key: cord-022200-hqc8r31t
authors: HYATT, ALEX D.
title: Protein A–Gold: Nonspecific Binding and Cross-Contamination
date: 2012-12-02
journal: Colloidal Gold
DOI: 10.1016/b978-0-12-333928-7.50007-1
sha: 
doc_id: 22200
cord_uid: hqc8r31t

nan

Colloidal gold has been a popular electron-dense marker in immunoelectron microscopy for the past decade. Over this period protein A-gold has been used extensively as a reagent for binding immunoglobulins from many mammalian species. Due to the broad reactivity of the probe and the ease of preparation, protein Α-gold remains a popular choice as an immunolabel. The aim of this chapter is to discuss the potential problems associated with protein Α-gold as stated in the literature and as recog nized in this laboratory. To understand the potential difficulties, it is nec essary first to discuss the techniques involved with the preparation of colloidal gold and protein Α-colloidal gold conjugates.

Colloidal gold is generally prepared by the reduction of tetrachloroauric acid, H(AuCl 4 )-4H 2 0, with a range of chemicals including phosphorus (Zsigmondy, 1905; Zsigmondy and Thiessen, 1925) , trisodium citrate (Frens, 1973) , ascorbic acid (Stathis and Fabrikanos, 1958) , sodium borohydride (Tschopp et aL, 1982) , and trisodium citrate and tannic acid (Slot and Geuze, 1985) . The gold colloid is formed from micromolecular units (nuclei) and the increase in particle size is dependent on the rate of forma tion of nuclei and the rate of crystal growth. Thus, by controlling the reduction of tetrachloroauric acid, gold probes of varying diameters can be produced. Reduction with white phosphorus, ascorbic acid, or triso dium citrate, for example, produces gold particles with average diameters of 5, 12, and 16 nm, respectively (Slot and Geuze, 1981) . For further de tails, see Handley (1989) .

An important surface property of colloidal gold is its negative charge. In an aqueous phase, colloidal gold will stay in suspension by electro static repulsion. If electrolytes are added, the ion layers of the particles are compressed and so the critical "cohesion" distance is reduced, thereby facilitating flocculation. It has been recognized since 1901 (Zsig mondy, 1901) that gold particles can be stabilized in solution by the addi tion of proteins. One of the more successful of such proteins is Staphylo coccus aureus coat protein A. The complexing of protein A (as with other proteins) to colloidal gold involves electrostatic interaction between the negatively charged surface of gold particles and positively charged groups of the proteins. It is generally accepted that the protein is attracted into the action radius of van der Waals-London attractive forces and there firmly binds to the surface of the gold particles.

The important parameters for successful preparation of stable and ac tive protein Α-gold complexes are the pH of the colloidal gold suspension (Roth, 1983) and the determination of the minimum amount of protein required for stabilization. Although the isoelectric point for protein A is 5.1, complexes are generally produced at a pH range of 5.9 to 6.2. How ever, adequate protein binding can be obtained at a pH between 5 and 6 (Slot and Geuze, 1981) ; the overall bioactivity and stabilization do not appear to be compromised in such complexes. The amount of protein A required for stabilization of colloidal gold is generally determined by the technique described by Zsigmondy and Thiessen (1925) . The test results in a color change of the colloidal gold solution from orange-red to blue when there is incomplete stabilization. The minimum amount of protein A required to prevent this effect is taken as an estimation for the stabiliza tion of the gold solution. It is common practice to add 10 to 100% excess protein A to the solution to ensure stabilization Geuze, 1981, 1985) . The protein A complex is then stored in a buffer (e.g., phosphatebuffered saline, PBS) containing an excess of unrelated protein such as bovine serum albumin (BSA); this further stabilizes the gold complex.

