Poly-N-acetyllactosaminyl saccharide chains of band 3 as determinants for anti-band 3 autoantibody binding to senescent and oxidized erythrocytes

Natural IgG antibodies to band 3 glycoprotein of erythrocyte membrane (anti-band 3 IgG) are known to bind to senescent erythrocytes, and believed to initiate antibody-dependent phagocytic removal of the senescent cells from blood circulation. Anti-band 3 antibodies also bind to the erythrocytes with hereditary hemoglobin abnormalities such as sickle, beta-thalassemic and hemoglobin K?ln erythrocytes. Oxidative stress appears to be responsible for the generation of senescent cell antigen on erythrocytes, to which anti-band 3 antibodies bind, since erythrocytes oxidized in vitro bind anti-band 3 IgG, and various oxidative modifications are observed in senescent erythrocytes as well as the erythrocytes with abnormal hemoglobin. Major antigenic sites of band 3 for anti-band 3 IgG autoantibodies are its sialylated poly-N-acetyllactosaminyl saccharide chains. The poly-N-acetyllactosaminyl saccharide chains on senescent erythrocytes are indeed involved in the antigenic sites on senescent erythrocytes. The finding of the carbohydrate epitopes and possible involvement of oxidative mechanism are compatible with the band 3 clustering hypothesis, in which clustering of band 3 molecules in erythrocyte membrane is supposed to be responsible for effective binding of anti-band 3 IgG to the cell surface, because the carbohydrate epitopes of band 3 can form multivalent epitopes on cell surface when band 3 molecules cluster, and oxidative stress can induce such clusters. Interestingly, poly-N-acetyllactosaminyl chains of band 3 on oxidized erythrocytes are also recognized by macrophages directly. Thus, poly-N-acetyllactosaminyl chains may play dual roles as determinants for recognition by anti-band 3 IgG and by macrophages.     

Role of the carboxyl-terminal lectin domain in self-association of galectin-3

Galectin-3 is a member of a large family of beta-galactoside-binding animal lectins and is composed of a carboxyl-terminal lectin domain connected to an amino-terminal nonlectin part. Previous experimental results suggest that, when bound to multivalent glycoconjugates, galectin-3 self-associates through intermolecular interactions involving the amino-terminal domain. In this study, we obtained evidence suggesting that the protein self-associates in the absence of its saccharide ligands, in a manner that is dependent on the carboxyl-terminal domain. This mode of self-association is inhibitable by the lectin's saccharide ligands. Specifically, recombinant human galectin-3 was found to bind to galectin-3C (the carboxyl-terminal domain fragment) conjugated to Sepharose 4B and the binding was inhibitable by lactose. In addition, biotinylated galectin-3 bound to galectin-3 immobilized on plastic surfaces and the binding could also be inhibited by various saccharide ligands of the lectin. A mutant with a tryptophan to leucine replacement in the carboxyl-terminal domain, which exhibited diminished carbohydrate-binding activity, did not bind to galectin-3C-Sepharose 4B. Furthermore, galectin-3C formed covalent homodimers when it was treated with a chemical cross-linker and the dimer formation was completely inhibited by lactose. Therefore, galectin-3 can self-associate through intermolecular interactions involving both the amino- and the carboxyl-terminal domains and the relative contribution of each depends on whether the lectin is bound to its saccharide ligands.     

Purification to apparent homogeneity and properties of pig kidney L-fucose kinase

L-Fucokinase was purified to apparent homogeneity from pig kidney cytosol. The molecular mass of the enzyme on a gel filtration column was 440 kDa, whereas on SDS gels a single protein band of 110 kDa was observed. This 110-kDa protein was labeled in a concentration-dependent manner by azido-[32P]ATP, and labeling was inhibited by cold ATP. The 110-kDa protein was subjected to endo-Lys-C digestion, and several peptides were sequenced. These showed very little similarity to other known protein sequences. The enzyme phosphorylated L-fucose using ATP to form beta-L-fucose-1-P. Of many sugars tested, the only other sugar phosphorylated by the purified enzyme was D-arabinose, at about 10% the rate of L-fucose. Many of the properties of the enzyme were determined and are described in this paper. This enzyme is part of a salvage pathway for reutilization of L-fucose and is also a valuable biochemical tool to prepare activated L-fucose derivatives for fucosylation reactions.     

