A Comprehensive Antibody Panel for Immunohistochemical Analysis of Formalin-Fixed,Paraffin-Embedded Hematopoietic Neoplasms of Mice: Analysis of Mouse Specific and Human Antibodies Cross-Reactive with Murine Tissue

Abstract

Immunohistochemistry is an indispensable tool in human pathology enabling immunophenotypic characterization of tumor cells. Immunohisto-chemical analyses of mouse models of human hematopoietic neoplasias have become an important aspect for comparison of murine entities with their human counterparts. The aim of this study was to establish a diagnostic antibody panel for analysis of murine lymphomas/leukemias, useful in formalin-fixed/paraffin-embedded tissue. Overall, 48 antibodies (4 rabbit monoclonal, 12 rabbit polyclonal, 2 goat polyclonal, 11 rat, and 19 mouse monoclonal), which were either mouse-specific (14) or cross-reactive with murine tissue (34) were tested for staining quality and diagnostic value in 468 murine hematopoietic neoplasms. Specific staining was achieved with 29 antibodies, of which 18 were human antibodies cross-reactive with murine tissue. Only 23 (B220, BCL-2, BCL-6, CD117, CD138 (2×), CD3 (2×), CD43, CD45, CD5, CD79αcy, cyclin D1, Ki-67 (2×), Mac-3, Mac-2, lysozyme, mast cell tryptase, MPO, Pax-5, TdT, and TER-119) were regarded as valuable for diagnostic evaluation. Immunohistochemistry was also established in an automated immunostainer for high throughput analysis. The antibody panel developed is useful for the classification of murine lymphomas and leukemias analyzed, and a valuable tool for human and veterinary pathologists involved in the diagnostic interpretation of murine models of hematopoietic neoplasias.

Introduction

Immunohistochemistry (IHC) has become an indispensable tool for the detailed immunophenotypic characterization of lymphomas and leukemias. In human pathology, the application of immunohistochemistry and immunofluorescence for the determination of the lineage of hematopoietic tumor cells began almost 30 years ago (Levy et al., 1977; Chu and MacDonald, 1979; Leong et al., 1979; Stein et al., 1980, 1981; Bhoopat et al., 1989; Jaffe et al., 1999; Harris et al., 2000a, 2000b).

With the increasing use of animal models for human hematopoietic diseases, a detailed analysis, including IHC, has become mandatory to characterize these entities in the model organism. The mouse is the best-studied mammalian genetic system for modeling human disease, and huge efforts have been undertaken to introduce and adopt standardized nomenclature for the classification of murine diseases (Goldforb et al., 1984; Yanagi et al., 1984; Cardiff, 2001; Boivin et al., 2003; Nikitin et al., 2004; Shappell et al., 2004). Veterinary and human pathologists recently recommended the use of a new classification for hematopoietic neoplasms in mice (Kogan et al., 2002; Morse et al., 2002) based on the latest World Health Organization (WHO) classification of human tumors of hematopoietic and lymphoid tissues (Jaffe et al., 2001).

The use of IHC for the accurate classification of mouse hematopoietic tumors has been identified as being increasingly important to facilitate the comparison between mouse and human hematopoietic neoplasms. Although many mouse-specific antibodies are commercially available for IHC in frozen tissue, cryostat sections are inferior in quality when compared to paraffin sections. During the last 20 years, pathologists engaged in the pathology of mice have been making efforts to apply immunohistochemistry in the diagnosis of mouse hematopoietic diseases (Fredrickson et al., 1985,1999; Frith et al., 1993; Ward et al., 1993; Qi et al. 1998, 2000; Sabourin et al., 2000; Morse et al., 2001; Eason et al., 2003; Lange et al., 2003; Utsuyama and Hirokawa, 2003; Canela et al., 2004; Huang et al., 2004; Miyakawa et al., 2004; Neeson and Paterson, 2004).

Recently, in a collaborative work, several institutions published their experience with a total of 196 antibodies that label paraffin-embedded mouse tissue (Mikaelian et al., 2004). However, only 3 of the antibodies tested in that study are useful for the classification of murine hematopoietic neoplasms (Mikaelian et al., 2004). Therefore, the aim of this study was to develop a standardized panel of antibodies reactive in paraffin-embedded tissue, useful for the classification of hematopoietic tumors. A panel of 48 commercially available antibodies was tested in 468 hematopoietic mouse tumors for specific staining and diagnostic value.

