APMIS 100: 1001-1007, 1992
The use of Dost-embeddinszimmunoelectron microscoDv in the diagnosis of glomerular diseases Comparison of immunoelectron microscopic and immunofluorescence studies FREDRIK J. SKJORTEN, SVERRE-HENNING BRORSON, BORGHILD ROALD, ERIK H. STROM and BJ0RG LUND Department of Pathology, Ullevaal Hospital, and the University of Oslo, Oslo, Norway
Skjerrten, F. J., Brorson, S.-H., Roald, B., Strerm, E. H. & Lund, B. The use of post-embedding immunoelectron microscopy in the diagnosis of glomerular diseases. Comparison of immunoelectron microscopic and immunofluorescence studies. APMIS 100: 1001-1007, 1992. Fifty renal biopsies were studied by immunoelectron microscopy after embedding in a partly hydrophilic polyacrylic resin (LR White). Immunofluorescence studies were carried out on frozen sections of parallel tissue samples. Polyacrylic embedding gave good preservation of the renal ultrastructure and precise localization of immunoglobulin and C3c antibodies within glomerular electron-dense deposits. Non-specific staining of plasma proteins within vascular lumina could easily be detected. There was good correlation between immunoelectron and immunofluorescence microscopy. Immunoelectron microscopy is a very sensitive method, which can detect small amounts of antigen. More cases were, however, positive by immunofluorescence than by immunoelectron microscopy. This discrepancy may be explained by difference in sample size, and by difference in resolution of morphological details (electron microscopy versus fluorescence microscopy).
Key words: Renal biopsies; immunoelectron microscopy; immunofluorescence microscopy. F. J. Skjerrten, Department of Pathology, Ullevaal Hospital, 0407 Oslo 4, Norway.
For a proper diagnosis of renal biopsies it is considered essential to know if, and what sort of, immunoglobulins are deposited in the renal tissue (4). This information is commonly obtained by immunofluorescence studies (IF) of frozen sections. Similar results may be obtained from studies on paraffin-embedded tissues, using immunoenzyme (most commonly immunoperoxidase) methods. Both methods have their shortcomings, according to the techniques employed. I F microscopy has lower resolution than ordinary light microscopy, and this may at times lead to diagnostic problems related to the position of the deposits within the glomeruli. The immunoperoxidase method applied to sec-
Received May 25, 1992. Accepted August 14, 1992.
tions of paraffin-embedded tissue may cause difficulties in the differentiation of positively reacting intraluminal plasma from true immunoglobulin deposits in the capillary wall (3). Electron microscopy (EM) is commonly used as an adjunct in the diagnosis of renal biopsies. The detection of immunoglobulins by immunoelectron microscopy (IM-EM) has been attempted in many laboratories, often with limited success. A report by Herrera (1989) inspired us to change from epoxy embedding of renal biopsies to embedding in a partly hydrophilic acrylic polymer (9). With a post-embedding immune technique on thin sections of this material, we have obtained reproducible immune reactions with antibodies to immunoglobulins, while retaining acceptable ultrastructural morphology of the renal tissue. The aim of the present study was to correlate the results of 1001
SKJBRTEN et ul.
routine I F studies of renal biopsies with those of immunoelectron microscopy (IM-EM) of the same material.
MATERIALS AND METHODS From 1. Jan. 1990 to 30. Dec. 1991 a total of 50 renal biopsies were studied by light microscopy, IM-EM, and IF. The clinical diagnoses as given at the time of biopsy are shown in Table 1. Renal biopsy material was immediately divided into three parts: for LM, IM-EM, and IF. IM-EM Procedure The tissue was immediately fixed in phosphatebuffered 4 per cent paraformaldehyde and brought to the laboratory, where it was minced and fixed in buffered paraformaldehyde for &20 h. The tissue was then postfixed in 1% osmium tetroxide for 20 min, dehydrated in graded alcohols, infiltrated with acrylic monomer, and embedded in acrylic resin according to the manufacturer’s recommended procedure, without accelerator (9). The resin was polymerized at 54°C. Semi-thin sections were cut from all blocks, stained with methylene-blue basic fuchsin, and used for a preliminary light microscopic report sent to the clinician. Two representative glomeruli were selected for IM-EM. Ultrathin sections were cut from these blocks, supported on nickel grids, and left to dry overnight. Following etching in saturated NaI04 for 30 min, the sections were incubated with primary antibodies (Table 2) in 5% BSA overnight in a refrigerator (grids floating on drops in a closed chamber). Sections were then washed three times in PBS and incubated with gold-conjugated secondary antibody for one h (Table 2) at room temperature, rinsed in distilled water, and stained with uranyl acetate and lead citrate according to standard EM procedures. Control sections were similarly treated, but incubation with primary antibody was omitted. Manipulation during the immune procedure had to be very gentle in order not to damage the sections. Micrographs had to be taken at
TABLE 1, Clinical diannoses Diagnoses Number of cases Acute glomerulonephritis 3 2 Subacute glomerulonephritis Chronic glomerulonephritis 11 7 Nephrotic syndrome Renal disease NOS 18 Vasculitis 2 2 Amyloidosis Hvoertension 5 Total 50
1002
TABLE 2. Antibody reagents used Antibody reagents M / P Working Source* dilution FITC-IgG** P 1/10 DAKO FITC-IgM** P 1/10 DAKO FITC-IgA** P 1/10 DAKO FITC-C3c** P 115 DAKO IgG*** P 1/200 DAKO IgM*** P 1/200 DAKO IgA*** P 1/200 DAKO c3c*** P 1/200 DAKO Amyloid AA M 1/40 DAKO Kappa light chain P 1/400 DAKO Lambda light chain P 1/400 DAKO Gold-conjugated P 1/50 Amersham anti-rabbit Gold-conjugated P 1/30 Amersham anti-mouse M = Monoclonal; P = Polyclonal * Sources: DAKO A/S, Copenhagen, Denmark Amersham International, Amersham, UK ** Primary rabbit antisera conjugated with fluorescein isothiocyanate (FITC) *** Primary rabbit anti-human antiserum.
