Sep 21, 1988 - MERETE ALBRECHTSEN. AND MICHAEL A. KERR Department of Pathology, University of Dundee,. Ninewells Hospital and Medical School.
British )ournut of Hnernntofugy. 1989. 7 2 , 3 12-320
Characterization of human neutrophil glycoproteins expressing the CD15 differentiation antigen ( 3 -fucosyl-N-acetyllactosamine) MERETEALBRECHTSEN A N D MICHAELA. KERRDepartment of Pathology, University of Dundee, Ninewells Hospital and Medical School. Dundee Received 21 September 1988: accepted for publication 24 January 1989
Summary. The expression of the CD15 antigen. 3-fucosyl N-acetyllactosamine. on neutrophil glycoproteins has been studied by SDS gel electrophoresis and immunoblotting. The antigen is expressed on several glycoproteins. both intracellularly and on the cell surface. Each subcellular compartment appears to contain a specific antigen. A soluble, granule glycoprotein (M,80-90K) probably accounts for most of the intracellular staining detected immunohistochemically. Membrane glycoproteins of M,. 8 5-90K and 2 5K are associated with granule membranes, the latter being a n integral membrane protein. The CDl5 antigen is expressed on several cell surface glycoproteins with M, in the range of 165K and
105K. These antigens are also contained in a n intracellular pool which is brought to the surface on activation of the cells with chemotactic peptides. The 165K and 105K antigens show identical electrophoretic mobility to two of the major glycoproteins detectable by PAS or protein staining of gels of detergent extracts of cell membranes. These glycoproteins include the complement receptor, CK3. The beta chain of CR3 (105K) and to a lesser extent the alpha chain (165K) express CD15: however, most of the CD15 antigen is associated with other glycoproteins of these molecular masses.
The neutrophil plays a major role in the defence against invading microorganisms. as well as in removing damaged or aged autologous tissue. It is the first cell type to accumulate at sites of infection or other sites of inflammation, and deficiencies of neutrophil function are associated with recurrent infections. Some of the cell surface glycoproteins involved in neutrophil functions such as the recognition and phagocytosis of foreign organisms have been identified: these include receptors for the complement fragments C3b and C3bi (Ross, 1986: Sim& Walport. 1987) and receptors for the Fc region of immunoglobulins IgG (Anderson 8- Looney. 1986) and IgA (Albrechtsen et nl. 1988). Most of the neutrophil surface proteins hare yet to be characterized structurally and functionally. The majority of the neutrophil surface proteins that have been characterized were first identified by means of monoclonal antibodies. Many of these antibodies react with other leucocytes in addition to neutrophils. suggesting functions for the proteins that are shared by several cell types. A number of 'neutrophil-specific' monoclonal antibodies have
been reported, nearly all of which are of the IgM class and recognize the carbohydrate antigen 3-fucosyl-N-acetyllactosamine, sometimes referred to as LeXor X-hapten (Huang et al. 1983: Gooi rt al. 1983). Antibodies with these characteristics were classified as Cluster of Differentiation (CD)15 at the 3rd International Leucocyte Typing Workshop (1987). These antibodies are widely available commercially (e.g. Leu M1, Dako Ml).In addition to recognizing neutrophils in blood and tissues, these antibodies are useful histochemically in the recognition of certain leucocyte malignancies such as Hodgkin's disease and histiocytosis X (Hall & Ardenne, 1987: Santamaria et al, 1988) and also malignancies of nonhaemopoietic cells (Kerr 8- McCarthy, 1985; Sanders et al, 1988). Anti-CD1 5 antibodies recognize both glycolipids and glycoproteins from neutrophil membranes. Most have been found to immunoprecipitate glycoproteins of M, 140-1 80K and 95-1 1 0 K from detergent extracts of surface '2iI-labelled neutrophils (Skubitz et al. 1983: Tetteroo et al, 1984). Melnickrtal(1985)andSkubitz&Snook( 1987) haveshown CDl5 antibodies to inhibit a number of neutrophil functions such as phagocytosis, enzyme release and the respiratory burst. They also demonstrated that these antibodies reacted
Correspondence: L)r M. A. Kerr. Department of Pathology. University ofDundee. Ninewells Hospital and Medical School. Dundee DD1 9SY.
