Beta2 integrins target Rap GTPases to the plasma membrane by means of degranulation

We report the novel observation that engagement of β2 integrins on human neutrophils is accompanied by increased levels of the small GTPases Rap1 and Rap2 in a membrane-enriched fraction and a concomitant decrease of these proteins in a granule-enriched fraction. In parallel, we observed a similar time-dependent decrease of gelatinase B (a marker of specific and gelatinase B-containing granules) but not myeloperoxidase (a marker of azurophil granules) in the granule fraction, and release of lactoferrin (a marker of specific granules) in the extracellular medium. Furthermore, inhibition of Src tyrosine kinases, or phosphoinositide 3-kinase with PP1 or LY294002, respectively, blocked β2 integrin-induced degranulation and the redistribution of Rap1 and Rap2 to a membrane-enriched fraction. Consequently, the β2 integrin-dependent exocytosis of specific and gelatinase B-containing granules occurs via a Src tyrosine kinase/phosphoinositide 3-kinase signaling pathway and is responsible for the translocation of Rap1 and Rap2 to the plasma membrane in human neutrophils.     

Isoprenylation of rap2 proteins in platelets and human erythroleukemia cells

The covalent modification of proteins by isoprenoid derivatives of mevalonic acid was investigated in human platelets, cells that lack the ability to synthesize endogenous cholesterol, and human erythroleukemia (HEL) cells, cholesterol-producing cultured cells derived from megakaryocytes. When washed platelets or HEL cells were incubated with [3H]mevalonic acid, the radiolabel was incorporated into a distinct group of proteins with molecular masses between 21,000 and 28,000. We have identified one of these proteins as a ras-related rap2 protein based on its immunoreactivity with a polyclonal antiserum raised against purified recombinant rap2b. This anti-rap2 antiserum was used for two-dimensional immunoblotting analysis and immunoprecipitation of mevalonate-labeled rap2 from platelets and HEL cells. These results suggest that rap2 may undergo a series of carboxyl-terminal modifications similar to the p21ras proteins. In addition, it is shown that non-cholesterol-producing cells are capable of incorporating isoprenyl groups into specific proteins.     

Generation of specific antibodies against the rap1A, rap1B and rap2 small GTP-binding proteins. Analysis of rap and ras proteins in membranes from mammalian cells.

Specific antibodies against rap1A and rap1B small GTP-binding proteins were generated by immunization of rabbits with peptides derived from the C-terminus of the processed proteins. Immunoblot analysis of membranes from several mammalian cell lines and human thrombocytes with affinity-purified antibodies against rap1A or rap1B demonstrated the presence of multiple immunoreactive proteins in the 22-23 kDa range, although at strongly varying levels. Whereas both proteins were present in substantial amounts in membranes from myelocytic HL-60, K-562 and HEL cells, they were hardly detectable in membranes from lymphoma U-937 and S49.1 cyc- cells. Membranes from human thrombocytes and 3T3-Swiss Albino fibroblasts showed strong rap1B immunoreactivity, whereas rap1A protein was present in much lower amounts. In the cytosol of HL-60 cells, only small amounts of rap1A and rap1B proteins were detected, unless the cells were treated with lovastatin, an inhibitor of hydroxymethylglutaryl-coenzyme A reductase, suggesting that both proteins are isoprenylated. By comparison with recombinant proteins, the ratio of rap1A/ras proteins in membranes from HL-60 cells was estimated to be about 4:1. An antiserum directed against the C-terminus of rap2 reacted strongly with recombinant rap2, but not with membranes from tested mammalian cells. In conclusion, rap1A and rap1B proteins are distributed differentially among membranes from various mammalian cell types and are isoprenylated in HL-60 cells.     

Beta 2-microglobulin in neutrophils: an intragranular protein

The subcellular localization of beta2-microglobulin (beta 2m) in human neutrophils was determined by an enzyme-linked immunosorbent assay on subcellular fractions obtained by Percoll density gradient centrifugation of neutrophils disrupted by nitrogen cavitation. The neutrophils were found to contain 160 ng beta 2m/mg protein. Approximately two-thirds co-located with the markers for specific granules and was released from intact cells during degranulation, whereas one-third of the beta 2m was located together with markers of the plasma membrane. This fraction was not further enriched during degranulation. These results indicate that beta 2m cannot be universally used as a plasma membrane marker as hitherto assumed, but beta 2m may serve as an indicator of neutrophil degranulation.     