9,100 x primary magnification for the gold particles to be readily visible (particle size 15 nm). Generally, there was little non-specific background precipitation of gold particles. Only particles localized over electron-dense deposits, and in clearly increased numbers above the non-specific background, were accepted as positive. Reagents Buffer: 0.15 M phosphate buffer, pH 7.4 Fixatives: 4% paraformaldehyde in buffer 2% osmium tetroxide in buffer Resin: LR White, medium grade (BIORAD, Micro Science Div., Heme1 Hempstead, Hertfordshire) Immunojluorescence microscopy One part of the fresh renal biopsy was transported to the laboratory in PBS, snap frozen in dried ice in liquid isopentane, and embedded in O.C.T. compound (Tissue-Tek). Serial cryosections (5 pm) underwent a 5 min fixation in cold acetone and blocking with 10% BSA for 30 min before application of the immunological reagents. The sections were subjected to a 30 min incubation at room temperature with fluorescein-conjugated primary antisera to IgG, IgM, IgA, and C3c (working dilutions see Table 2). The sections were then rinsed in PBS, air-dried, and mounted in buffered polyvinyl alcohol medium. A set of serial sections were additionally stained with haematoxylin and eosin for morphological tissue identification. A Leitz Aristoplan microscope with a
IMMUNOELECTRON MICROSCOPY OF RENAL BIOPSIES
12 V lOOW mercury lamp and dry objectives was used. Narrow-band excitation and selective filtration of green (fluorescein)emission colour was registered, especially when related to the glomeruli. The mean number of glomeruli per sample was 9.2 (median 13.5). Observations were partly recorded on Ektachrome P800 (daylight).
RESULTS The majority of cases had symptoms indicative of glomerulonephritis (Table 1). The final pathological diagnoses are given in Table 3. Glomerulonephritis was diagnosed in 74% of the cases, while 15% had hypertensive renal disease as the primary diagnosis. In 15% of the glomerulonephritis cases there were also signs of hypertensive vascular damage, at times with glomerular hyperperfusion injury. The ultrastructural morphology of the renal tissue was well preserved in LR White-embedded control sections in which the immune procedure had not been carried out. Membranes and cellular organelles appeared essentially the same as after epoxy embedding, and electrondense deposits along the basement membranes were shown equally well. There was some loss of contrast following the immune procedure. Cellular details were still easily recognizable and electron-dense deposits retained most of their density (Fig. 1). Non-specific background precipitation of gold particles was minimal. Only particles localized over electron-dense deposits, and in clearly increased numbers over the background, were accepted as positive. Table 3 compares the results of IM-EM with IF. Positivity is defined as a positive reaction for one or more of the antibodies tested. We found discrepancies
in 12 cases (24%). The discrepancies were equally distributed between positive IM-EM only and positive I F only. Two IM-EM-negative cases had been fixed in unbuffered formalin by accident. In two cases which were recorded as IFpositive, the positivity was found to be restricted to intravascular plasma by IM-EM (Fig. 2). Table 4 gives the results for individual antibodies. There was variable correlation between IM-EM and I F results, poor for IgG, considerably better for IgM, IgA, and C3c. Thus, for IgG, one of four IM-EM-positive samples were positive by IF (25%), while 39 of 46 IM-EM cases were also negative by IF (85%). For IgA, I0 and 13 IM-EM-positive cases were also positive by I F (77%), and 36 of 37 IM-EM-negative cases were also negative by I F (97%). For IgM and C ~ Cthe , results were similar to IgA. The presence of electron-dense deposits demonstrated by EM, located either along the peripheral basement membrane, or in paramesangial areas, was considered to be diagnostic of glomerulonephritis (Fig. 1). The increased density and typical granularity of IgA deposits, as seen in epoxy resin-embedded tissue, was preserved after LR White embedding. In one case of glomerulonephritis, both IF and IM-EM were negative in spite of the presence of characteristic electron-dense deposits. No case lacking electron-dense deposits was considered to show evidence of glomerulonephritis. In hypertensive vascular and glomerular disease, electron-dense material was also often shown to be present in the capillary walls by EM. This material was patchily distributed in the paramesangial areas, and/or band-like along the basement membrane, on the endothelial side. It was usually positive for C3c and/
TABLE 3. Results according to final pathology diagnosis for immunoglobulin and/or C3c detection Diagnosis IF pos. IF pos. IF neg. IF neg. Total IM-EM neg. IM-EM pos. IM-EM pos. IM-EM neg. 0 12 Diffuse mesangioproliferative g1.n. 2 2 16 Mesangioproliferativeg1.n. with focal 2 7 2 1 12
accentuation Membranous g1.n. Minimal change Focal glomerulosclerosis Benign/malignant nephrosclerosis Various non-inflammatory kidney diseases Total
1 1 0 2 0 6
1 0 1 4 3 28
0 0 0 1 1 6
0 0 0 2 5 10
2 1
I 9 9 50
g1.n. = glomerulonephritis.
