145
Journal of Immunological Methods, 139 (1991) 145-147
© 1991 ElsevierSciencePublishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100170J JIM 05932 L e t t e r to the editors
Failure to demonstrate TNFa-specific autoantibodies in human sera by ELISA and Western blot H a n s - G e o r g Leusch, G i s e l a Sitzler a n d S u s a n n e M a r k o s - P u s z t a i Institute of Medical Immunology, Medical Faculty, Pauwelsstrasse 30, 5100 Aachen, F.R.G.
(Received24 October 1990, revisedreceived24 January 1991, accepted 15 February 1991)
Dear Editors, As recently described (Bendtzen et al., 1989, 1990; Fomsgaard et al., 1989) autoantibodies t o T N F a may be detected by Western blot analysis in the sera of patients with Gram-negative septicemia, rheumatoid arthritis and in some healthy individuals. The specificity of the autoantibodies may be established by absorption of the sera with recombinant T N F a prior to the assay. Using a T N F a cytotoxicity bioassay (cell line L929) Seckinger et al. (1989) detected T N F a inhibitory factors in urine and Foley et al. (1990) found similar activity in sera of patients with sarcoidosis, tuberculosis and Crohn's disease. The specificity of the T N F a inhibitor in urine samples was shown by neutralizing the inhibitory activity with increasing amounts of human recombinant TNFa. In contrast to the findings with human urine Foley et al. (1990) found that there was no correlation between the inhibitory effect of the sera in the L929 bioassay and the ability of the sera to compete with the binding of an enzyme-conjugated T N F a monoclonal antibody to immobilized T N F a in an ELISA system. Therefore, these results contradict the findings of Fomsgaard and the existence of specific autoantibodies binding directly to T N F a remains, at the very least, controversial.
Correspondence to: H.G. Leusch, Institute of Medical Immunology, Medical Faculty, Pauwelsstrasse 30, 5100 Aachen, F.R.G.
Proceeding on the assumption that autoantibodies to cytokines such as T N F a may have a regulatory effect in the inflammatory cascade we have sought evidence for specific antibodies to T N F a in several immunoglobulin preparations. In addition, we analysed different human sera collected from patients with septicemia (n = 11), with chronic autoimmune disorders (n = 10) and from renal tumor patients (n = 4) undergoing treatment with r I F N a and rTNFa. ELISA and Western blot assays were used for these investigations. To test for ELISA reactivity microtiter plates (Nunc, immunoplates) were coated with 100 /~1 of monoclonal mouse anti-human r T N F a (clone 199, 1 : 20, Boehringer, Mannheim, F.R.G.) overnight at room temperature in 50 m M carbonate buffer pH 9.6 followed by blocking the wells with 1% BSA in the same buffer for 1 h at room temperature. Binding of the human T N F a (10 n g / m l ) (Knoll, Mannheim, F.R.G.) in incubation buffer (50 mM Hepes, 150 m M NaC1, 200 m M sodium potassium tartrate, 1% BSA, 0.5% Synperonic F68, 0.75% PEG 40,000, 0.01% phenol, p H 7.0) was performed at 4 ° C for 18 h. After washing the plates three times with PBS-Tween (0.1%) 100/zl of patients' sera, or the corresponding protein G isolated IgG fractions, or 100/zl of immunoglobulin preparation (final concentration of 4 mg I g G / m l ) were added and incubated overnight at 4 ° C. Afterwards, wells were washed again, filled with 100 /xl of a rabbit antih u m a n IgG-alkaline phosphatase conjugate (Sigma, Deisenhofen, F.R.G.) (1/2000 in incubation buffer) and incubated at room temperature
146