Subcellular localization of the b-cytochrome component of the human neutrophil microbicidal oxidase: translocation during activation

We describe a new method for subcellular fractionation of human neutrophils. Neutrophils were disrupted by nitrogen cavitation and the nuclei removed by centrifugation. The postnuclear supernatant was applied on top of a discontinuous Percoll density gradient. Centrifugation for 15 min at 48,000 g resulted in complete separation of plasma membranes, azurophil granules, and specific granules. As determined by ultrastructure and the distribution of biochemical markers of these organelles, approximately 90% of the b-cytochrome in unstimulated cells was recovered from the band containing the specific granules and was shown to be in or tightly associated with the membrane. During stimulation of intact neutrophils with phorbol myristate acetate or the ionophore A23187, we observed translocation of 40-75% of the b-cytochrome to the plasma membrane. The extent of this translocation closely paralleled release of the specific granule marker, vitamin B12-binding protein. These data indicate that the b-cytochrome is in the membrane of the specific granules of unstimulated neutrophils and that stimulus-induced fusion of these granules with the plasma membrane results in a translocation of the cytochrome. Our observations provide a basis for the assembly of the microbicidal oxidase of the human neutrophil.     

A novel type of cytoplasmic granule in bovine neutrophils

We obtained cell preparations containing greater than 95% neutrophils from freshly drawn bovine blood. The cells were suspended in sucrose and disrupted in a Dounce homogenizer, and the postnuclear supernate was fractionated by zonal differential sedimentation and by isopycnic equilibration. The subcellular fractions were characterized biochemically by testing for marker enzymes and other constituents known to occur in azurophil and specific granules of other species, and by electrophoretic analysis of extracts of the particulate material. In addition, each fraction was examined by random-sampling electron microscopy. We found that bovine neutrophils contain in addition to azurophil and specific granules a third type of granule, not known to occur in neutrophils of other species. These novel granules are larger, denser, and considerably more numerous than the two other types. Except for lactoferrin, they lack the characteristic constituents of azurophil granules (peroxidase, acid hydrolases, and neutral proteinases) and of specific granules (vitamin B12-binding protein). Instead, they contain a group of highly cationic proteins not found in the other granules, and they are the exclusive stores of powerful oxygen-independent bactericidal agents. We studied the fate of the large granules in bovine neutrophils exposed to opsonized particles, the ionophore A 23187, or phorbol myristate acetate. The appearance in the cell-free media of antibacterial activity and of the characteristic highly cationic proteins as revealed by electrophoresis was monitored and compared with the release of azurophil and specific granule markers. In addition, changes of the relative size of the large granule compartment induced by phagocytosis were assessed by morphometry. The results show that exocytosis of the large granules occurs following both phagocytosis and exposure to soluble stimuli. Like the specific granules, the large granules appear to be discharged by true secretion under conditions where the azurophil granules are fully retained.     

Kinetics of fusion of the cytoplasmic granules with phagocytic vacuoles in human polymorphonuclear leukocytes. Biochemical and morphological studies

This study on human neutrophils was conducted to measure the kinetics of degranulation of the different cytoplasmic granules into phagocytic vacuoles, and to relate the timing of these events to the burst of respiration that accompanies phagocytosis by these cells. Purified neutrophils were incubated with latex particles opsonized with human immunoglobulin (Ig)G, and phagocytosis was stopped at timed intervals. The cells were examined by electron microscopy to document the sequence of degranulation of the cytoplasmic granules. The azurophil granules and lyosomes were identified by histochemical staining for peroxidase and acid phosphatase, respectively. Phagocytic vacuoles were separated from cell homogenates by floatation on sucrose gradients and assayed for contained lactoferrin, myeloperoxidase, and acid hydrolases. The conclusions drawn from the biochemical and morphological studies were in agreement and indicated: particle uptake and vacuole closure can be completed within 20 s; both the specific and azurophil granules fuse with the phagocytic vacuole much earlier than is generally appreciated, with half-saturation times of 39 s (99% confidence limits, 15-72); oxygen consumption has kinetics similar to those of the fusion of these granules with the phagosome; degranulation of the acid hydrolases beta- glucuronidase, N-acetyl-beta-glucosaminidase (biochemical assays), and acid phosphatase (biochemical assay and electron microscopic cytochemistry) have kinetics of degranulation that are similar to each other but totally different from and much slower than that of myeloperoxidase with half-saturation times of between 354 and 682 s (99% confidence limits, 246-883). This suggests that the acid hydrolases are not co-located with myeloperoxidase in the azurophil granule but are contained in distinct lysosomes, or "tertiary granules".     