1003
SKJ0RTEN rt at.
or IgM (Fig. 3). Capillary deposits in glomerular hyperperfusion injury (1) were also frequently positive for IgM and C ~ Crarely , for IgG. Two cases with congo-positive glomerular deposits were positive for AA protein (Fig. 4). A case with fibrillar, congo-negative deposits was negative for all antibodies tested, including kappa and lambda, and was considered not to represent amyloid. In comparison with our established procedures for the processing of renal biopsies, an attempt was made to estimate the extra cost that the introduction of IM-EM entailed. No new equipment and no extra reagents were necessary. Our previous standard procedure, including conventional EM and IF, required about 17 h preparative work per sample. IM-EM is more labour intensive, and requires at least an extra four h of preparative work per sample.
DISCUSSION IM-EM has been applied to the study of a number of tissues for the detection of many antigens, employing different immune techniques and different embedding media. IM-EM of cryosections permits the detection of most antigens, but the preservation of morphological details is variable (1 1). Pre-embedding IM-EM has been successfully applied to the study of renal tissues in selected cases ( 6 ) . The pre-embedding technique, however, requires diffusible antibodies and markers for adequate tissue penetration. Post-embedding IM-EM on glutaraldehydefixed, epoxy-embedded tissue is well suited for the study of antigens which are preserved using these procedures, for example peptide hormones ( 7 , 12). Like most researchers, we have been unable to detect immunoglobulins with this procedure. This may be explained partly by reduc-
tion of available epitopes on immunoglobulin molecules by aldehyde crosslinking, and partly because epoxy resin is impenetrable to water and proteins. Only immunoglobulin epitopes on the surface of the sections are therefore available for binding to antibodies. Recently, Herrera (5) successfully demonstrated immunoglobulins by IM-EM using a post-embedding technique on polyacrylic resin sections. Following his recommendations, we obtained reproducible results with this technique. There was good preservation of the ultrastructural morphology when paraformaldehyde-osmium fixation was maintained, similar to, but not quite as good as, the tissue preservation obtained after aldehyde-osmium fixation and epoxy embedding. It is essential for the preservation of immunoglobulin antigenicity that the temperature of the tissue does not exceed 56 "C during polymerization of the embedding medium. Polymerization therefore has to be carried out in a stable thermostat, without catalyst, since catalytic polymerization may render this procedure exothermic (9). Immune incubation had to be preceded by NaI04 etching in order to remove Os04 from the surface of the thin sections. Without etching, the immune reactions were weak or negative. If osmium fixation was omitted, contrast was poor and morphology was inadequate for ultrastructural diagnostic work. Antibodies have been found to penetrate acrylic polymers to a depth of about 200 nm in 30 min (10). Gold particles with a diameter of 15 nm do not penetrate the resin at all. Therefore, only antibodies at the surface are visualized by gold particles of this size. We are not aware of other papers offering detailed comparisons of IM-EM and I F examination of renal biopsies. Herrera (1989) studied 53 renal biopsies embedded in LR White by IMEM, and stated that there was good correlation
Fig. 1. Massive electron-dense paramesangial deposits, with gold-labelled IgA (negligible background staining). Good ultrastructural preservation. 24,000 x . B: basement membrane. D: deposit. N: nucleus. Fig. 2. Gold-labelled IgA in intravascular plasma; non-specific reaction. Suboptimal ultrastructural preservation. 21,000 x . B: basement membrane. E: epithelium. N: nucleus. Fig. 3. Gold-labelled C3c in glomerular hyperperfusion injury. 17,000 x . B: basement membrane. D: deposits. Fig. 4. Gold-labelled AA amyloid. Note precise localization of gold particles on individual amyloid fibrils. 75,000 x . AA: amyloid. N: nucleus.
