Timing, Targeting and Sorting of Azurophil Granule Proteins in Human Myeloid Cells K Arnljots, O Sørensen, K Lollike and N Borregaard

Home , Defensin, Myeloperoxidase, Specific granule

Leukemia (1998) 12, 1789–1795  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu Timing, targeting and sorting of azurophil granule proteins in human myeloid cells K Arnljots, O Sørensen, K Lollike and N Borregaard

The Granulocyte Research Laboratory, Department of Hematology, University Hospital, Rigshospitalet, Copenhagen, Denmark

Biosynthesis of 3 human granule proteins, myeloperoxidase, ules can be separated into large and smaller granules that dif- defensin and lysozyme, all present in azurophil granules, was fer not only in size but also with respect to the important anti- investigated in normal bone marrow cells and in the promyelo- bacterial peptides, defensins, which are found exclusively in cytic cell line HL-60 to see whether differences in timing of 5 biosynthesis could explain the well established differences in the large peroxidase-positive granules. Likewise, the peroxi- their subcellular localization in the mature neutrophil dase negative-granules on the basis of their content of lactofer- (targeting), and whether differences exist in the efficiencies by rin and gelatinase have been divided into three subsets: gran- which granule proteins are retained in cells (sorting). Normal ules that contain lactoferrin but no gelatinase (15%), granules human bone marrow cells were separated into three bands by that contain lactoferrin and gelatinase (60%), and granules density gradient centrifugation. Band 1 contains band and seg- that contain gelatinase but no lactoferrin (25% of peroxidase- mented cells, band 2 mainly myelocytes, metamyelocytes and 6 some band cells, and band 3 myeloblasts and promyelocytes negative granules). Furthermore, these subsets of peroxidase- in addition to megakaryocytes and proerythroblasts. Cells from negative granules also differ in the extent to which they these bands, as well as undifferentiated HL-60 cells, were release their content by any given stimulus.7 This granule het- pulsed with radiolabeled cysteine and methionine, and erogeneity is important, since it provides the neutrophil with biosynthesis of granule proteins was subsequently evaluated the ability to differentiate its release of granule proteins, which by immunoprecipitation and quantified by phosphorimaging. Myeloperoxidase synthesis was maximal in cells from band 3 are needed at different steps during the journey of the neutro- while defensin biosynthesis was maximal in cells from band phil from circulation into tissues to meet, engulf and kill 2. Lysozyme was synthesized in cells from all bands but was microorganisms.8 Furthermore, granule heterogeneity separ- maximal in cells from band 2. These results are in agreement ates proteins that cannot co-exist in the same granule.9 If gran- with our hypothesis that timing of biosynthesis determines the ule proteins were synthesized at the same time, a very com- localization of individual granule proteins. While myeloperoxi- plex targeting and sorting mechanism would need to be dase and defensins were efficiently retained in immature cells (band 3), a significant fraction of lysozyme was routed out of invoked to explain this granule heterogeneity. If, on the other the cells, showing that differences exist in the sorting of gran- hand, proteins localized to different granule subsets were syn- ule proteins between constitutive and regulated secretion. In thesized at different stages of cell maturation, no targeting addition, defensin was less efficiently retained in cells from would be needed to explain the granule heterogeneity. This band 2 than from band 3, indicating that sorting mechanisms could be explained solely by differences in timing of biosynth- may depend on the stage of cell maturation. esis. We have shown that the granule heterogeneity of peroxi- Keywords: sorting; biosynthesis; defensins; lysozyme; myeloperoxidase; human neutrophils dase-negative granules can be explained by differences in timing of biosynthesis of the proteins localized in granules,10 and we have shown that the localization of a granule protein will change if the time of its biosynthesis is changed, eg the protein Introduction NGAL normally localized in peroxidase-negative granules becomes localized to azurophil granules, if expressed at the The human neutrophil contains a variety of granules that differ promyelocytic stage in HL-60 cells.11 both with respect to content and propensity for exocytosis.1 We wished to examine this hypothesis further by investigat- Traditionally, the granules have been classified as peroxidase ing the biosynthesis of three granule proteins that differ in their positive and peroxidase negative on the basis of their content localization: defensins, myeloperoxidase and lysozyme. of the enzyme myeloperoxidase. This classification has proven While defensins are restricted to a subset of peroxidase-posi- useful since it not only separates granules that are formed at tive granules, the large peroxidase-positive granules, and mye- different stages of maturation of the neutrophil precursors, but loperoxidase defines the peroxidase-positive granules, lyso- also distinguishes between granules that differ with respect to zyme is found in all granules of the human neutrophil, exocytosis. Peroxidase-positive granules (also named azuro- including gelatinase granules.12 Examination of the stage of phil or primary granules) are formed at the promyelocyte stage myeloid cell maturation at which these proteins are expressed while the peroxidase-negative granules (also termed second- would therefore provide information on the relation between ary or specific granules) are formed at the myelocyte and met- time of biosynthesis and subcellular localization. A further amyelocyte stage.2 In contrast to peroxidase-positive granules, aim of this study was to gain information on whether there is peroxidase-negative granules fuse readily with the plasma an active sorting of proteins between constitutive secretion membrane and empty their content to the exterior. In recog- and retention, and whether any such mechanism is restricted nition of this, peroxidase-negative granules are also named the to a particular stage of cell maturation. secretory granules of the neutrophil.3 The two granule subsets differ greatly in their protein profiles.4 In addition, it has been established that major differences Materials and methods exist within these granule subsets. Peroxidase-positive gran- Cell culture Correspondence: N Borregaard, The Granulocyte Research Laboratory, Department of Hematology, Rigshospitalet L-4042, 9 HL-60 cells were grown in RPMI 1640 medium (Gibco-BRL, Blegdamsvej, DK-2100 Copenhagen, Denmark; Fax: 45 3545 4295 Life Technologies, Gaithersburg, MD, USA) supplemented Received 30 March 1998; accepted 6 August 1998 with 10% heat-inactivated fetal bovine serum (FBS), 100 U/ml Azurophil granule proteins in human myeloid cells K Arnljots et al 1790 penicillin, 100 ␮g/ml streptomycin at 37°C in a humidified 1% (v/v) Triton X-100, 1% (w/v) sodium deoxycholate, 0.1%