Different glycosphingolipid composition in human neutrophil subcellular compartments

The binding of a number of carbohydrate-recognizing ligands to glycosphingolipids and polyglycosylceramides of human neutrophil subcellular fractions (plasma membranes/secretory vesicles of resting and ionomycin-stimulated cells, specific and azurophil granules) was examined using the chromatogram binding assay. Several organelle-related differences in glycosphingolipid content were observed. The most prominent difference was a decreased content of the GM3 ganglioside in plasma membranes of activated neutrophils. Gangliosides recognized by anti-VIM-2 antibodies were detected mainly in the acid fractions of azurophil and specific granules. Slow-migrating gangliosides and polyglycosylceramides with Helicobacter pylori-binding activity were found in all acid fractions. A non-acid triglycosylceramide, recognized by Gal4Gal-binding Escherichia coli, was detected in the plasma membrane/secretory vesicles but not in the azurophil and specific granules.Although no defined roles of glycosphingolipids have yet been conclusively established with respect to neutrophil function, the fact that many of the identified glycosphingolipids are stored in granules, is in agreement with their role as receptor structures that are exposed on the neutrophil cell surface upon fusion of granules with the plasma membrane. Accordingly, we show that neutrophil granules store specific carbohydrate epitopes that are upregulated to the plasma membrane upon cell activation.     

Differential up-regulation of specific and azurophilic granule membrane markers in electropermeabilized neutrophils.

We have developed an alternative method to study the degranulation in electropermeabilized human neutrophils by measuring the up-regulation of the specific membrane markers CD63 (residing in the azurophilic granules of resting neutrophils) and CD67 (present in specific granules). The expression of these marker proteins was measured by the binding of specific antibodies to paraformaldehyde-fixed cells and subsequent flow cytometry. We first investigated whether the changes in CD63 and CD67 expression after stimulation of intact cells were comparable with earlier measurements of neutrophil degranulation, in which the release of soluble marker proteins was measured. These experiments indicated that this new method compares favourably with earlier studies, both with respect to kinetics and stimulus dependency. Subsequently, we applied this method (which does not include centrifugation of the cells) to study degranulation in electropermeabilized neutrophils. In permeabilized neutrophils, a clear up-regulation of the specific granule marker CD67 was observed upon incubation with a free Ca2+ concentration of 1 microM, a value of the cytosolic free Ca2+ concentration occurring in formylmethionyl-leucyl-phenylalanine (FMLP)-activated neutrophils. The azurophilic granule marker CD63 required GTP-gamma-S besides 1 microM Ca2+ for a significant up-regulation. Hence, our study indicates a different requirement for intracellular signals of the two main types of granules in human neutrophils.     

Stimuli which provoke secretion of azurophil enzymes from human neutrophils induce increments in adenosine cyclic 3'-5'-monophosphate

1. Stimuli for human neutrophils were divided into two classes on the basis of their ability to induce degranulation: complete secretagogues provoked release of both azurophil and specific granules, while incomplete secretagogues only induced release of specific granules. 2. Complete secretagogues, which possessed the ability to induce secretion of azurophil granules, also induced transient increments in total cellular cyclic AMP levels: incomplete secretagogues did not. 3. Complete secretagogues, unlike the incomplete variety, also induced further increments of cyclic AMP in prostaglandin E1-pretreated neutrophils. 4. Inhibition of lysosomal enzyme release by prostaglandin E1 was closely correlated with elevated levels of cyclic AMP induced by the prostaglandin alone, than with the much higher transient increment in cyclic AMP produced by stimulation of prostaglandin E1-treated cells. 5. Our results describe the first biochemical difference between neutrophil responses associated with secretion of azurophil granules, as opposed to specific granules: transient increments in cyclic AMP.     

Localization of chymotrypsin-like cationic protein, collagenase and elastase in azurophil granules of human neutrophilic polymorphonuclear leukocytes.

The subcellular localization of granulocyte collagenase, elastase and chymotrypsin-like cationic protein was determined using velocity centrifugation of cytoplasmic granules of human polymorphonuclear leukocytes. The proteases were assayed by immunochemical and enzymatic methods. Measurements of lactoferrin and myeloperoxidase distinguish exactly between constituents of specific and azurophil granules. Collagenase, elastase and chymotrypsin-like cationic proteins showed an almost identical sharp and unimodal distribution. They co-sedimented with myeloperoxidase demonstrating that these enzymes are localized exclusively in the azurophil granules.     