1004
IMMIJNOELECTRON MICROSCOPY OF RENAL BIOPSIES
1005
SKJBRTEN
TABLE 4. Immunoelectron microscopy results for individual antibodies (Parenthesis: uositive bv IF) IEG IEM IEA C3c Total + 4 (1) 12 (10) 13 (10) 22 (17) 52 (38) 46 (7) 38 (6) 37 (1) 28 (4) 149 (18) Total 50 f8) 50 (16) 50 (11) 50 (21) 200 (561
with IF findings, but gave no details. In our study, two thirds of the glomerulonephritis-like conditions were positive by both methods. An equal number of cases (12%) were IF-positive and IM-EM-negative, or IF-negative and IMEM-positive. In two of the IF-positive and IMEM-negative cases, the I F positivity was restricted to plasma in capillary lumina, and was thus without biological significance. In these cases, IM-EM therefore demonstrated a higher diagnostic specificity than IF. Nine per cent were negative by both methods. The best correlations between IM-EM and I F were found for IgA and C3c. IgA was detected more often by IM-EM than by IF, The number of IgG-positive cases was low, with a large discrepancy between I F and IM-EM. This may be explained by nonspecific staining of intraluminal plasma IgG by IF, IgG being the immunoglobulin present in the highest concentration in plasma. In the application of IF to demonstrate tissuebound immunoglobulins, extensive washing of the tissue has been recommended before the immune procedure is performed, in order to reduce non-specific binding of antibodies to passively retained immunoglobulins (2). Such washing cannot be applied to tissue which is intended for IM-EM since the washing procedure will destroy the fine structure of the tissue. IF of frozen sections is a sensitive method for the detection of the antibodies in question. IMEM appears to have a somewhat lower sensitivity, probably related to the number of glomeruli in the sample: only two glomeruli are studied per sample by IM-EM, versus 9.2 for IF. Another factor is the number of epitopes available for binding, which is low for IM-EM due to inactivation by crosslinking, and the minute amount of tissue available for antibody visualization by gold particles; while antibodies penetrate the sections in both instances, gold particles do not penetrate plastic sections. Fluorochrome-conjugated antibodies, however, 1006
et
al.
penetrate frozen sections. On the other hand, the specificity of IM-EM is very high, due to the unique possibility for correlation of antibody binding with ultrastructural morphology. This makes it possible to exclude non-specific staining. In the present study, we chose to let the diagnosis of glomerulonephritis rest on the demonstration of electron-dense deposits with a morphology conforming to the accepted picture of immune complexes (4).The demonstration of immunoglobulins and/ or complement in these deposits, in our opinion, strengthened the diagnosis of glomerulonephritis. Hypertensive renal damage may also lead to deposition of immunoglobulins and/or complement in vessel walls (1). We find that these deposits differ morphologically from those of glomerulonephritis. IM-EM of amyloid deposits offers an opportunity for their subclassification into secondary, lambda, kappa, and other types (8). Labelled antibodies localized on the fibrils give a specific diagnosis. In conclusion, IM-EM is a very sensitive method for the detection of immunoglobulins in renal biopsies, it has the capacity to detect very small quantities of antigens. The combination of immune technique and ultrastructure adds an extra dimension to the diagnostic specificity, which is greater than that of IF alone. The increased specificity has diagnostic importance, for instance in the differentiation of tissue changes caused by immune complex disease or hypertension. In laboratories which routinely use EM in the study of renal biopsies, the moderate extra cost of IM-EM is justified by the increased specificity afforded by this method.
REFERENCES 1. Bohle, A., Biwer, E. & Christensen, J. A . : Hyperperfusion injury of the human kidney in different glomerular diseases. Am. J. Nephrol. 8: 179-186, 1988. 2 . Brandtmg, f!: Mucosal and glandular distribution of immunoglobulin components. Immunology 26: 1101-1 114, 1974. 3. Brandtmg, I(, Huitfeldt, H. S., Oppedal, B. R. & Rognum, I: 0.:Letter to the Editor. Re: Demonstration of lymphocyte surface antigens in paraffin-embedded human tissue. J. Immunol. Methods 87: 289-290. 1986.
IMMUNOELECTRON MICROSCOPY OF RENAL BIOPSIES
4. Churg, J. & Sobin, L. H.: Renal disease. Classification and atlas of glomerular diseases. IgakuShoin, Tokyo, New York 1982. 5. Herrera, G. A , : Ultrastructural postembedding immunogold labelling: applications to diagnostic pathology. Ultrastruct. Pathol. 13: 485499, 1989. 6. Hinglais, N., Kazatchkine, M . D., Bhakdi, S., Appay, M.-D., Mandet, C . , Grossetete, J. & Bariety, J.: Immunohistochemical study of the C5b-9 complex of complement in human kidneys. Kidney Int. 30: 399410, 1986. 7. Holm, R., Farrants, G. W , Nesland, J. M . , Sobrinho-Simoes. M., Jmgensen, 0. G. & Johannessen, J. Y ; Ultrastructural and electron immunohistochemical features of medullary thyroid carcinoma. Virchows Arch. Path. Anat. A 414: 375-384, 1989. 8. Linke B. R,Huhn, D., Casanova, S. & Donini, U.: Methods in laboratory investigation. Immunoelectron microscopic identification of human
9.
10. 1 1.
12.
AA-type amyloid: exploration of various monoclonal AA-antibodies, methods of fixation, embedding and of other parameters for the proteinA gold method. Lab. Invest. 61; 691-697, 1989. “LR White” product information sheet. Polaron Instruments Inc., 2293 Amber Drive, Line Lexington Industrial Park, Hatfield, PA 19440, U.S.A., 1989. Newman, G. R. & Hobot. J. A . : Modern acrylics for post-embedding immunostaining techniques. J. Histochem. Cytochem. 35; 971-981, 1987. Roos, N . & Morgan, A . J.; Cryopreparation of thin biological specimens for electron microscopy: methods and applications. Microscopy Handbooks 21, Oxford University Press. Royal Microscopical Society, 1990. Wang, B.-L. & Larsson, L.-I.;Simultaneous demonstration of multiple antigens by indirect immunofluorescence or immunogold staining. Histochemistry 85: 47-56, 1985.
