From the Beatrice and Samuel A. Seaver Laboratory, Division of ...... Ding, A. H., Porteu, F., Sanchez, E., and Nathan, C. F. (1990) J. Exp. Med. Med. 163,221- ...
THEJOURNAL IC)
OF
BIOLOGICAL CHEMISTRY
Vol. 266, No. 28, Issue of October 5, pp. 18846-18853,1991 Printed in U.S.A.
1991 hy The American Society for Biochemistry and Molecular Biology, Inc
Human Neutrophil Elastase Releases a Ligand-binding Fragment from the 75-kDa Tumor Necrosis Factor (TNF) Receptor COMPARISONWITHTHEPROTEOLYTICACTIVITYRESPONSIBLEFORSHEDDINGOFTNF RECEPTORSFROMSTIMULATEDNEUTROPHILS* (Received for publication, April 22, 1991)
Franqoise PorteuSS, Manfred Brockhausv, DavidWallachII , Hartmut Engelmann[[, and Carl F. Nathan** From the Beatrice and Samuel A. Seaver Laboratory, Division of Hematology-Oncology, Department of Medicine, Cornell University Medical College, New York, N e w York 10021, Whe Central Research Units, F. HoffmannLa Roche Ltd., CH-4002 Basel, Switzerland, and11 the Department of Molecular Genetics and Virology, The Weizmann Instituteof Science, Rehovot 76100, Israel
To localize the protease(s) involved in shedding of Tumor necrosis factor a (TNF)’ is a cytokine released by tumor necrosis factor receptors (TNF-R) from actimonocytes, macrophages, T cells, NK cells, mast cells, and vated neutrophils (PMN) (Porteu, F., and C. Nathan neutrophils (PMN) in response to infection, antigen, or injury (1990) J. Exp. Med. 172, 599-607), we tested subcel- (1-3). Discovered for its ability to induce hemorrhagic necrolular fractions from PMN for their ability to cause loss sis of tumors, T N F has more recently been implicated as a of TNF-R from intact cells. Exposure of PMN to soni- mediator of inflammation, septic shock, andcachexia (1).Its cated azurophil granulesa t 3 7“C resulted in inhibition profound effects onneutrophils may playarolein these of lZ5I-TNF binding; 50%inhibition ensued when PMN responses. T N F increases PMN phagocytic activity and cywere treated for1min with azurophil granules equivtotoxicity (4, 5), promotes their adhesion to endothelialcells alent to 2-3 PMN per indicator cell. The TNF-R-de- (6), and allows PMN adherent to biological surfaces t o degrading activity in azurophil granules was identified granulate (7) andundergo a prolonged respiratory burst (8). as elastase by its sensitivity to diisopropyl fluorophosBiological responses to T N F are initiated by its binding to phate (DFP), al-antitrypsin and N-methoxysuccinylspecific cell surface receptors (TNF-R). In binding studies,a Ala-Ala-Pro-Val chloromethyl ketone(MSAAPV-CK), single class of TNF-R with Kd ranging from lo-’ to 5 X lo-” and by the abilityof purified elastase to reproduce the M has been identified on all nucleated cell types analyzed, effect of azurophil granules. Elastase preferentially including PMN (1, 9-11). Two distinct TNF-R havebeen acted on the 75-kDa TNF-R, reducing by 85-96% the isolated (12, 13) and their genes cloned (14-20). These TNFbinding of lZ5I-TNF to mononuclear cells expressing R, named A (type 11) and B (type I), both seem biologically predominantly this receptor, while having no effect on active (21-23). However, they differ in M,, immunoreactivity, endothelial cells expressing almost exclusively the 55- and distribution. The type A TNF-R is a glycoprotein of -75 kDa TNF-R. Elastase-treated PMN released a 32-kDa kDa and is themajor TNF-R species found on the surface of soluble fragment of p75 TNF-R that bound TNF and several cellsof myeloid origin including the humanhistiocytic reacted with anti-TNF-R monoclonalantibodies. In lymphoma line U937 and blood monocytes, whereas the 55contrast, fMet-Leu-Phe-activatedPMN shed a 42-kDa kDa type B TNF-R seems to be expressed preferentially on fragment from p75 TNF-R, along with similar amounts epithelial cells (12, 13, 22-24). Based on their susceptibility of a 28-kDa fragment from p55 TNF-R. Shedding of to inhibition of “’I-TNF binding by anti-type A or type B both TNF-Rsby intact activatedPMN was more extenTNF-R monoclonal antibodies (mAbs), PMN are believed to sive than shedding caused by elastase and was comdisplay similar amounts of the two TNF-R types on their pletely resistant to DFP and MSAAPV-CK. Thus, the surface (12). On cells other than PMN, the expression of TNF-R-releasing activity of azurophil granules is disreceptors A and B can be regulated independently (17, 23). tinct from that operative in intact stimulated PMN and could provide an additionalmechanism for the control Soluble ligand-bindingforms of TNF-R A and B of M , -30,000 have been isolated from urine and serum (24-26). of cellular responses to TNF at sites of inflammation. Inanearlierstudy we showed thatwithinminutes of exposure to the chemotactic factors Wet-Leu-Phe (fMLP) and C5a, PMN release their surface TNF-R (11).Later studies (18-20, 27) showed that the two types of TNF-R could be
-
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $.Permanent address:INSERM U90, Hopital Necker, Paris, France. To whom correspondence should be addressed: Cornell University Medical College, Box 57, 1300 York Ave., New York, NY 10021. Tel.: 212-746-2986; Fax: 212-746-8536. ** Supported by National Institutes of Health Grant CA45218.