Material and Methods

Tissues

Tissues from 468 mice of both genders, of different background strains (C57BL/6J (96), mixed 129S4;C57BL/6J (168), C3HeB/FeJ (2), NMRI/Nh (200), Balb/cAnPt (2)) and of different ages (22 weeks to 2 years) were analyzed. Most animals had hematopoietic neoplasms previously confirmed with Southern blot analysis using TCRβ-specific probes (J1 and J2), immunoglobulin (Ig) heavy and kappa light chain-specific probes (IgH-J11 and Ig(κ)), and/or flow cytometry (Rawat et al., 2004; Sørensen et al., 2004, 2005; Schessl et al., 2005). Thirty-two cases were morphologically not classified as lymphoma (nonhematopoietic tumors (3/32), splenic hyperplasia (25/32), no changes (4/32)). Tumor samples were collected from retrovirally infected mice (n = 200), transgenic mice (n = 70), irradiated Blm^{− / −} mice (Luo et al., 2000) (n = 168), irradiated mice (n = 26), and untreated wild-type mice (n = 4). Spleen, thymus, liver, and lymph nodes were examined histologically. Spleen, thymus, lymph nodes, intestinal tract, lung, skin, and liver of unaffected C57BL/6J, C3HeB/FeJ, NMRI/Nh, and Balb/cAnPt mice, as well as human tonsils, bone marrow, Burkitt lymphoma, Mantle cell lymphoma, and colon carcinoma were used as control tissues.

Histology and Histochemistry

Tumor samples were fixed in 10% buffered formalin or 4% paraformaldehyde (PFA) for at least 24 hours. For histology and immunohistochemistry 2–5 μm thick sections (Microtom, Microm, Walldorf, Germany) were cut from paraffin blocks and mounted on SuperFrost Plus slides (Roth, Karlsruhe, Germany). Slides were stained with hematoxylin and eosin (H&E), periodic acid-Schiff reaction (PAS), chloroacetate esterase (CAE), and/or Giemsa when indicated.

Immunohistochemistry

Immunohistochemistry was performed on an automated immunostainer with a Basic DAB or iVIEW DAB detection kit (Ventana Medical System, Inc., Tucson, AZ) according to the company’s protocols for open procedures with slight modifications. After deparaffinization and rehydration, the slides were placed in a microwave pressure cooker in 0.01 mol/L citrate buffer (pH = 6.0) containing Tween 20, and heated in a microwave oven at maximum power for 30 minutes. After heat-induced antigen retrieval the slides were allowed to cool down in 0.05 M Tris-HCL saline (TBS, pH = 7.6), containing 3% goat or rabbit serum (GIBCO, Karlsruhe, Germany) for 20 minutes. The antibody panel used is summarized in Table 1. All mono- and polyclonal antibodies (AB) were commercially available and raised against human (33/48 AB), murine (14/48 AB), or bovine antigens (1/48 AB).

For all antibodies incubation with primary antibodies was performed overnight in a moist chamber at room temperature. The procedure was completed on the next day either manually with an ABC-Kit (Vector, Burlingame, Canada), and a DAB Peroxidase Substrate Tablet Set (Sigma-Aldrich, Munich, Germany), following the manufacturer’s protocol, and counterstained with hematoxylin or by applying the semiautomated procedure with the immunostainer. For primary rat, rabbit, and goat antibodies a secondary antibody rabbit anti-rat, biotinylated IgG (Dako, Hamburg, Germany), goat anti-rabbit, biotinylated IgG (Vector, Burlingame, Canada), or donkey anti-goat, biotinylated IgG (Jackson Immunore-search Laboratories, West Grove, PA) were used at a dilution of 1:300, 1:750, or 1:500, respectively.

