Transblot studies with biotin-labeled proteins: Electrophoretic mobilities and detection limits

ANALYTICAL

BIOCHEMISTR\’

Transblot

MICHAEL

156, 76-80

( 1986)

Studies with Biotin-Labeled Proteins: Electrophoretic Mobilities and Detection Limits NEUMAIER,

URSULA

Received

FENCER,

October

AND

CHRISTOPH

WAGENER

15, 1985

Molecular weight standard proteins. mouse IgG as well as several antigens cross-reacting with the carcinoembryonic antigen (CEA). were biotin labeled. submitted to sodium dodecyl sulfatepolyacrylamide gel electrophoresis. and transferred to nitrocellulose. The bound proteins w’err revealed by the use of avidin-peroxidase conjugates and a suitable substrate. The ratio of N-hydrosuccinimido biotin (NHSB) to protein yielding the lowest detection limit was determined. At an optimal NHSB/protein ratio. 0.33 ng of IgG heavy chains and 0. I7 ng of IgG light chains could be visualized. With the exception of human albumin and ovalbumin. the increase in apparent molecular weight after biotin labeling was less than 10% for the proteins tested. The method has proven to be a valuable addition to Western blots performed with CEA-related antigens and monoclonal anti-CEA antibodies. 80 1986 Academic PWSS. inc. KEY WORDS: biotin label: protein blotting: protein staining: nitrocellulose: electrophoresis.

The electrophoretic transfer of proteins from polyacrylamide gelsto nitrocellulose (1) has made possiblea highly sensitive detection of proteins when probes, e.g., antibodies or lectins, are used(L3). An equally sensitive detection of transferred proteins without the use of specific ligands, however, is difficult to assess.Protein stains such as amido black or Coomassieblue are not sensitive enough by far. Improvements in sensitivity were achieved by the introduction of Indian ink (4) and, very recently, of colloidal metals (5). Here we describe an alternative method for the visualization of proteins on nitrocellulose. We show that by the use of biotin-labeled proteins and avidin-peroxidase conjugates, it is possibleto detect proteins in sensitivity ranges comparable to those of immunochemical techniques. The majority of proteins tested showed only slight deviations in electrophoretic mobility at biotin substitution rates yielding the highest sensitivities of detection.

0003~2697/86

$3.00

Copyright Q 1986 by Academic Press. Inc. All rkbts of reproduction m any form reserved.

MATERIALS

AND

METHODS

hlaterials. High-molecular-weight standards were from Sigma, Taufkirchen. FRG: cytochrome c was from Boehringer-Mannheim. FRG. Avidin-peroxidase conjugate was from Sigma. Mouse IgG’ of subclass2a was isolated from mouse ascitic fluid according to Ey et al. (6). The subclasswas determined by enzyme immunoassay as described (7). CEA and related antigens were isolated by differential immunoaffinity chromatography using monoclonal antibodies with different epitope specificities (8). Biotinylation procedure. The biotinylation procedure followed the protocol of Guesdon r Abbreviations used: BGP-I. biliary glycoprotein I; CEA. carcinoembryonic antigen; DMF, dimethylformamide: NHSB. N-hydrosuccinimido biotin: IgG. immunoglobulin G; RT. room temperature: PBS. phosphatebuffered saline: SDS-PAGE. sodium dodecyl sulfatepolyacrylamide gel electrophoresis; BSA, bovine serum albumin; TBS. Tris-buffered saline.

