Sage Journals: Discover world-class research

Abstract

Immunohistochemistry is a widely accepted tool to investigate the presence and immunolocalization of cytokines in tissue sections at the protein level. We have tested the specificity and reproducibility of IFNγ immunohistochemistry on tissue sections with a large panel of anti-IFNγ antibodies. Thirteen different commercially available anti-IFNγ antibodies, including seven advertised and/or regularly applied for immunohistochemistry/-cytochemistry, were tested using a three-step streptavidin–biotin–peroxidase technique and a two-step immunofluorescence (FACS) analysis. Immunoenzyme double staining was used to identify the IFNγ-positive cells. Serial cryostat sections were used of human reactive hyperplastic tonsils, rheumatoid synovium, and inflammatory abdominal aortic aneurysms, known to possess a prominent Th1-type immune response. In vitro phorbol myristate acetate/ionomycin-stimulated T-cells served as positive control; unstimulated cells served as negative control. Cultured T-cells were used adhered to glass slides (immunocytochemistry), in suspension (FACS), or snap-frozen and sectioned (immunohistochemistry). Immunocytochemistry and FACS analysis on stimulated cultured T-cells showed positive staining results with 12 of 13 anti-IFNγ antibodies. However, immunohistochemistry of sectioned stimulated T-cells was negative with all. Unstimulated cells were consistently negative. IFNγ immunohistochemical single- and double staining analysis of the tissue sections showed huge variations in staining patterns, including positivity for smooth muscle cells (n = 8), endothelial cells (n = 4), extracellular matrix (n = 4), and CD138+ plasma cells (n = 12). Specific staining of T-cells, as the sole positive staining, was not achieved with any of the 13 antibodies. IFNIFNγ-immunohistochemistry appears unreliable because of lack of specificity to stain T-cells in situ. In fact, depending on the type of anti-IFNγ antibody used, a variety of different cell constituents were nonspecifically stained. Consequently, data based on IFNγ-immunohistochemistry must be interpreted with great caution.

Keywords

IFNγ protein immunohistochemistry immunocytochemistry T-cells plasma cells

Interferon-gamma (IFNγ) is a 34–50-kD cytokine exclusively produced by T-cells and natural killer cells (Farrar and Schreiber 1993), which plays a key regulatory role in specific immunity. For this reason, IFNγ has been widely investigated in immunological research. In addition to in vitro studies, many investigators also attempted to demonstrate IFNγ protein immunohistochemically in healthy and diseased tissues. For example, studies have been undertaken to localize IFNγ in rheumatoid arthritis (Ulfgren et al. 1995; Smeets et al. 1998a,b), atherosclerosis (Hansson et al. 1989; Frostegard et al. 1999), and inflammatory bowel disease (Camoglio et al. 1998). Cytokine immunohistochemistry is frequently used in combination with molecular biological techniques such as in situ hybridization (ISH) and polymerase chain reaction (RT-PCR) to gain insight into the biological role of cytokines in diseased tissues. However, our own immunohistochemical experiences with several anti-IFNγ antibodies have raised doubts concerning the specificity and the consistency of the staining results obtained. This provides the background for the present study, in which we verified the specificity of anti-IFNγ immunohistochemical staining using a panel of commercially available poly- and monoclonal antibodies.

Materials and Methods

Human Tissue Samples

Tonsils with hyperplastic reactive changes (n = 3, age 3–6 years), inflammatory abdominal aortic aneurysms (IAAA) (n = 2, age 60 and 75 years), and synovial tissue specimens from knee joint with rheumatoid arthritis (RA synovium) (n = 3, age 62–77 years) were obtained at surgery. These tissue samples were used as positive controls for IFNγ staining, based on the expression of IFNγ mRNA as detected with RT-PCR (Ramshaw et al. 1994; Harabuchi et al. 1996; Kotake et al. 1997). Moreover, IFNγ protein was detected with immunohistochemistry in tonsil and RA synovium (Hoefakker et al. 1993; Andersson et al. 1994; Dolhain et al. 1996; Smeets et al. 1998a,b) and ELISA in IAAA (Szekanecz et al. 1994). Non-atherosclerotic vessel segments (n = 4, age 10–14 years) were used as controls. All tissue samples were snap-frozen in liquid nitrogen and stored at –80C. Six-μm cryostat serial sections were cut, and stored at –80C.