The scavenger receptor serves as a route for internalization of lysophosphatidylcholine in oxidized low density lipoprotein-induced macrophage proliferation

We have recently demonstrated that the growth of murine macrophages is induced by oxidized low density lipoprotein (Ox-LDL) and that lysophosphatidylcholine (lyso-PC), a major phospholipid component of Ox-LDL, plays an essential role in its mitogenic effect. The present study was undertaken to further characterize the role of the macrophage scavenger receptor (MSR) in Ox-LDL-induced macrophage growth. The growth-stimulating effect of Ox-LDL on murine resident peritoneal macrophages was inhibited by maleylated bovine serum albumin (maleyl-BSA), a non-lipoprotein ligand for MSR but a poor carrier of lyso-PC, while maleyl-BSA itself failed to induce macrophage growth even in the presence of lyso-PC. Moreover, it competitively inhibited the endocytic uptake of 125I-Ox-LDL and the specific uptake of lyso-PC by MSR, whereas nonspecific lyso-PC transfer to cells was not affected. Furthermore, the Ox-LDL-induced cell growth of peritoneal macrophages obtained from MSR knockout mice was significantly weaker than that of macrophages obtained from their wild-type littermates. Our results suggest that the MSR is an important and efficient internalization pathway for lyso-PC in Ox-LDL-induced macrophage growth.     

Antibiotic properties of bovine lactoferrin on Helicobacter pylori

To investigate a potential new treatment for gastric Helicobacter pylori infection, we have examined the use of the natural antibiotic lactoferrin, found in bovine milk, for activity against Helicobacter species both in vitro and in vivo. Lactoferrin was bacteriostatic to H. pylori when cultured at concentrations > or =0.5 mg/ml. Growth of H. pylori was not inhibited by another milk constituent, lysozyme, or by a metabolite of lactoferrin, lactoferricin B, but growth was inhibited by the iron chelator deferoxamine mesylate. Lactoferrin inhibition of growth could be reversed by addition of excess iron to the medium. Lactoferrin in retail dairy milk was found to be more stable intragastrically than unbuffered, purified lactoferrin. Treatment of H. felis-infected mice with lactoferrin partially reversed mucosal disease manifestations. It is concluded that bovine lactoferrin has significant antimicrobial activity against Helicobacter species in vitro and in vivo. Bovine lactoferrin should be further investigated for possible use in H. pylori infections in man.     

CD36 is required for phagocytosis of apoptotic cells by human macrophages that use either a phosphatidylserine receptor or the vitronectin receptor (alpha v beta 3)

In vivo, apoptotic cells are efficiently removed by professional or nonprofessional phagocytes, a process thought to be essential for tissue remodeling and resolution of inflammation. Macrophages recognize apoptotic cells by several mechanisms, including recognition of exposed phosphatidylserine (PS); however, PS recognition on apoptotic cells has not been identified as a feature of human macrophages. The purpose of this study was to determine whether human monocyte-derived macrophages could be stimulated to recognize PS, defined as inhibition of phagocytosis by PS-containing liposomes. We also assessed the potential roles for scavenger receptors, CD14, and lectins. Uptake of apoptotic neutrophils into unstimulated macrophages was blocked about 50% by Arg-Gly-Asp-Ser and anti-alpha(v), and up to 20% by oxidized low density lipoprotein and N-acetylglucosamine, implying a major role for integrin and minor roles for scavenger and lectin receptors. Uptake into macrophages stimulated with beta-1,3-glucan was blocked 50% by PS liposomes and 40% by oxidized low density lipoprotein, suggesting that the macrophages had switched from using integrin to recognition of PS. MEM-18 and 61D3 (anti-CD14 mAbs) were poor inhibitors of apoptotic neutrophil uptake, but good inhibitors of apoptotic lymphocyte uptake. The switch to PS recognition was accompanied by down-regulation of alpha(v)beta3 expression and function. Anti-CD36 blocked uptake into unstimulated or stimulated macrophages, suggesting CD36 involvement not only with the alpha(v)beta3 integrin mechanism (as previously reported) but also with PS recognition. A maximum of 70% inhibition was achieved by combining anti-CD36 with either anti-a(v) or PS liposomes.     

