Inhibition of endogenous peroxidase for the immunocytochemical demonstration of intermediate filament proteins (IFP)

Journal of Immunological Methods, 116 (1989) 199-205

199

Elsevier JIM 05022

Inhibition of endogenous peroxidase for the immunocytochemical demonstration of intermediate filament proteins (IFP) A n t o n Hittmair and K u r t Werner Schmid Department of Pathology, University of lnnsbruck, lnnsbruck, Austria

(Received19 April 1988,revisedreceived17 June 1988,accepted22 August1988)

The necessity for minimally fixed and processed cell and tissue preparations for immunocytochemical studies of sensitive antigens such as lymphocyte surface markers is well recognised. In order to avoid methanol and hydrogen peroxide, which have been shown to be deleterious for certain antigens, various compounds have been proposed for blocking endogenous peroxidase activity (EPA) in tissue preparations which are to be used in immunoperoxidase reactions. In the present study the deleterious effect of methanol/H20 2 on intermediate filament proteins was demonstrated in both frozen sections and paraffin-embedded tissue. The use of alternative reagents for the non-deleterious blocking of EPA is recommended for immunocytochemical staining with antibodies against intermediate filaments. Key words: Intermediatefilamentprotein; Endogenousperoxidaseactivity;Immunocytochemistry;Cyclopropanonehydrate;Sodium

azide; Propyl-thiouracil

Introduction

In diagnostic immunocytochemistry antibodies against intermediate filament proteins (IFP) are most frequently used in an immunoperoxidase system. For the immunocytochemical demonstration of weakly expressed a n d / o r delicate antigens, e.g., lymphocyte surface markers, minimally fixed and processed cell and tissue preparations are required (MacMillan et al., 1982; Dienes et al., 1984; Holgate et al., 1986). In the case of the immunoperoxidase system this has resulted in an over-preservation of unwanted tissue elements including endogenous enzyme activity capable of drastically masking any specific staining due to the applied reagents. To overcome this a variety of compounds capable of selectively reacting chem-

Correspondence to: K.W. Schmid, Department of Pathology,Miillerstrasse44, A-6020 Innsbruck,Austria.

ically with the haem porphyrin IX ring structure were investigated for non-deleterious blocking of endogenous peroxidase activity (EPA). For this purpose compounds such as phenylhydrazine (Saito and Itano, 1981; Wood and Warnke, 1982; Jasani et al., 1984), sodium azide (Bolscher et al., 1984; Andrew and Jasani, 1987) and cyclopropanone hydrate (Wiseman et al., 1982; Schmid et al., 1988) have already been successfully employed. The aim of this study was to compare conventionally fixed and processed tissue preparations (both frozen sections and paraffin-embedded tissue) with tissue treated according to the recently developed procedures for the non-deleterious blocking of EPA, viz., sodium azide/H20 2 (Andrew and Jasani, 1987), cyclopropanone hydrate/H20 2 (Schmid et al., 1988) as well as propyl-thiouracil alone (Engler et al., 1982). In particular the immunolabelling of intermediate filament proteins has been studied for evidence of any loss of antigenicity during fixation and

0022-1759/89/$03.50 © 1989 ElsevierSciencePublishers B.V.(BiomedicalDivision)

200 processing. The avoidance or minimisation of such a loss is considered of central importance in the provision of diagnostically reliable immunohistochemical data.

Materials and methods

Tissue preparations (i) Frozen sections. Tissues from normal human cerebral cortex and cerebellum were obtained 6 h post mortem and human thyroid and uterine tissues were provided as surgical specimens. These were snap frozen using OCT (Tissue Tek) as a cryoprotectant and liquid nitrogen as the coolant. Sections were cut on a Cryostat 1720 (Leitz, F.R.G.) set to a cutting thickness of 5 /~m. The sections were taken onto alcohol cleaned chromegel-coated glass slides, air-dried and then stored at - 7 0 ° C until used. Prior to staining the sections were removed from - 7 0 ° C storage, air-dried for 20 rain to allow evaporation of moisture and fixed in absolute acetone (Merck, F.R.G.) for 10 rain, followed by 10 rain air-drying. (ii) Paraffin embedded tissue. Tissue blocks obtained adjacent to the material used for frozen sections were fixed for 18 h in 10% buffered formalin and routinely paraffin embedded. Sections were mounted on chrome-gel-coated glass slides. Prior to immunocytochemical staining the sections were dewaxed in xylene and rehydrated.