Neutrophil CD7 5 Antigen with molecules belonging to the LFAl/CR3/p150,95 family of leucocyte glycoproteins which are involved in adhesion and phagocytosis. This family of proteins share a common beta-subunit (apparent M, 100K) which forms non-covalent heterodimers with alpha-subunits of approximate M, 185K, 165K and 150K, respectively (Sanchez-Madrid et al, 1983). The data concerning the expression of CD15 on members of the LFA family of proteins are, however, somewhat confusing since Harlan et a1 (1985) and Skubitz & Snook (1987) have reported that neutrophils from individuals genetically deficient in the LFA family of molecules still express surface CD15. The nature of the molecules expressing the CD15 antigen is further confused by the fact that in tissue sections, the antigen is readily detected in granules as well as on the cell surface. Since it is well recognized that the LFA molecules are also stored as an intracellular pool, it is of interest to know whether this pool contributes to the granular staining or whether it is due to other, specifically intracellular proteins. At the molecular level, it is unclear whether both the alpha and beta chains of the LFA molecules express the antigen or just one, because immunoprecipitation techniques were used to detect the presence of the CD15 antigen on the LFA molecules and the alpha chain is precipitated by the anti-beta chain antibodies and vice versa. MC1, a neutrophil specific IgM monoclonal antibody raised in this laboratory, recognizes the LeXdeterminant and thus belongs to the CD15 family of antibodies (McCarthy et al, 1985).This antibody was selected from a panel of anti-CD15 antibodies because of its relatively high affinity. We have previously reported that this antibody recognizes glycoproteins of M, around 105K and 165K as detected by immunoblotting of neutrophil membrane detergent extracts (McCarthyet al, 1985).In addition to glycoproteins, MC1 has been found to react with the same neutrophil glycolipid antigens recognized by other anti-CD15 monoclonal antibodies (Group 1 in Kniep & Muhlradt, 1987; Dr B. Kniep, personal communication). We have now extended the protein immunoblotting studies to allow a further characterisation of the MC1 glycoprotein antigens. MATERIALS AND METHODS
Monoclonal antibodies. The monoclonal antibody MC 1 was produced after immunization of mice with whole human neutrophils as previously described (McCarthy et al. 1985). The monoclonal antibodies 3.9 (anti-p150,95 alpha-chain) and 44 (anti-CR3 alpha-chain) were kindly donated by Dr N. Hogg (Malhotra et al, 1986). MHM24 (anti-LFAl alphachain) and H52 (anti-CR3 beta-chain) were gifts from Dr A. J. McMichael (Hildreth & August, 1985; Hildreth et al, 1983). Isolation of neutrophils. Neutrophils were isolated from heparinized blood from healthy volunteers by a scaled up version of the discontinuous density gradient method described by English & Andersen (1 9 74). The average yield from 96 preparations over 2 years was 2.5 x lo9 neutrophils/l blood. The average purity, defined by volume spectroscopy (Potts et al, 1980) as the percentage of nucleated cells being neutrophils, was 92%; erythrocytes constituted less than 5% of the total number of cells in most preparations.
Cells not needed immediately were suspended at a concentration of l o 8 cells/ml in 0.25 M sucrose, 100 mM Hepes, 5 mM iodoacetamide. The protease inhibitor diisopropylfluorophosphate (DFP) was added to a final concentration of 2 mM and the cells stored at -20°C. The sucrose medium was chosen because it prevented lysis of the nuclei on freezing and thawing: cells frozen in buffer alone released their DNA on thawing, making the thawed suspension extremely viscous and difficult to handle. Neutrophil subfractionation. 1.25 x lo8 neutrophils stored at - 2OoCwere thawed and centrifuged for 30 min at 30 000 g. 5°C. The supernatant (SN1) was removed and the pellet ( P l ) was resuspended in 12 ml Washing Buffer (25 mM TrisHCI pH 7.4, 150 mM NaCl. 1 mM MgS04, 1 mM PMSF). The suspension was homogenised with eight strokes in a PotterElvehjem homogenizer with motor-driven pestle and centrifuged 10min at 100g. The pellet was re-homogenized in 5 ml fresh Washing Buffer and again centrifuged 10 min at 100 g. The resulting pellet (P2) was resuspended in 4 ml of 0.25 M sucrose, 2 5 mM Tris-HC1pH 7.4. The combined supernatants from the low-speed spins (SN2) were centrifuged at 30 000 g for 150 min. The supernatant (SN3) was then removed. The pellet (P3) was resuspended in 12 ml of 0.25 M sucrose, 25 mM Tris-HCI pH 7.4, 1%(w/v) Triton X-100. After 60 min at 4°C with gentle agitation, the suspension was centrifuged for 150 min at 30 000 g. The supernatant (SN4) was removed and the pellet resuspended in 2 ml 0.25 M sucrose, 25 mM Tris-HCI pH 7.4 (P4). Release of peripheral membrane proteins. To determine whether proteins sedimenting with the cell membranes were integral membrane proteins or merely absorbed to the membrane peripherally, pellet P3, from a neutrophil subfractionation (see above), was resuspended in 4 ml 100 mM Na2C03(pH 11.5). After 30 min at 4% the suspension was centrifuged at 30 000 g for 150 min. The supernatant was removed and the pellet resuspended in 4 ml KCI buffer (0.5 M KCI, 50 mM Tris-HC1 pH 7.4, 0.25 M sucrose), then centrifuged for 150 min at 30 000 g. The supernatant was removed and the pellet extracted with Triton X-100 as described above. Simpl$ed Triton extraction procedure. Neutrophils stored at - 20°C were thawed and centrifuged 3 min at 11 600 g in a Microcentaur microfuge. The supernatant was removed and the pellet resuspended in 1%(w/v)Triton X-100 in PBS, or in 1%(w/v) Nonidet P-40 (NP-40) in PBS. Phenylmethylsulphonylfluoride (PMSF) was added to a final concentration of 1 mM. The suspension was incubated at 4°C for 60 min with end-over-end mixing, then centrifuged 3 min at 11 600 g. The resulting supernatant is referred to as 'Neutrophil membrane Triton extract'. lndirect immunoprecipitation. Goat antibodies specific for mouse IgG whole molecule (GalgG)or for mouse IgM heavy chain (GaM)(both from Sigma) were coupled to Sepharose 4B at about 0.8 mg goat antibodies per ml Sepharose. 0.5 ml neutrophil or lymphocyte, membrane Triton extracts were incubated with 1 5 p1 monoclonal-antibody containing mouse ascites for 2 h at 4"C, with end-over-end agitation. 200 p1 of a 1:1 suspension of GAG- or GAM-Sepharose were added and incubation continued for 30 min. The suspension
Merete Albrechtsen and Michael A. Kerr