76

TRANSBLOT

STUDIES

WITH

et uf. (9). For the biotinylation of high-molecular-weight standards, 6 mg of the protein mixture was dissolved in 0.1 M sodium bicarbonate buffer. pH 8.3, at a final concentration of 12 g/liter. A 0.0 1 M solution ( 12-300 ~1) of NHSB in DMF (3.4 1 g/liter) was added to 100 ~1 of the protein sample. After incubation for 1 h at RT, the sampleswere made up to 1.2 ml and dialyzed exhaustively against PBS. Mouse IgG and CEA-related antigens were biotin labeled using a protein concentration of 10 mg/ml and a volume ratio of protein to NHSB solutions of 2: 1. SDS-R4G‘E and W‘esterublots. SDS-PAGE was performed under reducing conditions according to Laemmli (10). The acrylamide concentration of the separation gel was 7.5% for the runs performed with molecular weight standards and CEA-related antigens. For the biotin-labeled mouse IgG, an acrylamide concentration of 10% was used. The electrophoretie transfer of proteins to nitrocellulose was performed essentially according to Towbin el (I/. (1) at 10-l 1 V/cm for 5 h. Biotir~jVaticm of’proteins afier tran.$~r to nitrocel/do.re. To determine the detection limit of proteins biotinylated after transfer to nitrocellulose. diKerent amounts of molecular weight standard proteins (Sigma) were dissolved in 0.1 kt sodium bicarbonate buffer. pH 8.3. and soaked into nitrocellulose. Twentyfive-microliters resulted in dots with diameters of approximately 4 mm. Strips of nitrocellulose were incubated in a solution of 6 ~1of 0.1 M NHSB in DMF and 100 ~1 ofO.1 M sodium bicarbonate buffer, pH 8.3, for 1 h at RT. Staikng qf proteins. After protein transfer. the unspecific binding sites of the nitrocellulose were blocked by a solution of 5% BSA in 0.0 1 M Tris-HC1 containing 0.15 mol NaCl/ liter. pH 7.4 (BSA-TBS) for 30 min at 40°C. The nitrocellulose wasthen transferred to a 1: 200 dilution of avidin-peroxidase conjugate in BSA-TBS containing 10% normal rabbit serum (1 h, RT). After the nitrocellulose was washed with 0.05 M Tris-HCl. pH 7.4, the bound peroxidase was incubated with 0.05 M

BIOTIN-LABELED

77

PROTEINS

Tris-HCl, containing 1.26 mmol 3,3’-diaminobenzidine tetrahydrochloride and 0.005% H202 per liter (8). The staining reaction was stopped by rinsing the nitrocellulose with 0.05 M Tris-HCl, pH 7.4. The immunochemical staining of CEA and related antigens was performed as described (8). RESULTS

Optimal NHSB/protein ratio. The ratio of NHSB and protein solutions yielding maximum sensitivity for the detection of proteins after transfer to nitrocellulose was tested by using a mixture of high-molecular-weight standards. To a constant volume of 100 ~1 of protein solution ( 12 g/liter), variable volumes of a 0.01 M NHSB solution in DMF were added. The highest sensitivity was obtained at NHSB to protein ratios (w/w) of 0.17 to 0.33 (Fig. 1). At a ratio of 0.83, the sensitivity de-

200000

-

66000

-

45000

t

I 12

I

I

I

I

I

I

3

4

5

6

78

I

!

I

I

91oll

FIG. I. Transblots of molecular weight standards biotin labeled at different NHSBiprotein ratios (milligrams NHSB/milligram protein in the reaction mixture used for the biotinylation procedure). Lanes l-4: NHSB/protein = 0.17. 1. 2500 ng: 3. 150 ng: 3, 25 ng; 4, 2.5 ng. Lanes 5-8: NHSB/protein = 0.33. 5. 2.5 ng; 6, 25 ng: 7, 250 ng: 8, 2500 ng. Lanes 9-l I: NHSB/protein = 0.83. 9. 2500 ng: IO. 250 ng: I I. 25 ng.

78

NEUMAIER.

FENGER,

clined (Fig. 1. lanes 9- 11). The same was true when ratios lower than 0.17 were used (blots not shown). The introduction of biotin residues into proteins may alter their electrophoretic mobility. For this reason, the apparent molecular weights of the molecular weight standards were calculated relative to unlabeled standards for different NHSB/protein ratios (Table 1). For albumin, there is a significant increase in apparent molecular weight up to a NHSB/protein ratio of 0.17. In the molecular weight range of ovalbumin, two bands corresponding to 46,000 and 53,000, respectively, are present. The electrophoretic mobility of the remaining biotinylated standard proteins deviates only slightly from the unlabeled standards (Table 1). The same finding applies to the apparent molecular weights of the biotin-labeled heavy chain (54.000) and light chain (27,000) of mouse IgG (Fig. 2).

Sensitivity of’ the detection yf‘hiotin-labeled l,qY. Monoclonal mouse IgG was biotin labeled at an NHSB/protein ratio of 0.17. Transblots were performed with decreasing amounts of IgG. The least amount detected is 0.5 ng corresponding to 0.33 ng of heavy chains and 0.17 ng of light chains (Fig. 2).

TABLE APPARENT

MOLECULAR BIOTIN-LABELED

I

WEIGHTS

OF UNLABELED

STANDARD Apparent

AND

PROTEINS

molecular

weight

(kD)

NHSB/protein ratio” Standard proteins Myosin Catalase Phosphorylase Albumin Ovalbumin

Unlabeled proteins

h

200 116 93 66 45

0.08

0.17-0.83

>200 121

101 IS 46

52 Carboanhydrase

31

a Milligrams protein.