The abbreviations used are: TNF, tumor necrosis factor a;TNFR, T N F receptors;PMN,neutrophils;MNC,mononuclear cells; HUVEC, human umbilical vein endothelial cells; fMLP, formylmethionylleucylphenylalanine; MSAAPV-CK, N-methoxysuccinyl-AlaAla-Pro-Val chloromethyl ketone; TPCK, N-tosyl-L-phenylalanine chloromethyl ketone; TLCK, N“-p-tosyl-L-lysine chloromethyl ketone; DFP, diisopropyl fluorophosphate; mAb, monoclonal antibody; PBS, phosphate-buffered saline; FCS, fetal calf serum; SDS, sodium dodecyl sulfate; PAGE,polyacrylamide gel electrophoresis; Pipes, 1,4piperazinediethanesulfonic acid EGTA, [ethylenebis(oxyethy1enenitrilo)ltetraacetic acid.
18846
T N F Receptor-releasing Activitiesof Human Neutrophils
18847
monitored by measuringlightscattering a t 450 nm. The pooled fractions were centrifuged a t 180,000 x g for3h a t 4 "Cin an ultracentrifuge(BeckmanInstruments) to removePercoll, resuspended in relaxation buffer, frozen in aliquots by dipping in liquid nitrogen, and stored at-70 "C until use. In some experiments, isolated PMN were resuspended in KrebsRinger phosphate buffer with glucose (KRPG; pH 7.4, 300 mOSM), separated in two equal batches, and incubated for 10 min a t 37 "C with M fMLP or buffer alone,prior to disruptionandfractionation. Assays for Marker Enzymes and Protein Subcellular in FractionsSamples of Percoll gradient fractions and of pooled fractions were prepared for enzyme activity analysis by dilution in phosphate buffered-saline (PBS) containing 0.1% Triton X-100. Alkaline phosphataseandvitamin B1,-binding proteins were used as markers for plasma membrane andspecific granules, respectively, and were measured as described (32). The distribution of azurophil granules was determined by measuring myeloperoxidase by the colorimetric method of Andrews and Krinsky (33). Protein determination was carried out by the method of Lowry et al. (34) in the presence of 0.1% sodium dodecyl sulfate (SDS) to prevent interference of Triton X100 in the assay (35) and usingbovine serum albumin asa standard. Treatment of Cells with Subcellular Fractions of P M N a n dPurified Proteases-Freshly isolated PMN and MNC were resuspended a t 107/ml in KRPG containing 0.2% heat-inactivated fetal calf serum (FCS, Hyclone Laboratories, Logan, UT) and incubated in 1.5-ml polypropylene microcentrifuge tubes (Brinkman Instruments Co.) a t 37 "C with various concentrations of plasma membrane- or granuleenriched fractions, purified elastase, cathepsin G , or medium alone, while rotating end-over-end. Aftera 1-to 40-min incubation, thecells MATERIALS ANDMETHODS were washed twice with cold KRPG, resuspended in binding buffer Reagents-Pure recombinant human TNF (5.6 X lo7 units/mg) (KRPG containing 5% FCS), and testedfor TNF-R. In some experproduced in Escherichia coli was a gift of Genentech (South San iments, the reaction was stopped by adding DFP (4 mM final) and Francisco, CA). Iodinated T N F ( T I - T N Fwas ) prepared asdescribed incubating thecells 15 min onice before washing and bindingassays. (11)and had a specific activity of 200-400 Ci/mmol monomer. Inexperiments designed to characterizetheenzymatic activities, Purified elastase and cathepsin G from human leukocytes were azurophil granule-enriched fractions or, as a control, purified propurchased from Calbiochem. Lyophilized enzymes were dissolved in teases were preincubated with various protease inhibitors for 15 min distilled waterat 1 mg/ml, storedat -70 "C, and thawed immediately a t 4 "C followed by 10 min at room temperature before addition to before use. fhlLP, phorbol 12-myristate 13-acetate, the calcium ion- the cells. ophore A23187 (stored a t -70 "C as concentrated stock solutions in To test the effect of proteases on HUVEC,monolayers of adherent dimethyl sulfoxide), N-methoxysuccinyl-Ala-Ala-Pro-Valchloro- cells in 24-well plates were overlaid with 300 p1 of minimum essential methylketone(MSAAPV-CK),N-tosyl-L-phenylalaninechloromedium-a (Data Packaging, Cambridge, MA) with or without PMN methylketone(TPCK), N"-p-tosyl-L-lysine chloromethylketone azurophilgranules,or purified elastase, and incubated 30 min at (TLCK), leupeptin, and DFPwere from Sigma. al-antitrypsin was a 37 "C. The reaction was stopped by incubating with 4 mM DFP for generous gift of Dr. Nyoun So0 Kwon (Cornell University Medical 15 min a t 4 "C. "'1-TNF binding was measured after three washes College) and was purified as described (29). with medium. [57Co]Cyanocobalamin (100-300 pCi/pg) and "'I-streptavidin (20Binding Assays-Binding of "'I-TNF to suspensions of PMN and 40 pCi/pg) were purchased from Amersham Corp. of adherent HUVEC (36)was measured MNC (11)and to monolayers Antibodies-Purified mouse IgG mAbs specific for the p55 TNF-R exactly as described. In competition experiments thecells were prein(Htr-9, Htr-20, and clone 20) or the p75 TNF-R (Utr-1, Utr-4, and