cells incubated at 37° without neuraminidase.

Even more striking, however, were the effects of neuraminidase treatment on mAb 13.22 staining in the mononuclear cell populations. Most notably, a subset of lymphocytes was now stained with mAb 13.22, suggesting that neuraminidase treatment exposed a mAb 13.22 neoepitope on these cells (mean fluorescence intensity of 73 versus 3 for untreated cells) (Fig. 4). Note that exposure of this neoepitope was dependent on the presence of neuraminidase, and no mAb13.22 staining was observed on cells incubated at 37° without neuraminidase (mean fluorescence intensity of 2 versus 3 for cells maintained at 4°). Neuraminidase treatment also dramatically enhanced the staining on monocytes (mean fluorescence intensity of 291 versus 16 for untreated cells) (Fig. 4). Again the up-regulation of mAb 13.22 staining on monocytes was primarily dependent on the presence of neuraminidase, and only a modest increase in mAb 13.22 staining was observed on cells incubated at 37° without neuraminidase (mean fluorescence intensity of 48 versus 16 for cells maintained at 4°). Note that the entire population of monocytes shifted in staining intensity after neuraminidase treatment. In contrast, only a subset of neuraminidase-treated lymphocytes stained brightly with mAb 13.22 (Fig. 4). Further analysis by two-colour flow cytometry showed that many (62%), but not all of the 13.22⁺ lymphocytes expressed moderate levels of CD11b (Fig. 4). Based on the results shown in Fig. 3, it is likely that the remaining lymphocytes (CD11b^–/13.22⁺) express unidentified surface molecules containing the mAb 13.22 carbohydrate epitope. Analysis of monocyte staining patterns showed that essentially all of the 13.22-positive cells also stained brightly for CD11b (Fig. 5), indicating that the mAb 13.22 epitope is expressed primarily on CD11b/CD18 in these leucocytes. Overall, these results were surprising, considering that mAb 13.22 failed to stain lymphocytes and only showed dull staining of monocytes in untreated cell populations.

Figure 5.

Two-colour flow cytometric analysis of the expression of CD11b and the mAb 13.22. Neoepitope on neuraminidase-treated monocytes and lymphocytes. Human peripheral blood leucocytes were isolated and treated for 1 hr with 100 mU/ml neuraminidase. Untreated (a, c) and treated (b.d) cells were then analysed by two-colour flow cytometry using PE-anti-CD11b versus mAb 13.22 indirectly stained with a FITC-second-stage reagent. Two-colour plots for gated lymphocytes (a, b) and monocytes (c, d), which were identified by their distinctive forward and side light scatter profiles, are shown. The upper level of background staining seen with negative controls is indicated by the quadrant markers. The percentage of gated cells staining positive for each antigen is indicated. Representative of six independent experiments.

To define the lymphocyte populations stained by mAb 13.22 further, we used three- and four-colour flow cytometric analysis of neuraminidase-treated cells. As shown in Fig. 6, approximately two-thirds of the lymphocytes stained by mAb 13.22 after neuraminidase treatment were T cells (∼65% of the 13.22⁺ cells were CD3⁺ Fig. 6b). Among these 13.22⁺ T cells, ∼15% were γδ T cells (Fig. 6i). The remaining third of 13.22⁺ cells were NK cells as shown by staining for CD16 (FcγRIII) (∼36% of the 13.22⁺ cells were CD16⁺ Fig. 6e). Interestingly, most of the 13.22⁺ NK cells showed dull staining for CD16, as compared to the 13.22^– NK cells, suggesting that a unique NK cell subset is stained by mAb 13.22. Negligible levels of staining for CD19 (< 5% of the 13.22⁺ cells) suggested that very few, if any, of the 13.22⁺ cells were B cells (Fig. 6f). Thus, mAb 13.22 recognizes a carbohydrate neoepitope that is present primarily on T-cell and NK cell subsets after treatment with neuraminidase.

Figure 6.