Preparation of cyclopropanone hydrate 1-ethoxycyclopropanol was synthezised by Dr. Helmut Schmidhammer from the Institute of Organic and Pharmaceutical Chemistry, University of Innsbruck, Austria, using ethyl 3-chloropropionate as starting material. The first step was the formation of 1-ethoxy-trimethylsiloxy-cyclopropane (Rtihlmann, 1971), from which 1ethoxycyclopropanol was prepared by methanolysis (Salaiin, 1976). Hydrolysis at 9 6 ° C yielded approximately 98% cyclopropanone hydrate (Wiseman and Abeles, 1979).

(i)

1-ethoxy-l-trimethylsiloxycyclopropane.

Over a period of 1 h 29.7 ml (0.22 mol) ethyl 3-chloropropionate was added dropwise at room temperature to a mixture of 11.25 g (0.49 g-atom) sodium silicate, 30.9 ml (0.24 mol) chlorotrimeth-

ylsilane, and 100 ml anhydrous ether while stirring vigorously. After stirring for a further 30 min the inorganic material was filtered off and the filtrate distilled: the ether was first removed at normal pressure and then 1-ethoxy-l-trimethylsiloxycyclopropane was distilled at 15 mm. (ii) 1-ethoxycyclopropanol. A solution of 3.55 g (20.38 m m o l ) 1-ethoxy-l-trimethylsiloxycyclopropane in 30 ml methanol was stirred at 2 5 - 2 8 ° C in a waterbath for 8 h. The solvent was removed at room temperature on a rotary evaporator and then 1-ethoxycyclopropanol was distilled at 15 mm. The resulting pure liquid compound was stored at 4 ° C until used for the preparation of cyclopropanone hydrate. (iii) Cyclopropanone hydrate. 1 - e t h o x y cyclopropanol was hydrolysed to yield cyclopropanone hydrate by heating an equal volume of it with water at 96 ° C (bath temperature) for 4 rain. The conversion to cyclopropanone hydrate was monitored and confirmed by 1H-NMR spectroscopy.

Blocking of EPA Both frozen and paraffin-embedded tissues were either treated with a 0.5% H202 (Merck, 30% w / v ) / m e t h a n o l mixture for 20 min at room temperature or with an alternative inhibition solution consisting of glucose (10 mM)/glucose-oxidase (1 U / m l ) and (i) sodium azide (1 mM, Sigma), (ii) cyclopropanone hydrate (1.0 t~l/ml) or (iii) propyl-thiouracil (0.25 m g / m l , Sigma), respectively. For the alternative inhibition solution, the glucose-oxidase and r-D-glucose (both reagents obtained from Sigma) were mixed in phosphatebuffered saline (PBS, 0.01 M, p H 7.3) with sodium azide, cyclopropanone hydrate or propyl-thiouracil, respectively, and preheated for 15 min in a waterbath to 37 ° C before use in blocking experiments. Control sections (i.e., without any specific blocking of EPA) were prepared by placing them in PBS (0.01 M, p H 7.3) either for 20 min at room temperature or for 1 h at 37°C, respectively. Blocking with H 2 0 2 / m e t h a n o l was followed by rinsing in tap water for 10 rain, in distilled water for 20 min and 3 x 2 min in PBS. All the other blocking procedures were followed by washing in PBS for 3 X 2 min.