atmosphere of air/CO2 (19:1). Exponentially growing cells (w/v) SDS) supplemented with 1 mM phenylmethyl-sulphonyl- were used in all studies. ﬂuoride (Sigma Chemical, St Louis, MO, USA), 200 U/ml aprotinin (Trasylol; Bayer, Leverkusen, Germany), and 100 ␮g/ml leupeptin (Sigma). Undissolved material was pel- Isolation of bone marrow (BM) cells leted by ultracentrifugation (100 000 g for 20 min). The super- natants were stored at −20°C until immunoprecipitation. Fifteen milliliters of bone marrow were aspirated from the pos- Medium, from which the cells were pelleted, was diluted 5:1 terior superior iliac crest of healthy volunteer donors and with 5 × RIPA buffer before immunoprecipitation. mixed with 5 ml ACD (25 mM sodium citrate, 126 mM Immunoprecipitation was performed in a sequential man- glucose) present in the syringe used for aspiration. One vol- ner. First, the samples were preabsorbed with 20 ␮l non- ume of 2% Dextran T-500 (Pharmacia, Uppsala, Sweden) in immune rabbit serum (DAKO X0902) (DAKO, Glostrup, saline was added to aid the sedimentation of erythrocytes, and Denmark) and Protein A-Sepharose (Pharmacia). The appro- the supernatant was centrifuged at 200 g for 10 min. The pel- priate amounts of antibodies against MPO (DAKO A398), leted cells were resuspended in 10 ml phosphate-buffered lysozyme (DAKO A099), lactoferrin (DAKO A0186) and