cubated for 30 min a t 4 "C with saturating concentrations (10pg/ml) clone 32) were prepared against membrane (12) orsoluble forms (22) of mAb directed against the p75 or the p55 TNF-R before addition of TNF-R receptors as described. Purified mouse IgG2a mAb IB4 of '%TNF. For binding of mAbs, PMN or MNC (1 X 106/ml in 100 (anti-CD18) was a gift of Dr. Samuel Wright, the Rockefeller Uni- pl of KRPG, 10% FCS) were incubated with 10 pg/ml of either antiversity, New York. Purified mouse IgG was from Pierce ChemicalCo. CD18 mAb IB4 or control mouse IgG for 1 h a t 4 "C, washed twice Sheep anti-mouse IgG antibody conjugated to biotin was purchased with KRPG, 10% FCS, and reacted for 30 min a t 4 "C with biotinfrom Amersham Corp. conjugated sheep anti-mouse IgG antibody (1:200 final dilution).After Cells-PMN and MNC were isolated from heparinized blood of two washes, bound antibodies were detected by incubation with '% healthy adults by centrifugation on Neutrophil Isolation Medium streptavidin (-100,000 cpm) for 30 min at 4 "C. The cells were then (Los Alamos Diagnostics, Los Alamos, NM) as described (11). Second washed three times with cold saline and the pellets assayed for protein passage human umbilical vein endothelial cells (HUVEC) were the and radioactivity as described (11). gift of Dr. Eric Jaffe (Cornell University Medical College) and preGel Electrophoresis, Ligand Blot, and Western Blot Analysis-PMN pared as described (30). were incubated for 15 min at 37 "C with M L P , proteases, granuleSubcellular Fractionation of Neutrophils-PMN plasmamemenriched fractions, or buffer alone, as described above, and centribrane- and granule-enriched fractions were prepared by centrifuga- fuged a t 12,000 X g for 5 min at 4 "C. The cell-free supernatants were tion on Percoll density gradients as described by Borregaard et al. harvested and resolvedby SDS-polyacrylamide gel electrophoresis (31). Isolated PMN were resuspended at 2-5 X 107/ml in ice-cold (SDS-PAGE)accordingtothe procedure of Laemmli(37)under relaxation buffer containing 100 mM KCl, 3 mM NaCl, 1 mM ATP, nonreducing conditions. The proteinswere transferred to nitrocellu3.5 mM MgCI2, 10 mM Pipes, pH 7.3, anddisrupted by nitrogen lose membranes and probed with"'I-TNF in the presence or absence cavitation for 20 min at 400 p. s. i. and at 4 "C. The cavitate was of excess unlabeled TNF, as described (11).For Western blotting collected into a tube containing EGTA, pH 7.4 (1.25 mM final) and with anti-TNF-R mAb, the nitrocellulose sheets were blocked with centrifuged at 500 X g for 10min a t 4 "C to removenuclei and PBS containing 5% non-fat dry milk and 0.05% Tween 20 for 3 h at unbroken cells. Subcellular fractions were isolated by layering the room temperature and then incubated overnight with mAbs diluted post-nuclearcavitate over discontinuousPercolldensitygradients 1:2000 in the same buffer. After three washes the membranes were and centrifuging for 20 min a t 48,000 X g a t 4 "C. Fractions of 1 ml incubated for 1 h a t room temperature with biotinylated sheep antiwere collected, assayed for specific markers as decribed below, and mouse IgG antibody (1:4000), washed three times, and reactedfor 20 pooledaccordingly intoplasmamembrane-, specificgranule-, and min with '"I-streptavidin (0.01 pCi/ml) diluted in PBS containing azurophilgranule-enrichedfractions. The positions of thepeaks 3% bovine serum albumin. The membranes were then washed four corresponding to the different subcellular compartments were also times in PBS, dried, and autoradiographed.
shed by other types of cells upon activation. The mechanisms controlling TNF-R shedding have not been elucidated. The integral membrane protein nature of both p55 and p75 TNFRs and the absence of transcripts encoding soluble forms of these receptors (14-20) strongly suggest that shedding of TNF-Rs results from proteolysis of their extracellular domains. PMN contain numerous proteases in their lysosomes and plasma membrane which participate in hydrolysis of extracellular proteins and cell surface receptors (28). To localize the proteolytic activities involved in TNF-R shedding, we have fractionated resting and activated PMN and tested the various fractions for theirability to down-regulate TNF bindingand cause the release of antigenic and ligand-binding TNF-R from intact cells. We describe here the co-localization of a TNF-R-releasing activity to azurophil granules and its identification as elastase. This enzyme preferentially acts on the p75 TNF-R of PMNand mononuclear cells (MNC), releasing a soluble fragment of 32 kDa that retains its ability to bind both TNF and anti-TNF-R mAbs. By its selectivity for one type of TNF-R, the size of the fragments released, and its sensitivity to diisopropyl fluorophosphate (DFP), this proteolytic activity is distinct from that operative in intact fMLP-stimulated PMN.