Human neutrophil elastase releases a ligand-binding fragment from the 75-kDa tumor necrosis factor (TNF) receptor. Comparison with the proteolytic activity responsible for shedding of TNF receptors from stimulated neutrophils.

To localize the protease(s) involved in shedding of tumor necrosis factor receptors (TNF-R) from activated neutrophils (PMN) (Porteu, F., and C. Nathan (1990) J. Exp. Med. 172, 599-607), we tested subcellular fractions from PMN for their ability to cause loss of TNF-R from intact cells. Exposure of PMN to sonicated azurophil granules at 37 degrees C resulted in inhibition of 125I-TNF binding; 50% inhibition ensued when PMN were treated for approximately 1 min with azurophil granules equivalent to 2-3 PMN per indicator cell. The TNF-R-degrading activity in azurophil granules were identified as elastase by its sensitivity to diisopropyl fluorophosphate (DFP), alpha 1-antitrypsin and N-methoxysuccinyl-Ala-Ala-Pro-Val chloromethyl ketone (MSAAPV-CK), and by the ability of purified elastase to reproduce the effect of azurophil granules. Elastase preferentially acted on the 75-kDa TNF-R, reducing by 85-96% the binding of 125I-TNF to mononuclear cells expressing predominantly this receptor, while having no effect on endothelial cells expressing almost exclusively the 55-kDa TNF-R. Elastase-treated PMN released a 32-kDa soluble fragment of p75 TNF-R that bound TNF and reacted with anti-TNF-R monoclonal antibodies. In contrast, fMet-Leu-Phe-activated PMN shed a 42-kDa fragment from p75 TNF-R, along with similar amounts of a 28-kDa fragment from p55 TNF-R. Shedding of both TNF-Rs by intact activated PMN was more extensive than shedding caused by elastase and was completely resistant to DFP and MSAAPV-CK. Thus, the TNF-R-releasing activity of azurophil granules is distinct from that operative in intact stimulated PMN and could provide an additional mechanism for the control of cellular responses to TNF at sites of inflammation.     

Nonreceptor Protein-Tyrosine Kinases in Neutrophil Activation

Nonreceptor protein-tyrosine kinases are involved in the regulation of almost all neutrophil responses such as adhesion, chemotaxis, priming, oxidative burst, and degranulation. Here, we show that phagocytosis is also regulated by protein-tyrosine kinase activity. Using various protein-tyrosine kinase inhibitors, we further demonstrate that opsonized zymosan-induced degranulation of specific and azurophil granules is regulated by protein-tyrosine kinase activity, whereas phorbol ester-induced degranulation is not. Several of the nonreceptor protein-tyrosine kinases involving in neutrophil signal transduction are known, including Fgr, Hck, Lyn, Yes, and Syk. Among these, Hck and Fgr are localized on the azurophil and specific granules, suggesting the involvement of these two protein-tyrosine kinases in the regulation of degranulation. In this report, we characterize some of the molecular properties of Hck and Fgr. We discuss the methods generally used for the measurement of protein-tyrosine kinase activities in neutrophils highlighting precautions against proteolysis. In addition, we show that in subcellular fractions of retinoic acid-differentiated neutrophil-like NB4 cells, the 59- and 61-kDa forms of Hck are attached to the membranes of their respective compartments by different mechanisms. Finally, we discuss the functional roles of protein-tyrosine kinases in the regulation of neutrophil activation and speculate on the importance of their subcellular localization.     

Cytochrome b translocation to human neutrophil plasma membranes and superoxide release. Differential effects of N-formylmethionylleucylphenylalanine, phorbol myristate acetate, and A23187

The role of specific granules and cytochrome b in superoxide (O(2)) release was studied by comparing the effects of three different stimuli on normal human neutrophils, neutrophils congenitally deficient in specific granules, and granule-free normal neutrophil cytoplasts. Phorbol myristate acetate (PMA) stimulated normal neutrophils to release more O(2) than did N-formylmethionylleucylphenylalanine (fMet-Leu-Phe), which stimulated greater release than the calcium ionophore A23187. Neutrophils lacking specific granules produced variable amounts of O(2) in response to all stimuli. Stimulation with PMA, fMet-Leu-Phe, and A23187 produced maximal rates of O(2) release that were 32, 55, and 21% of that by normal cells. Likewise, granule-free neutrophil cytoplasts released 24, 20, and 0% of the O(2) released by intact cells. These data suggest that the stimuli require different mechanisms for activation. Three subcellular fractions (azurophil granule rich, specific granule rich, and plasma membrane rich) were separated by Percoll gradients from normal resting and stimulated neutrophils. In resting neutrophils, the cytochrome b content in the plasma membrane was 31% of the total, with the rest in the specific granule-rich fraction. Ten minutes after stimulation, PMA, fMet-Leu-Phe, and A23187 induced translocation of 27, 8, and 49%, respectively, of the cytochrome b from the specific granule-rich fraction to the plasma membrane. Although our data support a role for specific granule factors in A23187-induced O(2) release, there is no correlation between the amount of cytochrome b incorporated into the plasma membrane and the extent of O(2) production activated by the different stimuli.     

