Immunocytochemical Detection of Granzymes A and B in Peripheral Blood Lymphocytes from Healthy Individuals After Non-enzymatic Antigen Retrieval

Cytoplasmic granules of cytotoxic lymphocytes contain several constituents, including perforin, granzyme A (GrA), and granzyme B (GrB). After granule exocytosis, GrA and GrB are believed to enter the target cell through perforin-derived transmembrane channels to induce DNA fragmentation and apoptosis (Liu et al. 1995; Nakajima et al. 1995; Smyth and Trapani 1995; Heusel et al. 1994). Various reports have dealt with expression of GrA and GrB in peripheral blood lymphocytes from healthy individuals, but the findings are conflicting (Berthou et al. 1995; Kummer et al. 1995; Sunder-Plassmann et al. 1990). We addressed this matter through the application in immunocytochemistry of antigen retrieval methods routinely used in immunohistochemistry.

Cytocentrifuge preparations of lymphokine-activated killer (LAK) cells were fixed in 4% buffered formalin for 10 min. Part of the preparations were left untreated, and either enzymatic or non-enzymatic antigen retrieval was employed on the other part. Two enzymatic antigen retrieval methods were applied: digestion with 0.1 mg/ml pepsin A in 10 mM HCl, pH 2.0, at 37C, and digestion with 0.1 mg/ml protease Type XIV in PBS, pH 7.4, at 37C. In both cases, digestion was stopped after either 30 sec, 1 min, 2 min, or 4 min. Non-enzymatic antigen retrieval consisted of boiling the preparations in 10 mM sodium citrate, pH 6.0, for 10 min. Single-color stainings were performed with monoclonal antibodies (MAbs) GrA-6 and GrB-7, raised against recombinant human GrA and GrB proteins, respectively, which recognize GrA and GrB in paraffin-embedded, formalin-fixed tissue sections (Sanbio/Monosan; Uden, The Netherlands).

No staining was observed in the majority of untreated LAK cells (Figure 1 A). Enzymatic antigen retrieval did not have a beneficial effect, and extension of digestion time, especially in the pepsin A-containing antigen retrieval solution, led to loss of cell number and loss of morphology (not shown). After non-enzymatic antigen retrieval, the majority of LAK cells displayed a strong granular staining pattern, but cell loss and loss of morphology were not observed (Figure 1 B).

The mechanisms underlying the effects of enzymatic and non-enzymatic antigen retrieval differ, and a selective benefit from one treatment over the other has also been demonstrated in immunohistochemistry for other antigens (Cattoretti et al. 1993). The effect of enzymatic antigen retrieval is known to depend on proteolytic restoration of accessibility of antibodies to masked antigens. Masking of antigens is an artifact induced by formalin fixation and is believed to be due to formation of methylene crosslinks between reactive sites on proteins. The amount of crosslinking determines the ease with which high molecular weight antibodies diffuse towards their antigens. In contrast, the effect of non-enzymatic antigen retrieval appears to depend on protein denaturation, with heat accounting for most of this effect (Cattoretti et al. 1993).

MAbs GrA-6 and GrB-7 were raised against recombinant granzymes produced in a prokaryotic expression system, and the fact that reactivity of these MAbs is enhanced after non-enzymatic antigen retrieval suggests that they specifically recognize granzymes with a denatured conformation. Several factors can be pointed out that might have induced conformational changes in the recombinant granzymes and may have contributed to this specificity, e.g., incorrect folding of the recombinant granzymes in the prokaryotic expression system or partial degeneration of the recombinant granzymes after immunization. When we extended fixation time to 7 days, we still did not observe an influence on staining results when we employed non-enzymatic antigen retrieval (not shown). This observation implies that use of MAbs GrA-6 and GrB-7 in immunocytochemistry is not hampered by masking of the respective antigens due to formalin fixation, and explains the lack of beneficial effect of enzymatic antigen retrieval.