was centrifuged for 30 s at 2800 g and the supernatant removed. The Sepharose beads were washed with 2 x 1 ml 0.1% Triton X-100 in PBS. 1 x 1 ml (0.75 M NaCI. 0.1% Triton, 20 mM sodium phosphate pH 7.4), and 2 x 1 mlO.1% Triton in PBS. Bound proteins were eluted with 200 plO.5 M acetic acid, 0.1%Triton and the eluate neutralized immediately by addition of 75 p1 2 M Tris-HCI pH 8.7. SDS gel electrophoresis. Discontinuous SDS polyacrylamide gel electrophoresis was performed essentially as described by Laemmli (1970). Gels were either linear gradient gels from 5% to 1 5% acrylamide or 7.5% non-gradient gels. Samples were prepared for electrophoresis by addition of 1/2 to 1 volume of SDS-PAGE sample buffer: 0.1 M Tris-HC1 pH 8.0, 8 M urea, 2% SDS. For reducing gels, the sample buffer also contained 80 mM DTT: for non-reducing gels. 4 0 mM iodoacetamide. Samples were routinely boiled for 2 min: reduced samples were then alkylated by addition of 1 0 p1 1M iodoacetamide for every 100 pl sample buffer. Silver staining of SDS-gels was performed as described by Morrissey (1981). Gels were stained for carbohydrate using the periodic acid-Schiff (PAS)method as modified by LetarteMuirhead et a1 (1975).The Schiff reagent was prepared by bubbling sulphur dioxide through a 0.5% aqueous fuchsin solution. Neutrophils were surface labelled using the lactoperoxidase, periodate-borohydride (Gahmberg & Anderson, 1977) or galactose oxidase-borohydride techniques (Anderssen & Gahmberg, 1978) as described in Johnstone & Thorpe (1982). Immunoblotting. Proteins were transferred from SDS polyacrylamide gels to nitrocellulose membranes (0.45 pm pore size, Schleicher & Schuell) using the method ofTowbin et a1 (1979) except that the transfer buffer contained only 15% methanol, and a voltage gradient of 5 V/cm was applied for at least 5 h (max. 18 h). Immunostaining of the blots was performed with incubation times of 2 h at room temperature for the primary antibody and 1 h at room temperature for the secondary antibody (peroxidase-coupled sheep antibodies to mouse IgG, M. A and light chains, Serotec). The blots were developed using 4-chloro-1-naphtol as enzyme substrate. Stimulation of neutrophils with N-forniyl-methionyl-leucylphenylalunine (fMLP). 5 x 10' neutrophils were resuspended in 1 ml RPMI 1640 (GIBCO),20 mM Hepes buffer pH 7.0. 1 0 pl 10-j M fMLP in dimethylsulphoxide were added to give a final concentration of 1 O - i ~fMLP. The cells were incubated at 37°C for 25 min with occasional agitation, then pelleted for 5 min at 700 g and washed once with PBS. Cytofluorimetry. 5 x lob neutrophils were resuspended in 1 ml PBS containing 5% (v/v) normal human serum (PBS/ serum). 5 pl of an appropriate dilution of monoclonalantibody containing ascites fluid were added and the cells incubated 45 min on ice: every 10 min the cells were resuspended by gently tapping the tube. (The concentration of monoclonal antibody MC1 was limited to give detectable staining but insignificant aggregation of the cells.) 2 ml cold PBS/serum were added to each sample and the cells were pelleted at 700 g , then washed once more with 2 ml PBS/ serum. FITC-conjugated antibodies to mouse Ig (Unipath) were diluted 1: 3 5 in PBS/serum and the cells were resuspended in 100 pI of this solution. After 30 min on ice, with
gentle resuspension every 10 min. the cells were washed once with PBS/serum as above, the pellet was tapped loose and 1 ml propidium iodide solution (0.05 mg/ml propidium iodide in 0.1%sodium citrate) was added. Control samples were incubated with unspecific mouse ascites fluid, obtained by injection of non-Ig producing myeloma cells, instead of the specific primary antibody, followed by incubation with FITC-conjugated secondary antibodies as described above. The staining was analysed with an Ortho SOH cytofluorimeter with an 8W argon lamp. RESULTS
Identification of glycoproteins expressing the C D l 5 antigen using monoclonal antibody, MCl Antigens recognized by the monoclonal antibody, MC1, are present on the surface of neutrophils and can be detected by immunofluorescence staining of intact cells (see below). However, indirect immunoperoxidase staining of sections of fixed and paraffin-embedded tissues routinely shows the presence of MC1 antigens in intracellular organelles as well as on the cell surface. Electron microscopy of immunogoldlabelled sections has shown that these antigens are associated with intracellular granules of different size and density but not with the nucleus, cytoplasm, mitochondria or endoplasmic reticulum. When neutrophils were lysed and then fractionated as described above, glycoproteins carrying the MC1 determinant could be identified by immunoblotting in all fractions (Fig 1).The SN1 fraction, consisting of the soluble proteins which were released from the cells by freezing and thawing, contained the cytoplasmic proteins (9 7% of lactate dehydrogenase) and also most of the granule enzymes (93% of total cellular beta-glucuronidase and 68% of total lysozyme) released by the rupture of the granule membranes. Immunoblotting of this fraction produced bands in the M, 80-90K range and a faint band corresponding to M, l 0 0 K (Fig l B , lane 1). Immunoblotting of the pelleted, lysed cells (Pl), produced different bands corresponding to M, of around 165K. 105K, 25K and also several bands in the range 90-100K (Fig l B , lane 2). After homogenization of the cells, centrifugation at low speed gave a pellet (P2) containing nuclei, incompletely disrupted cells and large membrane fragments. This pellet contained about 25% of the MC1 antigen. Most of the antigen, but relatively little of the total protein, was associated with smaller membrane fragments which were pelleted only at high speed. Both pellets showed the same complex pattern on immunoblotting identical to that of the whole lysed cells. When the high-speed pellet was extracted with the detergent, Triton X-100, immunoblotting of the solubilized membranes identified two broad bands corresponding to glycoproteins of M, around 165K and 105K (Fig 1B. lane 4): on longer incubation in the peroxidase substrate minor bands of M, > 200K and approximately 8 5K were also observed. The 85K band was more prominent in extracts of cells which had not been treated with DFP, suggesting it is a proteolytic fragment of some of the larger antigens. However, the M, > 200K. 165K and 105K bands were still present in a
Neutrophil CD15 Antigen
Fig I. Analysis by SDS gel electrophoresis of the proteins found in each fraction produced by the neutrophil subcellular fractionation shown in Materials and Methods. Proteins were detected by (A)silver staining, or (B) immunoblotting with CD15 monoclonal antibody, MCI. Lane 1: SN1, lane 2: P1, lane 3: SN3, lane 4:SN4, lane 5: P4. Aliquots of the high-speed pellet, P3, were treated with Na2C03and then washed with 0.5M KCI prior to Triton extraction (see Materials and Methods for details). Lane 6 shows the material released by Na2C03:lane 7:the material released in the KCI wash: lane 8 shows the subsequent Triton extract: and lane 9:the final Triton insoluble material. L Low molecular mass markers (top to bottom: 66,45,36, 29,24,20,14 kDa). H:High molecular mass markers (top to bottom: 205, 116, 97,66, 45,29 kDa).