saline (PBS, 140 mM NaCl, 10 mM Na2HPO4NaH2PO4 pH defensins (a generous gift from Dr Tomas A Ganz, University 7.4) and layered on top of a gradient which contained 9 ml of California, Los Angeles, CA, USA) was added to Protein A- Percoll (Pharmacia) of density 1.065 on top of 9 ml Percoll of Sepharose (100 mg/ml) in binding buffer (500 mm NaCl, 3 mM

density 1.080. The desired densities of Percoll were obtained KCl, 8 mM Na2HPO4/KH2HPO4 pH 7.2) containing 1% bovine by mixing 10 × PBS with precalculated amounts of Percoll serum albumin (BSA) and rotated over night at 4°C. The

and distilled H2O; pH was adjusted to 7.4 by addition of 0.1 amount of each antibody added corresponds to 11 mg/g N HCl. The gradient was centrifuged in 50 ml centrifuge tubes Sepharose gel. The Sepharose particles were pelleted and at 1000 g for 20 min at 4°C. This resulted in fractionation of resuspended in an equal volume of RIPA buffer. One hundred BM cells into three groups that were harvested with a Pasteur microliters were then added to the cell extracts or medium pipette: band 1, a cell pellet at the bottom of the centrifuge and rotated for 2 h at 4°C. The Sepharose particles were tube; band 2, all cells between band 1 and a distinct band on washed ﬁve times in RIPA buffer and twice in PBS. It was top of the gradient, band 3.10 Band 1 was contaminated with ascertained that all antigen was effectively precipitated by the erythrocytes and subjected to 30 s of hypotonic lysis in dis- antibody-Proteinase A-Sepharose added. The immunoprecipi-

