The application of glucose oxidase-labeled antibodies for the detection of proteins on nitrocellulose

Journal of Immunological Methods, 93 (1986) 231-236

231

Elsevier JIM 04084

The application of glucose oxidase-labeled antibodies for the detection of proteins on nitrocellulose * William D. Geoghegan **, Dave Lee Williams and Robert E. Jordon Cutaneous lmmunopathology Unit and the Department of Dermatology, The University of Texas Health Science Center at Houston, TX, U.S.A (Received 4 March 1986, accepted 28 May 1986)

We have evaluated the sensitivity of immunostaining with glucose oxidase for the detection of monomeric human serum albumin (HSA) and monomeric human immunoglobulin G (IgG). A modification of a histochemical procedure was utilized by replacing phenazine methosulfate (PMS) with 1-methoxyphenazine methosulfate (mPMS) and by replacing Tris-HC1 with Tris-citrate to improve the solubility of the tetrazolium compounds tested, mPMS is less sensitive to light, and may be stored for long periods in solution; it is now used routinely by histochemists in place of PMS in dehydrogenase cytochemistry, pH values of 6.3-8.3 were tested, with the reaction at pH 8.3 providing a slight increase in sensitivity. The reaction rate increased markedly as the pH became more alkaline. The minimum quantity of HSA detected was 3 ng applied directly to nitrocellulose and 10 ng when blotted. Human IgG was routinely detected at 250 pg and occasionally at 100 pg when dotted on the nitrocellulose. Key words: Glucose oxidase; 1-Methoxyphenazine methosulfate; Blot; Isoelectric focusing; Protein detection

Introduction

Glucose oxidase (fl-D-glucose: oxygen 1oxidoreductase EC 1.1.3.4) isolated from the fungus Aspergillus niger has gained increasing attention in recent years in immunohistochemistry (Campbell and Bhatnagar, 1976; Clark et al., 1982; Rathles and Franks, 1982; Van Noorden et al., 1982) and as an immunoenzyme stain on nitrocellulose blots (Porter and Porter, 1984). The enzyme is highly selective for fl-D-glucose, converting it into D-glucono-S-lactone, which hydro-

* Supported in part by research Grants AM-32974 and AM35362 from the National Institutes of Health. ** Correspondence to: William D. Geoghegan, Ph.D., Department of Dermatology, The University of Texas Health Science Center at Houston, 6431 Fannin, Suite 1.204, Houston, TX 77030, U.S.A.

lyzes to gluconic acid (Pazur and Kleppe, 1964; Bowin et al., 1976). Glucose oxidase is a flavoenzyme whose flavin adenine dinucleotide (FAD) is reduced to FADH 2 in the production of the lactone. The FADH 2 in turn reduces 02 to H202 to regenerate the FAD (Bowin et al., 1976). When mPMS, a synthetic electron carrier, is introduced into the system it competes with the 02 for the FADH 2 resulting in the formation of mPMSH. The mPMSH in turn reduces the tetrazolium to the formazan (Ponti et al., 1978). Mammalian tissues lack glucose oxidase, thus this enzyme offers an alternative to other enzymes, such as alkaline phosphatase, which is present in many tissues, and peroxidase which must compete with a variety of other molecules for the substrate H202. A variety of peroxidase competitors, viz. hemoglobin, cytochromes, may also produce a false positive color reaction in the presence of H202

0022-1759/86/$03.50 © 1986 Elsevier Science Publishers B.V. (Biomedical Division)

232 and diaminobenzidine (or another suitable electron donor). Porter and Porter (1984) have published a method for the use of glucose oxidase-labeled second antibody to detect viral antigens separated by SDS polyacrylamide gel electrophoresis and blotted onto nitrocellulose. However, they did not determine the sensitivity of their method for the antigen. Instead, they reported that a positive reaction for the viral antigens was obtained when the primary antibody was diluted to as little as 1 t~g/ml. We have attempted to determine the lower limit of detectability for monomeric human serum albumin and monomeric human IgG dotted onto nitrocellulose. In addition, we have demonstrated the usefulness of the reaction as applied to blots of isoelectrically focused proteins.

