Separation of tumor necrosis factor α isoforms by two-dimensional polyacrylamide gel electrophoresis

1086

A. D. Watts

Electrophoresis 1997, 18, 1086-1091

el a/.

Alan D. Watts Nicholas H. Hunt Brett D. Hambly Geeta Chaudhri Department of Pathology, University of Sydney, Australia

Separation of tumor necrosis factor a isoforms by two-dimensional polyacrylamide gel electrophoresis The mouse macrophage cell-line RAW264.7, stimulated with lipopolysaccharide, was used as a model for the study of the production of tumor necrosis factor (TNF) isoforms. TNF is synthesised initially as a 26 kDa transmembrane precursor, which is then processed enzymatically by a protease to release a mature molecule of 17 kDa. Dose-dependent production of transmembrane TNF was assessed by fractionation of cell membranes and Western blot analysis followed by autoradiography and densitometry. Isoforms of both the precursor and mature molecules were separated using two-dimensional (2-D) electrophoresis with immobilised pH gradient 3-10 linear gels as the first dimension. After radiolabelling of cells with "S, both cell-associated and supernate-associated TNF isoforms were immunoprecipitated. A large number of protein spots were visualised on the 2-D gel map, for both the transmembrane and mature TNF species, more than have been detected previously using onedimensional sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE). The likelihood that these putative isoforms were the result of differential glycosylation was tested by preincubating the cells with tunicamycin. This had the effect of reducing the number of protein spots, notably the higher molecular weight species. There were a number of precursor TNF isoforms that were unchanged uDon tunicamvcin treatment and these uresumably reflect protein modifications other than glycosylation.

1 Introduction Tumor necrosis factor a (TNF) is a cytokine with a diverse array of immune and inflammatory functions. It is the principal mediator of septic shock, and is of major importance in rheumatoid arthritis, Crohn's disease and many autoimmune processes. It is thought to be the deregulated overexpression of TNF that contributes to the pathological state in these diseases. Hence, considerable efforts have been directed towards understanding the synthesis and processing of TNF in a number of cell types. Several approaches have been used in an attempt to negate the effects of TNF overexpression, including TNF neutralising agents and inhibitors of TNF processing such as the hydroxamic acid-based matrix metalloproteinase inhibitors [ 1-41. TNF is synthesised initially as a precursor, which is a type I1 integral membrane protein [5, 61 that is biologically active, and recent work has shown that it signals through a different mechanism to mature TNF [7]. The importance of glycosylation of cytokines is still unclear. A possible role for this type of modification is the alteration of cytokine action by changing the binding parameters for the specific cytokine receptors. Murine TNF has been shown previously to consist of differentially glycosylated isoforms [8, 91, and has one potential Correspondence: Professor Nicholas H. Hunt, Department of Pathology, Blackburn Building D06, University of Sydney, NSW 2006, Australia (Tel: +612-9351-2414; Fax: +612-9351-3429; E-mail: nhunta blackburn.med.su.oz.au) Nonstandard abbreviations: LPS, lipopolyasccharide; TNF, tumor necrosis factor a Keywords: Immobilized pH gradients / Two-dimensional polyacrylamide gel electrophoresis / Tumor necrosis factor a / Glycosylated

isoform

0 VCH

Verlagsgesellschaft mbH, 69451 Weinheim, 1997

N-glycosylation site [lo]. There are seven or eight isoforms of mature murine TNF, including the nonglycosylated 17 kDa form, that are resolved by one-dimensional SDS-PAGE. Very recently it has been shown that different glycosylated isoforms of TNF are expressed in mouse macrophages depending on the stimulus [ll], implicating the isoforms in differing biological roles. In the current study we have applied two-dimensional electrophoresis to the resolution of TNF isoforms.