Multi-colour flow cytometric analysis of neuraminidase-treated lymphocytes. Human peripheral blood mononuclear leucocytes were isolated and treated for 1 hr with 100 mU/ml neuraminidase and analysed by multicolour flow cytometry. Multi-colour plots for gated lymphocytes, which were identified by their distinctive forward and side light scatter profile, are shown. Staining patterns of FITC-anti-CD2 (a), biotin-anti-CD3 followed by avidin CyChrome (b), FITC-anti-CD4 (c), biotin-anti-CD16 followed by avidin CyChrome (e), APC-anti-CD19 (f), APC-anti-CD45RO (g), FITC-anti-CD45RA (h), and biotin anti-γ/δ TCR followed by avidin CyChrome (i), versus mAb 13.22 indirectly stained with a PE-second-stage reagent are shown. (d) and (j) show the staining pattern of PE-anti-CD8 and PE-anti-l-selectin, respectively, versus mAb 13.22 indirectly stained with a FITC-second-stage reagent. (k) and (l) show representative background staining due to FITC-, PE- and CyChrome second-stage reagents alone. Similar background staining was observed for the other second-stage reagents used. Furthermore, the same pattern of reactivity was seen when different combinations of the various directly conjugated mAbs were used in the same stain, further confirming the specificity of the staining reactions. The quadrant markers shown are based on the upper threshold of staining for negative controls and were used to determine percentage of cells staining with the indicated mAbs. The numbers shown in parentheses indicate the per cent of 13.22⁺ cells that were double-positive for 13.22 and the indicated antigen. Representative of nine independent experiments.

Further characterization of the 13.22⁺ lymphocyte populations showed that many were CD2⁺ (∼83% of the 13.22⁺ cells; Fig. 6a) and CD4⁺ (∼52% of the 13.22⁺ cells; Fig. 6c), while a smaller fraction was CD8⁺ (∼17% of the 13.22⁺ cells; Panel D). Of the NK cells stained by mAb 13.22, ∼80% were CD2⁺ while the remainder were CD2^– (not shown). Staining for CD45RO/RA suggested that the 13.22⁺ lymphocytes were split fairly evenly between CD45RO^high/RA^low (∼50% of the 13.22⁺ cells; Fig. 6g) and CD45RO^low/RA^high (∼40% of the 13.22⁺ cells; Fig. 6h). Interestingly, almost all (∼90%) of the 13.22⁺ lymphocytes also stained for l-selectin (Fig. 6j), a molecule intimately involved in lymphocyte homing to sites of inflammation.⁴⁶

The ability of neuraminidase treatment to expose or enhance expression of the mAb 13.22 epitope on lymphocytes and monocytes, respectively, suggested the possibility that this epitope might also be exposed in a physiologically relevant manner by neuraminidase released from activated leucocytes during an inflammatory response. Previously, Landolfi and co-workers^23,25 found that activation of T-lymphocyte cultures with Con A resulted in expression of endogenous T-cell neuraminidase activity and hyposialylation of Class I molecules. Therefore, we performed studies to investigate whether the mAb 13.22 neoepitope might be revealed under similar conditions. Lymphocytes were activated in culture with Con A and the cells were collected at various times for flow cytometric analysis. To monitor activation, cells were also stained for CD45RA/CD45RO, which provides an indication of the virgin/memory status of these cells.⁵⁰ As shown in Fig. 7, activation of lymphocytes with Con A induced the characteristic transition from CD45RA^high/RO^low (naive phenotype) to CD45RA^low/RO^high (memory/activated phenotype). Concomitant with the transition to a memory phenotype, increased staining of the mAb 13.22 epitope was observed in a subpopulation of lymphocytes after treatment with Con A (Fig. 7). This lymphocyte subpopulation represented ∼37% of the CD45RA^low/RO^high T lymphocytes. Similar results were observed for PHA-treated cells (data not shown). In comparison, lymphocytes activated with phorbol 12-myristate 13-acetate (PMA) or N-formyl-methionine-leucine-phenyalanine (fMLF) showed little or no up-regulation of the mAb 13.22 neoepitope, while typical activation-induced changes in CD11b/CD18 and l-selectin occurred in the lymphocyte, monocyte and neutrophil populations (Table 1). Specifically, monocytes and neutrophils showed striking up-regulation of CD18 and shedding of l-selectin during activation. Lymphocytes showed minimal up-regulation of CD18 but shed most of their l-selectin in response to PMA treatment. As expected, fMLF did not induce l-selectin shedding from lymphocytes (Table 1). Irrespective of the treatment condition, no changes in lymphocyte or monocyte mAb 13.22 antigen expression were observed. Thus, exposure of the 13.22 epitope is the result of a neuraminidase activity and cannot be explained by a simple increase or change in the size of the CD11b/CD18 pool during activation. In any case, these studies further support the hypothesis that T-cell activation results in an alteration of the carbohydrate decorations of lymphocyte membrane proteins, including CD11b and CD18, as shown here, and class I molecules, as reported by Landolfi et al.^23,25