Materials and methods

Equipment Electrofocusing was performed on an LKB Ultrophor (LKB, Rockville, MD) powered by an LKB 2197 power supply. Blotting was performed in a Bio-Rad Tran-Blot apparatus (Bio-Rad Labs., Richmond, CA). Supplies Iso-Gel ampholytes (3.5-10), Isogel agarose and Gel Bond were purchased (Marine Colloids, Rockland, ME). Tween 20, Trizma base (Sigma, St Louis, MO), fl-D-glucose (Calbiochem, La Jolla, CA), p-nitroblue tetrazolium chloride, ultra pure (pNBT), 1-methoxyphenazine methosulfate (mPMS), iodonitrotetrazolium violet (INT) (Research Organics, Cleveland, OH). Glucose oxidase conjugated to affinity-purified goat IgG specific for rabbit or human IgG (Jackson Immuno Research Labs., Avondale, PA). Antibodies to human serum albumin (HSA) were raised in New Zealand White rabbits immunized with HSA in complete Freunds adjuvant. Pooled sera were precipitated at 50% saturation with ammonium sulfate pH 6.8 followed by chromatography on DEAE cellulose (DE52, Whatman, Clifton, N J) equilibrated with 0.01 M sodium phosphate buffer pH 7.6. The unbound peak was

passed over HSA conjugated to Sepharose 4B by cyanogen bromide (Parikh et al., 1975). The column was washed with 0.1 M borate-buffered saline (pH 8.5 containing 1.0 M NaC1 and 0.1% Tween 20 (Smith et al., 1978). The colunm was washed with PBS until Tween 20 was no longer detected at 280 nm. Antibody was eluted with 0.5 M acetic acid and immediately neutralized with Tris base, concentrated, dialyzed against 0.01 M sodium phosphate buffer pH 7.4 and stored at - 2 0 ° C .

Monomeric human serum albumin (HSA) HSA (Sigma A2386) was dissolved in 0.0175 M sodium phosphate buffer pH 6.2, centrifuged at 20 000 rpm for 1 h, sterile filtered and passed over a DE52 (Whatman) column (Habeeb, 1977). The albumin used in these experiments was eluted with 0.0175 M sodium phosphate buffer plus 0.075 M NaC1. The eluting peak (mercapalbumin) was collected and passed over a G150 column (Janatova et al., 1968) and the third peak, monomeric HSA, was dialyzed against distilled water and lyophilized. Stock solutions were made up from the lyophilized material. Monomeric human IgG Human serum was precipitated at 50% saturation with ammonium sulfate followed by chromatography on DE52 equilibrated with 0.01 M sodium phosphate buffer pH 8.0. The unbound peak was dialyzed against 0.01% ammonium bicarbonate on lyophilized. IgG was dissolved in 0.01 M PB, 0.9% NaCI pH 7.3 and centrifuged for 30 min at 435 680 x g in a Beckman TL-100 ultracentrifuge using a TLA-100.2 fixed angle rotor to sediment IgG dimers. The protein concentration in the supernate was determined by absorbance at 280 nm (Miller and Metzger, 1965). Dilutions were made from this stock immediately and 5 /tl quantities applied directly to nitrocellulose and dried. Disclosing reagent This reagent was modified from Clark et al. (1982). A solution of 0.05 M Tris base was adjusted to pH 8.3 by the addition of I M citric acid before diluting to volume. Both p-NBT and INT were soluble in this buffer (INT required warming

233 to 37 ° C). All incubations were performed at room temperature, since an increase in temperature resuits in a decrease in the pH of Tris buffers. A stock solution of mPMS was made up at 10 m g / m l in H 2 0 and stored at 4°C (Hisada and Yagi, 1977). The disclosing reagent was made up to the desired volume immediately before use and contained 6.7 mg fl-D-glucose/ml, 0.67 mg p - N B T / m l (or INT) and 0.0167 mg m P M S / m l . Because the glucose is already in the fl form, it is not necessary to wait for 1-2 h for mutarotation to occur. Therefore, a fresh solution may be made and used immediately. All reactions should be kept in darkness to prevent non-specific formazan production (Van Noorden et al., 1982).