201 TABLE I PRIMARY ANTIBODIES USED ON VARIOUS TISSUES Antibody specificity G l i a l fibriUary acidic p r o t e i n (GFAP) Neurofilament (NF) Cytokeratin Vimentin Desmin

Tissue source

C e r e b r a l cortex, c e r e b e l l u m C e r e b r a l cortex, c e r e b e l l u m Thyroid gland C e r e b r a l cortex, cerebellum, t h y r o i d gland, uterus C e r e b r a l cortex, cerebellum, thyroid gland, uterus

Immunoperoxidase staining Acetone-fixed frozen sections and paraffin sections taken from all tissue specimens used in this study, either untreated or treated with one of the EPA inhibiting agents, were incubated in a humidified chamber for 20 rain at ambient temperature (18-20°C) with 10% normal rabbit or swine serum (DAKO) to reduce the non-specific background. The sections were then incubated for 1 h at room temperature either with non-immune mouse serum or 1/10, 1/25, 1/50, 1/100 and 1/200 dilutions of monoclonal anti-glial fibrillary acidic protein (GFAP) (Euro Diagnostic), mono-

clonal anti-neurofilaments (NF) (Euro Diagnostic, 70-200kD cocktail), monoclonal anti-cytokeratin (AE1-AE3, Hybritech), monoclonal anti-desmin (Heintel) and polyclonal anti-vimentin (Euro Diagnostic) antibodies (diluted in 0.6% bovine serum albumin in 0.01 M PBS, Sigma), respectively. The antibodies used on the various tissues are listed in Table I. The sections were then washed ( x 3) for 2 rain in PBS and incubated for 45 rain with a rabbit anti-mouse peroxidase conjugate (DAKO, monoclonal primary antibodies) or a swine antirabbit peroxidase conjugate (DAKO, polyclonal primary antibody), both diluted 1/100 in BSA/PBS. After three washes in PBS (2 min each) the enzyme reaction was developed in a freshly made solution of diaminobenzidine (DAB, Sigma, 0.5 mg/ml) and 0.01% H202 (Merck, 30% w/v) for 5-7 min at room temperature. In each staining experiment sections taken from frozen and paraffin-embedded blocks of a lymph node from a case of Hodgkin's disease with numerous infiltrating eosinophils were used as the positive control for testing the efficiency of EPA blocking. All tested sections were counterstained with haematoxylin, dehydrated, cleared in xylene and mounted for examination and subsequent storage.

T A B L E II BLOCKING OF EPA IN VARIOUS FROZEN AND PARAFFIN-EMBEDDED C o m p a r i s o n of m e t h a n o l / H 2 0 2 ,

TISSUES

s o d i u m azide, c y c l o p r o p a n o n e h y d r a t e a n d p r o p y l - t h i o u r a c i l O p t i m a l d i l u t i o n of a n t i b o d i e s used GFAP 1/100

NF 1/50

Cyto 1/100

Vim 1/100

Des 1/100

H202

f p

+ ++

+ ++

+ ++

+ + ++

+ ++

S o d i u m azide

f p

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

Cyclopropanone hydrate

f p

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

Propyl-thiouracil

f p

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

+ + + + +++

Key: + = w e a k staining; + + = m o d e r a t e staining; + + + = s t r o n g staining; + + + + = very s t r o n g staining; f = frozen section p r e p a r a t i o n ; p = p a r a f f i n - e m b e d d e d tissue sections.

202

Fig. 1. a: Immunohistochemical staining with antibodies against neurofilament (1/50) on a frozen section of human cerebellum, blocking EPA with methanol/H202. Note poor preservation of positively stained neurofibrils and damage of tissue (haematoxylin counterstain, oil immersion X 1000). b: A semi-adjacent section taken from the same tissue and immunostained as shown in a pretreated with sodium azide/glucose oxidase/glucose solution to block EPA (NF antibodies diluted 1/50, haematoxylin counterstain, x 400).