Cell Specimens

Cultured T-cell lines from human aortic atherosclerotic plaques were prepared as previously described (De Boer et al. 1999). Cells were cultured (5 × 10⁶ /ml) in Iscoves modification of Dulbecco's medium (IMDM; Gibco BRL Life Technologies, Paisley, Scotland) supplemented with 10% pooled human serum in the presence or absence of 10 ng/ml phorbol-12-myristate-13-acetate (PMA) (Sigma; St Louis, MO) and 100 ng ionomycin (Calbiochem; La Jolla, CA) in 25-cm² culture flasks. Monensin (Sigma) was added (2.88 μM/ml in dimethylsulfoxide) to prevent active cytokine excretion, thus resulting in intracellular accumulation of IFNγ. After 4 h, the cell suspensions were washed three times with PBS without protein additives and divided into three fractions. The first fraction was adhered overnight at 4C on Bio-Rad adhesion slides (Hercules, CA), 2 × 10⁴ cells per spot. After brief washing with PBS, these cell specimens were stored in PBS at 4C and used for immunocytochemistry within 2 days. The second fraction was concentrated in an Eppendorf tube in 100 μl PBS (2 × 10⁵ cells), mixed gently with 100 μl Tissue-Tek OCT compound (Sakura; Zoeterwoude, The Netherlands), and snap-frozen in liquid nitrogen. Six-μm cryostat sections were cut and handled as described for tissue cryostat sections. The remaining cells were used for FACS analysis.

Immunostaining Reagents

We employed 13 anti-human anti-IFNγ poly- and monoclonal antibodies, as listed in Table 1. As indicated in Table 1, seven anti-IFNγ antibodies were advertised and/or regularly applied for immunocytochemistry/-histochemistry. Other primary antibodies used in this study are listed in Table 2.

Biotinylated goat anti-mouse Ig (GAM/bio), biotinylated goat anti-rabbit Ig (GAR/bio), alkaline phosphatase-conjugated goat anti-rabbit Ig (GAR/AP), normal mouse serum (NMS), normal goat serum (NGS), normal swine serum (NSS), AP-conjugated streptavidin (streptavidin/AP), streptavidin–biotin complex with HRP (SABC/HRP), rabbit antifluorescein (rabbit anti-FITC), and endogenous biotin blocking kit were from DAKO (Glostrup, Denmark). Phycoerythrin (PE)-conjugated GAM, PE-conjugated GAR, biotinylated goat anti-mouse IgG2_a (GAM-IgG2_a/bio), and HRP-conjugated goat anti-mouse IgG₁ (GAM-IgG₁/HRP) were from Southern Biotechnology Associates (Birmingham, AL). Biotinylated and PE-conjugated swine anti-goat Ig (SAG) were from BioSource (Nivelles, Belgium). Rabbit anti-phycoerythrin (rabbit anti-PE) was from Biogenesis (Poole, UK). β-Galactosidase-conjugated streptavidin (streptavidin/GAL) was from Boehringer/Roche (Mannheim, Germany). PowerVision-AP-conjugated goat anti-rabbit Ig (PowerVision-GAR/AP) was from ImmunoVision Technologies (Daly City, CA). 3-Amino-9-ethylcarbazole (AEC), naphthol-AS-MX-phosphate, Fast Blue BB (cat. no. 3378), and saponin (cat. no. 7900) were from Sigma.