1007
The use of Dost-embeddinszimmunoelectron microscoDv in the diagnosis of glomerular diseases Comparison of immunoelectron microscopic and immunofluorescence studies FREDRIK J. SKJORTEN, SVERRE-HENNING BRORSON, BORGHILD ROALD, ERIK H. STROM and BJ0RG LUND Department of Pathology, Ullevaal Hospital, and the University of Oslo, Oslo, Norway
Skjerrten, F. J., Brorson, S.-H., Roald, B., Strerm, E. H. & Lund, B. The use of post-embedding immunoelectron microscopy in the diagnosis of glomerular diseases. Comparison of immunoelectron microscopic and immunofluorescence studies. APMIS 100: 1001-1007, 1992. Fifty renal biopsies were studied by immunoelectron microscopy after embedding in a partly hydrophilic polyacrylic resin (LR White). Immunofluorescence studies were carried out on frozen sections of parallel tissue samples. Polyacrylic embedding gave good preservation of the renal ultrastructure and precise localization of immunoglobulin and C3c antibodies within glomerular electron-dense deposits. Non-specific staining of plasma proteins within vascular lumina could easily be detected. There was good correlation between immunoelectron and immunofluorescence microscopy. Immunoelectron microscopy is a very sensitive method, which can detect small amounts of antigen. More cases were, however, positive by immunofluorescence than by immunoelectron microscopy. This discrepancy may be explained by difference in sample size, and by difference in resolution of morphological details (electron microscopy versus fluorescence microscopy).
Key words: Renal biopsies; immunoelectron microscopy; immunofluorescence microscopy. F. J. Skjerrten, Department of Pathology, Ullevaal Hospital, 0407 Oslo 4, Norway.
For a proper diagnosis of renal biopsies it is considered essential to know if, and what sort of, immunoglobulins are deposited in the renal tissue (4). This information is commonly obtained by immunofluorescence studies (IF) of frozen sections. Similar results may be obtained from studies on paraffin-embedded tissues, using immunoenzyme (most commonly immunoperoxidase) methods. Both methods have their shortcomings, according to the techniques employed. I F microscopy has lower resolution than ordinary light microscopy, and this may at times lead to diagnostic problems related to the position of the deposits within the glomeruli. The immunoperoxidase method applied to sec-
Received May 25, 1992. Accepted August 14, 1992.
tions of paraffin-embedded tissue may cause difficulties in the differentiation of positively reacting intraluminal plasma from true immunoglobulin deposits in the capillary wall (3). Electron microscopy (EM) is commonly used as an adjunct in the diagnosis of renal biopsies. The detection of immunoglobulins by immunoelectron microscopy (IM-EM) has been attempted in many laboratories, often with limited success. A report by Herrera (1989) inspired us to change from epoxy embedding of renal biopsies to embedding in a partly hydrophilic acrylic polymer (9). With a post-embedding immune technique on thin sections of this material, we have obtained reproducible immune reactions with antibodies to immunoglobulins, while retaining acceptable ultrastructural morphology of the renal tissue. The aim of the present study was to correlate the results of 1001
SKJBRTEN et ul.
routine I F studies of renal biopsies with those of immunoelectron microscopy (IM-EM) of the same material.
MATERIALS AND METHODS From 1. Jan. 1990 to 30. Dec. 1991 a total of 50 renal biopsies were studied by light microscopy, IM-EM, and IF. The clinical diagnoses as given at the time of biopsy are shown in Table 1. Renal biopsy material was immediately divided into three parts: for LM, IM-EM, and IF. IM-EM Procedure The tissue was immediately fixed in phosphatebuffered 4 per cent paraformaldehyde and brought to the laboratory, where it was minced and fixed in buffered paraformaldehyde for &20 h. The tissue was then postfixed in 1% osmium tetroxide for 20 min, dehydrated in graded alcohols, infiltrated with acrylic monomer, and embedded in acrylic resin according to the manufacturer’s recommended procedure, without accelerator (9). The resin was polymerized at 54°C. Semi-thin sections were cut from all blocks, stained with methylene-blue basic fuchsin, and used for a preliminary light microscopic report sent to the clinician. Two representative glomeruli were selected for IM-EM. Ultrathin sections were cut from these blocks, supported on nickel grids, and left to dry overnight. Following etching in saturated NaI04 for 30 min, the sections were incubated with primary antibodies (Table 2) in 5% BSA overnight in a refrigerator (grids floating on drops in a closed chamber). Sections were then washed three times in PBS and incubated with gold-conjugated secondary antibody for one h (Table 2) at room temperature, rinsed in distilled water, and stained with uranyl acetate and lead citrate according to standard EM procedures. Control sections were similarly treated, but incubation with primary antibody was omitted. Manipulation during the immune procedure had to be very gentle in order not to damage the sections. Micrographs had to be taken at
TABLE 1, Clinical diannoses Diagnoses Number of cases Acute glomerulonephritis 3 2 Subacute glomerulonephritis Chronic glomerulonephritis 11 7 Nephrotic syndrome Renal disease NOS 18 Vasculitis 2 2 Amyloidosis Hvoertension 5 Total 50
1002
TABLE 2. Antibody reagents used Antibody reagents M / P Working Source* dilution FITC-IgG** P 1/10 DAKO FITC-IgM** P 1/10 DAKO FITC-IgA** P 1/10 DAKO FITC-C3c** P 115 DAKO IgG*** P 1/200 DAKO IgM*** P 1/200 DAKO IgA*** P 1/200 DAKO c3c*** P 1/200 DAKO Amyloid AA M 1/40 DAKO Kappa light chain P 1/400 DAKO Lambda light chain P 1/400 DAKO Gold-conjugated P 1/50 Amersham anti-rabbit Gold-conjugated P 1/30 Amersham anti-mouse M = Monoclonal; P = Polyclonal * Sources: DAKO A/S, Copenhagen, Denmark Amersham International, Amersham, UK ** Primary rabbit antisera conjugated with fluorescein isothiocyanate (FITC) *** Primary rabbit anti-human antiserum.