203

Results

Complete blocking of EPA in the eosinophils of the Hodgkin's disease lymph node was achieved using sodium azide, cyclopropanone hydrate or propyl-thiouracil. Positive immunohistochemical staining with antibodies against GFAP was found in astrocytes and the radial glia in the molecular layer of the cerebellum. NF was found in the neurons of cerebral cortex, the cerebellum, the thyroid gland and the uterus. Vimentin and desmin could be demonstrated in myocytes of the uterus and in vessels of thyroid gland, cerebral cortex and cerebellum. Cytokeratin was identified in the follicular cells of the thyroid gland. As shown in Table II the antibodies used in this study consistently yielded very strong staining both with frozen sections and with paraffin-embedded tissue, following pretreatment with all of the alternative reagents used for EPA blocking. However, up to ten times higher concentrations of antibodies were required after H202 blocking of EPA in order to achieve approximately the same staining intensity when compared with blocking by one of the alternative reagents and this resulted in unacceptably high levels of non-specific background staining. Pretreatment of sections with methanol/H202 led to tissue damage in addition to loss of antigenicity both with frozen and paraffin-embedded tissues. This was particularly evident when using antibodies against NF, GFAP and cytokeratin. As shown in Fig. l a methanol/H202 pretreatment of sections resulted in breakage of neurofibrils whereas almost no such damage was observed following sodium azide/glucose oxidase/giucose treatment to block EPA (Fig. lb). The omission of EPA blocking produced unsatisfactory results since the associated high background staining prevented discrimination between specific and non-specific staining.

Discussion

The cytoskeleton of mammalian cells contains five subtypes of intermediate filament proteins: cytokeratin, vimentin, desmin, glial fibrillary acidic

protein (GFAP) and neurofilament (NF). Antibodies against intermediate filament proteins have recently become important in investigative and diagnostic immunocytochemistry. There is general agreement that the five subtypes of intermediate filaments show different and non-overlapping patterns of distribution in normal tissues. Antibodies against intermediate filaments are used as markers of tumour cell origin with important clinical implications for diagnosis and treatment. Several recent reports have indicated that the strict concept of the distribution of intermediate filament proteins may be disorganized in some neoplasms, such as renal carcinomas (Holthofer et al., 1983), adenoid cystic carcinomas (Caselitz et al., 1984), thyroid tumours (Miettinen et al., 1984), malignant mesothelioma (Churg, 1985; Jasani et al., 1985) and lung tumours (Gatter et al., 1986). Cytokeratins are a family of IFP. Thus cytokeratin filaments are represented in human tissues by at least 19 different cytokeratin polypeptides, which occur in cell type-specific combinations. Antibodies against cytokeratin are very useful in the identification of epithelial tumours and in distinguishing these from non-epithehal tumours. Vimentin is expressed predominantly in mesenchymal tissues and tumours whereas desmin is found in muscle tissues and muscle tumours. GFAP- and NF-specific antibodies are of considerable potential value in the study of tumours and neurodegenerative disorders both in the central (CNS) and peripheral (PNS) nervous system. In the mature normal CNS, GFAP is expressed by astrocytes, the radial glia in the molecular layer of the cerebellum and in ependymal cells (Eng and DeArmond, 1983). NF is composed of three different subunits (apparent molecular weights 70000, 150000 and 200000, respectively) and is expressed by mature neurons. Controversies regarding the expression of these intermediate filaments in different cells, such as GFAP in Schwann cells (Dahl et al., 1982), or the failure to detect NF and GFAP in the expected cell populations may reflect methodological limitations rather than a real absence of these antigens (Trojanowski, 1986). An improved staining procedure would help to resolve these difficulties. It is difficult to achieve efficient inhibition of EPA in the immunoperoxidase system. The uni-

204 versally most powerful method of EPA inhibition (methanolic H202) uses two compounds which are deleterious to sensitive antigens. In order to overcome the problem of endogenous versus specific peroxidase staining, other markers have been used (Erber et al., 1984). Since glucose oxidase does not occur in eukaryotic cells it offers substantial potential advantages over other enzymes (Rathley et al., 1981). However, we have found, like other investigators, that the reproducibility of the substrate conditions for this enzyme is critical. Furthermore, unlike the peroxidase mediated diaminobenzidine product, the enzyme product in the glucose oxidase system is unstable in organic solvents. Various attempts have been made to inhibit EPA in the immunoperoxidase system without adversely affecting delicate antigens. Phenylhydrazine, first recommended by Strauss (1972), was found to be more efficient in combination with hydrogen peroxide (Jasani et al., 1986; WynfordThomas et al., 1986). Kelly et al. (1987) used a low concentration of periodic acid to block EPA and this was particularly efficient when applied to frozen sections. Recently Andrew and Jasani (1987) described a procedure using sodium azide in combination with the more reliable nascent H202 produced by a glucose/glucose oxidase mixture. This was capable of non-deleterious inhibition of even highly resistant forms of EPA such as eosinophil peroxidase. F u r t h e r m o r e cyclopropanone hydrate has been found to be a potent agent for blocking EPA (Jasani et al., 1987; Schmid et al., 1988). Our study has shown that immunocytochemical studies with antibodies against I F P require satisfactory fixation and processing of the tissue preparations. The strongest staining was observed using frozen sections and blocking of EPA by one of the alternative inhibition reagents. This suggests that the embedding, formalin fixation, dewaxing, and rehydrating procedures are themselves deleterious to the antigens as is the blocking of EPA with methanol and H202. After blocking with m e t h a n o l / H 2 0 2 up to ten times higher concentrations of the antibodies were required to obtain approximately the same specific staining intensity. However, this was associated with inferior staining quality (high background staining