Lipid raft proteome of the human neutrophil azurophil granule

Detergent-resistant membrane domains (DRMs) are present in the membranes of azurophil granules in human neutrophils (Feuk-Lagerstedt et al., J. Leukoc. Biol. 2002, 72, 970). Using a proteomic approach, we have now identified 106 proteins in a DRM preparation from these granule membranes. Among these proteins were the lipid raft structural proteins flotillin-1 and -2, cytoskeletal proteins such as actin, vimentin and tubulin, and membrane fusion promoting proteins like annexins and dysferlin. Our results suggest that the azurophil granule membrane, in similarity to the plasma membrane, is an elaborate structure that takes part in intracellular signaling and functions other than the mere delivery of bactericidal effector molecules to the phagosome.     

Subcellular distribution and membrane association of human neutrophil substrates for ADP-ribosylation by pertussis toxin and cholera toxin

Neutrophil guanine nucleotide-binding proteins are important components of receptor-mediated cellular responses such as degranulation, chemotaxis, and superoxide production. Because the cytoplasmic granules of neutrophils serve as an intracellular store of receptors and NADPH oxidase components, we investigated the subcellular distribution of substrates for ADP-ribosylation by both pertussis and cholera toxins. Cholera toxin substrates of Mr 43 and 52 kDa were present only in the plasma membrane fraction. A 39-kDa pertussis toxin substrate was present in the plasma membrane, cytosol, and a specific granule-enriched fraction. There were no substrates for either toxin in the primary granules. Quantitative GTP-gamma-5 binding was localized predominantly to the plasma membrane fraction (47%), but significant portions were found in the specific granule-enriched fractions (13%) and cytosol (34%) as well. Two-dimensional gel electrophoresis and chymotryptic digests of the pertussis toxin substrate from these three subcellular fractions suggested that they are highly homologous. Triton X-114 phase partitioning was used to investigate the hydrophobicity of the toxin substrates. The pertussis toxin substrates in the plasma membrane and granule fractions behaved like integral membrane proteins, whereas the cytosolic substrate partitioned into both lipophilic and aqueous fractions. ADP-ribosylation converted the substrates to a somewhat less lipophilic form. These data suggest that the specific granules or an organelle of similar density serve as an intracellular store of a G protein with a 39-kDa alpha-subunit and that the cytosolic fraction of neutrophils contains free alpha-subunits of the same size.     

Specific granules of human eosinophils have lysosomal characteristics: presence of lysosome-associated membrane proteins and acidification upon cellular activation

Eosinophils possess characteristic specific granules. Their content may be important during host defense but it can also cause damage after release at sites of inflammation. We investigated possible lysosomal characteristics of these granules. Lysosome-associated membrane protein (LAMP)-1 and 2, were detected by Western blot, subcellular fractionation, and immunoelectron microscopy (IEM) and were localized to the membrane of specific granules and in vesicles of the cytoplasm, separate from secretory vesicles. No binding of mannose 6-phosphate receptor to proteins of specific granules could be detected, indicating that they are dephosphorylated and mature. Cellular activation by interleukin-5 caused acidification of specific granules, as detected by pH-dependent probes. The acidification was inhibited by concanamycin A (inhibitor of vacuolar H(+)-ATPase). Activation of eosinophils by serum-treated zymosan (STZ) caused degranulation into STZ-containing phagosomes and incorporation of LAMPs to their membranes. In conclusion, specific granules of eosinophils can be regarded as specialized primary lysosomes, a feature that may be important for their function and integrity.     