TNF Receptor-releasing Activities of Human Neutrophils
18848
disruption and fractionation. In this case, azurophil granule-, specific granule-, and plasma membrane-enriched fractions contained 78 f 4% of the myeloperoxidase, 90 f 5% of the vitamin BI2-binding protein, and 96 f 3% of the alkaline phosphatase, respectively (three experiments). No decrease in the contentof vitamin BI2-binding protein, in the specific granule peak, or myeloperoxidase, in the azurophil granule peak, was observed after activation of PMN, in agreement with previous studies (38) showing that in the absence of cytochalasin B, fMLP does not induce degranulation of these compartments from PMN insuspension. This stimulationis, however, sufficient to induce complete loss of surface TNF-R and to activate the proteolytic process responsible for shed0 i io is i5 ding of TNF-R (11). The fractions from Percoll gradients were pooled, centriFraction number fuged to remove Percoll, resuspendedby sonication, and tested FIG. 1. Distribution of marker enzymes in disrupted PMN fractionated on discontinuous Percoll gradients.Each fraction for their ability to decrease 9 - T N F binding to fresh PMN. As shown in Table I,exposure of PMN to the azurophil (1 ml) was assayedformyeloperoxidase (B), vitaminBls-binding granule pool from resting PMN at37 "C inhibited the subseprotein (0),and alkaline phosphatase (A),as specific markers for azurophil granules, specific granules, and plasma membrane,respec- quent binding of lZ5I-TNFa t 4 "C by 66 f 8%, ( n = 7 tively. (experiments)). The specific granule pool induced a smaller decrease in "'1-TNF binding (28 f 13%,n = 7), whereas the Assay for Soluble TNF-R-Soluble TNF-Rs were quantified incell- plasma membrane (TableI)and cytoplasmic pools (not free supernatants from PMN treated with proteases or fMLP by a shown) had no effect. No difference was observed using subspecific radioimmunoassay described indetail elsewhere.' Briefly, cellular fractionsprepared from PMNstimulated with M polyvinyl 96-well plates were coated with affinity-purified rabbit antifMLP before fractionation (Table I). murine IgG and incubated overnightat 4 "Cwith anti-p75 TNF-Ror Concentration-response studies showed that azurophil anti-p55 TNF-R mAb unable to block T N F binding, diluted in Trisbuffered saline containing 1% non-fat dry milk and 5 mM EDTA. granules were a potent inducer of TNF-R down-regulation The wells were washed with PBS, and PMN supernatant samples (2 (Fig. 2): 50% decrease in TNF binding followed exposure of X 10" cell equivalents) were added. After 3 h a t 4 "C, the plates were 3.5 X lo6 PMN for 40 min at 37 "C to -2.5 X lo6 cell washed in PBS and the TNF-R detected by incubation with "'I-TNF equivalents of this fraction. Maximal reduction in Iz5I-TNF for 2 h at 4 "C.The assay was calibrated by using purified recombinant binding was reached a t a ratio of 2-3 PMN equivalents of p55 TNF-R. granules to one PMN indicator cell (Fig. 2). However, inhibition was never complete; 25-40% (mean, 34 f 8%) specific RESULTS bindingremained on the cells even when theamount of Effects of Subcellular Fractions from Resting and Activated granules was increased further (Fig. 2) or the incubation time P M N o n T N F - R Expression-In a previous study (11) we prolonged for up to 1.5 h a t 37 "C (not shown). In contrast to theeffect observed with azurophil granules, demonstrated that human PMN rapidly shed their surface TNF-R when activated in suspension by a variety of stimuli the ability of specific granule-enriched fractions to downand suggested that thisrelease probably results from selective regulate TNF-R disappeared rapidly upon dilution (Fig. 2). proteolytic cleavage of the extracellular portion of TNF-R. In TABLE I an attempt tolocalize and identify the proteases involved in Effect of subcellular fractions fromresting or fMLP-actiuated PMN this reaction, we have fractionated both resting PMN and on "'I-TNF binding to PMN PMN activated under conditions leading to TNF-R shedding TNF specific binding and tested the subcellular fractions for their ability to induce Treatmentb TNF-R down-regulation at thesurface of fresh indicatorcells. Exp. 1 Exp. 2 Treated cells were washed twice prior to thebinding assay,so cprn/100 pgprotein that shed TNF-R could not interfere with the binding assay 6,696 f 77 6,810 f 384 Control by competing for '"I-TNF. Resting PMN fractions As shown in Fig. 1 for restingPMN,thefractionation 5,730 f 515 6,410 _t 242 Plasma membrane procedure resulted in efficient separation of three compartSpecific granules 5,320 f 40 4,460 f 49 -DFP ments. In three separateexperiments, 79 f 6%, 96 f 1%,and ND 7,040 f 195 +DFP 98 f 2% (mean f S.D.) of