2 Materials and methods 2.1 Materials

RAW264.7 cells were from the American Type Culture Collection (Rockville, MD, USA). Lipopolysaccharide (LPS) from E. coli serotype 055:B5 was a product of Sigma (Code No. L-2880). Tunicamycin (Code No. 654380) was from Calbiochem (La Jolla, CA, USA). The Multiphor I1 electrophoresis unit, Immobiline Dry-Strip, pH 3.0-10.0 linear (Code No. 18-1016-60) and ExcelGel SDS gradient 8-18% (code No. 80-1255-53) were products of Pharmacia (Uppsala, Sweden). The E752 5000 V power supply was a product of Consort (Turnhout, Belgium). the pH 3.0-10.0 ampholytes (Code No. 1631112) were from Bio-Rad (Hercules, CA, USA). CHAPS (Code No. C-9426) and iodoacetamide was from Sigma (Code No. 1-6125). TRAN33 LABEL ("'S-E. coli hydrolysate labelling reagent, containing 70 O/o ~-[.''S]methionine) was from ICN (Code No. 51006), as was methionine/cysteine-free DMEM (Code No. 16-424-54). Fetal bovine serum (Code No. 09701501) was from CSL (Parkville, Vic, Australia). Mini-Protean I1 apparatus and prestained low range molecular weight markers were from Bio-Rad. High tensile strength acrylamide (30 O/o acrylamide, 0.65% Bis), Duracryl (Code No. ELCR 2DC 070) was from Millipore. The polyclonal anti-serum was developed in rabbit against mature (17 kDa) mouse TNF [12]. Control 0173-0835/97/0707-1086

$17.50+.50/0

Separation of tumor necrosis factor a isoforms

Electrophoresis 1991, 18, 1086--1091

serum was rabbit anti-mouse immunoglobulins (Code No. Z109) from DAKO (Glostrup, Denmark). The secondary antibody, goat anti-rabbit IgG conjugated to horseradish peroxidase (Code No. 170-6463), was from BioRad, as was supported nitrocellulose Western transfer membrane (Code No. 162-0096). X-OMAT K film (Code No. 165 3211) was from Kodak (Rochester, NY, USA). The ECL detection system (Code No. RPN 2109) was from Amersham (Little Chalfont, Buckingham, UK). Molecular biology grade urea and RQI DNAse were from Promega (Madison, WI, USA). Soybean trypsin inhibitor (Code No. T-9003), aprotinin (Code No. A-1153), PMSF (Code No. P-7626) and Protein A-Sepharose CL4B fast flow were from Sigma (Code No. P-3391). DTE (Code No. 0425-256) was from Amresco (Solon, OH, USA).

a

A

B

C

D

E

1087

F

2.2 Cell culture and 35S-labelling

RAW264.7 cells were cultured in DMEM supplemented with 10% fetal bovine serum at 37°C and 5% CO, in a humidified atmosphere. For labelling experiments cells were seeded at 1 X lo6 per well in 6-well plates and allowed to adhere overnight for 18 h. Cells were stimulated for 3 h with lipopolysaccharide (10 pg/mL). In some wells tunicamycin (10 pg/mL) was also added. Tunicamycin is a known inhibitor of N-glycosylation [9]. The culture media was then removed and the confluent monolayers washed twice with warm methionine/cysteine-deficient DMEM and pre-starved with methioninel cysteine-deficient DMEM plus stimulants for 30 min. Immediately following this, the prestarving mix was aspirated and replaced with a labelling mix consisting of 100 pCi of T U N 3 % LABEL in 500 pL methionhe/ cysteine-free DMEM per well, plus stimulants. The cells were labelled for 30 min, then transferred straight to ice. For Western blot experiments, 1 X lo7 cells were added to 80 cm2 flasks and allowed to adhere overnight. 2.3 Preparation of cell membranes and Western blotting