For monoclonal mouse antibodies a mouse anti-mouse kit was used to reduce nonspecific staining following the company’s instructions (MoMap kit, Ventana Medical Systems). Briefly, before incubation of a primary antibody with the tissue, the diluted primary antibody was first allowed to react with its secondary, biotinylated IgG antibody (15 minutes, room temperature). Free binding sites were subsequently saturated with murine serum IgG for 30 minutes. The mixed solution was finally applied to the tissue section and incubated overnight in a moist chamber. The procedure was completed on the immunostainer, as recommended by the company. Appropriate positive and negative controls were used to confirm the adequacy of the staining, including a slide without the primary antibody to rule out the possibility of nonspecific binding.

Classification

Murine hematopoietic tumors were classified according to the Bethesda Proposals for classification of lymphoid and nonlymphoid hematopoietic neoplasms in mice (Kogan et al., 2002; Morse et al., 2002), and confirmed by supporting molecular techniques including Southern blot and flow cytometry.

Results

Staining Results and Diagnostic Value of Tested Antibodies

Forty-eight commercially available antibodies (Table 1) were selected to evaluate the staining quality and usefulness of IHC in the diagnosis of murine hematopoietic neoplasms in formalin-fixed and paraffin-embedded tissue sections. Fourteen were mouse-specific antibodies, including 11 rat monoclonal, 2 goat polyclonal, and 1 rabbit polyclonal. Eleven of these mouse-specific antibodies gave a crisp, strong and specific staining (Table 1). Only one mouse-specific antibody (CD79 clone HM79-12) gave a weak staining. Two mouse-specific antibodies, goat polyclonal anti-mouse IgG and IgM, gave strong specific staining, but also nonspecific background staining.

Of the remaining 34 antibodies, which were raised against nonmouse, i.e., human or bovine antigens, 19 were mouse monoclonal, 11 were rabbit polyclonal, and 4 were rabbit monoclonal antibodies. A reproducible and appropriate cross-reactivity was observed with 18 antibodies (5 mouse monoclonal, 4 rabbit monoclonal, and 9 rabbit polyclonal) (Table 1). The rabbit monoclonal antibodies gave excellent staining results, eliminating the problem of nonspecific background. Twelve antibodies tested failed to work in formalin-fixed, murine tissues, although specific staining was achieved with human control tissues (Table 1). In addition, 7 antibodies produced a weak specific staining and/or a strong background staining (Table 1). In summary, a strong, specific positivity could be demonstrated in 29 of 48 tested antibodies. However, only 23 of these 29 antibodies were useful for the diagnosis of murine lymphomas or leukemias (Table 2).

Antibodies Useful for Routine Diagnosis of Murine Hematopoietic Neoplasias

In general, Ki-67 (TEC-3 and SP6) was a good proliferation marker (Figure 1A).

Antibodies Useful for the Diagnosis of B-Cell Lymphomas

B220 (also known as CD45R), CD79αcy, Pax-5, CD138, and TdT were useful for the classification of B-cell lymphomas. B220 was performed in all cases. Only when B220 was negative and a B-cell lymphoma was suspected CD79αcy and/or Pax-5 were performed. CD79αcy and Pax-5 were performed only as a second line B-cell marker. In this series, several types of lymphomas belonging to the B-cell linage were diagnosed, including diffuse large B-cell lymphoma (54 cases), precursor lymphoblastic B-cell lymphoma (16 cases), Burkittlike B-cell lymphoma (5 cases) splenic marginal zone lymphoma (14 cases), follicular B-cell lymphoma (26 cases), small B-cell lymphoma (5 cases), plasmacytoma (94 cases), and B-cell lymphomas, not otherwise specified (16 cases).

1. Diffuse large B-cell lymphomas were always B220 positive (Figure 1B) and often accompanied by MAC-3 positive normal reactive histiocytes and a remarkable reactive proliferation of CD3 positive T-cells.

2. Precursor lymphoblastic B-cell lymphomas were CD45, TdT, and Pax-5 positive, but surface IgM and IgG negative. Most cases were also B220 positive.

3. Burkittlike lymphomas were TdT negative, CD45 (Figure 1C), B220, CD79αcy (Figure 1D) and Pax-5 positive (Figure 1E). Unlike the precursor lymphoblastic B-cell lymphomas, these lymphomas were uniformly positive for surface IgM and IgG, in keeping with their germinal center origin.