the myeloperoxidase, vitamin B12Azurophil granules bindingprotein and alkaline phosphatase were recovered, 2,690 f 226 1,924 k 146 -DFP respectively, in azurophilgranule-, specific granule-, and ND 7,408 f 99 +DFP plasma membrane-enrichedfractions. Azurophil granules Activated PMN 6,525 f 390 6,176 k 228 showed no contamination by markers of specific granules or Plasma membrane Specific granules plasma membrane, and no alkaline phosphatase was found 4,035 f 230 3,639 k 56 -DFP associated with specific granules. However, the specific gran7,252 -C 327 ND +DFP ule fraction showed some contamination by azurophil granAzurophil granules ules, in that 16 & 3% myeloperoxidase was recovered in the ND 2,048 f 175 -DFP second peak (Fig. 1). 7,277 & 286 ND +DFP A similar profile was obtained with PMN which had been ' Mean f S.D. for triplicates. stimulated for 10 min a t 37 "C with M fMLP prior to PMN (3.5 X lo6) were incubated for 40 min a t 37 "C with 2 X lo7
,1
W. Digel, F. Porzsolt, W. Lesslauer, and M. Brockhaus, manuscript submitted.
cell equivalents of each fraction, were or were not pretreated with 4 mM DFP, washed, and tested for "'11-TNF binding. Not determined.
18849
TNF Receptor-releasing Activities of Human Neutrophils
01 0
I
I
I
I
I
S
10
1s
20
2s
Granules (cell equivalents x l@)
FIG. 2. Concentration-dependent effect of azurophil- and specific granule-enriched fractions on the binding of ' T - T N F to PMN. PMN (3.5 X lo6) were incubated for 40 min at 37 "C with the indicated amounts of azurophil granules (0)or specific granules (O), before binding of "'1-TNF was measured at 4 "C. The results are mean f S.E. of triplicates. When error bars are not seen, they fall within the symbols.
Approximately 10 times more cell equivalents of specific granules than of azurophil granules were required to induce a 50% decrease in lZ5I-TNFbinding. This suggests that the activity presentin the specific granule fractions could be attributed to the12-20% contamination with azurophil granules, as detected by the myeloperoxidase assay (see Fig. 1). The decrease in TNF binding induced by both types of granules was completely prevented by pretreatment with the serine esterase inhibitor DFP (Table I). This suggested that serine proteases were responsible for the TNF-R down-regulating activities in both the azurophilic- and specific granuleenriched fractions. The complete inhibition of TNF-R downregulation by DFP also ruled out that the reduction in lz5ITNF binding could be due to competition with T - T N F for binding to its receptors. Kinetic Studies of TNF-R Modulation by Azurophil Granuleenriched Fractions-The decrease in binding of lZ5I-TNFto PMN induced by azurophil granules was maximal after -2 min at 37 "C (Fig. 3). These fractionswere also active at 4 "C, but longer incubation times were required (Fig. 3). In these experiments, after each incubation at 4 or 37 "C, the activity of azurophil granules was promptly stopped by addition of DFP before performing the binding assay at 4 "C. Selectivity of the Effect of Azurophil Granule-enrichedFractions on TNF-R-Down-regulation of TNF-R induced by azurophil granules did not reflect a generalized alteration of plasma membrane receptors, in that conditions which led to a decrease in '2,51-TNFbinding had no effect on theexpression of CD18 antigen as assessed by binding of mAb IB4 (Table 11). Nature of TNF-R Degrading Activity inAzurophil Granules and Effects of Purified Proteases and of Protease ZnhibitorsThe azurophil granules of PMN contain acid as well as neutral proteases (28, 39). The rapidity and potency of their effect at neutral pH and the complete inhibition of TNF-R downregulation by DFP strongly suggested the involvement of a neutral serine esterase. This prompted us to test theeffect of two known neutral serine esterases of azurophil granules of human PMN: elastase and cathepsin G (28, 39). As shown in Fig. 4, both elastase and cathepsin G induced a decrease in "'I-TNF binding at the surface of PMN in a concentrationdependentmanner.Elastase was much more potent than cathepsin G; 6 pg/ml elastase caused 50% reduction in TNF
20
0
I
I
I
I
I
I
5
10
15
20
25
30
T 35
Incubation time ( m i d FIG. 3. Time course of down-regulation of TNF-R on PMN by azurophil granules at 37 and 4 OC. PMN (3.5 X lo6) were incubated for the indicated times at 37 "C (0)or at 4 "C (0)with lo7 cell equivalents azurophil granule-enriched fractions or buffer alone. The reaction was stopped by adding DFP (4mM), and binding of '"1TNF was measured at 4 "C. The results are expressed as percent of specific binding seen with cells treated with buffer alone (5243 +- 530 cpm/100 Kg of protein) and are mean f S.E. of triplicates.