tilled H2O. Cells from all bands were than washed in saline tated proteins were subjected to SDS-PAGE by boiling and counted. Cytospin preparations were made by applying samples in 100 ␮l SDS sample buffer and separating on a 12% 2 × 105 cells in 100 ␮l saline on glass slides. These prep- polyacrylamide gel.13 Immunoprecipitated defensins were arations were stained by the May–Gru¨nwald-Giemsa tech- analyzed by SDS-Tricine-PAGE with 16.5% acrylamide in the nique and used for evaluating the quality of the separation. separating gel and 3.3% acrylamide in the stacking gel. The gels were stained by Coomassie blue, destained and subjected to fluorography using Amplify (Amersham, Bucks, UK) and Biosynthetic labeling of BM cells exposing the dried gels to Kodak X-Omatic AR at −80°C for 1–3 days. Quantification of radioactive labeling was done by Cells from each band were starved for 45 min at 37°C at a cell phosphorimaging on FUJI X Bas 2000, using the software concentration of 107/ml in methionine–cysteine-free medium TINA vers. 2.086 from Santax. (minimal essential medium; ICN Pharmaceuticals, Costa Mesa, CA, USA) supplemented with 100 U/ml penicillin, 100 ␮g/ml streptomycin and 10% (v/v) dialyzed FBS. The cells Subcellular fractionation were then pelleted by centrifugation and resuspended at 2 × 107 cells/ml in minimal essential medium, supplemented Isolated neutrophils were resuspended in Krebs Ringer phos- ␮ 35 35 with either 200 Ci/ml S-methionine/ S-cysteine phate buffer (KRP: 130 mM NaCl, 5 mM KCl, 1.27 mM CaCl2 35 35 (Trans S-label) (1211 Ci/mmol) or S-cysteine (1171 5mM glucose, 10 mM NaH2HPO4/Na2HPO4 pH 7.4) contain- Ci/mmol) (both ICN Pharmaceuticals) and pulsed for 60 or ing 5 mM diisopropyl fluorophosphate (Aldrich Chemical, Mil- 120 min. Subsequent chase was performed after washing the waukee, WI, USA), kept on ice for 10 min, pelleted and resus- cells by centrifugation and resuspension in RPMI 1640 pended in relaxation buffer (3–5 × 108 cells in 15 ml) and medium with 10% FBS at a density of 5 × 107 cells/ml and cavitated as described by Kjeldsen et al.14 Nuclei and incubation at 37°C for the times indicated. After incubation, unbroken cells were sedimented by centrifugation, and the the cells were pelleted and washed once in ice-cold saline postnuclear supernatant (10 ml) was loaded on a three-layer before solubilization and immunoprecipitation. Percoll (Pharmacia density gradient (1.05/1.09/1.12 g/ml, Metabolic labeling of HL-60 cells was performed by starv- gradient volume 27 ml) and fractionated. This resulted in a ing the cells for 30 min in methionine–cysteine-free medium gradient with four visible bands, from the bottom designated × 6 ␣ ␤ ␤ ␥ with 10% FBS at 2 10 cells/ml. The cells were then pulsed the -band, the 1-band, the 2-band and the -band. The for 3 h by incubation in methionine–cysteine-free medium to cytosol was present above the ␥-band on top of the Percoll. which 50 ␮Ci/ml of 35S-methionine/35 S-Cysteine had been The gradient was collected in fractions of 1 ml each by aspir- added. Subsequent chase was carried out as above. ation from the bottom of the tube. Azurophil granules were identified by their content of myeloperoxidase,15 specific granules by lactoferrin,15 and gelatinase granules by gelatin- Immunoprecipitation ase,14,16 all assayed by ELISA.17–19 Plasma membranes were identified by HLA,20 assayed by ELISA20 and secretory vesicles Cells were solubilized for 1 h at 4°C at a density of 107 by their content of latent alkaline phosphatase,21 assayed as cells/ml in RIPA buffer (150 mM NaCl, 30 mM Hepes pH 7.3, described.22 Azurophil granule proteins in human myeloid cells K Arnljots et al 1791 Results and discussion oxidase-negative granules.12 We have previously shown that the renowned marker for specific granules, lactoferrin, is syn- Timing of biosynthesis of granule proteins thesized almost exclusively in cells from band 2.10 This is con- firmed here. Thus, the bulk of defensins and lactoferrin is syn- A marked difference was observed in the profile of granule thesized by cells isolated in band 2. This could indicate that proteins synthesized from the cells of band 3, 2 and 1, as has defensins and lactoferrin are each specifically targeted to dif- been shown before.10 Myeloperoxidase is synthesized pre- ferent granules, since their localizations in granules differ dominantly in cells from band 3, which contains the most (Figure 2). It is, however, more likely that the apparent co- immature myeloid cells, while defensins are synthesized synthesis is not real, but reflects the fact that the cells in band mainly in band 2. This difference in biosynthesis can thus 2 represent a span of maturity of cells, of which the more explain why defensins are localized only in a subset of the immature synthesize defensin, while the more mature synthe- peroxidase-positive granules, since clearly the majority of size lactoferrin. immature cells synthesize myeloperoxidase, but not defensins.

This is visualized in Figure 1. 500 120 Lysozyme was synthesized mainly in cells from band 2, but 450 also in cells from band 3 and 1. This is consistent with its 400 100 350 subcellular localization in both peroxidase-positive and per- MPO 80 Lysozyme concentration 300 concentration µg/ml 250 60 µg/ml 200 a 40 MP0 1 lkiDa 150 100 20 50 0 0 ·+ 9,7 0 5 10 15 20 25

20 120 18 100 16 14 80 Lysozyme b Defensins 12 concentration Defens,in %of total 10 60 µg/ml µg/ml 8 40 6 4 20 -4--- 2 6 0 0 5 10 15 20

300 120

250 100

·C: 200 80 Ly.sozy1me Lactoferrin Lysozyme concentration concentration µg/ml 150 60 µg/ml 100 40