IV-hydrosuccinimidobiotin/milligram

32

2200

120 101 85 46 52 32

AND

WAGENER

200000

-

93000

-

66000

-

25000

-

18000

=

116000

I I I 123L56 FIG. 1. 2500 0.5 ng.

I

I

I

2. Transblot of biotin-labeled mouse IgG za. ng: 2, 500 ng; 3. 100 ng: 4. 20 ng; 5. 2.5 ng; 6.

Tramblots with biotin-labeled proteins as purit!> control oj‘CEA-related antigens. Antigens cross-reacting with CEA were isolated from human bile, spleen tissue, and meconium by perchloric acid extraction and immunoaffinity chromatography using monoclonal antibodies (8). The antigen preparations were biotin labeled at NHSB/protein ratios of approximately 0.17. As demonstrated in Fig. 3, lane 1, the bile protein of M, 84,000 which corresponds to the cross-reactive biliary glycoprotein I (BGP-I) is contaminated by a protein of JV~ 55,000 which did not bind the monoclonal antibodies in Western blots (8). Similar to the Western blot data, the CEArelated spleen protein shows up as a single major band (&fr 52,000) (Fig. 3, lane 2). In addition to the two CEA-related high-molecular-weight meconium antigens, a minor lowmolecular-weight contaminant is present which did not show immunoreactivity (Fig. 3, lane 3). The comparison of the apparent molecular weight of the biotin-labeled antigens

PRANSBLOT

200 000

m

116 000

w

93 000

-

66 ,300

-

STUDIES

WITH

S”

45 300

m

29 000

),

I

I

I

3

12

FIG. 3. Transblot of biotin-labeled CEA-related antigens from bile. spleen and meconium. I. Bile antigens. 3 pg: antigen. 1 pg. 2. spleen antigen. 3 fig: 3. meconium

and of the immunoblotted in Table 2. Stuhing ofprotcirzs tn rlitrol’L’lllrlo.,c’.

antigens is shown

biotirr?Yatd

qfirt. traru-

Molecular weight standards were immobilized on nitrocellulose at amounts of 3’0. 3, 0.3, and 0.03 pg. respectively. As DMF dissolutesnitrocellulose when applied at high concentrations, the maximum volume of Dh4F which could be added to IO0 ~1 of sodium bicarbonate buffer without affecting the integrity of the nitrocellulose amounted to 6 ~1. At concentrations higher than 0.1 mot/liter DMF, NHSB was not readily dissolved. The incubation volume was kept small in order to reach high protein labeling compared to hydrolysis of NHSB. Under theseoptimal experimental conditions, the detection limit was 3 pg protein per dot.

,/k

DISCUSSION Biotin-labeled proteins can be detected with high sensitivity after electrophoretic separation and transfer to nitrocellulose using avidinperoxidase conjugates and 3,3’-diaminobenzidinetetrahydrochloride as a substrate. A

BIOTIN-LABELED

79

PROTEINS

similar approach has been described by Anderson ef al. (1 1) who immunoprecipitated biotinylated viral proteins which were then separated by SDS-PAGE, transferred to nitrocellulose, and visualized by avidin-peroxidase conjugates. Here, we present studies on the effect of biotinylation on the electrophoretie mobilities of molecular weight standard proteins. Furthermore we determined the detection limits for biotinylated IgG heavy and light chains and compared the sensitivity of the blots performed with biotinylated proteins with the sensitivity of Western blots for which monoclonal antibodies were used as specific probes (8). By the technique described, IgG heavy and light chains were detected in the picogram range. Thus. the detection limit is at least as low as that reached by immunochemical and related techniques. As the method works with a variety of difl’erent proteins and glycoproteins. it seems to be of general applicability. The nntroduction of biotin residues affected the electrophoretic mobility of the various proteins to a different extent. At NHSB/protein ratios yielding highest sensitivities, the molecular weights of human albumin and ovalbumin increased significantly. For the other proteins tested, however, the increase was lessthan IO%).At higher NHSB/protein ratios. the molecular weight of the standard TABLE APPAREN r MOLECULAR BIOTIN-LARELED

WEIGHTS OF UNLABELED C&I-RELATED ANTIGENS Apparent

Source of antigen Meconium

2

molecular

Antigens detected with monoclonal anti-CEA antibodies” 165.000 I00.000

weight

Biotm-labeled antigens 168.000 106.000

Bile

8 1.OOO

84,000 53.000

Spleen

49,000

52.000

‘See Ref. (8).