Physical determinants of intermembrane protein transfer

Intermembrane protein transfer between erythrocytes and phospholipid vesicles was examined under a variety of conditions to investigate physical factors governing this process. Human erythrocytes were incubated with sonicated dimyristoylphosphatidylcholine vesicles containing trace [14C]dipalmitoylphosphatidylcholine. Protein-vesicle complexes were separated from cells and from membrane fragments by density gradient centrifugation. The yield of isolated protein vesicles was determined from the 14C-vesicle marker; protein compositions were analyzed by SDS-polyacrylamide gel electrophoresis. Enzymatic removal of portions of the cytoplasmic or exoplasmic domains of cell membrane proteins had little effect on the extent of protein transfer. Membrane additives such as cholate produced a 2-fold increase in protein-vesicle yield. The selectivity of protein transfer from erythrocytes was influenced by the lipid composition of recipient vesicles: inclusion of cholesterol increased band 3 content while the presence of anionic phospholipids reduced transfer. Proteins transferred from 32P-labeled cells differed in specific radioactivity from bulk cell proteins: glycophorin, highly phosphorylated in the cell membrane, showed no detectable labeling in the corresponding protein-vesicle band. These observations suggest that cell-to-vesicle protein transfer is insensitive to bulk steric and electrostatic properties of cell membranes, but enhanced by membrane defects. Recipient membrane composition influences the selectivity of transferred proteins and may reveal subtle differences in the membrane association of protein subpopulations.     

L-fucose residues on cellulose-based dialysis membranes: quantification of membrane-associated L-fucose and analysis of specific lectin binding

Contact of mononuclear human leukocytes with cellulose dialysis membranes may result in complement-independent cell activation, i.e. enhanced synthesis of cytokines, prostaglandins and an increase in beta 2-micro-globulin synthesis. Cellular contact activation is specifically inhibited by the monosaccharide L-fucose suggesting that dialysis membrane associated L-fucose residues are involved in leukocyte activation. In this study we have detected and quantitated L-fucose on commercially-available cellulose dialysis membranes using two approaches. A sensitive enzymatic fluorescence assay detected L-fucose after acid hydrolysis of flat sheet membranes. Values ranged from 79.3 +/- 3.6 to 90.2 +/- 5.0 pmol cm-2 for Hemophan or Cuprophan respectively. Enzymatic cleavage of terminal alpha-L-fucopyranoses with alpha-L-fucosidase yielded 7.7 +/- 3.3 pmol L-fucose per cm2 for Cuprophan. Enzymatic hydrolysis of the synthetic polymer membranes AN-69 and PC-PE did not yield detectable amounts of L-fucose. In a second approach, binding of the fucose specific lectins of Lotus tetragonolobus and Ulex europaeus (UEAI) demonstrated the presence of biologically accessible L-fucose on the surface of cellulose membranes. Specific binding was observed with Cuprophan, and up to 2.6 +/- 0.3 pmol L-fucose per cm2 was calculated to be present from Langmuir-type adsorption isotherms. The data presented are in line with the hypothesis that surface-associated L-fucose residues on cellulose dialysis membranes participate in leukocyte contact activation.     

Sulfhydryl oxidation of mutants with cysteine in place of acidic residues in the lactose permease

To examine further the role of charge-pair interactions in the structure and function of lactose permease, Asp237 (helix VII), Asp240 (helix VII), Glu126 (cytoplasmic loop IV/V), Glu269 (helix VIII), and Glu325 (helix X) were replaced individually with Cys in a functional mutant devoid of Cys residues. Each mutant was then oxidized with H2O2 in order to generate a sulfinic and/or sulfonic acid at these positions. Due to the isosteric relationship between aspartate and sulfinate, in particular, and the lower pKa of the sulfinic and sulfonic acid side chains, oxidized derivatives of Cys are useful probes for examining the role of carboxylates. Asp237-->Cys or Asp240-->Cys permease is inactive, as shown previously, but H2O2 oxidation restores activity to an extent similar to that observed when a negative charge is reintroduced by other means. Glu126-->Cys, Glu269-->Cys, or Glu325-->Cys permease is inactive, but oxidation does not restore active lactose transport. The data are consistent with previous observations indicating that Asp237 and Asp240 are not critical for active lactose transport, while Glu126, Glu269, and Glu325 are irreplaceable. Although Glu269-->Cys permease does not transport lactose, the oxidized mutant exhibits significant transport of beta,D-galactosylpyranosyl 1-thio-beta,D-galactopyranoside, a property observed with Glu269-->Asp permease. The observation supports the idea that an acidic residue at position 269 is important for substrate recognition. Finally, oxidized Glu325-->Cys permease catalyzes equilibrium exchange with an apparent pKa of about 6.5, more than a pH unit lower than that observed with Glu325-->Asp permease, thereby providing strong confirmatory evidence that a negative charge at position 325 determines the rate of translocation of the ternary complex between the permease, substrate, and H+.     