non-DFP treated sample, showing the considerable resistance to proteolysis exhibited by these molecules. The Triton insoluble fraction, P4,contained MC1-reactive glycoproteins of M, 85-90K and 25K (Fig l B , lane 5). In order to investigate whether the membrane antigens identified above were integral membrane proteins or periph-
era1 proteins, merely absorbed to the membranes, the highspeed pellet (P3) was treated with 100 mM Na2C03. a procedure which has been shown to release peripheral membrane proteins in other systems (Fujiki et al, 1982). The membranes were subsequently washed with 0.5 M KCI to ensure release of all peripheral proteins and the insoluble
Merete Albrechtseri and illichael A. Kerr residue was extracted with Triton. When these fractions were analysed by immunoblotting with MC1, the 165K and 1 0 5 K antigens were found in the Triton extract (Fig lB, lane 8)and must thus be classified as integral membrane proteins. The Triton-insoluble 8 5-90K antigens were released from the membranes by the NaLCOl treatment, suggesting that they are not integral membrane proteins. The 25K antigen was not released (Fig 1B. lane 6).
Fig 2. Keutrophil membrane proteins separated by SDS gel electrophoresis. Lane 1 : Neutrophil surface proteins "'iodinated by the lactoperoxidase technique, detected by autoradiography. Lanes 2 . 3 and 5 : glycoproteins of the Triton extract of neutrophil membranes detected by the PAS technique ( 2 ), silver stain ( 3 )or immunoblotting with MCl ( 5 ) . Lanes 4 and 6 : CDI 5 antigens affinity purified from neutrophil membrane Triton extract on MCI -Sepharose detected by Silver stain (4) or immunoblotting (61. In lane 4 the 8 0 K protein marked with a n arrow is the heavy chain of MCI. The protein marked with an arrow head is bovine serum albumin added to stop non-specific absorption to the resin. Both contaminants had leaked from the column.
Characterimtion of the 1 0 5 K and 165 K glycoproteins expressing CDI 5 The two major lh5K and 105K antigens detected by immunoblotting correspond to the CD15 antigens identified in many laboratories by immunoprecipitation of surfaceradioiodinated neutrophil membrane proteins. Our own results showed that these were the only glycoproteins precipitable from surface-labelled cells by monoclonal antibody. MC1. This suggests that the other glycoproteins detected by immunoblotting are components of the granule membranes. Neither the 165K or 105K glycoproteins were seen as particularly abundant bands when surface-iodinated proteins were analysed by gel electrophoresis and autoradiography (Fig 2 , lane 1). However, these bands showed identical electrophoretic mobilities to two of the major proteins of the Triton extract detectable by silver staining and to the major glycoproteins in the neutrophil membrane Triton extract identified by PAS-staining of a n SDS-gel (Fig 2 , lanes 2 and 3 ) . The same bands were also detected following surface labelling of neutrophils using the galactose oxidase/ borohydride technique. In order to confirm which antigens were associated with the plasma membrane and which were granular, enucleated cells or cytoplasts were prepared by the method of Roos et al 198 3 1. The cytoplast preparations, which lacked granules
Fig 3. EfFect of tMLP on neutrophil surface expression ofCR 3 alpha-chain and CD15 antigens. Aliquots of freshly isolated neutrophils were kept on ice (A. Cl or were stimulated with IO-;M LMLP for 3 0 min at 37OC (B. D1. The cells were labelled with MC1 (panels A and B) or OKM1, anti-CR3
alpha-chain, antibodies (panels C and Dl followed by FITC-coupled anti-mouse Ig antibodies. and their fluorescence quantitated by cytofluorimetry.
Neutrophil CD15 Antigen
Fig 4. Indirect immunoprecipitationsfrom neutrophil membrane Triton-extracts using the monoclonal antibodies MCI (lane 1). 44 (anti-CR3 alpha-chain;lane 2), MHM24 (anti-LFAl alpha-chain;lane 3), 3.9 (antip150,95 alpha-chain; lane 4), H52 (antiCR3/LFAl/pl50.95 beta-chain; lane 5). Control immunoprecipitationsin which the membrane extract was omitted are shown in lanes 7-1 1. Lane 6 shows molecular mass markers (205, 116, 97, 66, 45, 29 kDa). Lane 12 contained unfractionated neutrophil membrane Triton extract. (A) Silver stained SDS gel; (B) immunoblot stained using antibody MC1. detectable by electron microscopy, when subjected to SDS gel electrophoresis, showed only the 105K and 165K proteins to be detectable on immunoblotting with MC1. All of the other MCl positive bands could be detected in the organelle-rich pellet (results not shown). Since the surface expression of several neutrophil membrane proteins is known to be increased in response to the chemotactic peptide, fMLP (Springer et aI. 1984; Todd et af, 1984) demonstrating an intracellular pool of plasma membrane proteins, we studied the effect of fMLP on surface expression of the CD15 determinant using indirect immunofluorescence and cytofluorimetry. We observed that stimula~ in most cases resulted in a marked tion with 1 0 - 7fMLP increase in MCl staining (Fig 3 , A+B). The increase was comparable to that seen with antibodies specific for the CR3 alpha-chain (Fig 3 , C+D), or beta chain (results not shown)
confirming the presence of a n intracellular pool of MC1 antigens which can be mobilized to the surface on stimulation of the cell.