TABLEI1 Effect of azurophil granule-enriched fraction onthe expression of TNF-R and CD18 antieen on PMN Specific binding Treatmentb TNF
mAb anti-CD18
cpm/lOO pg protein
7,252 f 247 61,497 +- 7,404 Control 65,615 k 5,065 3,294 f 360 Azurophil granules Mean k S.D. for triplicates. PMN (3.5 X lofi)were incubated for 15 min at 37 "C with either buffer alone or azurophil-enriched fractions (lo7 cell equivalents) before binding of '"I-TNF and mAb IB4 as described under "Materials and Methods."
binding in 30 min at 37 "C, whereas >30 pg/ml cathepsin G was required for the same effect. Moreover, only elastase decreased TNF-R at4 "C (Fig. 4). This feature, together with the inability of elastase to cause complete TNF-R downregulation (range 60-75% maximal inhibition), simulated the action of azurophil granules themselves and suggested that elastase is the major enzyme responsible for theTNF-R decreasing activity detected in subcellular fractions of PMN. The major role of elastase was confirmed by testing the effects of protease inhibitors on the ability of azurophil granules to decrease '2sI-TNF binding. In addition to DFP, downregulation of TNF-R induced by azurophil granules was completely prevented by the specific elastase inhibitors q-antitrypsin and MSAAPV-CK (Table 111).The specificity of these reagents for elastase is shown in Table IV: they prevented TNF-R modulation induced by purified elastase, while having no effect on the decrease in '"I-TNF binding caused by leupeptin, and phencathepsin G. In contrast, TLCK, TPCK, anthroline, respectively inhibitors of trypsin, chymotrypsin,
TNF Receptor-releasing Activities of Human Nputrophils
18850 6000
n
, 6
, 10
, 15
. 20
. 25
. 90
S5
FIG. 4. Effect of p u r i f i e d e l a s t a s e a n d c a t h e p s i nG o n b i n d i n g o f""I-TNF to I'MN. I'MN were incuhated for 30 min at 37 "C ( / i / / d .s,vrnho/s) or at 4 " C (open qvrnhols) with the indicated concent rat inns of elastase ,.( 0 )o r cathepsin G (0,O), washed, a n d assayed Ibr '.'.'I-TNF hinding. Results are means S.E. for triplicates.
*
Inhihitor"
'."T-TNF hindin$
"I'MN (3.5 X 10") were incuhated for 15 min at 37 "C with nzurophil granules (6.5 X 10"cell equivalents), pretreated or not with the indicated protease inhihitors. The reaction was stopped by DFP followetl by washes and T N F hinding assays. Ethanol (0.5%),which was used as the diluent. for MSAAV1'-CK and TI'CK, had no effect o n granule action. " Results are expressed as a percent of specific binding seen with cells treated for with hul'fer alone, which averaged 12,619 ? 756 cpm/ 100 p g of protein and are mean ? S.D.for triplicate determinations.
TARI.E IV of purified neutrophil serine f,stcrnscs, in the presence or nhscJncr,of r/nstn.sc inhihitors. on thc nhilitv of P M N t o hind ''.'I-TNF Effect
" Purified proteases were preincuhated with the indicated inhihitor o r the diluent alone for 15 min at 4 "C and then I O min a t 20 "C
before addition to the cells. After a 20-min incuhation at 37 "C, the reaction was hlorked with DFI' and "'"I-TNF hinding assays performed. "'I'NF specific hinding is given as a percent of specific hinding seen with cells treated for 2 0 min a t 37 "C with huffer alone, which overaged 7,428 :1,603 cpm/l00 p g of protein. Results are mean ? S I ) . for the numher of experiments indicated in parentheses.