AND

80

NEUMAIER,

FENGER.

proteins remained constant. It may be concluded that the increase in apparent molecular weight is related to the number of free amino groups available for the reaction with NHSB. In addition to the actual increase in molecular weight, the hydrophobic biotin residues may affect the SDS-binding properties of proteins to a different extent. The Western blots with biotin-labeled proteins were used as a test for purity of CEArelated antigens previously purified by perchloric acid extraction and differential immunoaffinity chromatography from human bile, spleen, and meconium using monoclonal antibodies with different epitope specificities (8). The staining pattern of the spleen antigen is well comparable to that obtained with monoclonal anti-CEA antibodies. Thus, this preparation does not contain significant amounts of proteins nonreactive with the monoclonal antibodies. In the antigen preparations from human bile and meconium, additional bands in the range of blr 55,000 were detected by using the biotin-labeled antigen preparation. These bands did not bind monoclonal anti-CEA antibodies in Western blots (8): they may represent contaminants associated with the CEA-related antigens as described previously (12). Though CEA-related antigens were biotinlabeled after immunosorbent purification, we would anticipate that biotinylation of a protein mixture may be performed prior to the immunosorbent step as the biotinylation procedure does not affect the immunological properties of CEA ( 13). Recently, a procedure has been described by which haptens were coupled to proteins transferred to nitrocellulose ( 14). The proteins were then visualized by enzyme-conjugated anti-hapten antibodies. This technique does not affect molecular weight determinations. Using a similar approach, we biotinylated proteins immobilized on nitrocellulose. The maximum concentration of NHSB which could be reached for this labeling technique was limited by the effect of DMF on the nitrocellulose and by the limited solubility of

AND

WAGENER

NHSB. Under optimal conditions, the least detectable amount of protein was 3 pg. Lower NHSB concentrations resulted in lower detection limits. Thus, the sensitivity of this method is an order of magnitude lower than that obtained by the use of prelabeled proteins. In conclusion, biotin-labeled proteins can be detected in the subnanogram range after separation by SDS-PAGE and electrophoretic transfer to nitrocellulose using avidin-peroxidase conjugates and an appropriate substrate. The method has proven to be a valuable addition to Western blots performed with CEArelated antigens and monoclonal anti-CEA antibodies. X~“io/e u&l& in pvoc$ After submission of this manuscript. a method has been described by which surface proteins on leukocytes were identified using a biotin-avidin protein blotting technique (W. L. Hurley, E. Finkelstein. and B. D. Holst. 1985, J. Immz~nol. ,&f&l&s 85, 195-202.).

REFERENCES I. Towbin. H., Staehelin. T.. and Gordon. J. ( 1979) Proc. IWL .ld Sci. IW 16, 4350-4354. 2. Bumette, W. N. (198 I) And. Biochem. 112, 195-203. 3. Gershoni. J. M.. and Palade, G. E. (1983) .&ul. Bitd1etn. 131, I-15. 4. Hancock. K.. and Tsang. V. C. W. (1983) Atul. Biochem. 133, l57- 162. 5. Moeremans, M., Daneels. G., and De Mey, J. (1985) And Biochetn. 145, 3 15-32 I. 6. Ey, P. L.. Prowse. S. J., and Jenkin. C. R. (1978) Imnz~tt~ochrtn~rrr:l~ 15, 429-436. 7. Wagener. C.. Yang. Y. H. J., Crawford, F. G.. and Shively. J. E. ( 1983) J. Immunol.130, 2308-23 15. 8. Neumaier. M., Fenger. U.. and Wagener. C. (1985) J Immunol. 135, 3604-3609. 9. Guesdon. J.-L.. Temynck, T., and Avrameas. S. ( 1979) J. Hisfochem C’ymchem 27, 1 13 I -I 139. IO. Laemmli. U. K. ( 1970) Nuture (London) 227, 680685.

I I. Anderson. holzer,

L. J., Coombs. R. A.. Tsou. C.. and HierJ. C. (1984) J. Clin. Microhiol. 19, 934-

936.

12. Orjasaeter. H. (1976) Acta Pui/zol. .Uicrohio/. I?ntnwud. Scund (C) 84, 235-24 I 13. Shively. J. E.. Wagener. C., and Clark, B. R. ( 1986) in Methods in Enzymology. Academic Press, Orlando. Ra., in press. 14. Kittler, J. M.. Meisler, N. T., Viceps-Madore, D.. Cidlowski. J. A.. and Thanassi, J. W. (1984) At&. Biochem. 137,2 10-2 16.