The expression of CDl5 on the LFAI/CR3/pl50,95family oJ adhesion molecules The monoclonal antibodies 44 and H52. specific for the CR3 alpha-chain and the LFAl/CR3/pl50.95, common betachain, respectively, both immunoprecipitated from neutrophi1 membrane Triton extracts, polypeptides of approximate M, 165K and 105K as seen by SDS-PAGE and silver staining (Fig 4A. lanes 2 and 5 ) . A second absorption with the same antibodies precipitated no more antigen. Monoclonal antibodies MHM24 (anti-LFAl alpha-chain) and 3.9 (antipl50.9 5 alpha-chain) precipitated little if any detectable antigen from these extracts, reflecting the low level of
Merete Albrechtsen and Michael A. Kerr
expression of these molecules on neutrophils (Fig 4A. lanes 3 and 4). MC1 precipitated a barely visible band corresponding to a 165K protein (Fig 4A. lane 1). Immunoblotting with MCI showed strong staining of the 105K antigen precipitated by antibodies 44 and H52. as well as very weak staining of the corresponding 165K molecules (Fig 48, lanes 2 and 5). The antigens precipitated by MC1 were, as expected. strongly stained (Fig 4B. lanes 1 and 12). This experiment demonstrates that MC1 recognizes the LFAl common beta-chain and reacts weakly with the CR3 alpha-chain. However. most of the 165K MCI antigen appears to be distinct from the CR3 receptor. It is clear that the CR3 beta-chain can only account for part of the 105K MCl-reactive material since. the betachain is seen on Western blots as a well-defined band while MCl produces a broad band on immunoblotting of neutrophi1 membrane Triton extracts (compare lanes 2 and 5 in Fig 4B with 1 and 12). Furthermore, repeated absorptions of a neutrophil membrane Triton extract with the monoclonal antibody H52 (anti-beta-chain) failed to remove a significant proportion of the 105K MCI -reactive material as determined by immunoblotting. These observations were confirmed by affinity chromatography of a neutrophil membrane Triton extract on MC1Sepharose (see Fig 2 . lanes 4-6). This procedure isolated 165K and 105K proteins as shown by SDS-PAGE and silver staining. However, the 105K component was only barely visible on the silver-stained gel although immunoblotting of the isolated material with MC1 showed strong staining of both 165K and 105K proteins. Most of the material with M, 105K and 165K was not absorbed by the column. When indirect immunoprecipitations were performed from a mononuclear cell membrane Triton extract using the same monoclonal antibodies as above. SDS-PAGE and silver staining showed a 170K protein to be precipitated with the antiLFA-I alpha-chain antibody MHM24. and 170K, 165K, 150K and 105K proteins to be precipitated by the anti-betachain antibody H52. Immunoblotting of the precipitated proteins with MCl produced no staining at all, consistent with the lack of staining of lymphocytes and monocytes by MC1 in histochemical sections and showing clearly that the IFA1/CR3/p150395 beta-chain on mononuclear cells is glycosylated differently from the same polypeptide on neutrophils. DISCUSSION Analysis of the CD15 antigen immunoblotting has revealed greater complexity than was suggested by surface labelling and immunoprecipitation studies. The results show clearly that the carbohydrate antigen, 3-fucosyl N-acetylactosamine is expressed on several neutrophil glycoproteins. These glycoproteins are not only cell surface molecules but also soluble and membrane glycoproteins from the granules. The glycoproteins which are expressed on the cell surface are also stored in a n intracellular pool. Our results confirm and extend the observations of Melnick et al (1985) that CDl5 antibodies immunoprecipitated glycoproteins of M, 56, 80 and 1OOK from radiolabelled specific granule membranes and a glycoprotein of around 30K from azurophil granule membranes. Taken together these studies suggest that each
subcellular compartment contains a unique glycoprotein expressing the antigen. The 8 0 K soluble granule glycoprotein expressing the CD15 antigen, which probably accounts for the majority of the granular staining seen in tissue sections has not been previously recognized. Although we have not characterized the protein in detail, it is not lactoferrin, an approximately 8 0 K glycoprotein which when isolated from milk has been reported to carry the Le” determinant (Matsnmoto et al, 1982).None of the MCl antigens appeared to be expressed on neutrophil lactoferrin. since immunoblotting with a specific antiserum to lactoferrin produced bands at different positions from those produced by MCl (results not shown). The major glycoproteins of a Triton extract of neutrophil membranes identified by PAS staining of polyacrylamide gels are a group of proteins with molecular weights of around 165K and 105K. Proteins of the same molecular weights express the CD15 antigen. However, there is considerable heterogeneity in the carbohydrate chains associated with these glycoproteins since different populations of proteins with molecular weights in these ranges are recognized by different lectins and by the CD15 monoclonal antibody. Affinity chromatography on anti-CD15 antibody-Sepharose results in binding of only a small fraction of the 165K glycoprotein and even less 105K. However, as expected, all of the material reactive with anti-CDl5 antibody on immunoblotting was bound to the resin. Most of the 165K and 105K glycoproteins therefore do not express the antigen. In terms of the glycoproteins of the neutrophil cell surface, when judged by the intensity of staining of proteins from the membrane Triton extract after electrophoretic separation, the 165K and 105K glycoproteins are relatively abundant molecules. The fact that only a fraction express the CD15 antigen is consistent with our measurement of around 40 000 antigenic sites per cell by radiometric titration (McCarthy et al. 1985). Spooncer et aZ(1984) and Fukuda et a! (19 8 4 ) have shown by carbohydrate analysis that two of the major carbohydrate structures carried by neutrophil glycoproteins were a polyfucosylated lactosaminoglycan and a sialylated fucosyl-lactosaminoglycan both of which carried the 3-fucosyl N-acetylactosamine determinant recognized by the CD15 monoclonal antibodies. They reported that expression of these antigens was at least 3 times greater on glycoproteins than on glycolipids. Although they did not assign these chains to individual glycoproteins they did demonstrate by immunoblotting that glycoproteins of M, 130-1 70K. l O 0 K and a minor component of 8 0 K expressed the antigen, consistent with our own results. We have shown that the expression of the CD15 antigen is increased markedly upon stimulation of the neutrophil with the chemotactic peptide, fh4LP. The extent was similar to the increased expression of the complement receptor, CR3. Increased expression of both molecules is associated with marked increase in adhesiveness of the cells and marked susceptibility to aggregation. Analysis of the CD15 antigen following stimulation was made difficult by the even more marked susceptibility of the cells to aggregation following the addition of the anti-CD15 monoclonal antibody. Although a marked enhancement of expression of CD15
Neutrophil C D l 5 Antigen was observed for most experiments using cells from a large panel of donors, occasionally the staining intensity with MC1 seemed to decrease after fMLP stimulation, with a considerable proportion of cells showing only background levels of staining. In these cases, the apparent concentration of cells in the fMLP treated samples was found to be much reduced. Control experiments showed that incubation with the MC1 antibody can cause extensive aggregation of fMLP stimulated neutrophils (results not shown); consequently, the initially insignificant lymphocyte contamination (always < 10%) becomes an increasing proportion of the single-cell population. Since lymphocytes are not labelled by MC1 this could account for the increased proportion of unstained cells seen after fMLP stimulation. Data concerning the expression of CD15 on members of the LFAl family of proteins, including CR3, are conflicting. Harlan et a1 (1985) and Skubitz & Snook (1987) have reported that neutrophils from individuals genetically deficient in molecules of the LFAl family still express CD15. Furthermore, Hickstein et a1 (1986) have reported that the CD15 antigen did not cocap with CR3 when neutrophils were treated with anti-CR3 antibodies. However, immunoprecipitation studies by Melnick et ul(1985). Hickstein et a1 (1986) and by Skubitz & Snook (1987) have clearly demonstrated immunoprecipitation of CD15 antigens by anti-LFA antibodies. Hickstein et a1 (1986) showed that preclearing with anti-CD15 removed only a small fraction of each of the bands immunoprecipitated by the anti-beta chain monoclonal or by OKMl (an anti-CR3 alpha chain). In addition, preclearing with OKMl or an anti-beta chain monoclonal only minimally reduced the 170K and 1 1 0 K bands immunoprecipitated with CD15. Skubitz & Snook (1987), however, by a series of complex preclearing experiments (involving four absorption steps with each antibody), showed that the anti-CR3 alpha antibody absorbed all the CDl5 reactive 165K band and antiLFA beta absorbed all of the 165-185K band although neither absorbed all of the 1 0 5 K bands. Similarly, after seven preabsorptions with anti-CD15 all of the LFAl precipitable bands were apparently absorbed. These data were interpreted to show that all of the surface CD15 antigen was expressed on molecules of the LFAl family and that the repeated absorptions necessary to obtain this result was a reflection of the low affinity of the anti-CD15 monoclonal antibodies. Although these immunoprecipitation experiments suggest a heterogeneity in the expression of the CD15 antigen on members of the LFAl family of molecules, they cannot give any quantitative picture of the relative expression of the antigen on the alpha and beta chains or on other glycoproteins with the same molecular size. Our own results by the more direct technique of immunoblotting confirm the basic observations made in these studies, that CR3 beta and to a lesser extent alpha chain do express the CD15 antigen. However, in contrast to the results of Skubitz & Snook (1987). our own results suggest that CR3 can account for only a small proporion of the CD15 expression. Furthermore, only a small percentage of the CR3 molecules appear to express the CDl5 antigen. CR3 beta chain is common to a family of molecules expressed on all leucocyte types. The alpha chains are selectively expressed on different leucocyte types. The clear
lack of expression of the CD15 antigen on lymphocytes and possibly on monocytes shows differential glycosylation of the chains in different cell types. The function of the carbohydrate on these molecules remains unclear. Skubitz & Snook (1987) showed that the 165K and 105K glycoproteins immunoprecipitated with anti-CD15 monoclonal antibodies were far more resistant to proteolysis than the glycoproteins of the same molecular weight immunoprecipitated by antiLFAl beta chain monoclonals. We have confirmed these results (not shown).The carbohydrate might therefore play a protective role. It is also possible that the carbohydrate is involved in the control of cellular adhesion. The marked increase in adhesion and self-association of neutrophils after stimulation with fMLP has been attributed to the increased expression of CR3. We have now shown that CR3 cannot be the only protein that is brought to the surface on stimulation. Recently, Philips et a1 (1988) have shown that the increased adhesion of neutrophils resulting from stimulation