Table 1

Anti-human IFNγ antibodies used in this study

Clone	Type or species	Dilution	Final conc., source, and code a (μg/ml)		Indicated application b	IHC/ICC application, references
–	Rabbit	1:2000	0.5	Serotec AHP253	ELISA, IHC	Tested by Serotec
B-B1	Mouse IgG₁	1:100	10	Serotec MCA748	N, FACS, ELISA
25718.11	Mouse IgG_2a	1:100	5	RD Syst MAB285	N, WB, ELISA, IHC	Tested by RD Systems
–	Goat	1:200	0.5	RD Syst AF285	N, ELISA, WB
MMHG-1	Mouse IgG₁	1:250	4	Accurate YD21500	N, WB, IP, IHC, ELISA	Unreferenced
35B10G6	Mouse IgG₁	1:500	2	Biosource AHC4432	IHC, ELISA	Unreferenced
25723.11	Mouse IgG₁	1:3	2.5	Becton–Dickinson 340449	FACS
45–15	Mouse IgG₁	1:100	1	IQ Products 160P	FACS
–	Rabbit	1:10.000	?	Genzyme IP-500 c	WB, IHC/ICC	Haynes et al. 1993
H21	Mouse IgG_2a	1:200	5	Genzyme 1598–00 c	N, ELISA, IHC	Camoglio et al. 1998; Smeets et al. 1998a
4S.B3	Mouse IgG₁	1:20	5	Pharmingen 18901A	N, FACS
MD-2	Mouse IgG₁	1:100	5	Van der Meide et al. 1985	N, ELISA	Hoefakker et al. 1993; Dolhain et al. 1996; Thepen et al. 1996; Van Hoffen et al. 1996; Smeets et al. 1998b
F12	Mouse IgG₁	1:20	5	HyCult Biotech HM2003	N

^aSerotec (Poole, UK); RD Systems (Abington, UK); Accurate (Westbury, NY); Biosource (Nivelles, Belgium); Becton–Dickinson (San Jose, CA); IQ Products (Groningen, The Netherlands); Genzyme (see RD Systems); Pharmingen (see Becton–Dickinson); HyCult Biotech (Uden, The Netherlands).

^bELISA, enzyme-linked immunoenzyme solid assay; IHC, immunohistochemistry; ICC, immunocytochemistry; FACS, fluorescent activated cell sorter; WB, Western blotting; IP, immunoprecipitation; N, neutralization.

^cCurrently not available.

Table 2

Other primary antibodies used in this study

Antibody/clone	Type or species	Source	CD	Dilution	Incubation	Staining
MMHGR-1	Mouse IgG₁	Novus Molecular (San Diego, CA)	–	1:500	o.n., 4C	IFNγ receptor
Leu 4/SK1, FITC-conj	Mouse IgG₁	Becton–Dickinson	3	1:200	60 min, RT	T-cells
Leu 5b/S5.2, FITC-conj	Mouse IgG_2a	Becton–Dickinson	2	1:50	60 min, RT	T-cells, natural killer cell subset
Leu 19/MY31, PE-conj	Mouse IgG₁	Becton–Dickinson	56	1:20	60 min, RT	Natural killer cells
Syndecan-1/B-B4, FITC-conj	Mouse IgG₁	IQ Products	138	1:200	60 min, RT	B-cells, plasma cells
CD68, EBM11/biotin-conj	Mouse IgG₁	DAKO (van der Loos and Göbel 2000)	68	1:200	60 min, RT	Macrophages
von Willebrand factor	Rabbit	DAKO	–	1:10,000	60 min, RT	Endothelial cells
1A4	Mouse IgG_2a	DAKO	–	1:200	60 min, RT	α-actin, smooth muscle cells
HLA-DR, DP, DQ, CR3/43	Mouse IgG₁	DAKO	–	1:1000	60 min, RT	Major histocompatibility complex class II
Control IgG₁, DAK-GO1	Mouse IgG₁	DAKO	–			Isotype control
Control IgG_2a	Mouse IgG_2a	DAKO	–			Isotype control