4. Splenic marginal zone lymphomas were usually B220 negative but CD79αcy positive confirming the B-cell origin of these tumors. Since CD79αcy was usually positive, Pax-5 was not performed in this group. BCL-2 (Figure 1F) was positive in most cases.

5. Murine follicular B-cell lymphomas lacked BCL-2 and BCL-6 reactivity, although normal germinal centers in mice are BCL-6 positive (Figure 1G). B220 was always positive.

6. Small B-cell lymphomas were consistently positive for all B-cell markers. In addition, BCL-2 positivity and TdT negativity was observed.

7. CD138 (Figure 1H) was always positive in mature plasmacytomas; in contrast plasmablastic variants were CD138 negative or positive and sometimes also B220 positive. All plasmacytomas were positive for Igκ (Figure 1I). Few reactive Igλ positive plasma cells could be demonstrated (Figure 1J).

Antibodies Useful for the Diagnosis of T-Cell Lymphomas

CD3, CD5, CD43, and TdT were used for the diagnosis of T-cell lymphomas. In this series only 2 different types of murine T-cell lymphomas were diagnosed: precursor T-cell lymphoblastic (149 cases) and small T-cell lymphoma (3 cases). Lymphoblastic T-cell lymphomas were, in addition to T-cell markers, positive for TdT (Figure 1K). All T-cell lymphomas analyzed were positive for CD3 (Figure 1L), CD5 (Figure 1M) and CD43.

Antibodies Useful for the Diagnosis of Myeloid Leukemias

MPO was used to demonstrate intra-cytoplasmic granules in acute myeloid leukemia (21 cases, Figure 1N), and was helpful to distinguish acute myeloid leukemia without maturation from diffuse large B-cell lymphomas and Burkittlike lymphomas. The cytoplasmic staining pattern in myeloblasts was very characteristic with enhancement of the Golgi area. Ter-119 proved to be a useful marker for the diagnosis of erythroid leukemia (4 cases, Figure 1O). This marker recognizes early erythroblasts through mature erythrocytes.

Antibodies Useful for Histiocytic and Mast Cell Neoplasms

Mac-2, lysozyme, Mac-3, CD117, and mast cell tryptase were helpful in histiocytic and mast cell neoplasms. Only 3 cases were diagnosed as mast cells neoplasms. However, neoplastic mast cells stained positive for CD117 and mast cell tryptase. Although uncommon in other strain backgrounds, histiocytic sarcomas were seen as late onset tumors in irradiated Blm deficient mice on a mixed 129S4;C57BL/6 background (21 cases) and virus infected NMRI mice (5 cases). Histiocytic sarcomas stained positive for Mac-2, Mac-3, and lysozyme.

Problems Occurring with Automated IHC-Protocols for Murine Tissue

It was the aim of this study to establish a panel of antibodies for murine hematopoietic tumors in order to provide a useful diagnostic tool for any research laboratory. Therefore, manual procedures with commercially available antibodies were established first to evaluate the quality of stains and second to analyze their diagnostic value. As this task was successfully achieved with 29 antibodies, further optimization and automatization were pursued by using an automated immunostainer (Ventana Medical Systems). All antibodies were established using a basic DAB detection kit, which gave good results with all cross-reactive polyclonal primary antibodies raised against human epitopes, without nonspecific background (Figure 2B).

In contrast, 12 mouse monoclonal antibodies did not work at all, 4 showed only a weak staining, and 5 had a strong non-specific staining, which hampered the interpretation. However, during the course of this study, the basic DAB detection kit was discontinued and replaced by an iVIEW DAB detection kit. Both kits contain a secondary antibody mixture of rabbit-IgG and mouse-IgM and -IgG, but in different concentrations. Application of the iVIEW DAB kit resulted in a nonspecific staining of cell membranes of B-cells and cytoplasm of plasma cells when primary rabbit antibodies were used (Figure 2D).

To solve this problem, a diluted secondary antibody only against rabbit IgG was used together with the iVIEW DAB kit. Specific, crisp staining was again achieved without non-specific staining (Figure 2F). To avoid nonspecific staining of primary mouse monoclonal antibodies, a mouse anti-mouse kit was used as described above. Although fully automated staining procedure gave negative results, a semi-automated approach with manual blocking of nonspecific binding by applying goat serum to the slides prior to overnight incubation of the immune complex and completing the procedure in the immunostainer yielded specific results.