and a substantial damage of the tissue). In conclusion, in the present study using antibodies against intermediate filament proteins, satisfactory results were achieved both on frozen and paraffin-embedded material with non-deleterious blocking of EPA by means of sodium azide, cyclopropanone hydrate or propyl-thiouracil respectively, combined with minute amounts of nascent H202 provided by glucose-oxidase and glucose.

Acknowledgement The authors would like to express their sincere gratitude to Dr. Bharat Jasani, Department of Pathology, University of Wales College of Medicine, C a r d i f f / U K for his kind help and advice during the preparation of this paper.

References Andrew, S.M. and Jasani, B. (1987) An improved method for the inhibition of endogenous peroxidase non-deleterious to lymphocyte surface markers. Application to immunoperoxidase studies on eosinophil-rich tissue preparations. Histochem. J. 19, 426. Bolscher, B.G.J.M., Plat, H. and Wever, R. (1984) Some properties of human eosinophil peroxidase, a comparison with other peroxidases. Biochim. Biophys. Acta 784, 177. Caselitz, J., Becker, J., Seifert, G., Weber, K. and Osborn, M. (1984) Coexpression of keratin and vimentin filaments in adenoid cystic carcinomas of the salivary gland. Virchows Arch. Abt. A Pathol. Anat. Histopathol. 403, 337. Churg, A. (1985) Immunohistochemical staining for vimentin and keratin in malignant mesothelioma.Am. J. Surg. Pathol. 9, 360. Dahl, D., Chi, N.H., Miles, L.E., Nguyen, B.T. and Bignami, A. (1982) Flial fibrillary acidic (GFA) protein in Schwann cells: Fact or artifact. J. Histochem. Cytochem. 30, 912. Dienes, H.P., Essling, K. and John, H.D. (1984) Application of immunoelectron microscopy for identification of T lymphocyte subjects and HLA DR antigen positive cells in tissue. J. Histochem. Cytochem. 32, 547. Eng, L.F. and DeArmond, S.J. (1983) Immunohistochemistry of the glial fibrillary acidic protein. Prog. Neuropathol. 5, 19. Engler, H., Tautog, A. and Nakashima, T. (1982) Mechanism of inactivation of thyroid peroxidase by thioureylene drugs. Biochem. Pharmacol. 31, 3801. Erber, W.N., Pinching, A.J. and Mason, D.Y. (1984) Immunocytochemical detection of T and B cell populations in routine blood smears. Lancet i, 1024.