9,100 x primary magnification for the gold particles to be readily visible (particle size 15 nm). Generally, there was little non-specific background precipitation of gold particles. Only particles localized over electron-dense deposits, and in clearly increased numbers above the non-specific background, were accepted as positive. Reagents Buffer: 0.15 M phosphate buffer, pH 7.4 Fixatives: 4% paraformaldehyde in buffer 2% osmium tetroxide in buffer Resin: LR White, medium grade (BIORAD, Micro Science Div., Heme1 Hempstead, Hertfordshire) Immunojluorescence microscopy One part of the fresh renal biopsy was transported to the laboratory in PBS, snap frozen in dried ice in liquid isopentane, and embedded in O.C.T. compound (Tissue-Tek). Serial cryosections (5 pm) underwent a 5 min fixation in cold acetone and blocking with 10% BSA for 30 min before application of the immunological reagents. The sections were subjected to a 30 min incubation at room temperature with fluorescein-conjugated primary antisera to IgG, IgM, IgA, and C3c (working dilutions see Table 2). The sections were then rinsed in PBS, air-dried, and mounted in buffered polyvinyl alcohol medium. A set of serial sections were additionally stained with haematoxylin and eosin for morphological tissue identification. A Leitz Aristoplan microscope with a
IMMUNOELECTRON MICROSCOPY OF RENAL BIOPSIES
12 V lOOW mercury lamp and dry objectives was used. Narrow-band excitation and selective filtration of green (fluorescein)emission colour was registered, especially when related to the glomeruli. The mean number of glomeruli per sample was 9.2 (median 13.5). Observations were partly recorded on Ektachrome P800 (daylight).
RESULTS The majority of cases had symptoms indicative of glomerulonephritis (Table 1). The final pathological diagnoses are given in Table 3. Glomerulonephritis was diagnosed in 74% of the cases, while 15% had hypertensive renal disease as the primary diagnosis. In 15% of the glomerulonephritis cases there were also signs of hypertensive vascular damage, at times with glomerular hyperperfusion injury. The ultrastructural morphology of the renal tissue was well preserved in LR White-embedded control sections in which the immune procedure had not been carried out. Membranes and cellular organelles appeared essentially the same as after epoxy embedding, and electrondense deposits along the basement membranes were shown equally well. There was some loss of contrast following the immune procedure. Cellular details were still easily recognizable and electron-dense deposits retained most of their density (Fig. 1). Non-specific background precipitation of gold particles was minimal. Only particles localized over electron-dense deposits, and in clearly increased numbers over the background, were accepted as positive. Table 3 compares the results of IM-EM with IF. Positivity is defined as a positive reaction for one or more of the antibodies tested. We found discrepancies
in 12 cases (24%). The discrepancies were equally distributed between positive IM-EM only and positive I F only. Two IM-EM-negative cases had been fixed in unbuffered formalin by accident. In two cases which were recorded as IFpositive, the positivity was found to be restricted to intravascular plasma by IM-EM (Fig. 2). Table 4 gives the results for individual antibodies. There was variable correlation between IM-EM and I F results, poor for IgG, considerably better for IgM, IgA, and C3c. Thus, for IgG, one of four IM-EM-positive samples were positive by IF (25%), while 39 of 46 IM-EM cases were also negative by IF (85%). For IgA, I0 and 13 IM-EM-positive cases were also positive by I F (77%), and 36 of 37 IM-EM-negative cases were also negative by I F (97%). For IgM and C ~ Cthe , results were similar to IgA. The presence of electron-dense deposits demonstrated by EM, located either along the peripheral basement membrane, or in paramesangial areas, was considered to be diagnostic of glomerulonephritis (Fig. 1). The increased density and typical granularity of IgA deposits, as seen in epoxy resin-embedded tissue, was preserved after LR White embedding. In one case of glomerulonephritis, both IF and IM-EM were negative in spite of the presence of characteristic electron-dense deposits. No case lacking electron-dense deposits was considered to show evidence of glomerulonephritis. In hypertensive vascular and glomerular disease, electron-dense material was also often shown to be present in the capillary walls by EM. This material was patchily distributed in the paramesangial areas, and/or band-like along the basement membrane, on the endothelial side. It was usually positive for C3c and/
TABLE 3. Results according to final pathology diagnosis for immunoglobulin and/or C3c detection Diagnosis IF pos. IF pos. IF neg. IF neg. Total IM-EM neg. IM-EM pos. IM-EM pos. IM-EM neg. 0 12 Diffuse mesangioproliferative g1.n. 2 2 16 Mesangioproliferativeg1.n. with focal 2 7 2 1 12
accentuation Membranous g1.n. Minimal change Focal glomerulosclerosis Benign/malignant nephrosclerosis Various non-inflammatory kidney diseases Total
1 1 0 2 0 6
1 0 1 4 3 28
0 0 0 1 1 6
0 0 0 2 5 10
2 1
I 9 9 50
g1.n. = glomerulonephritis.