Isoelectric focusing 2.5% Iso-Gel ampholytes in 1% Isogel agarose were poured on Gel bond (2 mm final gel thickness). The 10/~1 samples were applied to an LKB sample application wick and extracted into the gel for 15 min at 240 constant V. The sample wicks were removed and the gel focused at 14 W constant power for 45 min. The gel was laid on top of a piece of nitrocellulose already wet and in a container of 0.01 M Tris-acetate pH 7.5. The Gel Bond was carefully separated from a corner and then peeled away. The gel and nitrocellulose were sandwiched between a pre-wet Whatman no. 1 filter paper, two pieces on each side, before being sandwiched between two sponge pads. These were placed in a Trans-Blot cassette with the nitrocellulose to the anode. The blot was run at 225 mA, constant current, for 18 h at 4°C. After the blot, the nitrocellulose was dried and blocked in 4 mM KH2PO4, 16 mM HPOaNaz, 115 mM NaC1, 0.05% Tween 20 p H 7.3 (Batteiger et al., 1982) for 1 h. The blot was then incubated in 0.02 M sodium phosphate buffer, 0.25 M NaC1 0.05% Tween 20 p H 7.6 (PBS Tween) containing 10 /~g/ml affinity-purified rabbit IgG anti-HSA for 1 h. The blot was then rinsed three times, 10 rain each, in the same buffer and transferred to the glucose oxidase-labeled goat IgG anti-rabbit IgG-Fc (10 /~g/ml) in PBS Tween for 1 h. The blot was rinsed three times as before and placed in disclosing reagent for 10 min to 1 h. After reacting for the desired time, the blot was rinsed three times in distilled water and dried in the dark.

Proteins dotted onto nitrocellulose were treated in the same manner.

Results

The pH optimum for glucose oxidase is generally reported to be acidic, but reports of the actual p H optimum vary (Pazur and Kelppe, 1964; Gestrelius et al., 1973; Suffin et al., 1979). Our initial studies began at p H 6.3 in 0.01 M phosphate buffer. At this pH, the reaction required an hour in the disclosing reagent and sensitivity by the simple indirect procedure employing a glucose oxidase-labeled anti-antibody to detect HSA was limited to 20 ng HSA. However, the background was white or nearly white. Campbell and Bhatnagar (1976) reported that p H 8.2-8.6 was optimal for light microscopic studies employing glucose oxidase-labeled goat antibody disclosed in a mixture of PMS, NBT and glucose. However, NBT in the range of 8.4-9.2 reportedly reduces to a reddish formazan in the presence of protein bound sulfhydryl (Deguchi, 1964). Because of the reported reaction of NBT with sulfhydryl groups at pH 8.4-9.2, we tested the specificity of the reaction using HSA and IgG (Fig. 1). The HSA utilized was purified to enrich for monomeric mercapalbumin (containing one free sulfhydryl per molecule) (Habeeb, 1977). At p H 8.3, no band for albumin was observed in lanes 2 and 4, although a strong reaction for IgG, identified by glucose oxidase-labeled goat anti-human IgG, was observed. In lanes 6 and 8, where a simple indirect method was used to identify HSA, the albumin was readily identified while no reaction for IgG could be detected. INT also failed to produce a colored formazan at pH 8.3 in the absence of glucose oxidase. In our hands, the reaction time for glucose oxidase shortened considerably at pH 8.3 to less than 30 min and the sensitivity of the blots increased to 10 ng HSA (Fig. 2). When HSA was applied directly to the nitrocellulose in a 5 /~1 volume, it could be detected at 3-5 ng as a faint blue color. One disadvantage of pH 8.3, however, was that sometimes a bluish background color developed on blots, especially when the blotting time was longer than 10-12 h. The background

234

Fig. 1. Isoelectric focusing. Lanes 1-12, even numbers 80 ng human serum albumin, odd numbers 55 ng human IgG. Lanes 1-8 blotted on nitrocellUlose; lanes 1-4 incubated with glucose oxidase-labeled goat anti-human IgG; lanes 5-8 incubated with rabbit anti-HSA, followed by glucose oxidase-labeled goat anti-rabbit IgG; lanes 9-12 stained with Coomassie blue. Note specificity and low background of glucose oxidase reactions in lanes 1-8. Albumin not detected by Coomassie blue.