Osmotic forces are not critical for Ca2+-induced secretion from permeabilized human neutrophils

In order to examine the role of osmotic forces in degranulation, the effects of solutes and osmolality on granule secretion were explored using both FMLP-stimulated, intact neutrophils and Ca2+-stimulated, permeabilized cells. We employed a HEPES-based buffer system which was supplemented with: a) permeant (KCl or NaCl) or impermeant (Na-isethionate or choline-Cl) ions, or b) permeant (urea) or impermeant (sucrose) uncharged solutes. Intact and permeabilized cells had significantly different solute requirements for degranulation. FMLP-stimulated release from intact cells was supported by NaCl or Na-isethionate greater than KCl greater than choline-Cl or sucrose greater than urea. In contrast, the rank order of Ca2+-stimulated release from permeabilized cells was choline-Cl greater than Na-isethionate, KCl, or NaCl greater than sucrose greater than urea. Hypo-osmotic conditions caused increased levels of background granule release from both intact and permeabilized neutrophils. However, hypo-osmolality inhibited both FMLP-stimulated degranulation from intact cells and Ca2+-induced release from permeabilized neutrophils. While hyperosmotic conditions inhibited stimulated release from intact cells, this inhibition was much less pronounced in permeabilized cells when the granules were directly exposed to these solutions. In fact, hyperosmotic sucrose greatly enhanced Ca2+-induced secretion. Although isolated specific and azurophil granules showed some lytic tendencies in hypo-osmotic buffers, the overall stability of the isolated granules did not indicate that swelling alone could effect degranulation. These results suggest that degranulation in permeabilized cells is neither due to nor driven by simple osmotic forces (under resting or stimulated conditions) and emphasize differences obtained by bathing both the granules and plasma membrane (as opposed to membranes alone) in various solutes.     

Human beta-adrenergic receptors. Simultaneous purification of beta 1- and beta 2-adrenergic-receptor peptides.

Monoclonal antibodies to insulin secretory granule membranes were obtained following immunization of mice with granule membranes purified from a rat transplantable insulinoma. The specificities of the antibodies were investigated by using binding assays with different insulinoma subcellular fractions, by indirect immunofluorescence studies with intact and permeabilized cells, and by immunoblotting of granule membrane proteins fractionated by SDS/polyacrylamide-gel electrophoresis. Fifty-six antibodies were characterized initially, and 21 representative cell lines were cloned. The antibodies fell into four categories: (1) binding preferentially to secretory granules, and reacting with a component of approx. 80,000 Da on immunoblots (antigen designated SGM 80); (2) binding preferentially to secretory granules, and reacting with components of approx. 110,000 and 50,000 Da on immunoblots (antigen designated SGM 110); (3) binding preferentially to secretory granules but unreactive on immunoblots; (4) binding to membrane antigen(s) with a widespread intracellular distribution which included granules and plasma membranes. The antigens SGM 80 and SGM 110 were studied in more detail and both were shown to be integral membrane glycoproteins with antigenic determinants located on the internal face of the secretory granule membrane. These antigens were also present in normal rat islets of Langerhans and similar components were detected by immunoblotting in secretory granules from anterior pituitary and adrenal medulla. Proteins which were immunologically related to SGM 80 and SGM 110, but distinct in molecular size, were also identified in liver. It is concluded that secretory granules contain specific components which are restricted in subcellular location but widespread in tissue distribution. The antibodies obtained will be valuable reagents in the further investigation of the biogenesis and turnover of insulin secretory granules.     

Subcellular distribution of glycosidases in human polymorphonuclear leucocytes.

The subcellular distribution of nine glycosidases were studied in fractions of homogenized human polymorphonuclear leucocytes (neutrophils) obtained by isopycnic centrifugation through linear sucrose density gradients. The substrates were 4-methylumbelliferyl glycosides. All nine glycosides were hydrolysed by enzymes in neutrophil cytosol fractions, and by enzymes in at least one granule population. alpha-Glucosidase activity sedimented in sucrose density gradients to a point (p = 1.180 g/ml) just above the specific granules, possibly the 'tertiary' granule population. The peak corresponding to alpha-glucosidase did not co-sediment with, but considerably overlapped, the peak corresponding to lactoferrin, a marker for specific granules (p = 1.187 g/ml). alpha-Galactosidase activity was found primarily in heavy azurophil granules (p = 1.222 g/ml). alpha-Mannosidase activity was found primarily in light azurophil granules (p = 1.206 g/ml), following the distribution of myeloperoxidase, the commonly used azurophil granule marker. beta-Glucosidase activity was concentrated in mitochondrial fractions (p = 1.160 g/ml). All other glycosidases presented complex distributions, with activities not restricted to one granule class. Granule-associated glycosidase activities were increased 2--38 times when measured in the presence of 0.05% Triton X-100, indicating latency of the enzymes within granules.