1003
SKJ0RTEN rt at.
or IgM (Fig. 3). Capillary deposits in glomerular hyperperfusion injury (1) were also frequently positive for IgM and C ~ Crarely , for IgG. Two cases with congo-positive glomerular deposits were positive for AA protein (Fig. 4). A case with fibrillar, congo-negative deposits was negative for all antibodies tested, including kappa and lambda, and was considered not to represent amyloid. In comparison with our established procedures for the processing of renal biopsies, an attempt was made to estimate the extra cost that the introduction of IM-EM entailed. No new equipment and no extra reagents were necessary. Our previous standard procedure, including conventional EM and IF, required about 17 h preparative work per sample. IM-EM is more labour intensive, and requires at least an extra four h of preparative work per sample.
DISCUSSION IM-EM has been applied to the study of a number of tissues for the detection of many antigens, employing different immune techniques and different embedding media. IM-EM of cryosections permits the detection of most antigens, but the preservation of morphological details is variable (1 1). Pre-embedding IM-EM has been successfully applied to the study of renal tissues in selected cases ( 6 ) . The pre-embedding technique, however, requires diffusible antibodies and markers for adequate tissue penetration. Post-embedding IM-EM on glutaraldehydefixed, epoxy-embedded tissue is well suited for the study of antigens which are preserved using these procedures, for example peptide hormones ( 7 , 12). Like most researchers, we have been unable to detect immunoglobulins with this procedure. This may be explained partly by reduc-
tion of available epitopes on immunoglobulin molecules by aldehyde crosslinking, and partly because epoxy resin is impenetrable to water and proteins. Only immunoglobulin epitopes on the surface of the sections are therefore available for binding to antibodies. Recently, Herrera (5) successfully demonstrated immunoglobulins by IM-EM using a post-embedding technique on polyacrylic resin sections. Following his recommendations, we obtained reproducible results with this technique. There was good preservation of the ultrastructural morphology when paraformaldehyde-osmium fixation was maintained, similar to, but not quite as good as, the tissue preservation obtained after aldehyde-osmium fixation and epoxy embedding. It is essential for the preservation of immunoglobulin antigenicity that the temperature of the tissue does not exceed 56 "C during polymerization of the embedding medium. Polymerization therefore has to be carried out in a stable thermostat, without catalyst, since catalytic polymerization may render this procedure exothermic (9). Immune incubation had to be preceded by NaI04 etching in order to remove Os04 from the surface of the thin sections. Without etching, the immune reactions were weak or negative. If osmium fixation was omitted, contrast was poor and morphology was inadequate for ultrastructural diagnostic work. Antibodies have been found to penetrate acrylic polymers to a depth of about 200 nm in 30 min (10). Gold particles with a diameter of 15 nm do not penetrate the resin at all. Therefore, only antibodies at the surface are visualized by gold particles of this size. We are not aware of other papers offering detailed comparisons of IM-EM and I F examination of renal biopsies. Herrera (1989) studied 53 renal biopsies embedded in LR White by IMEM, and stated that there was good correlation
Fig. 1. Massive electron-dense paramesangial deposits, with gold-labelled IgA (negligible background staining). Good ultrastructural preservation. 24,000 x . B: basement membrane. D: deposit. N: nucleus. Fig. 2. Gold-labelled IgA in intravascular plasma; non-specific reaction. Suboptimal ultrastructural preservation. 21,000 x . B: basement membrane. E: epithelium. N: nucleus. Fig. 3. Gold-labelled C3c in glomerular hyperperfusion injury. 17,000 x . B: basement membrane. D: deposits. Fig. 4. Gold-labelled AA amyloid. Note precise localization of gold particles on individual amyloid fibrils. 75,000 x . AA: amyloid. N: nucleus.