The protein Α-gold method is a two-stage immunolabeling procedure. The antigenic site within or on the specimen is revealed (indirectly) by the addition of a primary antibody. The excess antibody is removed and protein Α-gold added. Protein A has two functional Fc binding regions (Langone, 1982) , which interact predominantly at the CH 2 and CH 3 do mains of the Fc region of the antibody (Forsgren and Sjöquist, 1966) . The binding is strong in most IgG classes of several mammalian species such as human, rabbit, guinea pig, and dog; weaker with IgG classes from horse, cow, and mouse; and weakest in IgG from goat, sheep, rat, and chicken (Forsgren and Sjöquist, 1966; Kronvall et al., 1970 Kronvall et al., , 1974 Biberfeld et al., 1975; Lindmark et al., 1983) . Binding to different classes of IgG, human IgA, and IgE and IgM from various species has also been shown (Goudswaard et al., 1978; Langone, 1982; Lindmark et al., 1983) . It is obvious from the above that antibodies used in protein Α-gold proce dures should be generated from species which produce high-affinity anti body-protein A complexes. Furthermore, the antibodies should be of the one class, that is, they should be affinity-purified or monoclonal antibodies.

Labeling of antigens can occur prior to embedding (preembedding immunocytochemistry) or postembedding of whole or sectioned specimens.

Labeling can therefore occur either on or within tissue sections or whole mounts (for example, whole cells and complete cytoskeletal matrices). Irrespective of the mode of antigenic presentation, the labeling protocol generally follows a set format such as that outlined below. Labeling of structures in Fig. 2.1a ,b was produced by the same protocol, the details of which are shown in parentheses. The incubations were performed in plastic petri dishes and solutions changed with micropipettes ( Fig. 2. 2).

1. Wash (PBS, 3 min). 2. Fixation (0.1% glutaraldehyde in PBS, 2 min; when cytoskeletons are prepared the fixative includes 1% of the non-ionic detergent NP40).

3. Wash (PBS, 3x3 min). 4. Blocking step [PBS containing 1% BSA (PBS-BSA), 10 min]. 5. Primary antibody (1 : 10 dilution in PBS-BSA, 37°C (310 Κ), 1 hr). 6. Wash (PBS-BSA, 6x3 min). 7. Protein A-gold (1 : 20 in PBS-BSA, 37°C (310 Κ), 1 hr). 8. Wash (PBS, 6x3 min). 9. Stain (2% phosphotungstic acid adjusted to pH 6.8 with 1 Μ KOH).

Provided the antibodies are affinity purified or are monoclonal, the anti gen-antibody interactions will be specific and indicative of the presence and localization of the antigen. This assumes that the antigen has not been leached or translocated from or within its biological matrix due to inap propriate fixation. The possibility of "unwanted" antibodies, from whole sera, reacting with cell constituents is discussed in the following section.

Gold particles of varying sizes can be produced by different methodolo gies (Handley, 1989) . When the gold particles vary in size, i.e., their coef ficient of variation (CV) is greater than 15% (Slot and Geuze, 1985) , the solutions are considered polydispersive in size; when the CV is less than 15% they are referred to as monodispersive. The production of monodispersive colloidal gold is dependent on the method for synthesis (Han dley, 1989) and subsequent treatment (Bendayan, 1984) . The ability to produce solutions of monodispersive gold particles is of paramount im portance in multiple-labeling immunoelectron microscopy. The advantage of protein A is that it can be readily complexed to a gold probe of any size and possesses an affinity for most IgG of most mammals. Protein Α-gold probes therefore provide a potential means whereby colocalization of antigens within a single sample can be achieved. In the literature there are numerous reports which claim success in double labeling of anti gens within/on the one biological section. The techniques include colocal- ization of antigens on one face of a section with probes of different sizes (Geuze et al, 1981; Roth, 1982; Hisano et al, 1984) or the labeling of both faces of a section (Bendayan, 1982) .

In the first technique addition of each antibody is followed by protein Α-gold probes of specific sizes in a defined sequence. The first protein Α-gold probe to be applied is the smaller of the two (3-5 nm), followed by the larger probe (12-15 nm). When this sequence is followed, little cross-contamination occurs (Roth, 1982) . If free protein A is added during the last minutes of step 7 (see above) little cross-contamination is re ported to occur irrespective of which probe is applied first (Slot and Geuze, 1984) . If, however, the order of probes is reversed without the addition of free protein A, cross-contamination can occur (Roth, 1982) . The protocol for this form of double labeling is a simple extension of that outlined for single labeling. 1-7. Same as on p. 22.