50 20

-4!-I 114- 0 0 0 5 10 15 20 25

60 120

50 100 fl Lactoferri:n Gelatlnase 40 80 Lysozyme concentration conce~ration µg/ml 30 60 µg/ml 9,7' 20 40 + 66 10 20 0 0 0 5 10 15 20 25 3 2 '1 Fraction number Figure 2 Subcellular distribution of defensin, myeloperoxidase, Figure 1 Biosynthesis of myeloperoxidase, defensin, lysozyme lactoferrin, lysozyme and gelatinase in normal neutrophils from per- and lactoferrin in normal human bone marrow cells. Normal cells ipheral blood. 3 × 108 neutrophils were disrupted by nitrogen cavi- were fractionated into three bands: band 1 (segmented and bands), tation and fractionated on a three-layer Percoll density gradient. Frac- band 2 (myelocytes, metamyelocytes and some bands) and band 3 tions of 1 ml were aspirated from the bottom of the centrifuge tube (myeloblasts and promyelocytes).10 Cells were pulsed with 35S- and assayed for markers of azurophil granules (MPO)-specific gran- methionine/35S-cysteine for 2 h, washed and incubated in tracer-free ules (lactoferrin), gelatinase granules (gelatinase) and for lysozyme. medium for 2 h, washed, lysed and immunoprecipitated. (a) Precipi- Samples from each fraction were taken for SDS-PAGE and transferred tate with anti-MPO antibodies, (b) precipitate with anti-defensin anti- to nitrocellulose membranes and probed with a rabbit antibody bodies; (c) precipitate with anti-lysozyme antibody; and (d) precipitate against human neutrophil defensin. The intensity of the immunostain- with anti-lactoferrin antibody. ing was quantified by CREAM. Azurophil granule proteins in human myeloid cells K Arnljots et al 1792 This was addressed by changing the density of the medium the role of mannose 6-phosphate in sorting to lysosomes,23 used for separation of bands 2 and 3 from 1.080 g/ml to the mechanism(s) by which sorting is effected is elusive.24,25 1.081 g/ml. While this change shifted the major part of defen- Aggregation of proteins by either low pH or by high Ca2+ has sin synthesizing cells from band 2 to band 3, it had little effect been indicated as playing a role for some proteins.26–28 N- on the localization of lactoferrin synthesizing cells which linked carbohydrates may play a role in addition to mannose were still largely confined to band 2 (Figure 3). We therefore 6-phosphate-mediated sorting.29,30 Recently, hydrophobic conclude, that major differences exist in timing of biosynthesis domains were claimed to be important, but this has not been of myeloperoxidase, defensins and lactoferrin in relation to validated.31 Neutrophils contain an abundance of granules, as maturation of myeloid cells, and that this can explain the do other granulocytes, and these granules are formed within differences in subcellular localization of individual granule a quite narrow time frame of 3–5 days.2,32 It is therefore poss- proteins, including the heterogeneity of peroxidase-positive ible that formation of granules is so intense, that the bulk of granules. proteins entering the trans-Golgi network will go to granules, even without specific segregation from the route of constitutive secretion, which is usually considered the default route for Sorting of granule proteins between the constitutive proteins on the secretory pathway. If no specific retention of and regulated secretory pathways proteins to the regulated secretory pathway, ie to granules, exists, then all proteins that are synthesized at the same time It is generally assumed that sorting between these two path- should be retained equally efficiently in cells, reflecting the ways takes place in the trans-Golgi network, but apart from bulk flow of proteins to granules. This was investigated by examining the distribution of newly synthesized proteins between cells and medium at the different stages of maturity • MPO kDa represented by bands 3, 2 and 1 (Figures 4 and 5). Retention