1004
IMMIJNOELECTRON MICROSCOPY OF RENAL BIOPSIES
1005
SKJBRTEN
TABLE 4. Immunoelectron microscopy results for individual antibodies (Parenthesis: uositive bv IF) IEG IEM IEA C3c Total + 4 (1) 12 (10) 13 (10) 22 (17) 52 (38) 46 (7) 38 (6) 37 (1) 28 (4) 149 (18) Total 50 f8) 50 (16) 50 (11) 50 (21) 200 (561
with IF findings, but gave no details. In our study, two thirds of the glomerulonephritis-like conditions were positive by both methods. An equal number of cases (12%) were IF-positive and IM-EM-negative, or IF-negative and IMEM-positive. In two of the IF-positive and IMEM-negative cases, the I F positivity was restricted to plasma in capillary lumina, and was thus without biological significance. In these cases, IM-EM therefore demonstrated a higher diagnostic specificity than IF. Nine per cent were negative by both methods. The best correlations between IM-EM and I F were found for IgA and C3c. IgA was detected more often by IM-EM than by IF, The number of IgG-positive cases was low, with a large discrepancy between I F and IM-EM. This may be explained by nonspecific staining of intraluminal plasma IgG by IF, IgG being the immunoglobulin present in the highest concentration in plasma. In the application of IF to demonstrate tissuebound immunoglobulins, extensive washing of the tissue has been recommended before the immune procedure is performed, in order to reduce non-specific binding of antibodies to passively retained immunoglobulins (2). Such washing cannot be applied to tissue which is intended for IM-EM since the washing procedure will destroy the fine structure of the tissue. IF of frozen sections is a sensitive method for the detection of the antibodies in question. IMEM appears to have a somewhat lower sensitivity, probably related to the number of glomeruli in the sample: only two glomeruli are studied per sample by IM-EM, versus 9.2 for IF. Another factor is the number of epitopes available for binding, which is low for IM-EM due to inactivation by crosslinking, and the minute amount of tissue available for antibody visualization by gold particles; while antibodies penetrate the sections in both instances, gold particles do not penetrate plastic sections. Fluorochrome-conjugated antibodies, however, 1006
et
al.
penetrate frozen sections. On the other hand, the specificity of IM-EM is very high, due to the unique possibility for correlation of antibody binding with ultrastructural morphology. This makes it possible to exclude non-specific staining. In the present study, we chose to let the diagnosis of glomerulonephritis rest on the demonstration of electron-dense deposits with a morphology conforming to the accepted picture of immune complexes (4).The demonstration of immunoglobulins and/ or complement in these deposits, in our opinion, strengthened the diagnosis of glomerulonephritis. Hypertensive renal damage may also lead to deposition of immunoglobulins and/or complement in vessel walls (1). We find that these deposits differ morphologically from those of glomerulonephritis. IM-EM of amyloid deposits offers an opportunity for their subclassification into secondary, lambda, kappa, and other types (8). Labelled antibodies localized on the fibrils give a specific diagnosis. In conclusion, IM-EM is a very sensitive method for the detection of immunoglobulins in renal biopsies, it has the capacity to detect very small quantities of antigens. The combination of immune technique and ultrastructure adds an extra dimension to the diagnostic specificity, which is greater than that of IF alone. The increased specificity has diagnostic importance, for instance in the differentiation of tissue changes caused by immune complex disease or hypertension. In laboratories which routinely use EM in the study of renal biopsies, the moderate extra cost of IM-EM is justified by the increased specificity afforded by this method.
REFERENCES 1. Bohle, A., Biwer, E. & Christensen, J. A . : Hyperperfusion injury of the human kidney in different glomerular diseases. Am. J. Nephrol. 8: 179-186, 1988. 2 . Brandtmg, f!: Mucosal and glandular distribution of immunoglobulin components. Immunology 26: 1101-1 114, 1974. 3. Brandtmg, I(, Huitfeldt, H. S., Oppedal, B. R. & Rognum, I: 0.:Letter to the Editor. Re: Demonstration of lymphocyte surface antigens in paraffin-embedded human tissue. J. Immunol. Methods 87: 289-290. 1986.
IMMUNOELECTRON MICROSCOPY OF RENAL BIOPSIES
4. Churg, J. & Sobin, L. H.: Renal disease. Classification and atlas of glomerular diseases. IgakuShoin, Tokyo, New York 1982. 5. Herrera, G. A , : Ultrastructural postembedding immunogold labelling: applications to diagnostic pathology. Ultrastruct. Pathol. 13: 485499, 1989. 6. Hinglais, N., Kazatchkine, M . D., Bhakdi, S., Appay, M.-D., Mandet, C . , Grossetete, J. & Bariety, J.: Immunohistochemical study of the C5b-9 complex of complement in human kidneys. Kidney Int. 30: 399410, 1986. 7. Holm, R., Farrants, G. W , Nesland, J. M . , Sobrinho-Simoes. M., Jmgensen, 0. G. & Johannessen, J. Y ; Ultrastructural and electron immunohistochemical features of medullary thyroid carcinoma. Virchows Arch. Path. Anat. A 414: 375-384, 1989. 8. Linke B. R,Huhn, D., Casanova, S. & Donini, U.: Methods in laboratory investigation. Immunoelectron microscopic identification of human
9.
10. 1 1.
12.
AA-type amyloid: exploration of various monoclonal AA-antibodies, methods of fixation, embedding and of other parameters for the proteinA gold method. Lab. Invest. 61; 691-697, 1989. “LR White” product information sheet. Polaron Instruments Inc., 2293 Amber Drive, Line Lexington Industrial Park, Hatfield, PA 19440, U.S.A., 1989. Newman, G. R. & Hobot. J. A . : Modern acrylics for post-embedding immunostaining techniques. J. Histochem. Cytochem. 35; 971-981, 1987. Roos, N . & Morgan, A . J.; Cryopreparation of thin biological specimens for electron microscopy: methods and applications. Microscopy Handbooks 21, Oxford University Press. Royal Microscopical Society, 1990. Wang, B.-L. & Larsson, L.-I.;Simultaneous demonstration of multiple antigens by indirect immunofluorescence or immunogold staining. Histochemistry 85: 47-56, 1985.
1007