for 2 h. Color was developed with 100 /~1 onitrophenyl-phosphate in substrate buffer (Behring, Marburg, F.R.G.). The reaction was stopped with 100 t~l 1 M N a O H and absorbance measured at 405/492 nm. The results were evaluated by comparing the absorbance values of the patients' sera with the negative control lacking rTNFa. Binding of r T N F a to the immobilized mouse antiTNFc~ was tested using a rabbit polyclonal antirTNFc~ (Genezyme, Boston, U.S.A.). Binding was detected using a goat anti-rabbit alkaline-phosphatase conjugated antibody (Sigma, Deisenhofen, F.R.G.) at a dilution of 1/2000 in incubation buffer. In addition, we measured the T N F a concentration with a monoclonal TNFa-ELISA procedure (anti-TNFc~ MAK clone 199 and antiTNFc~ clone 195-horseradish peroxidase-conjugated) (Boehringer Mannheim, F.R.G.), but using the described incubation buffer instead of that recommended by the manufacturer. Only three patients showed a measurable TNFc~ concentration, two with septicemia (47 p g / m l and 720 p g / m l respectively) and one with a chronic autoimmune disorder (68 pg/ml). To rule out the possibility that anti-TNFc~ antibodies might be complexed in vivo with TNFc~, a C l q binding assay with immobilized C l q was performed according to the manufacturer's instructions (Bios, Munich, F.R.G.) using a polyclonal rabbit anti-TNFa antibody (1/500 in dilution buffer). As second antibody, the anti-rabbit IgG-alkaline phosphatase conjugate (1/1000) was used in dilution buffer. Anti-TNFa antibodies were neither found in patients' sera nor in the immunoglobulin preparations using the ELISA procedure. Furthermore, the T N F / a n t i - T N F a immunocomplex assay showed no differences between the sera of patients and the negative controls. In order to confirm the results obtained by ELISA, Western blot assays were performed with the human recombinant T N F a stabilized by human serum albumin. T N F a was boiled for 5 min in sample buffer with mercaptoethanol and SDS (Laemmli, 1970) (1600 U r T N F a in 20 /~1 sample buffer per lane) before application to SDS-PAGE (Laemmli, 1970). Proteins were transferred by semi-dry electroblotting to nitrocellulose (0.2 t~m) (Schleicher & Scht~ll, Dassel, F.R.G.) in
1
2
3
4
5
6
KD
94
67
40
30
20 14
Fig. 1. Reaction of V e n i m m u n and the sera of two patients with h u m a n r T N F and stabilization proteins in a Western blot blocked with Tween 20. Lane 1: standard proteins; lane 2: h u m a n rTNFa, both lanes stained with AuroDye; lane 3: Venimmun; lanes 4 and 5: sera of two patients; lane 6: polyclonal rabbit anti-human r T N F a (positive control) used in a parallel Western blot.
T r i s / g l y c i n e / S D S / m e t h a n o l buffer (48 mM Tris, 39 mM glycine, 0.0375% SDS, 20% methanol) for 1 h at 0.8 m A / c m 2. Blotted molecular weight markers (Pharmacia, Freiburg, F.R.G.) and the TNFc~ sample were stained using Auro Dye (Janssen, Beerse, Belgium). r T N F a was identified in monomeric form at the 17 kDa position (Fig. 1). Blocking and immunodetection were performed according to Fomsgaard et al. (1989) using Tween 20-Block. Patients' sera (diluted 1/5), the corresponding IgG fractions and the immunoglobulin preparations were applied at a concentration of 4 mg I g G / m l in blotting buffer (Fomsgaard et al., 1989) and incubated at room temperature overnight. The polyclonal rabbit anti-human r T N F a antibody was used as a positive control. The detector antibody, an HRPO-conjugated anti-human IgG and IgM antibody (Dianova, Hamburg, F.R.G.), was used at a dilution of 1/500 in incubation buffer. For the rabbit antibody an HRPO-conjugated anti-rabbit IgG antibody (BioRad, Munich, F.R.G.) was used at a dilution of 1/500 in incubation buffer. After incubation at
147 1
Ko
2
i~
94 67
40
30
20 14
Fig. 2. Non-specific reaction of Venimmun with standard proteins in Western blot blocked with Tween 20. Lane 1: standard proteins, AuroDye protein stain; lane 2: Venimmun used as parallel Western blot.