Cell membranes were prepared from multiple lots of 2 X lo7 RAW264.7 cells stimulated with LPS (1 ng/mL to 10 pg/mL) for 6 h, according to the method described in [13]. Briefly, cells were washed 3 times in ice-cold PBS and harvested by scraping. Cells were then collected by centrifugation at 1200 rpm, 5 min, 4°C. Pellets were washed twice in 1 mL lysis buffer (0.1 M sodium phosphate buffer, pH 7.4, 0.1 M MgC12, 2 M NaCl, 0.1 M DTT, 0.1 M PMSF, 2% w/v NaN,, 20 pL RQ1 DNAse), and resuspended in 2 mL lysis buffer. Cells were sonicated on ice, and cellular disruption was monitored using trypan blue. 1.2 mL of a 41% w/v sucrose solution in lysis buffer was poured into a 5 mL ultracentrifuge tube. The cell lysate was then carefully overlayed onto the sucrose solution. The samples were centrifuged at 95 000 X g for 1 h. The white interfacial band containing the membrane protein was then removed using a needle and syringe. The sucrose was diluted by adding a 2- to 3-fold excess of lysis buffer and the membranes pelleted at 95000 X g for 20 min. The supernates were discarded and the pellets washed twice in 0.5 mL lysis buffer. The membrane pellets were resuspended in membrane resus-

Figure 1. (a) Dose-dependent production of transmembrane TNF assessed by Western blot. A scanned image of an X-ray film exposure

to a filter developed by ECL. In lane (A) the membrane sample was from unstimulated cells, in (B), (C), (D), (E), (F) the membrane samples were from cells exposed to 0.001, 0.01, 0.1, 1, and 10 pglmL LPS, respectively. Note that the higher molecular mass glycosylated isoforms are only faintly detected despite the high intensity of the major 26 kDa form. (b) Densitometric analysis performed using ImageQuant software of the laser scanned exposure for 3 independent experiments as reported in (a). The relative quantities of protein are expressed as a percentage of the combined volume integration values for all 5 treatments (mean SEM). Clearly, the production of transmembrane TNF is LPS dose-dependent.

*

pension buffer (50 mM Tris-HC1, pH 7.4, 500 mM NaCl, 5 mM EDTA, pH 8.0, 0.5% v/v Triton X-100, 1 mM PMSF, 50 pg/mL soybean trypsin inhibitor). Samples were denatured at 100°C for 5 min. A total of 15 pg of membrane protein per sample was separated on a 12% SDS-PAGE gel at 140 V, using the Mini-PROTEAN I1 apparatus. Western transfer of the gels was onto supported nitrocellulose filters at 45 V for 1h, and the filters blocked overnight with 5% low fat dry milk, 5% fetal calf serum, and 1% bovine serum albumin at 4°C. The filters were probed with anti-mouse TNF serum for 2 h at room temperature and incubated with a goat anti-rabbit IgG horseradish peroxidase conjugate for 1h at room

1088

kDa 143

97

50

3s 30 22

Electrophoresis 1997, 18, 1086-1091

A. D. Watts et a/.

---

4.0

5.0

7.0

8.0

Figure 2. Separation of 1°F immunoprecipitates from whole cell lysates by 2-D electrophoresis revealed a number of resolved spots, as well as some more diffuse protein patterns. The major protein spots are numbered consecutively with the predominant 26 kDa form labelled No. 16 (gly refers to the higher molecular weight glycosylated isoforms). Superimposed upon the 2-D gel map is a section from a one-dimensional 12% SDS-PAGE gel run with the same samples for cross-reference. This gel image and all subsequent images were captured using a BASlOOO Phosphorimager.