Figure 7.

Exposure of the mAb 13.22 neoepitope on lymphocytes after mitogen activation. Human peripheral blood mononuclear cells were cultured for 7 days in the absence (Control; a, c) or presence (Con A; b, d) of 5 µg/ml Con A. The cells were then harvested and analysed by multi-colour flow cytometry. Panels labelled Control and Con A show the staining pattern of APC-anti-CD45RO versus FITC-anti-CD45RA, (a) and (b) show the staining pattern of APC-anti-CD45RO versus mAb 13.22 indirectly stained with a PE-conjugated second-stage antibody, and (c) and (d) show the staining pattern of FITC-anti-CD45RA versus mAb 13.22 indirectly stained with a PE-conjugated second-stage antibody. In (a)–(d), the upper level of background staining seen with negative controls and the percentage of gated cells staining above background are indicated. Representative of three independent experiments.

Table 1.

Analysis of mAb 13.22 antigen expression on leucocytes activated with PMA and fMLF: comparison with CD18 and l-selectin expression

	Lymphocytes			Monocytes			Neutrophils
Antigen	Control	PMA	fMLF	Control	PMA	fMLF	Control	PMA	fMLF
13.22	15	18	12	41	47	39	1498	3995	1459
CD18	357	467	429	371	723	1197	138	1333	361
l-selectin	257	41	297	705	22	144	352	3	28

Human peripheral blood leucocytes were isolated and activated in vitro with PMA (100 ng/ml for 15 min at 37°) and fMLF (1 µm for 5 min at 37°). The cells were stained for mAb 13.22 antigen, l-selectin (DREG 56), and CD18 (TS1/18 or IB4; TS1/18 is shown, but both gave similar results) and analysed by flow cytometry as described in the Materials and Methods. Control cells received no activating agent. Mean fluorescence staining intensity (arbitrary units) on gated lymphocytes, monocytes and neutrophils is shown for each treatment group. In all experiments, background staining of secondary anti-IgG and anti-IgM controls was minimal. The data from one of three independent experiments is shown. Note that all experiments gave similar results.

The results described above suggested that the mAb 13.22 neoepitope might be expressed on lymphocytes in sites of inflammation in vivo. To evaluate this possibility, we analysed lymphocytes present in synovial fluid aspirated from inflamed joints of patients with rheumatoid arthritis. We found that 18·1 ± 4·7% [mean ± SD; n = 5 (range 10·4–25·1%)] of the synovial fluid lymphocytes did indeed express the mAb 13.22 neoepitope (representative dot plots are shown in Fig. 8) and that these cells were primarily T cells and NK cells (as demonstrated by two-colour flow cytometric analysis, not shown). These results confirm that the mAb 13.22 neoepitope is a physiologically relevant marker, which is exposed on distinct subpopulations of inflammatory lymphocytes. Further studies are now in progress to evaluate the nature of this subset and determine if it varies with different types of arthritis.

Figure 8.

Expression of the mAb 13.22 antigen on synovial fluid leucocytes. Human synovial fluid leucocytes were isolated and stained with mAb 13.22, followed by a FITC-conjugated second-stage antibody. To control for background staining, cells were also incubated with the FITC-conjugated second-stage antibody alone (a, c). Expression of the mAb 13.22 antigen on gated lymphocytes (identified by their distinctive forward and side light scatter profile) is shown for two representative samples (b, d). The upper level of background staining seen with negative controls is indicated by the markers, and the percentage of gated cells staining above background is noted. Representative of samples collected from five individual patients.