-4- 97 a MPO kDa

+- 87

b Defensln Dafensin

+ • + 6

C Lactoferrin Lysozyme C + + 97 1, + 17

3 2 1 3 2 1 Figure 4 Immunoprecipitates of chase medium form normal bone Figure 3 Biosynthesis of myeloperoxidase (a), defensin (b) and lac- marrow cells; same experiment as in Figure 1. Cells were incubated toferrin (c) in bone marrow cells isolated on separating medium of for 2 or 6 h in tracer-free medium. The medium was then immuno- 1.081 g/ml similar experimental set-up as in Figure 1, but with a slight precipitated with antibodies against myeloperoxidase, defensins and change of density of the separating medium. lysozyme. Azurophil granule proteins in human myeloid cells K Arnljots et al 1793 Defensin medium. It is noted that the defensin found in the medium is 10 Thousands almost exclusively on unprocessed form (7 kDa). This agrees with the notion that processing is an event that takes place in azurophil granules.33 8 ------In contrast to myeloperoxidase and defensins which are efficiently retained in cells from band 3, lysozyme is not 6 ------efficiently retained in cells from any of the bands. This again argues for specific sorting between the pathways for constitutive and regulated secretion. The observation could also be 4 ------explained if lysozyme is as efficiently retained in granulocyte precursors as are defensins and myeloperoxidase, but in 2 addition is synthesized by other cells present in the bone marrow, eg monocyte precursors which do not efficiently retain the protein. To address this, we examined biosynthesis of 0 myeloperoxidase and lysozyme in the promyelocytic cell line Band 3 Band 2 Band 1 HL-60. These cells are arrested at the early promyelocytic stage of maturation. The cell line used by us was incapable of synthesizing defensins. It is clearly seen that the specific Myeloperoxidase sorting of granule proteins also took place in these cells since 5 Thousands the retention of myeloperoxidase was very effective while most lysozyme escaped to the medium (Figures 6 and 7). Thus, it is most likely that the difference observed in band 3 4 cells between myeloperoxidase and defensins on the one hand and lysozyme on the other is a reflection of sorting between constitutive secretion and storage. 3 The mechanism of sorting is not known. It is possible that the granule proteins bind to one or more carrier, that itself is 2 synthesized as a constituent of granules. The differences in sorting may have an implication in practical hematology. Pro- teins, such as MPO, which are efficiently retained in cells when synthesized are not likely to be useful parameters of myelopoietic activity when assayed in plasma. In contrast, 0 Band 3 Band 2 Band 1 MPO Lysozyme 1400 Cell Medirum kfia. 1200 +97 1000 800 600 400 200 0 Band 3 Band 2 Band 1