Immunocytochemistry and Immunohistochemistry

Cell specimens and cryostat tissue sections were fixed in either acetone (10 min, 4C) or 4% paraformaldehyde (PFA) in PBS (5 min, room temperature). When PFA fixation was used, 0.1% saponin was added for membrane permeabilization (Sander et al. 1991; Dolhain et al. 1993) to all incubation steps. Endogenous peroxidase activity was blocked with 0.1% sodium azide + 0.3% peroxide in 50 mM Tris-HCl-buffered saline, pH 7.8 (TBS) (20 min, RT) (Li et al. 1987). Endogenous biotin was blocked with subsequent incubations of 0.1% avidin and 0.01% d-biotin from the DAKO endogenous biotin blocking kit (twice for 15 min at RT). Primary antibodies and antibody-enzyme conjugates were diluted in TBS + 1% bovine serum albumin (BSA). Single immunostaining consisted of overnight incubation at 4C with the selected primary anti-IFNγ antibodies (Table 1) and standard SABC/HRP detection. HRP activity was visualized with AEC (0.5 mg/ml) and peroxide (0.01%) in acetate buffer (pH 5.2, 50 mM).

Controls consisted of replacing the primary antibody with a non-immune mouse antibody of identical subclass (DAKO). Specific IgG concentration and Ig isotype/subclass were matched with the specific primary antibody from the original single-staining experiment.

Immunoenzyme Multiple Staining of Tissue Sections

A double-staining procedure based on a multistep technique (van der Loos 1999) was applied for comparison of IFNγ immunostaining with different cellular markers. The staining procedure consisted of the following subsequent incubation steps: NGS (1:10, 15 min), IFNγ antibody (clones MMHG-1 1:100, MD-2 1:50 or rabbit antibody Genzyme IP-500 1:5000, all overnight at 4C), GAM/bio (1:200, 30 min) or GAR/bio (1:400, 30 min), SABC/HRP (1:100, 30 min), and NMS (1:10, 15 min). For the goat anti-IFNγ antibody, the first part of the double staining consisted of NSS (1:10, 15 min), IFNγ goat antibody (1:100, overnight at 4C), SAG/biotin (1:100, 30 min), SABC/HRP (1:100, 30 min), and NGS (1:10, 15 min). Then, a panel of directly conjugated primary antibodies was applied: FITC-conjugated anti-CD3, CD2, CD138, biotinylated anti-CD68 (van der Loos and Göbel 2000), and PE-conjugated CD56 (60 min) (Table 2). For the FITC- and PE-conjugated antibodies, further steps were as follows: rabbit anti-FITC (1:1000, 15 min), or rabbit anti-PE (1:200, 15 min) and PowerVision-GAR/AP (undiluted, 30 min). The biotinylated antibody was followed by streptavidin/AP (1:100, 30 min). The enzymatic activities of AP and HRP were visualized in blue (Fast Blue BB) and red (AEC), respectively (van der Loos 1999).

Two triple immunoenzyme stainings using β-galactosidase (turquoise), AP (red), and HRP (brown) as marker enzymes, and blue nuclear counterstain (van der Loos 1999), were performed to monitor the major cell types present in the tissue specimens. The first consisted of anti-α-actin (mouse IgG2a) and anti-CD68 (mouse IgG1), combined with FITC-conjugated anti-CD3, in a multistep staining procedure. Anti-α-actin and anti-CD68 were incubated in a cocktail, followed by GAM-IgG2_a/biotin (1:50), GAM-IgG1/HRP (1:50) in a cocktail (30 min) and streptavidin/GAL (1:40, 30 min). After a blocking step with normal mouse serum (1:10, 15 min), incubation was performed with FITC-conjugated anti-CD3, rabbit anti-FITC (1:1000, 15 min), and GAR/AP (1:20, 30 min). Smooth muscle cells, macrophages, and T-cells were immunostained in turquoise, brown, and red, respectively. The second consisted of anti-α-actin (mouse IgG2_a), anti-CD68 (mouse IgG1), and anti-von Willebrand factor (rabbit) in one cocktail, followed by GAM-IgG2_a/bio (1:50), GAM-IgG1/HRP (1:50), GAR/AP (1:20) in cocktail (30 min), and streptavidin/GAL (1:40, 30 min). Smooth muscle cells, macrophages, and endothelial cells were immunostained in turquoise, brown, and red, respectively.