8. Free protein A (= 0.05 mg/ml) in the last few minutes of step 7. 9. Wash (PBS, 6x3 min). 10. Repeat steps 5 to 7.

While this procedure has enjoyed some success, there have been nu merous reports of problems associated with it (Bendayan, 1982; Ben dayan and Stephens, 1984; Hyatt et al., 1988 ). An alternative procedure is described by Bendayan (1982) . In this procedure both sides of the sec tion are used; that is, there are t w o independent labeling steps. This sec ond technique avoids any cross-contamination b e t w e e n the different im munolabeling steps.

N o t all double labeling is performed on biological sections. W h e n only one aspect of a sample is available for labeling (for example, viruses ad sorbed to a grid substrate), then successful double labeling, as illustrated in Fig. 2 .3, may not be possible with either of the above m e t h o d s . H y a t t et al. (1988) , for example, found that protein Α -g o l d could not be used for the multiple epitope mapping of A k a b a n e virus as it resulted in colabeling of the primary antibody. Successful double labeling could be achieved only with specific IgG-gold probes (Fig. 2.3) . F o r additional in formation on multiple labeling, the reader is referred to Doerr-Schott (1989) .

Some potential problems associated with protein Α -g o l d labeling are also associated with other immunocytochemical techniques, namely non- specific staining (for example, nonspecific antibodies and electrostatic at tachment; Taylor, 1978; Behnke et al., 1986; Gosselin et al., 1986; Birrell et al., 1987) . Double immunolabeling with the protein A method can in duce further sources of error, namely cross-contamination of antibodies with protein Α-gold. The problems of cross-contamination and nonspe cific labeling are discussed below (also see Birrell and Griffith, 1989; Park etal., 1989) .

As stated previously, it is advisable to use monospecific antibodies raised in the appropriate mammalian species for the primary localization of antigens. From a practical viewpoint, affinity-purified and monoclonal antibodies generally yield specific labeling with a clean background ( Fig.  2.1a) . The labeling in Fig. 2 .1b, on the other hand, was produced with a whole porcine serum as the primary antibody. Although 1% BSA was used to minimize nonspecific labeling (as per Fig. 2.1a) , the background in Fig. 2 .1b was considerably greater. An explanation for this is that the use of whole serum (containing numerous natural antibodies, targeted an tibodies, and antibodies to any carrier molecules or extraneous material carried over or used in the immunization) with protein Α-gold produces many immunocomplexes. The overall effect can result in the adsorption of unwanted antibodies and subsequent labeling of various cellular con stituents in addition to the targeted antigen. If normal serum is used to block nonspecific binding of primary antibodies to the reactive sites of tissues, an effect similar to that described above may occur. In many instances the problem of such nonspecific binding may be reduced by decreasing the effective antibody concentration. If the studies involve in fectious agents, hormones, or the like (i.e., cause-and-effect experi ments), then nonspecific staining can be further reduced by preadsorption of the serum with the normal (nonexperimental) biological tissue.

Biological samples bathed in serum may also provide another source of error (for example, tissue culture cells and cells of the blood). It is not unreasonable to expect that during specimen preparation some antibodies may fortuitously bind to the sample surface. Such antibodies may provide additional binding sites for protein A. These samples should be thor oughly washed (e.g., with PBS) before labeling is attempted.

In double-labeling experiments cross-contamination is now recognized as a problem which arises from the affinity which protein A possesses for different regions of IgG antibodies, namely the Fc and Fab regions (Roth, 1982; Endresen, 1979; Zikan, 1980) . Protein A can bind two IgG mole cules; thus, after the primary antibody-protein A complex has been formed, free IgG binding sites can still be available on the bound protein A. Addition of the second antibody can result in its binding to the free sites on protein A molecules and target specified antigens. A second pro tein Α-gold probe can potentially bind not only to any protein A-gold free IgG antibodies from the primary incubation (the Fc region of the secondary IgG complexes) but also to the Fab regions of IgG antibodies.