*
thiol,andmetalloproteases,wereineffective in preventing TNF-R down-regulation hv azurophil granules fTahle111 antl data not shown). ElastasePrpfrrrntiallyInduccs thr Rrlrnsr of n Soluhlr Fragmmt from p7.5 TNF-X (Its Action I s I)i.stinc! f r o m T,VI.'R Shpdding upon Activation of I'MN with fMI,I')-To tletermine whether serine esterases fromP M N were able to induce the release of soluhle fragments of TNF-R, cell-free supernatants from P M N treated with azurophil granules, purified elastme, or buffer alone were suhjectedto SI)S-PAGE, transferred to nitrocellulose, and probed with either ':'.'I-TNF,a n t i p75 TNF-R, or anti-p55 T N F - R mAh. Recause the terminology of TNF-Rs is confusing (t.ype A = t?.pe 11, mAhs U t r - I and clone 32; t.ype R = t,ype I. mAhs Htr-9 and clone 20), in t h e following discussion we refer t o the TNF-Rs as p7.5 (t.ype A) and p55 (type B ) , even though their soluhle proteolytic fragments of course have smaller M , values. In s u p e r n a t a n t s from granule-or elastase-treated PMMN, anti-p75 TSF-R mAh reacted specifically and strongly with a 32"kDa protein (Fig. 5.4, lanes 9 and 4 ) . whereas anti-p.55 TNF-11 mAh reacted only very faintly wit,h a diffuse hand at 28-35 k I h (Fig. 5 A , Innm 7 and 8).Of these species. only small amounts of the 32-kDa fragment of p7.5 TNF-R were detected in supernatants from huffer-treated cells (Fig. SA, lanrs I and 5 ) . S o n e of the fragments reacted with control antibodies f Fig. SA, lnncs 911). The smearsof radioactivity in a high M, range seen with supernatants from granule-treated PMN after blotting with either '"-'I-TNF, anti-TNF-1%mAhs, o r control mouse IgG probably represent nonspecific trapping of radioart ive material by the large amount of granule proteins present in these supernatant,s and loaded on the gel. Only the 2R-kl)a soluble fragment of the p55 T N F - R specifically bound "I-TXF under the conditions of the ligand Mot experiments (Fig. SA, lnnrs 13-1.5). No "'.'I-TNF hinding protein correspondingto the 32-
18851
T N F Receptor-releasing Activitiesof Human Neutrophils kDa soluble fragment of p75 TNF-R was detected in ligand blots, even when autoradiography was prolonged from 4 days (Fig. 5) to 2 weeks (not shown). However, in an immunoassay, in which TNF-R molecules trapped by a mAb unable to block the TNF binding site were detected with "'I-TNF, the 32kDa soluble fragment cleaved from p75 TNF-R by elastase was able to bind lZ5I-TNF(1203 f 28, 1212 f 13, and 170 f 17 cpm bound in supernatants from PMN treated with elastase, granules, or buffer alone, respectively). This suggests that the soluble fragment of p75 TNF-R loses its ligandbinding activity after SDS-PAGE and transfer to nitrocellulose. We have shown previously (11)that fMLP triggers PMN to release a soluble 28-kDa protein that binds '"I-TNF specifically in ligand blots. The TNF-Rfragments shed in those conditions were further characterized here by Western blotting with anti-p75 TNF-R and anti-p55 TNF-R mAbs and compared with the species released upon exposure of the cells to elastase. The patterns of soluble TNF-R found in supernatants from fMLP- or elastase/granule-treated PMN were very different (Fig. 5). fMLP induced the release of large amounts of fragments derived from both TNF-R types, as shown by the strong signals observed with mAbs and with '"I-TNF. The soluble form of the p55 TNF-R appeared on both ligand blots and Western blots as a diffuse band of 2835 kDa (Fig. 5A, lunes 6 and 13), much more intense than that observed in elastase-induced supernatants. By contrast, fMLP induced the release of a very small amount of the 32kDa fragment reacting with anti-p75 TNF-R mAb. Instead, fMLP triggered the shedding of p75 TNF-R in a major form of42 kDa (Fig. 5A, lane 2 ) . The 42-kDa fragment retained the ability to bind "'I-TNF in the immunoassay (1113 f 54 cpm bound). The small amounts of the 32-kDa fragment of p75 TNF-R that could be detected in media of fMLP-activated PMN or control cells (Fig. 5A, lanes 1 and 2 ) were no longer visible when PMN were pretreated with DFP prior to activation (Fig. 5B). By contrast, DFP did not affect shedding of the 42-kDa form of the p75 TNF-R by fMLP-activated PMN (Fig. 5B). This suggests that the 32-kDa fragment is generated spontaneously by an elastase-like activity, present at the surface of PMN or released in very small quantities from the cells upon incubation at 37 "C. In contrast, the DFP-resistant protease activated by fMLP that leads to shedding of the 42-kDa fragment of p75 TNF-R is clearly distinct from elastase. Judging from the intensity of the signals obtained by using the anti-p75 TNF-R to blot the supernatants from either fMLP- or granule/elastase-treated PMN (Fig. 5A, compare lane 2 with lanes 3 and 4 ) , these agents appear to trigger the release of similar amounts of fragments of p75 TNF-R. This contrasts strikingly with what is observed for p55 TNF-R. Whereas fMLP induces the release of similar amounts of fragments of p75 and p55 TNF-Rs (Fig. 5A, compare lane 2 and lane 6 ) , p55 TNF-R appears to be relatively resistant to elastase. This is reflected in the weak intensity of the 28-kDa band detected by ligand blots or Western blots with anti-p55 TNF-R mAb in supernatants from elastase-treated PMN (Fig. 5A, compare lane 6 with lunes7-8, and lane 13 with lanes 14-15). Theseresults were confirmed by measuring soluble TNF-R fragments in a radioimmunoassay (not shown). Effect of Azurophil Granule-enriched Fractions and Purified Elastase on TNF-R of Cells Other than PMN-TNF-R on MNC was extremely sensitive to down-regulation by PMN neutral proteases. As shown in Fig. 6A, exposure of these cells to azurophil granules or purified elastase for 15 min at 37 "C
4j 2
Granules
Elastase
Activator
FIG. 6. Effects of azurophil granules, elastase,or activators of endogenous pathways on '''I-TNF binding to MNC and HUVEC. A , MNC (3.5 X lo7) were incubated for 15 min at 37 "C with either azurophil granules (lo7cell equivalents), elastase (20 pg/ ml), or calcium ionophoreA23187 (IO-' M),treated with DFP, washed, and assayed for binding of "'1-TNF at 4 "C. B, monolayers of confluentHUVEC were overlaid with medium containing azurophil granules, elastase, or phorbol myristate acetate (10 ng/ml). Binding of '"1-TNF was measured after 30 min at 37 "C and treatment with DFP. Results are mean f S.D. from three independent experiments. of binding obtained with They are expressed as the percent inhibition cells treated with buffer alone, which averaged 7580 f 1791 cpm on MNC and 5387 f 529 cpm on HUVEC.