FACS Analysis

All 13 anti-IFNγ antibodies were applied for indirect intracellular FACS staining. In vitro PMA/ionomycin-stimulated T-cells were fixed with 1% PFA in PBS (10 min, 4C) and treated with 0.1% saponin for permeabilization (Sander et al. 1991). Final concentrations of the primary antibodies diluted in PBS with 0.1% saponin and 1% BSA (60 min, 4C) were similar as mentioned for immunohistochemistry/cytochemistry in Table 1. Depending on the primary antibody, PE-conjugated GAM, GAR, or SAG (all 1:200) was used as second-step reagent (30 min, 4C). Fluorescence of 10⁴ cells was analyzed on a FACS Calibur (Becton Dickinson, Richmond, CA).

Leakage of IFNγ from Tissue Sections

From the frozen stimulated and unstimulated cultured T-cells, 20 6-μm sections were cut and mounted on organosilanecoated slides. Either unfixed, acetone-fixed (10 min, 4C, air-dried) or PFA-fixed (5 min, RT, briefly washed three times with TBS) sections were encircled with a wax pen and covered with 100 μl TBS for 30 min. Fifty-μl samples were taken from the buffer covering the tissue sections and subjected to IFNγ ELISA. A Pelikine IFNγ kit was used, following the instructions as supplied by the manufacturer (CLB; Amsterdam, The Netherlands).

Results

IFNγ Staining Characteristics with Cultured T-cells

Stimulated intact T-cells on adhesion slides showed a distinct signal in the majority of the cells with 12 of 13 anti-IFNγ antibodies (Figure 3E). Unstimulated cells were negative or occasionally weakly positive (Figure 3B). Stimulated T-cells showed a much stronger staining intensity after PFA fixation (Figure 3E) than after acetone fixation (Figure 3D). These immunocytochemical staining results were completely confirmed by FACS analysis (Table 3).

An interesting observation was that IFNγ immunohistochemistry was consistently negative after cryosectioning of stimulated T-cells, (Table 3; Figure 3F), even after raising the primary antibody concentration to 50 μg/ml. This result was found after both acetone and PFA fixation.

To test the possible leakage of IFNγ from either unfixed, acetone-, or PFA-fixed cultured T-cell sections, buffer covering the specimens was subjected to ELISA. Buffers covering unfixed sections from stimulated T-cells contained IFNγ protein at 14.4–49.3 pg/ml (n = 4) and concentrations below detection limit (± 10 pg/ml) for the unstimulated cells. After acetone or PFA fixation, only values close to or below detection limit were found. Performance of this leakage test with the other tissue specimens failed because IFNγ values with unfixed sections were too low.

IFNγ Staining Characteristics with Tissue Sections

Comparison of the IFNγ single staining with triple immunostainings marking the major cellular components in tonsil, RA synovium, and IAAA indicated positive IFNγ staining of various tissue components with a preference for smooth muscle cells (n = 8), endothelial cells (n = 4), extracellular matrix (n = 4), macrophages (n = 1), and nerve bundles (n = 2). Considering only those IFNγ antibodies that were advertised and/or regularly applied for immunohistochemistry (n = 7; Table 1), there was no consistent staining pattern either. The positive cell types included smooth muscle cells (n = 4), endothelial cells (n = 3), extracellular matrix (n = 2), macrophages (n = 1), and nerve bundles (n = 1). Moreover, distinct expression of IFNγ with T-cells was not observed. Remarkably, a strong IFNγ smooth muscle cell positivity was also found in the media of non-atherosclerotic and non-inflamed aortic segments (Figure 1K). Apart from these various cellular constituents a distinct, consistently positive cell population was observed with 12 out of 13 antibodies. This characteristic cell population was found exclusively positive without additional staining of other cellular elements with three anti-IFNγ antibodies: clones MMHG-1, 35B10G6, and RD Systems goat antibody (Table 3). These positive cells were found in relatively low amounts in tonsil and IAAA but were abundant in the RA synovium (Figures 2B-2D). Negative control experiments did not show any staining (Figures 1I and 2A). Typical results of anti-IFNγ immunohistochemistry of all three tissues are summarized in Table 3 and illustrated in Figures 1A-1H and Figures 2B-2D.