Evidence for a role for galectin-1 in pre-mRNA splicing

Galectins are a family of beta-galactoside-binding proteins that contain characteristic amino acid sequences in the carbohydrate recognition domain (CRD) of the polypeptide. The polypeptide of galectin-1 contains a single domain, the CRD. The polypeptide of galectin-3 has two domains, a carboxyl-terminal CRD fused onto a proline- and glycine-rich amino-terminal domain. In previous studies, we showed that galectin-3 is a required factor in the splicing of nuclear pre-mRNA, assayed in a cell-free system. We now document that (i) nuclear extracts derived from HeLa cells contain both galectins-1 and -3; (ii) depletion of both galectins from the nuclear extract either by lactose affinity adsorption or by double-antibody adsorption results in a concomitant loss of splicing activity; (iii) depletion of either galectin-1 or galectin-3 by specific antibody adsorption fails to remove all of the splicing activity, and the residual splicing activity is still saccharide inhibitable; (iv) either galectin-1 or galectin-3 alone is sufficient to reconstitute, at least partially, the splicing activity of nuclear extracts depleted of both galectins; and (v) although the carbohydrate recognition domain of galectin-3 (or galectin-1) is sufficient to restore splicing activity to a galectin-depleted nuclear extract, the concentration required for reconstitution is greater than that of the full-length galectin-3 polypeptide. Consistent with these functional results, double-immunofluorescence analyses show that within the nucleus, galectin-3 colocalizes with the speckled structures observed with splicing factor SC35. Similar results are also obtained with galectin-1, although in this case, there are areas of galectin-1 devoid of SC35 and vice versa. Thus, nuclear galectins exhibit functional redundancy in their splicing activity and partition, at least partially, in the nucleoplasm with another known splicing factor.     

Complete deficiency of glycophorin A in red blood cells from mice with targeted inactivation of the band 3 (AE1) gene

Glycophorin A is the major transmembrane sialoglycoprotein of red blood cells. It has been shown to contribute to the expression of the MN and Wright blood group antigens, to act as a receptor for the malaria parasite Plasmodium falciparum and Sendai virus, and along with the anion transporter, band 3, may contribute to the mechanical properties of the red blood cell membrane. Several lines of evidence suggest a close interaction between glycophorin A and band 3 during their biosynthesis. Recently, we have generated mice where the band 3 expression was completely eliminated by selective inactivation of the AE1 anion exchanger gene, thus allowing us to study the effect of band 3 on the expression of red blood cell membrane proteins. In this report, we show that the band 3 -/- red blood cells contain protein 4.1, adducin, dematin, p55, and glycophorin C. In contrast, the band 3 -/- red blood cells are completely devoid of glycophorin A (GPA), as assessed by Western blot and immunocytochemistry techniques, whereas the polymerase chain reaction (PCR) confirmed the presence of GPA mRNA. Pulse-label and pulse-chase experiments show that GPA is not incorporated in the membrane and is rapidly degraded in the cytoplasm. Based on these findings and other published evidence, we propose that band 3 plays a chaperone-like role, which is necessary for the recruitment of GPA to the red blood cell plasma membrane.     

Inhibition of tumor necrosis factor and subsequent endotoxin shock by pirfenidone

