Protease Antigen Recovery Decreases the Specificity of Bromodeoxyuridine Detection In Formalin-fixed Tissue

Abstract

Incorporation of halogenated nucleotide analogues is often used to assess DNA synthesis and to quantitate cellular proliferation. Multiple antibodies have been developed to bromodeoxyuridine (BrdUrd) and it is the most frequently utilized substrate. Because the immunodetection of incorporated BrdUrd requires DNA denaturation or nuclease digestion, most of these antibodies are not reactive in tissues or cells fixed with crosslinking agents. Antigen retrieval techniques utilizing protease digestion restore BrdUrd antigenicity and permit the detection of BrdUrd in formalin-fixed tissue. However, during the development of a double label immunohistochemical protocol to quantitate proliferating alveolar Type II cells, we noted nucleus-specific staining in lung sections from animals that had not received BrdUrd. Therefore, we systematically analyzed the specificity of the immunohistochemical detection of incorporated BrdUrd in formalin-fixed tissue after protease digestion. Enzymatic antigen recovery diminished the specificity of the BrdUrd reaction product and caused false-positive staining with the BU-1, B44, and BR3 monoclonal antibodies. Staining was less prominent with Bu20a but was more specific. Protease antigen recovery may decrease the specificity of BrdUrd immunodetection. Appropriate controls are required when enzymatic digestion is used to detect incorporated BrdUrd in formalin-fixed tissue. The type and duration of fixation, antibody to BrdUrd, protease, and tissue may affect the specificity of the staining pattern. (J Histochem Cytochem 45:1165–1170, 1997)

B romodeoxyuridine (BrdUrd) is a halogenated pyrimidine that may substitute for thymidine in newly synthesized DNA. Various monoclonal antibodies (MAbs) to BrdUrd have been developed and used in immunoassays to detect cell proliferation, cell cycle kinetics, sister chromatid exchange, and to isolate nascent DNA (Van Furth and Van Zwet 1988; Cohn and Lieberman 1984; Dean et al. 1984; Morstyn et al. 1983; Gratzner 1982). However, substituted DNA must be denatured or cleaved for the detection of incorporated BrdUrd. Because formalin fixation may prevent DNA denaturation or mask the antigenicity of incorporated BrdUrd, most anti-BrdUrd antibodies are not reactive in specimens fixed with crosslinking agents. To identify incorporated BrdUrd in formalin-fixed tissue, several investigators have shown that protease antigen retrieval may restore BrdUrd antigenicity and allow its detection (Hayashi et al. 1988; Shutte et al. 1987; Sugihara et al. 1986). The mechanisms of protease antigen retrieval are uncertain but may involve protein cleavage to expose epitopes that become hidden during protein folding caused by fixation or embedding, digestion of macromolecular complexes that block antigen recognition, transformation of non-antigenic precursors to immunoreactive products, exposure of crossreactive sites, or elimination of bonds between plastic monomers and antigens that form during embedding (Larsson 1988). Although protease antigen retrieval may be highly successful, it is usually necessary to determine empirically the best enzyme and the optimal concentration and duration of digestion, which may depend on the duration of fixation, type of tissue, and antigen being detected (Taylor and Shi 1994). There is also a critical period during which the antigen may be detected but before the tissue is digested and degraded. Finally, enzymatic recovery techniques may cause false-positive results, especially with DNAse digestion to detect nuclear antigens (Taylor and Shi 1994).

In the development of a double label immunohis-tochemical staining protocol to quantitate proliferating alveolar Type II cells, we found that antigen retrieval with several different proteases decreased the specificity of various MAbs for BrdUrd (Panos et al. 1995, 1996). Therefore, in this study we systematically determined the effect of various protease antigen retrieval methods on the detection of BrdUrd incorporated into tissue fixed in formalin. We show that protease antigen recovery may decrease the sensitivity of antibodies to BrdUrd, causing false-positive results. Therefore, appropriate negative controls should be utilized in immunohistochemical protocols using enzymatic antigen recovery to detect BrdUrd in formalin-fixed tissue sections.