205 Gatter, K.C., DunniU, M.S., Van Muijen, G.N.P. and Mason, D.Y. (1986) Human lung tumours may coexpress different classes of intermediate filaments. J. Clin. Pathol. 39, 950. Holgate, C.S., Jackson, P., Pollard, K., Lunny, D. and Bird, C.C. (1986) Effect of fixation on T and B lymphocyte surface membrane antigen demonstrated in paraffin processed tissue. J. Pathol. 149, 293. Holthofer, H., Miettinen, A., Passivuo, R., et al. (1983) Cellular origin and differentiation of renal carcinomas: a fluorescence microscopic study with kidney-specific antibodies, anti-intermediate filament antibodies and lectins. Lab. Invest. 49, 317. Jasani, B., Hallam, L.A., Newman, G.R. and Williams, E.D. (1984) Non-deleterious inhibition of endogenous peroxidase in immunolocalisation studies involving the use of monoclonal antibodies. J. Pathol. 142, 46A. Jasani, B., Edwards, R.E., Thomas, N.D. and Gibbs, A.R. (1985) The use of vimentin antibodies in the diagnosis of malignant mesothelioma. Virchows Arch. Abt. A Pathol. Anat. Histopathol. 406, 441. Jasani, B., Wynford-Thomas, D., Thomas, N.D. and Newman, G.R. (1986) Broad spectrum non-deleterious inhibition of endogenous peroxidase, light and electron microscope immunocytochemical application. Histochem. J. 18, 56A. Jasani, B., Andrew, S.M., Schmid, K.W., Thomas, N.D. and Schmidhammer, H. (1987) A haem specific reagent system for non-deleterious inhibition of endogenous peroxidatic activity (EPA): Application to immunoperoxidase staining to delicate antigens in EPA-rich tissues. Proceedings of th,." 38th Annual Meeting of the American Histochemical Society, New Orleans, J. Histochem. Cytochem. 35, 1031. Kelly, J., Whelan, C.A., Weir, D.G. and Feigherty, C. (1987) Removal of endogenous peroxidase activity from cryostat sections for immunoperoxidase visualisation of monoclonal antibodies. J. Immunol. Methods 96, 127. McMillan, E.M., Wasik, R. and Everett, M.A. (1982) OKTpositive cells. Their demonstration in human thymus, and the effect of fixation on the immunoperoxidase reaction. Arch. Pathol. Lab. Med. 106, 9. Miettinen, M., Franssila, K.O., Lehto, V.-P., et al. (1984)

Expression of intermediate filament protein in thyroid gland and in thyroid tumors. Lab. Invest. 50, 262. Rathley, T., Hocko, J.M., Franks, G.F., Suffin, S.C., O'Donnell, C.M. and Porter, D.D. (1981) Glucose oxidase enzyme methodology as a substitute for fluorescence microscopy in the clinical laboratory. Clin. Chem. 27, 1513. Riahlmann, K. (1971) Die Umsetzung yon Carbons~iureestern mit Natrium in Gegenwart von Trimethylchlorsilan. Synthesis, p. 236. Saito, S. and Itano, H.A. (1981) fl-Meso-phenylbiliverdin IXa and N-phenylprotoporphyrin IX products of the reaction of phenylhydrazine with oxyhaemproteins. Proc. Natl. Acad. Sci. U.S.A. 78, 5508. Salaiin, J. (1976) Preparation and substituent effect in the solvolysis of 1-ethylcyclopropyl tosylates. J. Org. Chem. 41, 1237. Schmid, K.W., Hittmair, A., Schmidhammer, H. and Jasani, B. (1988) Non-deleterious inhibition of endogenous peroxidatic activity (EPA) by cyclopropanone hydrate: a deftnitive approach. J. Histochem. Cytochem., submitted. Strauss, W. (1972) Phenylhydrazine as inhibitor of horseradish peroxidase for use in immunoperoxidase procedures. J. Histochem. Cytochem. 20, 949. Trojanowski, J.Q. (1986) Neurofilaments and glial filaments in neuropathology. In: J.M. Polak and S. Van Noorden (Eds.), Immunocytochemistry. Modern Methods and Applications, 2nd edn. Wrigth, Bristol, p. 414. Wiseman, J.S. and Abeles, R.H. (1979) Mechanism of inhibition of aldehyde dehydrogenase by cyclopropanone hydrate and mushroom toxin coprine. Biochemistry 18, 427. Wiseman, J.S., Nichols, J.S. and Kolpak, M.S. (1982) Mechanism of inhibition of horsradish peroxidase by cyclopropanone hydrate. J. Biol. Chem. 257, 6328. Wood, G.S. and Warnke, R.A. (1982) The immunologic phenotyping of bone marrow biopsies and aspirates: Frozen section techniques. Blood 59, 913. Wynford-Thomas, D., Jasani, B. and Newman, G.R. (1986) Immunohistochemical localisation of cell surface receptors using a noval method permitting simple, rapid and reliable L M / E M correlation. Histochem. J. 18, 387.