by chemotactic peptides cannot be due to CR3 since increased adhesion occurred much more rapidly than upregulation of CR3. The glycoproteins expressing the CDl5 antigens are clearly candidates for the mediators of these adhesion reactions. ACKNOWLEDGMENTS We are grateful to Dr Nancy Hogg, Dr A. Law and Professor A. J. McMichael for the gift of monoclonal antibodies. This work was supported by grants from the Danish Science Research Council and the Scottish Home and Health department. M.A. was the recipient of a NATO Science Fellowship. We thank Mr R. Fawkes and colleagues for photography and assistance in the preparation of this manuscript. REFERENCES Albrechtsen. M., Yeaman. G.R. & Kerr. M.A. (1988) Characterization of the IgA receptor from human polymorphonuclear leucocytes. Immunology, 64,201-205. Anderson, C.L. & Looney, R.J. (1986) Human leukocyte IgG Fc receptors. Immunology Today. 7, 264-266. Anderson, L.C. & Gahmberg, C.G. ( 1 978) Surface glycoproteins of human white blood cells. Analysis by surface labelling. Blood, 52. 5 7-6 7. English, D. & Anderson, B.R. (1974) Single step separation of red blood cells, granulocytes and mononuclear leukocytes on discontinuous density gradients ofFicoll-Hypaque.lournal of Imrnunological Methods, 5, 249-254. Fujiki, Y., Hubbard. H.L., Fowler, S. & Lazarow, P.B. (1982) Isolation of intracellular membranes by means of sodium carbonate treatment: Application to endoplasmic reticulum. journal of Cell Biology, 9 3 , 97-103. Fukuda, M., Spooncer, E., Oates. J.E., Dell, A. & Klock, J.C. (1984) Structure of sialylated fucosyl lactosaminoglycan isolated from human granulocytes. Journal of Biological Chemistry. 259, 1092510935. Gahmberg, C.G. & Andersson. L.C. (1977) Selective labelling of cell surface sialoglycoproteins by periodate-tritiated borohydride. Journal of Biological Chemistry. 252, 5888-5894. Gooi, H.C.. Thorpe, S.J., Hounsell, E.F., Rumpold. H.. Kraft. D., Forster. 0. & Feizi. T. (1983) Marker of peripheral blood granulocytes and monocytes of man recognized by two monoclonal antibodies VEP 8 and VEP 9 involves the trisaccharaide 3 fucosyl-
Merete Albrechtsen and Michael A. Krrr
N-acetyllactosamine. Europeuri [ourrial of Zmrriuriologg, 1 3 . 306312. Hall. P.A. 8: Ardenne. A.J.D. 119871 The value of CD15 immunostaining in diagnosing Hodgkin's disease: a review of the published literature. lourrial oJ'Cliriica1 f'atliology. 40, 1298- I 304. Harlan. J.M.. Killen. P.D.. Senecal. FM.. Schwartxs. B.R.. Yee, E.K.. Taylor.R.F..Beatty.P.G..Price.T.H. &Ochs. H.D. (1985)Therole of neutrophil membrane glycoprotein GP150 in neutrophil adherence to endothelium in vitro. Blood. 66. 167-1 78. Hickstein. D.D.. Locksley. R.M.. Beatty. P.G.. Smith. A,. Stone. D.M. ti Root. R.K. (1986) Monoclonal antibodies binding to the human neutrophil C3bi receptor have disparate functional effects. Blood. 67. 1054-1062. Hildreth. J.E.K. & August. J.T. 11985) The human lymphocyte function associated (HLFA) antigen and related macrophage differentiation antigen iHMac- 1 1: Function. effects of subunitspecific monoclonal antibodies. Joiirrial oflrnm~mology,134, 32723280. Hildreth. J.E.K.. Gotch. F.M.. Hildreth. P.D.K. d Mchiichael. A,]. (19 8 3 ) A human lymphocyte associated antigen involved in cell mediated lympholysis. Europeuri journal of ltrtrriuriolog~~. 13. 202208. Hogg, N. & Horton. M.A. 11987) .Vyeloid antigens: new and previously defined clusters. I,eucoc!p Tgpitig I I I (ed. by A . 1. McMichael). pp. 5 76-602. Oxford 1;niversity Press. Huang. L.C., Civin. C.I.. Magnani. J.L.. Shaper. J.H. & Ginsburg. \'. 11983) My-1, the human myeloid specific antigen detected by mouse monoclonal antibodies is a sugar sequence found in lactoN-fucopentaose 111. Blood. 61, 1020-102 3. Johnstone. A. & Thorpe. R. I 1 9 8 2 ) Irrirriirnorft~~rnisrr~~ in Pracriru. Blackwelt Scientific Publications. Oxford. Kerr. M.A. & McCarthy. N.C. (1985)A carbohydrate differentiation antigen of granulocytes. brain and many tumours. Biocf~erriical Societg Transactions. 13. 424-426. Kniep. €3. & Muhlradt. K. (1987) Monoclonal antibodies in the myeloid panel recognising glycosphingolipids. kucoryte Tgping I I I (ed. by A. J. McMichael). pp. 619-62 I . Oxford L'niversity Press. Laeminli. U.K. ( 1 9701 Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Yatrrre. 227, 680685. Letarte-Muirhead. M.. Barclay. A.N. & Williams. A.F. I 19 i5 1 Purification of the Thy-1 molecule. a major cell surface glycoprotein of rat thymocytes. Biochrrriical Journal. 1 5 1 , 685-h95. Malhotra. V.. Hogg. N . & Sim. R.B. (1986) Ligand binding by the ~ 1 5 0 . 9 5antigen of 11937 cells: properties in common with complement receptor type 3 (CR3). European journal of Imniunoloyy. 16, 1117-1123. Matsumoto. A,, Yoshima. S.. Takasaki. S. & Kobata. A. ( 1 9 8 2 ) Structural study of the sugar chains of human lactoferrin: finding of four novel complex type asparagine linked sugar chains. [ourrial of Biochemistry, 91, 143-1 54, McCarthy. N.C., Albrechtsen. M.T. & Kerr. M.A. (1985)Characterization of a human granulocyte differentiation antigen (CDwli1 commonly recognised by monoclonal antibodies. Bioscierirv Reporrs. 5. 933-941. iMelnick. D.A.. Nauseef. W.M.. Markowitz. S.11.. Gardner. J.P. 8 Malech. H.L. ( 1985) Biochemicaf anafysisand subcellularlocalization of a neutrophil specific antigen. PMN-7. involved in the respiratory burst. journal of Immunoloyy. 1 34, 3 346- 3 3 5 5. Morrissey, J.H. ( 19x1 ) Silver stain for proteins in acrylamide gels: a modified procedure with enhanced uniform sensitivity. Annlgtical Biocfiemisfry, 1 1 7, 307- 3 1 3 . Philips. M.R.. Buyon. J.P.. Winchester. R.. Weissman. G. & Abramson. S.B. (19881 tipregulation ofthe iC3b receptor (CR3)is neither necessary nor sufficient to promote neutrophil aggregation. Jourrial of Clinical Investigation. 82, 4 9 5-501.