Nonspecific staining was reduced with exception of some weak staining in few germinal center cells. In addition, primary goat polyclonal antibodies gave negative results when diluted in a commercially diluent (Ventana Medical System). To avoid this problem, these antibodies were diluted in PBS buffer, resulting in specific staining.

Discussion

Due to the sequencing of the mouse genome (Waterston et al., 2002) and other advances in genetic technology (van der Weyden et al., 2002) an increasing number of mouse models of human cancer are being created. A large number of these develop lymphomas and leukemias, either by design or coincidentally (Tadesse-Heath and Morse, 2000). Because of this, human and veterinary pathologists are increasingly confronted with the analysis and characterization of murine hematopoietic neoplasms.

Due to the need to validate mouse models, there is an increasing need for reliable antibodies to aid in the correct classification and comparison of murine hematopoietic tumors with their human counterpart (Rawat et al., 2004; Sørensen et al. 2004, 2005; Schessl et al. 2005). In this study, 48 antibodies were tested for their specific reactivity or cross-reactivity with murine tissue and for their usefulness in the classification of murine lymphomas or leukemias. The antibodies selected were based on the list of antibodies known to be most useful in the diagnosis of human hematologic malignancies.

Of the 48 antibodies tested, 29 worked on paraffin-embedded murine tissue. However, not all of them were helpful for the diagnosis of murine hematopoietic neoplasms. Some markers were of special interest only for specific studies (Hahn et al., 2000) (e.g., cyclin D1, p21, p27), whereas others were used frequently in routine diagnostics (i.e., B220, CD79αcy, CD138, TdT, CD3, MPO, Mac-2, lysozyme, Ter-119, and Ki-67), and are proposed as a basic panel for the immunophenotype of murine hematopoietic neoplasms.

The most reliable murine pan-T-cell marker is CD3. However, detailed analysis of T-cell subtypes was not possible because of unavailability of CD4 and CD8 antibodies reactive in formalin-fixed tissue. Currently, there seems to be no alternative to flow cytometry or IHC in frozen tissue for further T-cell characterization. CD 5 was positive in most lymphoblastic lymphomas and small T-cell lymphomas in this study, but showed weaker reactivity than CD3.

B220 turned out to be the best pan-B-cell marker although it was negative in some cases of B-lymphoblastic lymphoma and most cases of splenic marginal B-cell lymphoma, which required other B-cell markers like CD79αcy and/or Pax-5 to document the B-cell phenotype of these neoplasms. CD138 positivity was helpful to confirm the plasma cell origin of a B-cell neoplasm.

BCL-6 expression in reactive lymphoid hyperplasia is confined exclusively to germinal center cells (Hans et al., 2004). Human follicular lymphomas are derived from germinal center cells; accordingly, neoplastic follicular center B-cells express BCL-6. The most common cytogenetic abnormality, t(14;18)(q32;q21), which is present in 70–95% of human follicular lymphomas results in constitutive overexpression of BCL-2. The simultaneous expression of BCL-2 and BCL-6 characterizes neoplastic follicles. Despite its name, the relationship of murine “follicular” lymphoma to its human counterpart is controversial. Murine “follicular” B-cell lymphomas may arise from any compartment of the follicle (mantle area or germinal center) and mostly show a diffuse growth pattern (Morse et al., 2002). Furthermore, BCL-2 and BCL-6 were not expressed in murine “follicular” B-cell lymphomas.