Tumor necrosis factor-alpha (TNF) is an extremely potent cytokine which is involved in the pathogenesis of a number of diseases. Interruption of its synthesis can result in a reduction of inflammation and subsequent pathology. A new experimental drug pirfenidone (5-methyl-L-phenyl-2-(1H)-pyridone, trade name: Deskar) has been reported to have beneficial effects for the treatment of certain fibrotic diseases. The present study describes the inhibition of TNF in vitro as well as the inhibition of circulating TNF in vivo by pirfenidone. Isolated, thioglycollate-induced peritoneal macrophages (Mphi) from C57BL/6 mice were exposed to either lipopolysaccharide (LPS) or mannosylated bovine serum albumin then incubated with 0.1-0.9 mg/ml of pirfenidone. This substance inhibited the production of TNF in a dose-dependent manner as measured by ELISA. One i.p. injection of either 100 or 200 mg/kg pirfenidone inhibited the induction of circulating TNF following a single i.v. injection of LPS. Endotoxin shock was induced in mice using an i.p. injection of galactosamine and LPS. The higher dose of pirfenidone (200 mg/kg) completely inhibited shock and subsequent mortality. Lower doses of pirfenidone or administration either prior to or post challenge only partially inhibited symptoms. These results indicate that pirfenidone is able to inhibit both TNF induction and subsequent endotoxin shock. Additional studies are warranted to establish this drug as a potential treatment for diseases where TNF plays a major role.     

Involvement of reactive oxygen intermediates in tumor necrosis factor alpha-dependent bacteriostasis of Mycobacterium avium

We studied the involvement of reactive oxygen intermediates and reactive nitrogen intermediates in the bacteriostasis of two Mycobacterium avium strains differing in virulence by resident peritoneal macrophages. We found that both the highly virulent strain (25291) and the low-virulence strain (1983) of M. avium induced superoxide production but inhibited nitrite production in vitro. This inhibition was due to the production of superoxide, a nitric oxide scavenger. The stimulation of superoxide production was two- to fivefold higher in strain 1983-infected than in strain 25291-infected resident peritoneal macrophages and was independent of contaminating T cells or NK cells. Superoxide secretion was dependent on the tumor necrosis factor (TNF) produced endogenously by the macrophages. This was also true when macrophages were isolated from infected mice. Addition of TNF to the infected resident peritoneal macrophages caused only a slight, albeit significant, increase in superoxide production by strain 25291-infected macrophages. Incubation of resident peritoneal macrophages with different scavengers of reactive oxygen intermediates showed that strain 1983 was susceptible to hydrogen peroxide produced by resident peritoneal macrophages. Strain 25291 was shown to decrease superoxide secretion inside heavily infected bone marrow-derived macrophages. This strain was also shown to be a better trigger for production of reactive oxygen intermediates than strain 1983. In summary, strain 1983 induced high levels of TNF synthesis that acted in an autocrine fashion to stimulate production of reactive oxygen intermediates by macrophages leading to growth restriction mediated by hydrogen peroxide. The highly virulent strain 25291 induced low levels of TNF synthesis, and therefore little reactive oxygen intermediate production, and could also inhibit superoxide production by the infected macrophages.     

MRL/lpr and MRL+/+ macrophage DNA synthesis in the absence and the presence of colony-stimulating factor-1 and granulocyte-macrophage colony-stimulating factor

Macrophage accumulation and proliferation as well as altered macrophage properties have been observed in autoimmune MRL mice. To determine whether there might be innate differences in the proliferative responses, we examined the DNA synthesis responses of peritoneal macrophages and macrophages derived in vitro from bone marrow precursors (bone marrow-derived macrophages (BMM)). Murine peritoneal exudate macrophages normally require the addition of macrophage CSF (CSF-1) to enter cell cycle in vitro. In contrast, we have found that many thioglycollate-induced adherent peritoneal macrophages, but not resident peritoneal macrophages, from both MRL/lpr and MRL+/+ mice atypically underwent DNA synthesis even in the absence of added CSF-1. They also responded very well to granulocyte-macrophage CSF. These findings may help to explain the appearance of increased macrophage numbers in MRL lesions. In contrast to a previous report, it was found that MRL/lpr and MRL+/+ BMM did not have an enhanced response to CSF-1 and that modulation of CSF-1 receptor expression was not more rapid in MRL BMM. We also found no evidence for abnormal CSF-1 internalization and degradation or for the lpr mutation to have any enhanced effect on BMM survival in the absence of CSF-1. TNF-alpha lowered the DNA synthesis response to CSF-1 of MRL/lpr BMM rather than enhanced it, as has been reported. Our data suggest that the enhanced accumulation of macrophages in the MRL/lpr kidney cannot be explained by a proposed model of enhanced responsiveness of MRL/lpr BMM to CSF-1, including a contribution by TNF-alpha.     