Discussion

Leucocytes modulate the expression of cell surface antigens during specific developmental and functional stages. For example, activation of T lymphocytes results in the up-regulation of cell surface Pgp-1 or CD44,^31,51 Ta_1,⁵² VLA-4,⁵³ IL-2 receptor,^21,22 and in the shedding of l-selectin.³⁶ In addition, modification of the carbohydrate expressed on CD45 (via alternative splicing of exons 4–6) results in the expression of a low-molecular weight isoform (CD45RO), and the conversion from CD45RA (high-molecular weight isoform) to CD45RO coincides with the transformation of a naïve to a memory phenotype.^54,55 Activation also induces modulation of cell surface antigens on myeloid cells. For example, activation of neutrophils has been reported to induce changes in the surface expression of a number of antigens, including integrin β₂, alkaline phosphatase, chemotactic receptors, flavocytochrome b, l-selectin, CD53, etc. [e.g. see refs ^{14,15,56–58}]. In addition, mononuclear phagocytes recruited to sites of inflammation have been shown to express a pro-inflammatory (P) phenotype, which is characterized by an up-regulated expression of integrins β₁ and β₂.⁵⁹

In the present studies, we characterize a novel carbohydrate neoepitope that is exposed on a subset of T lymphocytes after neuraminidase treatment or by activation in vitro with mitogens. This carbohydrate epitope was also up-regulated ∼18-fold on monocytes after neuraminidase treatment. Thus, the expression of a carbohydrate neoepitope on mononuclear leucocytes under these conditions suggests the exciting possibility that the mAb 13.22 epitope might be a novel marker for cells exposed to inflammatory conditions in vivo. Indeed, our analysis showing expression of the mAb 13.22 neoepitope on a subset of lymphocytes in synovial fluid from patients with rheumatoid arthritis supports this conclusion and demonstrates exposure of this epitope in a physiological setting.

The differential glycosylation of cell-surface molecules appears to play an important role in leucocyte function, and there is strong evidence that adhesiveness is correlated with cell-surface sialylation [reviewed in ref. 6]. For example, cell surface sialic acid has been shown to play a role in lymphocyte homing, and neuraminidase-treated cells exhibit abnormal migration patterns.^60,61 Sialic acid content and sialyltransferase activity are both lower in immature thymocytes compared to mature resting splenic T cells,⁶² suggesting that carbohydrate decoration could play a role in lymphocyte maturation. In addition, differential sialylation of cell surface glycoproteins has been reported to coincide with T-cell activation.²³ Studies on tumour cells have shown that the metastatic potential of tumour cells is altered by hyposialylation, which leads to greater adhesiveness and a decreased ability to migrate.⁶³ Stamatos et al.³⁰ recently found that desialylation of peripheral blood mononuclear cells promoted the growth of HIV-1, indicating that the removal of sialic acid from mononuclear cell surface carbohydrates resulted in enhanced binding or interaction of the virus with these cells. Landolfi and Cook²³ reported that class I molecules were hyposialylated on T lymphocytes compared to other lymphocyte populations. Interestingly, this group also found that activation of T cells resulted in an increase in endogenous neuraminidase activity,²⁵ which may participate in removing sialic acid moieties from cell surface glycoproteins, such as integrin β₂, during lymphocyte activation. The importance of sialic acid content in immune responses is further confirmed by studies showing that specific immune responses were restored by decreasing sialic acid content of surface molecules on antigen-presenting cells²⁴ and cytotoxic T lymphocytes.⁶⁴ Furthermore, neuraminidase treatment of macrophages has been shown to block the response to migration inhibitory factor/macrophage activating factor (MIF/MAF),⁶⁵ resulting in a decreased cytotoxic response.⁶⁶ Finally, neuraminidase treatment of leucocytes has been shown to promote neutrophil, lymphocyte and monocyte infiltration into the kidney,⁶⁷ and studies by Marín et al.⁶⁸ suggest that leucocyte desialylation due to elevated serum neuraminidase activity contributes to leucocyte infiltration during acute post-streptococcal glomerulonephritis.