Potts. R.C.. Gibbs. J.H.. Robertson. A.1.. Brown, R.A. & Beck, J.S. ( 19801 A simple method for determining the extent of cellular contamination in peripheral blood lymphocyte preparations. joiirrrnl 01Imnirrnological Methods. 3 5 , 177-1 87. Roos. D.. Voetman. A.A. & Meerhof. L.J. (1983) Functional activity of enucleated human polymorphonuclear leucocytes. Journal of Cell Biology. 9 7 , 368-377. Ross, G.D. (1986) Opsonization and membrane complement receptors. Ir~im~tnobiology of the Complement Sgstem (ed. by G. D. Ross), pp. Xi-1 14. Academic Press. Orlando. Sanchtx-Madrid. F., Nagy. J.A., Robbins. E., Simon, P. & Springer, T.A. ( 198 3 I A human leukocyte differentiation antigen family with distinct a subunits and a common b subunit. The lymphocyte function-associated antigen (LFA-1). the C3bi complement receptor (OKMI/Macl) and the p l 5 0 , 9 5 molecule. Journal of Experimeritrtl Medicine. 158, 1785-1803. Sanders. U.S.A.. Kerr. M.A., Hopwood. D., Coghill, G. & Milne, G.A. ( 1988) Expression of the 3-fucosyl N-acetyl1actosamoine (CDl5) antigen in normal. metaplastic. dysplastic and neoplastic squamous epithelia. journal of Pathology. 154, 25 5-266. Llamas, L., Ree. H.J., Sheibani. K.. Ho. Y-S.. Su. 1-1. & Santamaria. M.. Hsu. S-M. ( 1988) Expression of sialylated Leu M1 in histiocytosisX. American ]ourrial of Cljnical Pathology, 89, 21 1-21 6 . Sim. R.H. & Walport, M.J. (1987) C3 receptors. Complement in Health nrrd Disease (ed. by K. Whaley). pp. 125-162. MTP Press. Lancaster. Skubitz. K.M.. Pessano. S.. Bottero, L.. Ferrerro. D.. Rovera. G. & August. J.T. (198 3) Human granulocyte surface molecules identified by murine monoclonal antibodies. Journulo~lmrriunalogy.1 3 I , 1882-1 888. Skubitz. K.M. & Snook. R.W. (1987) Monoclonal antibodies that recognize Lacto-N-Fucopentaose 111 (CDl5) react with the adhesion promoting glycoprotein family (LFA-l/HMac-I/ GP150.95) and CR1 o n human neutrophils. Journal of Immutiologll. 1 39, 1 6 3 1-1 6 39. Spooncer. E.. Fukuda, M.. Klock. J.C.. Oates. J.E. & Dell, A. (1984) Isolation and characterization of polyfucosylated lactosaminoglycan from human granulocytes. Journal of Biological Chemistry. 259. 4792-4801. Springer. T.A.. Thompson. W.S.. Miller. L.J.. Schmalstieg, F.C. & Andersson. D.C. ( 1984) Inherited deficiency of the Mac-1, LFA-I , p l i 0 . 9 5 . glycoprotein family and its molecular basis. Journal of Experimental Medicine. 160, 1901-1918. Tetteroo. P.A.T.. Mulder. A,. Lansdorp. P.M., Zola. H.. Baker, D.A., Visser, F.J. & von dem Borne. A.E.G. Kr. (1984)Myeloid associated antigen. 3 fucosyl N-acetyllactosamine (FAL);location o n various granulocyte membrane glycoproteins and masking upon monocytic differentiation. European Journal of Immunology, 14, 10891095. Todd. A.F., Arnaout. M.A.. Rosin, R.E., Crow1ey.C.A.. Peters, W.A. & Babior. B.M. (19 8 4 ) Subcellular localisation of the large subunit of Mol (Mola; formerly g p l 1 0 ) a surface glycoprotein associated with neutrophil adhesion. Journal of Clinical Investigation. 74, 1280-1290. Towbin. H.. Staehelin. T. &Gordon,J. (1 979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proceedirigs of tire National Acndemy ojSciences ojthe United States of America. 7 6 , 4 3 50-43 57. van der Schoot. C.E.. Visser. F.J., Bos, M.J.E.. Tetteroo, P.A.T. & von dem Borne, A.E.G. K. ( 1 9 8 7 ) Serological and biochemical characterization of myeloid specific antibodies of the myeloid panel. Leucocyre Typing I I I (ed. by A. J. McMichael). pp. 603-61 1. Oxford hiversity Press.