room temperature for 2 h color was developed as described by Young et al. (1989). Bands were observed in the position of rTNFet (17 kDa) with all of the IgG fractions of patients' sera and the immunoglobulin preparations. Some of the diluted sera produced a band at 17 kDa together with a band at 68 kDa showing the position of the stabilization protein (Fig. 1). To confirm these resuits, we also used the non-fat dry milk block method according to the manufacturers instructions (Biozyme, Hameln, F.R.G.) because nonspecific binding of immunoglobulins to protein on nitrocellulose when using detergent blocking (as done by Fomsgaard et al., 1989) has been documented (Batteiger et al., 1989). After blocking with non-fat dry milk no bands at 17 kDa or at the position of the stabilization proteins (68 kDa) were observed in the Western blot assay when analysing patients' sera or the immunoglobulin preparations, although a positive reaction was seen with the polyclonal anti-human rTNFet-immune serum in the 17 kDa position. To demonstrate non-specific binding when
using the Tween 20 detergent block, we tested the reaction of the patients sera and immunoglobulin preparations with the blotted standard proteins. In fact, under these experimental conditions all of the standard proteins showed non-specific binding with human IgG or IgM (Fig. 2). In conclusion, using two immunochemical methods we failed to establish the presence of specific antibodies to T N F a in sera of human origin. Using the Western blot method described by Fomsgaard's group we found non-specific binding with the stabilization proteins and even with standard proteins. We cannot exclude the possibility that the binding of specific T N F a antibodies may be masked by the non-specific binding of immunoglobulins when using the Tween 20 block because Fomsgaard et al. (1989) were able to eliminate the reaction by absorption with soluble TNFa. Nevertheless, caution should be exercised when interpreting Western blot results after blocking with Tween 20. Furthermore, it will be necessary to consider which techniques should be used to assay naturally occurring specific antiT N F a autoantibodies other than the inhibition of T N F a in biological systems. References Batteiger, B.E. (1988) Blocking of Immunoblots. Handbook of Immunoblotting of Proteins, Vol. 1. CRC Press, Boca Raton, FL, p. 145. Bendtzen, K., Svenson, M., Fomsgaard, A. and Poulsen, LK. (1989) Native inhibitors (autoantibodies) of IL-la and T N F [letter]. Immunol. Today 10, 222. Bendtzen, K., Svenson, M., Jonsson, V. and Hippe, E. (1990) Autoantibodies to cytokines - Friends or Foes? Immunol. Today 11,167. Foley, N., Lambert, C., McNicol, M., Johnson, N. and Rook, G.A.W. (1990) An inhibitor of the toxicity of tumor necrosis factor in the serum of patients with sarcoidosis, tuberculosis and Crohn's disease. Clin. Exp. Immunol. 80, 395. Fomsgaard, A., Svenson, M. and Bendtzen, K. (1989) Autoantibodies to tumor necrosis factor a in healthy humans and patients with inflammatory diseases and gram-negative bacterial infections. Scand. J. Immunol. 30, 219. Laemmli, U.I. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680. Seckinger, P., Isaaz, S. and Dayer, J.-M. (1989) Purification and biologic characterization of a specific tumor necrosis factor a inhibitor. J. Biol. Chem. 20, 11966. Young, P.R. (1989) An improved method for the detection o f peroxidase-conjugated antibodies on immunoblots. J. Viroi. Methods 24, 227.
Journal of Immunological Methods, 139 (1991) 145-147
© 1991 ElsevierSciencePublishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100170J JIM 05932 L e t t e r to the editors
Failure to demonstrate TNFa-specific autoantibodies in human sera by ELISA and Western blot H a n s - G e o r g Leusch, G i s e l a Sitzler a n d S u s a n n e M a r k o s - P u s z t a i Institute of Medical Immunology, Medical Faculty, Pauwelsstrasse 30, 5100 Aachen, F.R.G.
(Received24 October 1990, revisedreceived24 January 1991, accepted 15 February 1991)
Dear Editors, As recently described (Bendtzen et al., 1989, 1990; Fomsgaard et al., 1989) autoantibodies t o T N F a may be detected by Western blot analysis in the sera of patients with Gram-negative septicemia, rheumatoid arthritis and in some healthy individuals. The specificity of the autoantibodies may be established by absorption of the sera with recombinant T N F a prior to the assay. Using a T N F a cytotoxicity bioassay (cell line L929) Seckinger et al. (1989) detected T N F a inhibitory factors in urine and Foley et al. (1990) found similar activity in sera of patients with sarcoidosis, tuberculosis and Crohn's disease. The specificity of the T N F a inhibitor in urine samples was shown by neutralizing the inhibitory activity with increasing amounts of human recombinant TNFa. In contrast to the findings with human urine Foley et al. (1990) found that there was no correlation between the inhibitory effect of the sera in the L929 bioassay and the ability of the sera to compete with the binding of an enzyme-conjugated T N F a monoclonal antibody to immobilized T N F a in an ELISA system. Therefore, these results contradict the findings of Fomsgaard and the existence of specific autoantibodies binding directly to T N F a remains, at the very least, controversial.
Correspondence to: H.G. Leusch, Institute of Medical Immunology, Medical Faculty, Pauwelsstrasse 30, 5100 Aachen, F.R.G.