Figure 1

Labeling scores for tissue sections from animals treated with BrdUrd (solid bars) or PBS (open bars) and fixed for 18 hr. Enzyme concentrations for antigen recovery, B44 (A), BR3 (B), Bu-1 (C), and Bu20a (D), and the immunohistochemical staining protocols for the various antibodies are presented in Materials and Methods. ^∗ p>0.05.

Materials and Methods

Male Sprague–Dawley rats (225–250 g) were injected ip with 100 mg/kg BrdUrd dissolved in sterile, endotoxin-free PBS at a concentration of 10 mg/ml. Eighteen hr later the animals were sacrificed and the testes removed and fixed in 10% phosphate-buffered formalin. The testes were then embedded in paraffin and 4-μm sections were mounted on 3-aminopropyltriethoxysilane coated slides to enhance adherence. After air-drying, the slides were heated overnight at 50C. They were dewaxed in xylene and rehydrated through a graded series of ethanol solutions (100%, 100%, 70%, 50% v/v) and finally washed in PBS. Endogenous peroxidase activity was inhibited by treating the slides with 5% H₂O₂ in methanol twice for 5 min. The sections were then washed in PBS and permeabilized by incubating in 0.1% Triton X-100 dissolved in PBS for 30 min at room temperature (RT). After washing in PBS, the sections were incubated with the various proteases under these conditions: pepsin (Sigma; St Louis, MO), 0.1% solution (w/v) dissolved in 0.1 N HCl for 30 min at 37C; pronase E (Sigma), 100 μg/ml dissolved in 0.02 M Tris-HCl, pH 7.6, containing 20 mM CaCl₂-2·H₂O for 30 min at 37C; protease Type XXIV (Sigma), 0.1% solution (w/v) dissolved in 0.1 M phosphate buffer, pH 7.4, for 15 min at RT; trypsin Type III (Sigma), 0.1% solution (w/v) dissolved in 0.05 M Tris-HCl, pH 7.6, containing 0.1% CaCl₂ 2 · H₂O for 30 min at 37C. For the experiments determining the effect of the duration of fixation, the concentrations of the enzymes were increased to 0.2% (w/v) pepsin, 500 μg/ml pronase E, 0.2% (w/v) protease Type XXIV, and 0.2% (w/v) trypsin Type III. The slides were then washed in PBS and nuclear histones were extracted from DNA by incubating the sections in cold 0.1 N HCl for 10 min. Double-stranded DNA was denatured with 2 N HCl for 30 min at 37C. [This step was omitted for sections exposed to the BU-1 antibody (Amersham; Arlington Heights, IL), which was applied with a proprietary mixture of nucleases.] HCl was neutralized with 0.1 M borax, pH 8.5, for 10 min at RT. After washing in PBS, the sections were incubated with 3% horse serum in PBS for 30 min at RT. Excess blocking solution was blotted off and primary antibodies applied under the following conditions. BU-1 (Amersham) proprietary antibody/nuclease solution was applied undiluted for 1 hr at RT. B44 (Becton–Dickson; San Jose, CA) was applied at a 1:4 dilution (6.25 μg/ml) in 3% horse serum containing 0.1% Triton X-100 overnight at 4C. BR3 (Caltag Laboratories; So. San Francisco, CA) was applied at a 1:200 dilution (5.0 μg/ml)) in 3% horse serum containing 0.1% Triton X-100 overnight at 4C. Bu20a (Dako; Carpinteria, CA) was applied at a 1:50 dilution (6.0 μg/ml)) in 3% horse serum containing 0.1% Triton X-100 overnight at 4C. After incubation with the primary antibody, the sections were washed in PBS twice for 5 min each.