Enzymatic and chemical footprinting of anthracycline antitumor antibiotics and related saccharide side chains

DNase I and three DNA chemical footprinting agents were used to compare the DNA binding properties of the anthracycline antitumor antibiotics daunomycin, aclacinomycin A, and ditrisarubicin B. These anthracyclines contain a tetracyclic chromophore which intercalates into DNA and a monosaccharide, trisaccharide, and two trisaccharide side chains, respectively. These side chains consist of between one and three 2,6-dideoxy, 1,4-diaxially linked sugars. Three chemical probes, fotemustine, dimethyl sulfate, 4-(2'-bromoethyl)phenol, and the enzymic probe DNase I were used in the footprinting experiments. The chemical probes provided a clear picture of the binding pattern at 37 degrees C and more detailed information than that obtained using the standard DNase I footprinting assay. All three anthracyclines showed preferred binding to 5'-GT-3' sequences in both the chemical and enzymatic footprinting. DNase I footprinting showed that the number of base pairs of DNA protected from cleavage increased with the number of saccharide groups present at particular sites and is consistent with DNA binding of the saccharide side chains. Alkylation of runs of guanine by fotemustine was inhibited by all three anthracyclines, while alkylation by dimethyl sulfate was enhanced for most guanines. The probe 4-(2'-bromoethyl)phenol showed that all three anthracyclines completely protected all of the adenines in the minor groove from alkylation, and enhanced major groove guanine alkylation was observed with aclacinomycin A, daunomycin, and, to a much lesser extent, ditrisarubicin B. These results are consistent with intercalation of the aglycone ring and binding of the rigid, hydrophobic saccharide side chains in the minor groove. Footprinting of four methyl glycosides related to the anthracyclines showed no evidence of DNA binding with any of the agents studied.     

Oxidative damage and erythrocyte membrane transport abnormalities in thalassemias

Oxidative damage induced by free globin chains has been implicated in the pathogenesis of the membrane abnormalities observed in alpha and beta thalassemia. We have evaluated transport of Na+ and K+ in erythrocytes of patients with thalassemias as well as in two experimental models that use normal human red blood cells, one for alpha thalassemia (methylhydrazine treatment, alpha thalassemia like) and one for beta thalassemia (phenylhydrazine treatment, beta thalassemia like). With the exception of the Na-K pump, similar alterations in membrane transport were observed in thalassemia and thalassemia-like erythrocytes. These were: increased K-Cl cotransport, Na-Li countertransport and reduced Na-K-Cl cotransport. The Na-K pump was reduced in thalassemia-like cells, whereas it was increased in severe alpha thalassemia and in beta thalassemia cells. The increased K-Cl cotransport activity could be observed in light and dense fractions of beta-thalassemic cells. K-Cl cotransport in thalassemic and thalassemia-like erythrocytes was partially inhibited by [(dihydro-indenyl) oxy] alkanoic acid and completely abolished by dithiothreitol. Thus, oxidative damage represents an important factor in the increased activity of the K-Cl cotransport observed in thalassemias, and of the K+ loss observed in beta-thalassemia erythrocytes.     

Absence of the major sialoglycoprotein in the membrane of human En(a--) erythrocytes and increased glycosylation of band 3

The human En(a-) blood group is a rare recessive trait. These erythrocytes lack the major membrane sialoglycoprotein (PAS1, MN protein, or glycophorin) and contain a decreased amount of PAS2 as demonstrated by radiolabeling of surface porteins and chemical techniques. A third glycoprotein, Band 3, contains two labeled oligosaccharide chains; the more complex oligosaccharide has a higher molecular weight in En(1-) cells than in normal cells. A fourth glycoprotein, PAS3, is present in usual amounts in En(a-) cells. Cells heterozygous for En(a) are intermediate in these respects. The sialic acid is decreased in En(a-) cells, but the total carbohydrate is similar in the different membranes. The glycolipids are present in normal amounts but are much more exposed to galactose oxidase in En(a-) cells than in normal cells.     

Brain cytochrome oxidase activity after kindled seizures: a quantitative histochemical mapping study

A Japanese male patient with myelodysplastic syndrome (MDS) was shown to have associated Tn syndrome; the first report of Tn syndrome with MDS. The Tn expression was demonstrated on erythrocytes, granulocytes, monocytes, platelets, and lymphocytes by flow cytometric analysis using a lectin and an antibody. Electrophoresis of erythrocyte membrane proteins revealed slower mobility of glycophorin B from the patient than that from normal individuals, suggesting a glycophorin B molecular abnormality.