Figure 5 Intracellular and extracellular accumulation of newly synthesized proteins in normal human bone marrow cells fractionated by density centrifugation into three bands (band 1, most mature cells; + 14 band 2, cells of intermediate maturity; and band 3, most immature cells). Immunoprecipitates from cell lysates and medium were quantified by phosphorimaging. ᭿, cells; ᮀ, medium. 0 11 3 6 of myeloperoxidase and defensin is efficient in cells from band Figure 6 Biosynthesis of myeloperoxidase and lysozyme in undif- 3, but less efficient in cells from band 2. This is most pro- ferentiated HL-60 cells. HL-60 cells were metabolically labeled with 35S-methionine/35S-cysteine and incubated for 2 h. The cells nounced for defensin of which the vast majority of newly syn- were than chased for up to 6 h and aliquots were sampled at the time thesized protein is found in the medium. Furthermore, all points indicated. Cells and medium were separated and immunpreci- defensin synthesized in cells from band 1 is localized to the pitated with antibodies against myeloperoxidase and lysozyme. Azurophil granule proteins in human myeloid cells K Arnljots et al 1794 Myeloperoxidase References 1 Borregaard N, Cowland JB Granules of the human neutrophilic polymorphonuclear leukocyte. Blood 1997; 89: 3503–3521. 2 Bainton DF, Ullyot JL, Farquhar M. The development of neutrophilic polymorphonuclear leukocytes in human bone marrow. J Exp Med 1971; 143: 907–934. 3 Gallin JI. Neutrophil specific granules: a fuse that ignites the inflammatory response. Clin Res 1984; 32: 320–328. 4 Borregaard N, Lollike K, Kjeldsen L, Sengeløv H, Bastholm L, Nielsen MH, Bainton DF. Human neutrophil granules and secretory vesicles. Eur J Haematol 1993; 51: 187–198. 5 Rice WG, Ganz T, Kinkade IM, Selsted ME, Lehrer RI, Parmley RT. Defensin-rich dense granules of human neutrophils. Blood 1987; 70: 757–765. 6 Kjeldsen L, Bainton DF, Sengeløv H, Borregaard N. Structural and functional heterogeneity among peroxidase-negative granules in human neutrophils: identification of a distinct gelatinase containing granule subset by combined immunocytochemistry and subcellular fractionation. Blood 1993; 82: 3183–3191. 7 Sengeløv H, Kjeldsen L, Borregaard N. Control of exocytosis in early neutrophil activation. J Immunol 1993; 150: 1535–1543. 0 8 Sengeløv H, Follin P, Kjeldsen L, Lollike K, Dahlgren C, Borre- Oh 1 h 3h 6h gaard N. Mobilization of granules and secretory vesicles during in vivo exudation of human neutrophils. J Immunol 1995; 154: 4157–4165. 9 Le Cabec V, Calafat J, Borregaard N. Sorting of the specific granule protein, NGAL during granulocytic maturation of HL-60 cells. Lysozyme Blood 1997; 89: 2113–2121. 300 10 Borregaard N, Sehested M, Nielsen BS, Sengeløv H, Kjeldsen L. Biosynthesis of granule proteins in normal human bone marrow cells. Gelatinase is a marker of terminal neutrophil differentiation. 250 Blood 1995; 85: 812–817. 11 Le Cabec V, Cowland JB, Calafat J, Borregaard N. Targeting of 200 proteins to granule subsets determined by timing not by sorting: the specific granule protein NGAL is localized to azurophil granules when expressed in HL-60 cells. Proc Natl Acad Sci USA 150 1996; 93: 6454–6457. 12 Lollike K, Kjeldsen L, Sengeløv H, Borregaard N. Lysozyme in 100 human neutrophils and plasma. A parameter of myelopoietic activity. Leukemia 1995; 9: 159–164. 13 Laemmli UK. Cleavage of structural proteins during the assembly 50 of the head of bacteriophage T4. Nature 1970; 227: 680–685. 14 Kjeldsen L, Sengeløv H, Lollike K, Nielsen MH, Borregaard N. 0 Isolation and characterization of gelatinase granules from human 3h 6h neutrophils. Blood 1994; 83: 1640–1649. Oh 1 h 15 Cramer E, Pryzwansky KB, Villeval JL, Testa U, Breton-Gorius J. Ultrastructural localization of lactoferrin and myeloperoxidase in Figure 7 Quantification of intracellular and extracellular sorting of human neutrophils by immunogold. Blood 1985; 65: 423–432. metabolically labeled myeloperoxidase and lysozyme in HL-60 cells. 