Tris, 0.01 O/o w/v bromophenol blue). IEF was performed with precast IPG gels, pH 3-10 linear gradient, using a Multiphor I1 system. The IPG strips were rehydrated in 8 M urea, 33 mM CHAPS, 10 mM DTE, O.O0lo/o w/v bromophenol blue, 0.6% v/v carrier ampholytes 3-10 for at least 6h. A 50 pL volume of sample was loaded at the cathodic end of the IPG strip. The IEF was carried out at 300 V for 4.5 h and 2000 V for 11.5 h with an E752 2.4 Immunoprecipitation power supply, and at 20°C using a thermostatic cooler. The strips were then used immediately or stored at All procedures were carried out on ice or at 4°C. The -86°C until use. Each strip was treated with a two-step supernates were retained, and the cells washed 3 times equilibration process before being applied to a secondwith ice-cold PBS. Following this, cells were lysed for dimension gel. First, the strips were incubated in 10 mM 10 min with mild agitation in lysis buffer and harvested. Tris-C1, pH 6.8, 8 M urea, 36% v/v glycerol, 0.5% w/v The lysates were vortexed for lOs, centrifuged 10 rnin at SDS, 130 mM DTE for 10 rnin at room temperature with 14000 X g, and the lysate supernates retained. The cell agitation. Second, the strips were transferred to 10 inM supernates were also centrifuged, then diluted 1 in 2 in Tris-CI, pH 6.8, 8 M urea, 36% v/v glycerol, OS0/o w/v lysis buffer. Anti-TNF serum (3 ILL) was added to all SDS, 135 mM iodoacetamide, 0.001 O/o w/v bromophenol samples, which were incubated for 16h on a rotating blue for an additional 10 min. The strips were allowed to device. Immunocomplexes were collected by incubation drain for 10 min and the placed onto an ExcelGel SDS with 10 pL protein A-Sepharose for 3 h on the rotating gradient 8-18 O/o precast gel. Prestained molecular weight device. The Sepharose beads were washed twice in markers were also applied to the gel via a piece of filter radioimmunoprecipitation assay (RIPA) buffer (10 mM paper. The SDS-PAGE was run according to the manuTris-Cl, 5 mM EDTA, 1O/o v/v Triton X-100, 1O/o w/v deo- facturer's instructions. Briefly, the proteins were separxycholic acid, 500 mM NaCI) and two times in RIPA ated at 20 mA for 30 min prior to removal of the TPG buffer without NaC1. Each sample was resuspended in strips, and 50 mA for 75 min. The gels were cooled to 30 pL SDS-PAGE sample buffer and denatured at 100°C 15"C during the run. The completed second-dimensional for 5 min. gels were fixed for 1h in 50% v/v methanol, 10% acetic acid and 10% glycerol, allowed to air-dry for l h , and 2.5 Tho-dimensional electrophoresis of TNF exposed to a Phosphorimager plate for seveal days. Scanned gel images were obtained using a BASlOOO immunoprecipitates Phosphorimager and MacBAS version 2.0 software Immunoprecipitates from RAW264.7 cells were diluted (Fujix). The isoelectric point scale for each gel was esti3-fold with an IPG sample buffer (9 M urea, 65 r n M DTE, mated from the position of the IPG strip on the second 1O/o v/v carrier ampholytes 3-10, 65 rnM CHAPS, 35 mM dimension gel.

temperature. The blots were developed with enhanced chemiluminescence reagents and exposed to X-ray film. The exposures were scanned using a laser densitometer (Molecular Dynamics) and volume integration densitometry was performed using ImageQuant software (Molecular Dynamics).

Electrophoresis 1997, 18, 1086-1091

Separation of tumor necrosis factor a isoforms

1089

kDa

50 35

30 22

-4.0

5.0

6.0

7.0

3 Results 3.1 Expression of TNF Relative expression levels of transmembrane TNF were determined by volume integration of the bands present on the X-ray film exposed to the Western blots. The assay was repeated three times and the data normalised by expressing the value for each treatment as a percentage of the combined total volume integration for each gel. The linear response of the X-ray film was checked by analysing serial dilutions of samples (data not shown). Expression of transmembrane TNF was found to be dose-dependent (see Figs. l a and b), and there was a detectable level of protein present with as little as 0.001 ng/mL LPS. Peak levels of transmembrane TNF were expressed with 1-10 pg/mL LPS. This is consistent with the LPS-induced production of secreted TNF measured by WEHI-I64 [I41 bioassay (unpublished observations). The glycosylated isoforms of transmembrane TNF were barely detectable by the Western blot technique, which may be a consequence of the membrane extraction procedure or lower affinity to anti-TNF serum. Radiolabelling of cells with TRAN3’S LABEL followed by immunoprecipitation was found to afford greater sensitivity for the study of glycosylated isoforms.