color faded away as the nitrocellulose dried overnight in the dark. HSA detectability was limited to 3-5 ng. When human IgG was the antigen, however, the limit of detection was considerably lower (Table I). Two buffer systems and two different tetrazolium compounds were tested. INT in Tris-citrate was slightly more sensitive than p-NBT in either buffer. But the difference faded by the next day. Nitrocellulose exposed to NBT must be stored in the dark. When left in the light, the sheets slowly develop a blue-purple background color. Dried and kept in the dark, they remain white indefinitely. Background color development occurs because the entire nitrocellulose sheet, even after blocking with Tween 20, is coated with a colorless layer of NBT. Squares of nitrocellulose were blocked with Tween 20, dried, incubated in NBT and rinsed exhaustively with water. When

TABLE I COMPARISON OF BUFFERS A N D TETRAZOLIUMS IN T H E DETECTION OF IgG Tris-HCl pH 8.3

Tris-citrate pH 8.3

p-NBT

INT a

p-NBT

INT

20 ng

+

+

+

+

10 n g

+

+

+

+

1 ng

+

+

+

+

0.5 ng 250 pg 100 pg 50 pg 25 pg

+ + -. .

+ ND b ND . .

+ + +

+ + +

. .

. .

a I N T did not dissolve completely in Tris-HCl even after warming to 37°C. It dissolved completely in Tris-citrate after warming to 37°C for 10 min: p-NBT was soluble in both buffers. b Not done.

,235

Fig. 2. Isoelectric focusing of purified HSA. HSA concentration: lane 1, 160 ng; 2, 80 ng; 3, 40 ng; 4, 20 ng; and 5, 10 ng. Reaction time in disclosing reagent at pH 8.3 is 20 min. Note the rapid decrease in intensity with dilution of the HSA. Circles around the two lanes mark the position of the faint reaction product. The reaction product is more easily observed on the white nitrocellulose.

these squares were exposed to alkaline ascorbic acid, pH 13, the entire square turned blue (Altman, 1976).

Discussion We have modified the histochemical procedure of Clark et al. (1982) by substituting mPMS for

PMS. mPMS, synthesized by Hisada and Yagi, (1977) is stable in scattered light, reduces more quickly than PMS and can be stored in a stock solution without loss due to photochemical deterioration. It has also been reported to be superior to PMS and all other electron carriers (Van Noorden et al., 1982; Van Noorden and Tas, 1982; Van Noorden and Butcher, 1984). In addition, mPMS does not stain cellular components (Van Noorden et al., 1982) and in our hands did not react with sulfhydryl or disulfide bonds to produce a visible color. We prefer to run the reaction at p H 8.3 because the reaction time is much shorter and the sensitivity is slightly enhanced. The disadvantage of using p H 8.3 is that a non-specific background color, which fades away overnight, sometimes develops on blots but was never observed on dots. The cause of this temporary blush of color is not known, but it is likely to be related to the substantivity of N B T (Altman, 1976). NBT when added to nitrocellulose, blocked with Tween 20, binds evenly over the entire surface. In areas where the reduced form of mPMS is deposited, NBT is reduced to the formazan and a permanently colored compound is generated (Raap, 1983). A quantitative study of the sensitivity of the glucose oxidase procedure applied to protein adsorbed to nitrocellulose blots has not been published. Porter and Porter (1984) reported that a minimum of 1 /xg/ml of primary antibody could be used to detect an unknown quantity of viral antigen. To put the detection limits in perspective with respect to the antigen, we chose to study monomeric HSA and monomeric human IgG. HSA is a medium sized protein (MW 65000) which reportedly adsorbs side on (Janatova et al., 1968), while IgG is larger (MW 146000), and reportedly adsorbs side on at low concentrations and end on at higher concentrations (Morrissey and Han, 1978). The role of molecular orientation and detectability on nitrocellulose has not been investigated, but the selection of an antibody to antigenic structures on the non-adsorbed surface of the protein would probably increase their detectability. We found that the detection limit varied with the antigen. IgG could be detected at 100 pg, while HSA was not detectable below 3-5 ng. We