Subsequently, non-enzymatic antigen retrieval was performed on formalin-fixed cytocentrifuge preparations of peripheral blood mononuclear cells from 10 healthy individuals. Two-color stainings were performed combining polyclonal rabbit anti- human CD3 antibodies with MAb GrA-6 or GrB-7. Two observers independently scored 300 staining cells per slide. Cells staining doubly positive were clearly distinguishable from cells staining singly positive (Figure 1 C). Identical staining patterns were observed for MAbs GrA-6 and GrB-7. Table 1 shows the distribution of CD3⁺ granzyme^-, CD3⁺ granzyme⁺, and CD3^- granzyme⁺ cells among staining cells. Our data imply that the same cells express both GrA and GrB. This assumption is supported by the observation that expression of both granzymes, as well as expression of perforin, is upregulated after activation of T-cells in vitro (Liu et al. 1989).

Figure 1.

(A,B) Lymphokine-activated killer cells stained with monoclonal antibody GrB-7. (A) Staining result when cells were not treated with an antigen retrieval method. A similar result was observed after enzymatic antigen retrieval. (B) Staining result after non-enzymatic antigen retrieval. (C) Two-color staining with polyclonal anti-CD3 antibodies (detected in red) and MAb GrB-7 (detected in blue) of peripheral blood mononuclear cells from a healthy individual showing cells staining doubly positive for granzyme B and CD3 and cells staining singly positive for either granzyme B or CD3. A similar staining pattern was observed with monoclonal antibody GrA-6. Bars = 3 μm.

GrA and GrB were expressed by 12 ± 1% (mean ± SEM) and by 9 ± 1% of CD3⁺ T-cells, respectively. Analogous to the distribution pattern of perforin (Berthou et al. 1995; Nakata et al. 1992), we suggest that granzyme-expressing T-cells reside primarily in the CD8⁺ T-cell subset. We did not perform two-color stainings with anti-CD8 antibodies and anti-granzyme MAb, because CD8 is also expressed by a subset of natural killer (NK)-cells (Lanier et al. 1986). Assuming the granzyme-positive T-cells to reside in the CD8⁺ T-cell subset and the CD4⁺ and CD8⁺ T-cells in these 10 healthy individuals to be normally distributed (Reichert et al. 1991), it can be estimated that GrA and GrB are expressed by 15–35% of CD8⁺ T-cells. This estimation is in the same range as values reported for expression of perforin by CD8⁺ T-cells (Berthou et al. 1995; Nakata et al. 1992).

Table 1

Distribution of CD3⁺ granzyme^-, CD3⁺ granzyme⁺, and CD3^- granzyme⁺ cells among peripheral blood mononuclear cells staining with polyclonal anti-CD3 antibodies and anti-granzyme monoclonal antibodies a

	CD3+ granzyme-	CD3+ granzyme+	CD3- granzyme+
Granzyme A	80 ± 1%	11 ± 1%	8 ± 1%
Granzyme B	84 ± 2%	8 ± 1%	8 ± 1%

^aData represent mean ± SEM from 10 healthy individuals.

Granzyme-expressing CD3^- cells most likely represent NK cells. The majority of cells in this subset have a granular morphology and have been shown to express perforin (Berthou et al. 1995; Nagata et al. 1992; Lebow and Bonavita 1990). Two-color stainings with anti-granzyme MAb and anti-CD16 or anti-CD56 antibodies were, however, not performed, because such stainings do not discriminate NK-cells from CD8⁺CD11a⁺ T-cells, which have also been shown to express CD16 and CD56 (Kern et al. 1994). CD8⁺CD11a⁺ T-cells can, in addition, be expected to express GrA and GrB on the basis of the expression of perforin by these cells (Berthou et al. 1995). The ratio of CD3^- granzyme⁺ cells to CD3⁺ granzyme⁺ cells was 0.9 ± 0.2 and 1.1 ± 0.2 for GrA and GrB, respectively. We feel that this ratio reflects the distribution of NK-cells and T-cells with cytotoxic potential in peripheral blood of healthy individuals. Follow-up of this ratio in cytocentrifuge preparations from peripheral blood and fine-needle aspiration biopsies might prove useful for demonstration of the contributions of NK-cells and T-cells to various cellular immune responses.

In conclusion, we have immunocytochemically demonstrated expression of GrA and GrB in peripheral blood lymphocytes from healthy individuals using non-enzymatic antigen retrieval of formalin-fixed cells. Application in immunocytochemistry of antigen retrieval methods routinely used in immunohistochemistry may prove useful for detection of other antigens in cytocentrifuge preparations and cell smears.