Proceeding on the assumption that autoantibodies to cytokines such as T N F a may have a regulatory effect in the inflammatory cascade we have sought evidence for specific antibodies to T N F a in several immunoglobulin preparations. In addition, we analysed different human sera collected from patients with septicemia (n = 11), with chronic autoimmune disorders (n = 10) and from renal tumor patients (n = 4) undergoing treatment with r I F N a and rTNFa. ELISA and Western blot assays were used for these investigations. To test for ELISA reactivity microtiter plates (Nunc, immunoplates) were coated with 100 /~1 of monoclonal mouse anti-human r T N F a (clone 199, 1 : 20, Boehringer, Mannheim, F.R.G.) overnight at room temperature in 50 m M carbonate buffer pH 9.6 followed by blocking the wells with 1% BSA in the same buffer for 1 h at room temperature. Binding of the human T N F a (10 n g / m l ) (Knoll, Mannheim, F.R.G.) in incubation buffer (50 mM Hepes, 150 m M NaC1, 200 m M sodium potassium tartrate, 1% BSA, 0.5% Synperonic F68, 0.75% PEG 40,000, 0.01% phenol, p H 7.0) was performed at 4 ° C for 18 h. After washing the plates three times with PBS-Tween (0.1%) 100/zl of patients' sera, or the corresponding protein G isolated IgG fractions, or 100/zl of immunoglobulin preparation (final concentration of 4 mg I g G / m l ) were added and incubated overnight at 4 ° C. Afterwards, wells were washed again, filled with 100 /xl of a rabbit antih u m a n IgG-alkaline phosphatase conjugate (Sigma, Deisenhofen, F.R.G.) (1/2000 in incubation buffer) and incubated at room temperature
146
for 2 h. Color was developed with 100 /~1 onitrophenyl-phosphate in substrate buffer (Behring, Marburg, F.R.G.). The reaction was stopped with 100 t~l 1 M N a O H and absorbance measured at 405/492 nm. The results were evaluated by comparing the absorbance values of the patients' sera with the negative control lacking rTNFa. Binding of r T N F a to the immobilized mouse antiTNFc~ was tested using a rabbit polyclonal antirTNFc~ (Genezyme, Boston, U.S.A.). Binding was detected using a goat anti-rabbit alkaline-phosphatase conjugated antibody (Sigma, Deisenhofen, F.R.G.) at a dilution of 1/2000 in incubation buffer. In addition, we measured the T N F a concentration with a monoclonal TNFa-ELISA procedure (anti-TNFc~ MAK clone 199 and antiTNFc~ clone 195-horseradish peroxidase-conjugated) (Boehringer Mannheim, F.R.G.), but using the described incubation buffer instead of that recommended by the manufacturer. Only three patients showed a measurable TNFc~ concentration, two with septicemia (47 p g / m l and 720 p g / m l respectively) and one with a chronic autoimmune disorder (68 pg/ml). To rule out the possibility that anti-TNFc~ antibodies might be complexed in vivo with TNFc~, a C l q binding assay with immobilized C l q was performed according to the manufacturer's instructions (Bios, Munich, F.R.G.) using a polyclonal rabbit anti-TNFa antibody (1/500 in dilution buffer). As second antibody, the anti-rabbit IgG-alkaline phosphatase conjugate (1/1000) was used in dilution buffer. Anti-TNFa antibodies were neither found in patients' sera nor in the immunoglobulin preparations using the ELISA procedure. Furthermore, the T N F / a n t i - T N F a immunocomplex assay showed no differences between the sera of patients and the negative controls. In order to confirm the results obtained by ELISA, Western blot assays were performed with the human recombinant T N F a stabilized by human serum albumin. T N F a was boiled for 5 min in sample buffer with mercaptoethanol and SDS (Laemmli, 1970) (1600 U r T N F a in 20 /~1 sample buffer per lane) before application to SDS-PAGE (Laemmli, 1970). Proteins were transferred by semi-dry electroblotting to nitrocellulose (0.2 t~m) (Schleicher & Scht~ll, Dassel, F.R.G.) in
1
2
3
4
5
6
KD
94
67
40
30
20 14
Fig. 1. Reaction of V e n i m m u n and the sera of two patients with h u m a n r T N F and stabilization proteins in a Western blot blocked with Tween 20. Lane 1: standard proteins; lane 2: h u m a n rTNFa, both lanes stained with AuroDye; lane 3: Venimmun; lanes 4 and 5: sera of two patients; lane 6: polyclonal rabbit anti-human r T N F a (positive control) used in a parallel Western blot.