In agreement with these observations are data from Suzuki and colleagues, who used high throughput inverse PCR to analyze the sequences of retroviral integration sites from a large tumor panel, including five “follicular” B-cell lymphomas. BCL-2 insertion was not identified in any of the cases examined (Suzuki et al., 2002). Thus, the lack of phenotypic overexpression of BCL-2 suggests that murine “follicular” lymphoma lacks the hallmark molecular alteration of human follicular lymphoma, implying that what is called “follicular” lymphoma in mice is a different entity. Furthermore, the criteria of “follicular” lymphoma in mice requires the presence of a mixture of centroblasts and centrocytes in a usually diffuse growth pattern, which makes it difficult for human pathologists to distinguish between “follicular” lymphomas in mice and human diffuse large B-cell lymphomas. It is also possible that what is called follicular lymphomas in mice is more related to human diffuse large B-cell lymphomas. Further studies are needed to clarify the relationship of human and murine follicular B-cell lymphomas, and human diffuse large B-cell lymphomas and murine follicular B-cell lymphomas. Antibodies against immunoglobulin light chains are used in human hematopathology to detect monoclonal B-cell populations. The ratio of B-cells expressing Igκ or Igλ in man is 0.7–2.8. Ratios of <0.5 and >4.0 are indicative of monoclonal B-cell proliferations (van Dongen et al., 2003). Although appropriate cross-reactivity was observed with antibodies against Igκ and Igλ, IHC is not useful for the detection of monoclonal plasma cell or B-cell proliferations in mice, because 95 % of all nonneoplastic murine B-cells/plasma cells express Igκ (Woloschak and Krco, 1987).

In contrast, IgM and IgG were valuable for the discrimination of precursor B-cell lymphoblastic lymphomas from mature B-cell lymphomas. The combination of IgG positivity and TdT negativity were helpful in the diagnosis of Burkitt-like lymphomas, a murine neoplasm composed of mature B-cells with a lymphoblastic morphology, which seems to be unique in mice and lacks a human counterpart.

IHC stains were successfully established for a large panel of antibodies after heat-induced antigen retrieval. In general, changes in antibody reactivity in paraffin-embedded material may relate to fixation time (Jacobsen et al., 1980). Surprisingly, in this study IHC stains of sufficient quality were achieved also in tissues that were fixed in formalin for several weeks. Another interesting observation was a poor IHC in several tissue samples that were fixed in PFA. Our results do not support previous reports suggesting that 2% to 4% PFA is the best fixative to preserve antigens for IHC in paraffin-embedded tissues (Kerr et al., 1988; Grootenhuis et al., 1996). Previous reports also described poor staining results using formalin as fixative (Arnold et al., 1996; Mikaelian et al., 2004). However, heat-mediated antigen retrieval techniques have overcome most of these problems. Moreover, formalin is the best fixative for successful molecular analysis, when frozen tissue is not available. Nevertheless, due to differences in fixation time and quality of tissues among laboratories, it is always recommended to perform titration of all antibodies and to include positive and negative controls in each run to avoid misinterpretation of the staining. In this study, all immunohistochemical stains were first established manually. Since in many instances the IHC of mouse tissue is performed in pathology departments, where automated immunostainers are available, we were interested to see whether it was possible to establish the same panel of antibodies in an automated immunostainer. Although we had some problems with nonspecific staining of murine B- and plasma cells due to the mix of secondary anti-mouse and anti-rabbit antibodies supplied in the detection kit, the problem was resolved using the secondary antibodies separately.

The use of monoclonal mouse antibodies for IHC on mouse tissue is challenging because secondary anti-mouse antibodies naturally result in binding to endogenous mouse immunoglobulins. A mouse anti-mouse kit was used to avoid this problem. Interestingly, B-cells of primary follicles and marginal zone cells were completely blocked with this procedure, whereas germinal center cells showed very weak membranous staining. Nonetheless, depending on the expression site of the target epitope (e.g., nuclear localization), an interpretation may be possible if one is aware of a possible nonspecific, membranous staining of B-cells.

In conclusion, the availability of antibodies for IHC analysis of formalin-fixed, paraffin-embedded tissue is growing as is the diagnostic experience of classifying murine lymphomas and leukemias. Here we demonstrated the useful application of 23 antibodies reactive in paraffin-embedded murine tissue. Additional useful antibodies might emerge in the future. Automated staining of murine formalin-fixed, paraffin-embedded tissue enables a fast throughput and a stronger and crisper signal when compared to manual procedures. The possibility to immunophenotype and, thereby characterize murine neoplastic hematopoietic cells in paraffin sections is considered to be a remarkable tool for the mouse pathologist and enables a direct comparison with tumor cells of their human counterparts.