8.0

Figure 3. Separation of TNF immunoprecipitates from lysates of cells that were treated with tunicamycin. There is a complete absence of the diffuse higher molecular weight isoforms (compare with Fig. 2), and the lower molecular weight species appear to be correspondingly more intense.

molecular mass (kDa) represent the positions of the prestained molecular weight markers, and are only an approximate guide since the prestaining process alters the true mobility of the marker proteins. The major form of transmembrane TNF is of molecular mass 26 kDa (spot 16).There are a number of species of higher molecular weight which appear quite diffuse (labelled ‘gly’). When the cells were treated with tunicamycin these species disappeared (Fig. 3). A number of lower molecular weight species are also present which were not affected by tunicamycin (spots 10-15). There are also species that do not appear to be localised to the p l restricted ‘cluster’ of TNF proteins and are not numbered. These proteins were not apparent on 2-D gel maps of immunoprecipitates from unstimulated cells, or stimulated cells immunoprecipitated with control serum (data not shown). Possibly, they represent associating proteins, or artefacts of heating during sample denaturation. However, all steps involving urea were conducted at room temperature, and for as short a time as possible to ensure minimal protein carbamylation.

The immunoprecipitate of cell supernates from LPSstimulated RAW264.7 cells containing secreted TNF isoforms was also separated by 2-D electrophoresis (Fig. 4). Treatment of the R4W264.7 cells with tunicamycin completely abolished spots 1-9, leaving only spots 10-12 (Fig. 5). These three spots were of identical mo3.2 Two-dimensional separation of TNF isoforms blity with respect to charge and molecular weight to three spots on the gel maps shown in Figs. 2 and 3, and The immunoprecipitate from the whole cell-lysate of ’3- are thus assumed to be identical. The isoelectric point of labelled RAW264.7 cells separated by IPG-2D electro- the various TNF isoforms varied from approximately pH phoresis yielded a spectrum of TNF isoforms (Fig. 2). 4.8-5.8, estimated from the linear gradient in the IPG Superimposed upon the 2-D gel map is a corresponding strips. This is in approximate agreement with the theoportion from a one-dimensional SDS-PAGE gel run retical p l of 4.96 for mouse 17 kDa TNF listed in the with the same sample. The horizontal bars indicating SWI SS-2DPAGE database [151.

1090

A. D. watts

Elecfrophoresis 1997, Ii?, 1086-1091

a/.

4 Discussion

kDa 50 35

30 22

5.0

6.0

Figure 4. The two-dimensional map for TNF immunoprecipitates from cell supernates reveals a total of approximately 12 discrete species, including a novel spot (No. 10) that may be a charge variant of mature 17 kDa TNF. The higher molecular mass isoforms migrate close together and are difficult to distinguish.

kDa 50 35

30 22

4.0

5.0

6.0

Figure 5. Tunicamycin completely prevented the formation of the high molecular weight cluster of isoforms, and only 3 spots (No. 10-12) remain visible compared with Fig. 4.