decreased the subsequent binding of "'I-TNF by 87-96%. This almost complete TNF-R down-regulation contrasted with the effect described above with PMN (mean maximum decrease of 66 k 8% in response to azurophil granules). On the other hand, a 30-min preincubation with either azurophil granules or purified elastase had little or no effect on "'ITNF binding to HUVEC (Fig. 6B).The minor effect of elastase observed here probably reflected damage to the HUVEC monolayer which was consistently observed after incubation with this enzyme (20 pg/ml) at 37 "C, rather than to a specific effect on surface TNF-R. Indeed, when the incubation was performed at 4 "C for 1.5 h (conditions in which elastase is fully capable of decreasing TNF-R on PMN (Fig. 4)), elastase-induced reduction in Iz5I-TNFbinding and the cytotoxic effect on HUVEC both disappeared (not shown). The absence of effect of elastase and azurophil granules on "'I-TNF binding to HUVEC ruled out that the decrease in binding to PMN and MNC could be the result of proteolytic degradation of 12'I-TNF (40) during the binding assay. Contrasting with this difference in response to PMNserine esterases, both MNC and HUVEC were able to down-regulate their surface TNF-R in response to activators of endogenous pathways (Fig. 6, A and B ) in a way similar to activated PMN (11). The VariableProportions of p75 and p55 T N F - R at the Surface of PMN, MNC, and HUVEC Determine the Sensitiuity of Their TNF-R to Elastase-The differential down-regulation of binding of 12'I-TNF induced by elastase on the various cell types studied could be explained most readily by the greater sensitivity of p75 TNF-R thanp55 TNF-R to this enzyme and by the variable proportions of these receptors at the cell surface. Alternatively, there may be intrinsic differences in sensitivity of the same gene product on cells of different histologic types. To distinguish between these pos-
TNF Receptor-releasing Activities
18852
of Human Neutrophils
TABLEV sibilities, we assessed the presence of p75 and p55 TNF-Rs at Comparative effects of azurophil granule enzymes and the surface of PMN and MNC by the ability of mAbs against of fMLP-inducible protease(s) onTNF-R each of these receptors to inhibit the binding of '"I-TNF. By Conditions flMLP-activated Azurophil this test, PMN display similar proportions of p75 and p55 Parameter protease granules TNF-Rs on their surface (Fig. 7). By contrast, anti-p75 TNFPercent inhibition of 4 "C 0 50-70 R mAb abolished by more than 80% the binding of "'I-TNF '"I-TNF binding 37 "C 100 50-70 to MNC, and anti-p55 TNF-R had only a small inhibitory +DFP 100 0 effect, suggesting that p75 is the major TNF-R species on +MSAAPV-CK 100 0 these cells. The reverse situation was seen with HUVEC, Species affected which express mostly p55 TNF-R, as shown by the near ++ ++ p75 TNF-R complete inhibition of '"I-TNF binding by anti-p55 mAb ++ +/p55 TNF-R (Fig. 7). Thus, the amount of p75 TNF-R on PMN, MNC, Fragments released (kDa) p75 TNF-R 42 32 and HUVEC corresponded to the extent of TNF-R downp55 TNF-R 28 (28) regulation observed following treatment with elastase. DISCUSSION
of PMN with fMLPisinsentitive tothe serine esterase inhibitor DFP and involves both p55 and p75 TNF-Rs, generating soluble fragments of 42 and -28 kDa, respectively. By contrast, the TNF-R releasing activity of azurophil granules is a serine esterase preferentially affecting the p75 TNF-R, from which it releases a 32-kDa fragment. The relative resistance of the p55 TNF-R todegradation by azurophil enzymes is shown both by the small amounts of fragments of this receptor found in supernatants from granule-treated PMN and by the inability of these fractions to reduce lZ5I-TNF binding to HUVEC, which expressed almost exclusively the p55 TNF-R. The TNF-R-shedding protease of azurophil granules was identified as elastase. Pure elastase induced TNF-R downregulation, and down-regulation by azurophil granules was abolished by the elastase inhibitors MSAAPV-CK and alantitrypsin (49, 50). Neither cathepsin G nor proteinase 3, the third known serine esterase of human PMN azurophil granules, are sensitive to inhibition by MSAAPV-CK (49,51). This suggests that neithercathepsin G nor proteinase3 contribute substantially to the TNF-R-releasing activity of azurophil granules. Fifty percent reduction in TNF binding could be obtained with