T r i s / g l y c i n e / S D S / m e t h a n o l buffer (48 mM Tris, 39 mM glycine, 0.0375% SDS, 20% methanol) for 1 h at 0.8 m A / c m 2. Blotted molecular weight markers (Pharmacia, Freiburg, F.R.G.) and the TNFc~ sample were stained using Auro Dye (Janssen, Beerse, Belgium). r T N F a was identified in monomeric form at the 17 kDa position (Fig. 1). Blocking and immunodetection were performed according to Fomsgaard et al. (1989) using Tween 20-Block. Patients' sera (diluted 1/5), the corresponding IgG fractions and the immunoglobulin preparations were applied at a concentration of 4 mg I g G / m l in blotting buffer (Fomsgaard et al., 1989) and incubated at room temperature overnight. The polyclonal rabbit anti-human r T N F a antibody was used as a positive control. The detector antibody, an HRPO-conjugated anti-human IgG and IgM antibody (Dianova, Hamburg, F.R.G.), was used at a dilution of 1/500 in incubation buffer. For the rabbit antibody an HRPO-conjugated anti-rabbit IgG antibody (BioRad, Munich, F.R.G.) was used at a dilution of 1/500 in incubation buffer. After incubation at
147 1
Ko
2
i~
94 67
40
30
20 14
Fig. 2. Non-specific reaction of Venimmun with standard proteins in Western blot blocked with Tween 20. Lane 1: standard proteins, AuroDye protein stain; lane 2: Venimmun used as parallel Western blot.
room temperature for 2 h color was developed as described by Young et al. (1989). Bands were observed in the position of rTNFet (17 kDa) with all of the IgG fractions of patients' sera and the immunoglobulin preparations. Some of the diluted sera produced a band at 17 kDa together with a band at 68 kDa showing the position of the stabilization protein (Fig. 1). To confirm these resuits, we also used the non-fat dry milk block method according to the manufacturers instructions (Biozyme, Hameln, F.R.G.) because nonspecific binding of immunoglobulins to protein on nitrocellulose when using detergent blocking (as done by Fomsgaard et al., 1989) has been documented (Batteiger et al., 1989). After blocking with non-fat dry milk no bands at 17 kDa or at the position of the stabilization proteins (68 kDa) were observed in the Western blot assay when analysing patients' sera or the immunoglobulin preparations, although a positive reaction was seen with the polyclonal anti-human rTNFet-immune serum in the 17 kDa position. To demonstrate non-specific binding when
using the Tween 20 detergent block, we tested the reaction of the patients sera and immunoglobulin preparations with the blotted standard proteins. In fact, under these experimental conditions all of the standard proteins showed non-specific binding with human IgG or IgM (Fig. 2). In conclusion, using two immunochemical methods we failed to establish the presence of specific antibodies to T N F a in sera of human origin. Using the Western blot method described by Fomsgaard's group we found non-specific binding with the stabilization proteins and even with standard proteins. We cannot exclude the possibility that the binding of specific T N F a antibodies may be masked by the non-specific binding of immunoglobulins when using the Tween 20 block because Fomsgaard et al. (1989) were able to eliminate the reaction by absorption with soluble TNFa. Nevertheless, caution should be exercised when interpreting Western blot results after blocking with Tween 20. Furthermore, it will be necessary to consider which techniques should be used to assay naturally occurring specific antiT N F a autoantibodies other than the inhibition of T N F a in biological systems. References Batteiger, B.E. (1988) Blocking of Immunoblots. Handbook of Immunoblotting of Proteins, Vol. 1. CRC Press, Boca Raton, FL, p. 145. Bendtzen, K., Svenson, M., Fomsgaard, A. and Poulsen, LK. (1989) Native inhibitors (autoantibodies) of IL-la and T N F [letter]. Immunol. Today 10, 222. Bendtzen, K., Svenson, M., Jonsson, V. and Hippe, E. (1990) Autoantibodies to cytokines - Friends or Foes? Immunol. Today 11,167. Foley, N., Lambert, C., McNicol, M., Johnson, N. and Rook, G.A.W. (1990) An inhibitor of the toxicity of tumor necrosis factor in the serum of patients with sarcoidosis, tuberculosis and Crohn's disease. Clin. Exp. Immunol. 80, 395. Fomsgaard, A., Svenson, M. and Bendtzen, K. (1989) Autoantibodies to tumor necrosis factor a in healthy humans and patients with inflammatory diseases and gram-negative bacterial infections. Scand. J. Immunol. 30, 219. Laemmli, U.I. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680. Seckinger, P., Isaaz, S. and Dayer, J.-M. (1989) Purification and biologic characterization of a specific tumor necrosis factor a inhibitor. J. Biol. Chem. 20, 11966. Young, P.R. (1989) An improved method for the detection o f peroxidase-conjugated antibodies on immunoblots. J. Viroi. Methods 24, 227.