The optimal level of LPS for production of TNF was determined by Western blot of preparations of RAW264.7 cell membranes and a thorough densitometric approach. Immunoprecipitation of both cell lysates and cell supernates from RAW264.7 cells stimulated with 10 pg/mL LPS, followed by 2-D electrophoresis, allowed visualisation of a total of 18 discrete species that are likely to contain TNF protein sequence. The probable origin of these isoforms is discussed. The 2-D gel maps of the TNF precursor molecule show a mixture of diffuse and discrete patterns of protein spots (Figs. 4 and 5). Tunicamycin had the general effect of preventing the formation of higher molecular weight species, and increasing the intensity of the species that remained. There are three protein spots (10-12) common to the 2-D gel maps from cell lysates and cell supernates. The isoform (spot 11) that migrates directly above the major 17 kDa (spot 12) mature form (see Figs. 2-5) has been identified previously [16] as a product of cleavage of the transmembrane precursor at an alternate site. This results in a secreted isoform that has an additional ten amino acids and an apparent molecular weight of 18.5 kDa. These additional ten amino acids render the protein biologically inactive with respect to cytotoxicity towards a TNF-sensitive cell line [16]. Spot 13 may be a degradation product of one of the higher molecular weight isoforms. It is conceivable that this may represent the portion of the transmembrane TNF molecule that remains in the membrane following TNF processing and migrates at an apparent molecular weight of 14 kDa. However, the anti-TNF serum used in this study was directed against the mature TNF molecule, and therefore the only mechanism by which this 14 kDa product could be formed during immunoprecipitation would be if traces of the TNF processing enzyme were still active. It has been shown that the transmembrane TNF molecule is synthesised as a trimer prior to localisation to the cell membrane [17], and hence it is possible that one or two subunits of the trimer are cleaved during immunoprecipitation, but remain associated with the immune complex via the remaining noncleaved subunith until sample denaturation. It is also possible that the transmembrane TNF trimer is partially cut by the processing enzyme prior to cell lysis, and is preserved in that state throughout the entire protocol. It is unlikely that any specific type of cleavage event, involving a TNF processing enzyme, could occur under denaturing conditions. Alternatively, the products could be formed by a nonspecific type of degradation following sample denaturation. Spots 13-15 (Figs. 2 and 3) potentially fall into this category. Spot 18 appears to be absent following treatment with tunicamycin and thus may represent a glycosylated TNF cleavage or degradation product. The possibility that these spots represent other proteins coimmunoprecipitating with transmembrane TNF should not be excluded until further analysis is conducted. By comparison of gel maps from TNF proteins immunoprecipitated from both intact cells and lysed cells, we have resolved an isoform (spot 17) that appears not to

Electrophoresis

1997, 18, 1086-1091

be expressed on the cell surface (data not shown). This isoform may represent some step in the synthesis of precursor TNF, or a form that is incapable of proper translocation to the cell membrane. There are classes of modification of the TNF precursor other than glycosylation. The human molecule has been shown to be serine phosphorylated on the cytoplasmic domain [18], although the biological significance of this is still unclear. Interestingly, the cytoplasmic domain of the precursor TNF molecule does not appear to be necessary for proper expression of transmembrane TNF 1191. Furthermore, the human molecules has been shown to be myristyl acylated on lysine residues adjacent to the cytoplasmic region [20]. This type of modification probably plays a role in proper insertion of the precursor TNF trimer into the cell membrane and/or anchoring. It is therefore possible that the higher molecular weight isoforms (in the region labelled ‘gly’) present in Fig. 2 are weakly detected and diffuse due to a combination of types of modification that alter both charge and apparent molecular weight of the major 26 kDa precursor molecule. The cell supernates yielded a total of 12 distinguishable protein spots (Fig. 4). The resolution of spots 1-9 may be improved by using narrower pH range IPG strips and larger second-dimensional SDS-gels, and would likely provide a more accurate estimation of P I of secreted TNF isoforms. Spots 1-9 are glycosylated isoforms and are not formed by RAW264.7 cells treated with tunicamycin (Fig. 5). Spots 11 and 12 are easily resolved by one-dimensional electrophoresis; however, spot 10 represents a new species, and is an obvious candidate for further analysis.

In conclusion, the separation of cytokines by IPG 2-D electrophoresis is a more thorough method of identifying isoforms that migrate closely on conventional onedimensional SDS-PAGE systems, and will allow more complete studies of relative expression levels of each isoform in different cellular settings. The next step in this work will be to conduct post-separation analysis of the resolved isoforms using preparative IPG 2-D electrophoresis. We would like to thank Simon Myers (Heart Research Institute), Ben Herbert (Australian Proteome Anaiysis Facifiq), and Clive Seymour (AMRAD Pharmacia Biotech) for their kind technical advice, Assoc. ProJ Cris dos Remedios for help with densitometric analysis, and also the Wenkart Foundation and the National Health and Medical Research Council f o r financial support. Received December 6, 1996