Reports which claim success with the "one-face" protein A-gold method use either a defined sequence of protein A-gold probes or free protein A (described above). If free protein A is not incorporated, then the success of the method depends on the smaller of the protein A-gold probes being used first. The small probes (= 3 nm) have approximately one associated protein A molecule, whereas larger probes (= 15 nm) can adsorb -60 such molecules (Roth, 1982) . If the larger of the gold probes is used for the visualization of the first antigen, many potential Fc binding sites (unoccupied protein A) are available for the second immunoglobulin and significant cross-contamination may therefore result. If excess free protein A is added, the primary antibody-protein A interactions may reach saturation and so minimize interactions with secondary protein A-gold probes. The binding of secondary antibodies to the existing pro tein A Fc binding sites should not be recognized by the secondary protein A-gold probe as it is assumed that IgG can bind only one protein A mole cule. Cross-contamination can still occur, however, via interactions with the Fab regions of IgG. If cross-contamination is a continued problem, alternative techniques to the protein A method should be investigated. Such techniques generally involve direct labeling and species-specific antibody-gold indirect labeling techniques.

The use of small (3-6-nm) gold probes has the advantage of improving immunocytochemical resolution. Anyone who has worked with small gold probes has observed them (in some specimens) to be extremely "sticky" (Fig. 2.4a,b) . Behnke et al. (1986 ), Birrell et al. (1987 ), and Birrell and Griffith (1989 have reported the problem to be largely associated with exposed areas (areas not covered with protein) on the gold particles themselves and secondarily with the method by which the colloidal gold is produced. Since small protein A-gold complexes are used widely, the problems associated with small probes merit discussion.

Flocculation of colloidal gold can be prevented by partial saturation of the particles (Goodman et al., 1980; Horisberger and Vauthey, 1984) . The stabilization of colloidal gold therefore does not ensure its saturation. It is because of this that excess unrelated protein (for example, BSA) is (1 : 10) . N o stabilizer was included within the added to these protein-gold complexes; the subsequent complex is then assumed to be saturated. However, it would appear that some immuno gold labeling patterns are independent of protein-protein interactions and result from interactions between specimen constituents and the gold par ticles (Behnke et al., 1986) . The occurrence of such nonspecific binding is obvious in whole cytoskeletons (Fig. 2.4a ) and thick polyethylene glycolextracted sections (A. D. Hyatt, unpublished observations) . Strategies in colloidal gold labeling of cytoskeletal elements are discussed at length by Birrell and Griffith (1989) . Nonspecific binding is attributed to electro static interactions between exposed areas on the gold particles (areas of negative charge) and cationic structures within the specimen (Behnke et al., 1986) .

The method used to prepare colloidal gold has also been considered a source for nonspecific labeling (Birrell et al., 1987) . Birrell et al. found that 5-nm gold particles produced by the trisodium citrate-tannic acid method exhibited a higher degree of nonspecificity than particles of simi lar size produced by other methods. This nonspecificity was attributed to the chemical agents used in colloidal gold production (Birrell et al., 1987) ; for example, colloidal gold produced by the white phosphorus method was found to contain phosphate groups. Tannic acid may therefore also be a source of contamination for colloidal gold produced by the trisodium citrate-tannic acid procedure (Birrell et al., 1987; Birrell and Griffith, 1989) . Such contaminants on gold particles may, as with tannic acid, form a series of complex interactions with biological specimens, particularly those fixed with an aldehyde and/or Os0 4 (Simionescu and Simionescu, 1976) . Observations in this laboratory and by Birrell et al. (1987) and Bir rell and Griffith (1989) that nonspecific labeling is more pronounced with smaller probes may be explained by the greater accessibility of these probes to tissue constituents within open and/or permeabilized specimens.