The sections were then incubated with biotinylated horse anti-mouse (heavy and light chain-specific) IgG (Vector Laboratories; Burlingame, CA) diluted 1:200 in 3% horse serum/0.1% Triton X-100 for 1 hr at RT and then washed in two changes of PBS for 5 min each. Biotin–streptavidin–horseradish peroxidase diluted 1:100 in PBS was added for 30 min at RT. After washing in two changes of 0.1 M phosphate buffer, pH 7.4, the horseradish peroxidase activity was detected using Ni/Co enhancement of the diaminobenzidine reaction product (Adams 1981). The sections were then washed twice in ddH₂O, dehydrated by passing through a graded series of ethanol solutions (30%, 50%, 70%, 100%, 100%, v/v), cleared in xylene, and mounted with Permount.

All sections were coded and independently graded by both investigators, who rated the slides on a scale of 0 to 5: 0, no staining; 1, sporadic nuclear staining in basal cells; 2, nuclear staining in all basal cells; 3, staining in basal cells throughout the section and extending to the intermediate cells intermittently; 4, staining in basal cells and extending to the intermediate cells throughout the section; 5, nuclear staining throughout the section. Data are expressed as the mean ± SEM of three separate experiments. Statistical analysis was performed using an unpaired t-test (StatView; Abacus Concepts) and p>0.05 was considered significant.

Figure 2

Labeling scores for tissue sections from animals treated with BrdUrd (solid bars) or PBS (open bars) and fixed for 31 (A), 175 (B), and 510 hr (C). The Bu20a antibody was used for all sections. Enzyme concentrations for antigen recovery and the immunohistochemical staining protocols are presented in Materials and Methods. ^∗ p>0.05.

Results

Antigen recovery with the different enzymes significantly affected the detection of BrdUrd (Figure 1). With both the B44 and BR-3 anti-BrdUrd MAbs, nuclear staining was present in unlabeled tissues that were not treated with proteases as well as in the enzymatically digested specimens. There were no significant differences in the labeling scores between specimens from BrdUrd-treated and control animals. No nuclear labeling was present in tissues from saline-instilled rats using the BU-1 MAb and either no antigen recovery or trypsin treatment. Pronase E, protease, and pepsin digestion produced equivalent labeling scores in BrdUrd-labeled and unlabeled specimens.

Bu20a was the most specific MAb and accurately discriminated BrdUrd-labeled tissues from unlabeled sections after either no treatment, pronase E, or trypsin digestion. However, after protease or pepsin antigen recovery, there were no significant differences between BrdUrd-treated and control animals.

Figure 3

(A) Section of testis from an animal given BrdUrd. The immunodetection protocol used the Bu-20a antibody and no enzymatic treatment. No nuclear staining is present. (B) Section from an animal given BrdUrd. The immunodetection protocol used the Bu-20a antibody with trypsin antigen recovery. There is sporadic nuclear staining within the basal cells. (C) Section from an animal given saline. The immunodetection protocol used the BR3 antibody with trypsin treatment. Nuclear staining is present throughout the basal cells. (D) Section from an animal given saline. The immunodetection protocol used the BR3 antibody with pepsin antigen recovery. There is nuclear staining in the basal cells and extending to the intermediate cells throughout the section. Bar = 200 μm.

To determine the effect of fixation duration on Bu20a detection of incorporated BrdUrd, tissue was fixed for 31–510 hr in buffered formalin (Figure 2). After 510 hr of fixation, no staining was detected in untreated sections, whereas at earlier time points Bu20a accurately discriminated between labeled and unlabeled tissues. Nuclear staining was present in unlabeled tissues with all the enzymatic treatments except trypsin. The highest levels of nonspecific staining occurred with protease and pepsin antigen recovery (Figure 3).