16 Dewald B, Bretz U, Baggiolini M. Release of gelatinase from a The gels photographed in Figure 6 were quantified by phosphor- novel secretory compartment of human neutrophils. J Clin Invest ᭿ ᮀ imaging , cells; , medium. 1982; 70: 518–525. 17 Kjeldsen L, Bainton DF, Sengeløv H, Borregaard N. Identification of neutrophil gelatinase-associated lipocalin as a novel matrix protein of specific granules in human neutrophils. Blood 1994; 83: 799–807. due to inefficient sorting, the level of lysozyme in plasma may 18 Kjeldsen L, Bjerrum OW, Askaa J, Borregaard N. Subcellular local- accurately reflect the myelopoietic activity of the bone mar- ization and release of human neutrophil gelatinase, confirming the row and at the same time support a high level of free circulat- existence of separate gelatinase-containing granules. Biochem J ing antibiotic protein.12 A similar mechanism may explain the 1992; 287: 603–610. 19 Kjeldsen L, Bjerrum OW, Hovgaard D, Johnsen AH, Sehested M, high levels of defensins in plasma found in patients with Borregaard N. Human neutrophil gelatinase: a marker for circulat- 34 septicemia and bacterial meningitis. ing blood neutrophils. Purification and quantitation by enzyme linked immunosorbent assay. Eur J Haematol 1992; 49: 180–191. 20 Bjerrum OW, Borregaard N. Mixed enzyme-linked immunosorbent assay (MELISA) for HLA class I antigen: a plasma membrane Acknowledgements marker. Scand J Immunol 1990; 31: 305–313. 21 Borregaard N, Miller L, Springer TA. Chemoattractant-regulated KA was supported by a post-doc fellowship from the Alfred mobilization of a novel intracellular compartment in human neu- Benzon Foundation, NB was supported by an elite fellowship trophils. Science 1987; 237: 1204–1206. from the Alfred Benzon Foundation. NB and KL were sup- 22 DeChatelet LR, Cooper MR. A modified procedure for the determi- nation of leukocyte alkaline phosphatase. Biochem Med 1970; 4: ported by the Danish Medical Reserarch Council. The expert 61–68. technical essistence of Mrs Charlotte Horn is gratefully 23 Johnson KF, Kornfeld S. The cytoplasmic tail of the mannose 6- acknowledged. phosphate/insulin-like growth factor-II receptor has two signals for Azurophil granule proteins in human myeloid cells K Arnljots et al 1795 lysosomal enzyme sorting in the Golgi. J Cell Biol 1992; 119: membrane association of procathepsin L is mediated by a 9-resi- 249–257. due sequence within the propeptide. J Biol Chem 1994; 269: 24 Sossin WS, Fisher JM, Scheller RH. Sorting within the regulated 567–572. secretory pathway occurs in the trans-Golgi network. J Cell Biol 29 Fiedler K, Simons K. The role of N-glycans in the secretory path- 1990; 110: 1–12. way. Cell 1995; 81: 309–312. 25 Harter C, Wieland F. The secretory pathway: mechanisms of pro- 30 Ponnambalam S, Banting G. Protein secretion: sorting sweet sort- tein sorting and transport. Bba Rev Biomembranes 1996; 1286: ing. Curr Biol 1996; 6: 1076–1078. 75–93. 31 Gorr SU, Darling DS. An N-terminal hydrophobic peak is the sort- 26 Yoo SH. pH- and Ca2+-dependent aggregation property of ing signal of regulated secretory proteins. FEBS Lett 1995; 361: secretory vesicle matrix proteins and the potential role of chromo- 8–12. granins a and b in secretory vesicle biogenesis. J Biol Chem 1996; 32 Bainton DF. Neutrophil granules. Br J Haematol 1975; 29: 17–22. 271: 1558–1565. 33 Valore EV, Ganz T. Posttranslational processing of defensins in 27 Canaff L, Brechler V, Reudelhuber TL, Thibault G. Secretory gran- immature human myeloid cells. Blood 1992; 79: 1538–1544. ule targeting of atrial natriuretic peptide correlates with its cal- 34 Panyutich AV, Panyutich EA, Krapivin VA, Baturevich EA, Ganz cium-mediated aggregation. Proc Natl Acad Sci USA 1996; 93: T. Plasma defensin concentrations are elevated in patients with 9483–9487. septicemia or bacterial meningitis. J Lab Clin Med 1993; 122: 28 McIntyre GF, Godbold GD, Erickson AH. The pH-dependent 202–207.