236 were unable to identify similar reports where the detectability of two carefully purified m o n o m e r i c antigens were compared. Several investigators have given values for the detection of I g G (not specifically depleted of I g G polymers which would increase the density of antigenic sites facilitating the development of color on nitrocellulose or an image on X-ray film). Hawkes et al. (1982) reported that 100 pg of h u m a n I g G could be detected with horseradish peroxidase (HRP)-conjugated goat anti-human I g G (equal to the m i n i m u m quantity of I g G detected by glucose oxidase). Porter and Porter (1984) stated that glucose oxidase provided a sensitivity similar to H R P . D a o (1985) was able to detect 12 pg of alkaline phosphatase-conjugated goat I g G dotted directly onto nitrocellulose. Finally, T o w b i n and G o r d o n (1984) state that radiolabeled antibodies can detect 1 pg of I g G (no reference given). A carefully controlled study utilizing the identical a n t i b o d y conjugated with the different enzymes or 1251 against the same preparation of antigen would resolve the question of sensitivity. Like Porter and Porter (1984), we f o u n d that the three-stage glucose oxidase anti-glucose oxidase ( G A G ) methods result in a 10-fold increase in sensitivity and an increase in b a c k g r o u n d color development. We did not pursue this rout because of the expense and because of the increased b a c k g r o u n d a c c o m p a n y i n g the use of G A G . We did not explore the reasons for the difference in detectability between H S A and IgG, however, the n u m b e r and spacing of available antigenic sites on the adsorbed proteins is probably important. The major weakness of glucose oxidase is the increasing faintness of the reaction p r o d u c t as one approaches the limit of detectability of the antigen. Its major strengths are low b a c k g r o u n d color and the absence, in m a m m a l i a n systems, of competing substances.

References Altman, F.P., 1976, Prog. Histochem. Cytochem. 6, 1. Batteiger, B., W.J. Newhall and R.B. Jones, 1982, J. Immunol. Methods 55,297. Bowin, J.C., M.T. Atallah and H.O. Hultin, 1976, in: Methods in Enzymology, Vol. 44, ed. K. Mosbach (Academic Press, New York) p. 478. Campbell, G.T. and A.S. Bhatnagar, 1976, J. Histochem. Cytochem. 24, 448. Clark, C.A., E.C. Downs and F.J. Primus, 1982, J. Histochem. Cytochem. 30, 27. Dao, M.L., 1985, J. Immunol. Methods 82, 225. Deguchi, Y., 1964, J. Histochem. Cytochem. 12, 261. Gestrelius, S., B. Mattisasson and K. Mosbach, 1973, Eur. J. Biochem. 36, 89. Habeeb, A.F.S.A., 1977, in: Immunochemistry of Protein, Vol. 3, ed. M.Z. Atassi (Plenum Press, New York) p. 197. Hawkes, R., E. Niday and J. Gordon, 1982, Anal. Biochem. 119, 142. Hisada, R. and T. Yagi, 1977, J. Biochem. 82, 1469. Janatova, J., J.K. Fuller and M.J. Hunter, 1968, J. Biol. Chem. 243, 3612. Miller, F. and H. Metzger, 1965, J. Biol. Chem. 240, 3325. Morrissey B.W. and C.C. Han, 1978, J. Colloid Interface Sci. 65, 423. Parikh, I., S. March and P. Cuatrecasas, 1975, in: Methods in Enzymology, Vol. 34, eds. W.B. Jakoby and M. Wilchek (Academic Press, New York) p. 77. Pazur, J.H. and K. Kleppe, 1964, Biochemistry 3, 578. Ponti, V., M.U. Dianzani, K. Cheeseman and T.F. Slater, 1978, Chem. Biol. Interactions 23,281. Porter, D.D. and H.G. Porter, 1984, J. Immunol. Methods 72, 1. Raap, A.K., 1983, Histochem. J. 15, 977. Rathles, T. and G.F. Franks, 1982, Am. J. Clin. Pathol. 77, 705. Smith, J.A., J.G.R. Hurrell and S.J. Leach, 1978, Analyt. Biochem. 87, 229. Soderquist, M.E. and A.G. Walton, 1980, Science 75, 386. Suffin, S.C., K.B. Much, J.C. Young, K. Lewin and D.D. Porter, 1979, Am. J. Clin. Pathol. 71,492. Towbin, H. and J. Gordon, 1984, J. Immunol. Methods 72, 313. Van Noorden, CJ.F. and R.G. Butcher, 1984, J. Histochem. Cytochem. 32, 998. Van Noorden, CJ.F. and J. Tas, 1982, Histochem. J. 14, 837. Van Noorden, C.J.F., F. Tas and J. Tas, 1982, J. Histochem. Cytochem. 30, 12.