Discussion

Many MAbs to BrdUrd have been developed. A halogenated ribouracil conjugated to a carrier protein has been the immunogen used in the creation of many of these antibodies. They are generally of low affinity (K_a approximately 10⁶ M^-1 for BrdUrd in solution) and detect the halogenated nucleotide only in single-stranded but not in double-stranded DNA (Adams 1981). Most methods for the immunodetection of incorporated BrdUrd require tissue to be frozen or fixed in dehydrating agents such as methanol/acetic acid (Gratzner 1982), acetone (Morstyn et al. 1983), or ethanol (Vanderlaan and Thomas 1985), and that DNA be denatured by acid hydrolysis, exposure to strong bases or formamide, or degraded by nuclease digestion. BrdUrd detection is prevented or significantly impaired by fixation in crosslinking agents such as formalin owing to prevention of DNA denaturation or irreversible crosslinking of nuclear proteins to DNA, thus blocking BrdUrd recognition. DNA containing BrdUrd or IdUrd has higher affinity for chromatin proteins, including histones and regulatory proteins, which may increase crosslinking of nuclear protein to areas of DNA containing halogenated nucleotides (Hamada 1985; Lin 1978). Enzymatic digestion of formalin-fixed tissue has been shown to restore BrdUrd antigenicity and permit immunodetection of BrdUrd within formalin-fixed tissue (Hayashi et al. 1988; Shutte et al. 1987; Sugihara et al. 1986).

We utilized protease antigen recovery in the development of a double label immunohistochemical technique to quantitate proliferating alveolar Type II cells (Panos et al. 1995, 1996). During the development of this protocol, we noted decreased specificity of BrdUrd detection and occasional nuclear staining in specimens that had not been exposed to BrdUrd. In this study, we systematically analyzed the specificity of the immunohistochemical reaction product in formalin-fixed tissue after protease digestion. Enzymatic antigen recovery diminished the specificity of the BrdUrd reaction product and caused false-positive staining. The BR3 MAb, which uses nuclease digestion to expose incorporated BrdUrd, produced the most intense staining and the greatest nonspecificity. In contrast, the Bu20a MAb yielded the lightest staining but was the most specific.

The cause of the false-positive staining is uncertain. Other investigators have suggested that enzymatic digestion may unmask or create antigenic epitopes common to unrelated tissue constituents (Heyderman 1979). Therefore, it is possible that proteolytic antigen recovery might produce common epitopes between halogenated and nonhalogenated nucleotides. In an analysis of IU-1, IU-2, and B-44 MAbs to BrdUrd, Vanderlaan and co-workers (Vanderlaan and Thomas 1985) demonstrated reactivity with BrdUrd incorporated into single-stranded but not double-stranded DNA. These antibodies also bound weakly to thymidine and to unsubstituted single-stranded DNA. In our study, staining was localized to the nucleus in specimens that were not labeled with BrdUrd, suggesting nucleus-specific rather than nonspecific background staining. Finally, the specificity of BrdUrd detection after protease antigen recovery may differ for various MAbs to BrdUrd. DNAse treatment does not permit BrdUrd detection with B44 but does work with the BU-1 MAb (Gonchoroff et al. 1986).

In a study comparing BrdUrd with [³H]-thymidine labeling in spermatagonial stem cells and Leydig cells in the testes and small intestinal crypt cells, Thoolen (Thoolen 1990) used microwave antigen recovery after fixation in Carnoy's fluid and found that 90% of the BrdUrd-labeled cells were also identified by [³H]-thymidine autoradiography. These studies used the B44 anti-BrdUrd MAb. Although the correlation between the BrdUrd and [³H]-thymdine labeling indices was highly significant statistically, no explanation was given for the increased recognition of proliferating cells with microwave antigen recovery and BrdUrd immunohistochemistry compared with [³H]-thymdine autoradiography.

Immunohistochemical detection of BrdUrd has been widely used to determine cell proliferation in cells and tissues. Enzymatic antigen recovery allows BrdUrd immunodetection in tissues preserved with aldehyde-containing fixatives. Careful and appropriate controls are required when protease digestion is used to determine a proliferative index based on BrdUrd immunodetection in formalin-fixed tissue specimens.