Nonspecific binding due to protein-protein and electrostatic interac tions can be minimized by including in the washing and incubation steps a noncompeting protein which has a high affinity for gold particles. These proteins (e.g., bovine skin gelatin, fish gelatin, and BSA) will bind to pro- tein-reactive constituents within the specimen and will minimize electro static interactions by covering the exposed areas on the gold particles. The choice of an appropriate stabilizer will significantly reduce the level of nonspecific labeling, as is illustrated in Fig. 2 .4c (compare with Fig.  2.4b ). An alternative or concomitant procedure would involve the incuba tion of protein Α-gold with the biological sample.

Immunolabeling with protein Α-gold has several distinct advantages:

(1) it is easy and quick to prepare complexes of various sizes; (2) the complexes interact with antibodies from many mammalian species; and (3) the method is sensitive, clean, and can yield valuable information pro vided the parameters of immunolabeling are understood. In other words, the investigator must be familiar with the characteristics of the primary antibody, protein Α-gold probes, and the specimen itself.

Generally, the use of the protein Α-gold method should involve (1) the use of highly specific antibodies; (2) the use of such antibodies at their lowest effective concentration; (3) the absence of normal serum as a blocker; and (4), if using open matrix specimens, inclusion of an effective stabilizer in the incubation steps. These observations, if followed, will reduce the level of nonspecific labeling and cross-contamination. It should also be noted that at no stage during the labeling protocol should the sample be allowed to dry. In addition, the sample should be thor oughly rinsed between antibody and protein Α-gold steps; this will re move unbound antibodies and protein A. Failure to comply with the above precautions will result in high backgrounds.

There are no detailed protocols for protein Α-gold labeling which have universal applicability. For each different experiment, detailed protocols should be developed to produce optimum labeling characteristics. These protocols must be based on results of control immunocytochemical ex periments. If protein Α-gold is the label, and nonspecific binding and cross-contamination are persistent problems in spite of the addition of excess free protein A, high-affinity stabilizers, adsorption of contaminat ing antibodies, and colloidal gold, then alternative labeling techniques should be pursued and evaluated.

The author gratefully acknowledges Drs. Β. T. Eaton and A. R. Gould for reading the manuscript and Mr. T. Wise for skillful technical assistance.

Non-specific binding of protein-stabilized gold sols as a source of error in immunocytochemistry

Double immunocytochemical labeling applying the protein A-gold technique

Protein Α-gold electron microscopic immunocytochemistry: Meth ods, applications, and limitations

Double labelling cytochemistry applying the pro tein Α-gold technique. Immunolabelling for Electron Microscopy

Demonstration and assaying of IgG anti bodies in tissues and on cells by labeled staphylococcal protein A

The grid-cell-culture technique: The direc examination of virus-infected cells and progeny viruses

Antibody competition studies witl gold-labelling immunoelectron microscopy

Phylogenetic insight int< evolution of mammalian Fc fragment of 7G globulin using staphylococcal protein A

Protein A of Staphylococcus aureus and related immunoglobulin re ceptors produced by streptococci and pneumococci

Binding of immunoglobulins t< protein A and immunoglobulin levels in mammalian sera

Factors affecting the staining with colloida gold

The preparation of protein Α-gold complexes with 3nm and 15nm gok particles and their use in labelling multiple antigens on ultra-thin sections

The colloidal gold marker system for light and electron microscopic cyto chemistry

Galloylglucoses of low molecular weight a: mordant in electron microscopy. I. Procedure, and evidence for mordanting effect

Sizing of protein Α-colloidal gold probes for immu noelectron microscopy

Gold markers for single and double immunolabellinj of ultrathin cryosections

Preparation of colloidal gold

Immunoperoxidase techniques: Practical and theoretical aspects

Ultrastructure of the mem brane attack complex of complement: Detection of the tetramolecular C9-polymeriz ing complex C5b-8

Interactions of pig Fab gamma fragments with protein A from Staphylococ cus aureus

Das kolloidale gold