The roles of leucocyte carbohydrate decorations are still not completely defined. It has been shown previously that CD11/CD18 polypeptides are N-glycosylated but not O-glycosylated [reviewed in ref. 45], and analysis of the oligosaccharides from leucocyte integrin β₂ indicates that there is an enrichment of high-mannose-type carbohydrate structures and a mixture of complex oligosaccharides.⁶⁹ Takeda⁷⁰ reported that LFA-1 (CD11a/CD18) exhibited lineage-dependent variability in sialylation patterns on T and B lymphocytes; however, the function of this variability has not been determined. Gahmberg and Tolvanen⁷¹ proposed that a major function of the peripheral portions of leucocyte cell-surface oligosaccharides was to act as ligands. For example, previous studies by Gbarah et al.⁷² have shown that integrin high-mannose-type oligosaccharides act as major ligands for type I fimbriated Escherichia coli. Furthermore, the ligands for E-, L-, and P-selectin molecules have all been found to contain sialyl Lewis-X structures [reviewed in ref. 73]. The importance of cell-surface sialylation in leucocyte function is further demonstrated by a genetic abnormality known as carbohydrate-deficient glycoprotein syndrome IA (CDGS IA).⁷⁴ Lymphocytes from these individuals have a lower level of sialylated glycans, which is suggested to impair their host defence capabilities.⁷⁴ On the other hand, sialidase activity is also essential for correct leucocyte function in host defence, and Chen et al.⁷⁵ found that IL-4 gene transcription and protein production by splenic T cells and purified IL-4-primed T cells are dependent on the enzymatic activity of T-cell-associated neu-1 sialidase. In addition, Yamamoto and Kumashiro⁷⁶ found that the T-cell neu-1 sialidase was required for conversion of vitamin D₃-binding protein to macrophage-activating factor, a potent inflammatory modulator. Neutrophils also express a mobilizable intracellular store of sialidase that plays a role in stimulated adhesion responses of these cells.⁷⁷ Furthermore, Katoh et al.⁷⁸ recently reported that the hyaluronic acid-binding ability of CD44-expressing monocytes was regulated by an inducible sialidase activity in these cells. Thus, it is clear that modulation of the level of leucocyte cell-surface sialylation by endogenous and/or exogenous sialidase activity does play an important role in modulating adhesion-dependent host defence functions in these cells. In addition, the studies reported here suggest that sialidase-mediated modification of CD11b and CD18 carbohydrate epitopes on leucocytes may play a role in modulating the local inflammatory response, such as in the inflamed synovium.

The functional role of integrin carbohydrates in mediating adhesion raises the interesting question of how modifications in these oligosaccharide structures might affect this adhesive function. Recently, Stamatos et al.³⁰ found that activation of CD4⁺ helper T cells resulted in desialylation of lymphocyte proteins and promoted a permissive state of growth for HIV-1 in vitro. Two explanations proposed to account for this effect were that non-specific removal of repulsive negative charge from the cell surface promoted virus–cell interactions, resulting in a higher level of infection, or that desialylation of specific HIV-1 targets on the cell surface might facilitate viral attachment and/or entry. Thus, in a similar manner, it is possible that desialylation of CD11b/CD18 and other leucocyte glycoproteins in the microenvironment of the inflammatory site, resulting in exposure of the mAb 13.22 antigen, might be important for mediating enhanced adhesive interactions with an invading pathogen. This possibility is supported by a recent study showing that desialylation of T lymphocytes results in an increased ability of these cells to bind pokeweed mitogen, thereby overcoming the monocyte requirement for pokeweed mitogen to induce T-cell mitogenesis.⁷⁹ Furthermore, Galvan et al.⁸⁰ found that virus-specific memory CD8⁺ T cells were hyposialylated and could be distinguished from naïve CD8⁺ T cells based on peanut agglutinin binding, which is increased on activated T cells. Finally, Soler et al.⁸¹ recently reported that leucosialin (CD43) was hyposialylated in monocyte THP-1 cells stimulated by interferon-γ and tumor necrosis factor-α and that this correlated with increased heterotypic adhesion. The conclusion of these authors was that the CD43 anti-adhesive effect was related to the level of sialylation.⁸¹ Whether exposure of the mAb 13.22 antigen plays a similar role in modulating adhesion is not know, and further studies are in progress to determine its physiological role in vivo.