Separation of tumor necrosis factor a isoforms

1091

5 References Gearing, A. J., Beckett, P., Christodoulou, M., Churchill, M., Clements, J., Davidson, A. H., Drummond, A. H., Galloway, w. A., Gilbert, R., Gordon, J. L., Leber, T. M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S . , Thomas, W., Wells, G., Wood, L. M., Woolley, K., Nature 1994, 270, 555-557. Gearing, A., Beckett, P., Christodoulou, M., Churchill, M., Clements, J. M., Crimmin, M., Davidson, A. H., Drummond, A. H., Galloway, W. A., Gilbert, R., Gordon, J. L., Leber, T. M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S., Thomas, W., Wells, G., Wood, L. M., Woolley, K., J. Leuko. B i d . 1995, 57, 174-777. McGeehan, G. M., Becherer, J . D., Bast, R. J., Boyer, C. M., Champion, B., Connolly, K. M., Conway, J. G., Furdon, P., Karp, S., Kidao, S., Mcelroy, A. B., Nichols, J., Pryzwansky, K. M., Schoenen, F., Sekut, L., Truesdale, A., Verghese, M., Warner, J., Ways, J. P., Nature 1994, 370, 558-561. Mohler, K. M., Sleath, P. R., Fitzner, J. N., Cerretti, D. P., Alderson, M., Kerwar, S. S., Torrance, D. S., Otten, E. C., Greenstreet, T., Weerawarna, K., Kronheim, s. R., Petersen, M., Gerhart, M., Kozlosky, C. J., March, C. J., Black, R. A., Nature 1994, 370, 218-220. Kriegler, M., Perez, C., DeFay, K., Albert, I., Lu, S. D., Cell 1988, 53, 45-53. Perez, C., Albert, I., DeFay, K., Zachariades, N., Gooding, L., Kriegler, M., Cell 1990, 63, 251-258. Grell, M., Douni, E., Wajant, H., Lohden, M., Clauss, M., Maxeiner, B., Georgopoulos, S., Lesslauer, W., Kollias, G., Pfizenmaier, K., Scheurich, P., Cell 1995, 83, 793-802. Jue, D.-M., Sherry, B., Luedke, C . , Manogue, K. R., Cerami, A,, Biochemistry 1990, 29, 8371-8377. Sherry, B. D.-M., Zentella, A., Cerami, A., Biochern. Biophys. Res. Commun. 1990, 173, 1072-1078. Pennica, D., Hayflick, J. S., Bringman, T. S., Palladino, M. A., Goeddel, D. V., Proc. Natl. Acad. Sci. USA 1985, 82, 6060-6064. Branch, D. R., Guilbert, L. J., I n f . J. Biochem. Cell Biol. 1996, 28, 949-955. Doherty, G. M., Lange, J. R., Langstein, H. N., Alexander, H. R., Buresh, C. M., Norton, J. A,, J. Immunol. 1992, 149, 1666-1670. Maeda, T., Balakrishnan, K., Mehdi, S . Q., Biochimi. Biophys. Acta 1983, 731, 115-120. Espevik, T., Nissen, M. J., J. Immunol. Methods 1986, 95, 99-105. Appel, R. D., Sanchez, J.-C., Bairoch, A,, Golaz, O., Miu, M., Vargas, J. R., Hochstrasser, D. F., Electrophoresis 1993, 14, 1232-1238. Cseh, K., Beutler, B., J. B i d . Chem. 1989, 264, 16256-16260. Tang, P., Hung, M.-C., Klostergaard, J., Biochemistry 1996, 35, 8216-8225. Pocsik, E., Duda, E., Wallach, D., J. InJ. 1995, 45, 152-160. Utsumi, T., Akimura, K., Kawabata, Z., Levitan, A., Tokunaga, T., Tang, P., Ide, A., Hung, M.-C., Klostergaard, J., Mol. Cell. Biol. 1995, IS, 6398-6405. [20] Stevenson, F. T., Bursten, S. L., Locksley, R. M., Lovett, D. H., J . Exp. Med. 1992, 176, 1053-1062.