Table 3

Anti-human IFNγ antibodies applied in FACS, ICC, and IHC a

		T-cells adh. slides b		T-cells sections b		T-cells FACS c		Tissue staining seen in RA synovium, inflam. aortic aneurysm, tonsil d , e
Clone/host	Source	Stim	Unstim.	Stim.	Unstim.	Stim.	Unstim	PC	T-cell	SMC	Peric	EC	MΦ	ECM	Nerve
Rabbit	Serotec AHP	–	–	–	–	–	–		No specific staining
B-B1	Serotec MCA	+	–	–	–	+	–	+	–	±	–	–	–	+	–
25718.11	RD Syst	+	–	–	–	+	–	+	–	±	–	–	–	–	–
Goat	RD Syst	+	–	–	–	+	–	+	–	–	–	–	–	–	–
MMHG-1	Accurate	+	–	–	–	+	–	+	–	–	–	–	–	–	–
35B10G6	Biosource	+	–	–	–	+	–	+	–	–	–	–	–	–	–
25723.11	Becton–Dickinson	±	–	–	–	±	–	±	–	–	–	–	–	+	–
45–15	IQ Products	+	–	–	–	+	–	+	–	±	++	+	–	–	±
Rabbit	Genzyme IP-500	+	–	–	–	+	–	±	–	+	+	±	++	±	±
H21	Genzyme 1598–00	+	–	–	–	+	–	+	–	+ +	+	+	–	+	–
4S.B3	Pharmingen	+	–	–	–	+	–	+	–	±	±	–	–	–	–
MD-2	Van der Meide et al. 1985	+	–	–	–	+	–	+	–	+	+	+	–	–	–
F12	HyCult Biotech	±	–	–	–	±	–	±	–	+	±	–	–	–	–
Control IgG₁	DAKO	–	–	–	–	–	–	–	–	–	–	–	–	–	–
Control IgG_2a	DAKO	–	–	–	–	–	–	–	–	–	–	–	–	–	–

^aTissue staining is expressed as the intensity observed in the individual cellular constituents: –, negative; ±, weakly positive; +, moderately positive; ++, strongly positive.

^b4% PFA-fixed cells.

^c1% PFA-fixed cells.

^deither 4% PFA- or acetone-fixed.

^ePC, subpopulation of CD138+ plasma cells; SMC, smooth muscle cells; Peric, pericytes; EC, endothelial cells; MΦ, macrophages; ECM, extracellular matrix; Nerve, nerve bundles.

Comparison of acetone and PFA as fixatives before IFNγ immunostaining showed similar results with respect to localization and staining intensity. Raising the antibody concentration (clones MMHG-1, MD-2, Genzyme rabbit antibody, and RD Systems goat antibody) up to 50 μg/ml, revealed over-stained images with intensely positive cells as described above.

To investigate the possible binding of IFNγ to its concomitant receptor, tissue localization of IFNγ immunostaining was compared with IFNγ receptor staining in serial sections. In RA synovium, IAAA, and tonsil, the anti-IFNγ receptor antibody showed distinct positive staining of endothelial cells, macrophages, and weak staining of smooth muscle cells (Figure 1J), which could only partly be co-localized with the anti-IFNγ staining (Figures 1B-1H). Moreover, medial smooth muscle cells in non-atherosclerotic aortas were IFNγ receptor (Figure 1L) -negative, which contrasted with the strong IFNγ positivity (Figure 1K).

Cellular Specificity of Anti-IFNγ Antibodies

To reveal the characteristic cell type showing positive staining with almost all anti-IFNγ antibodies, clones MMHG-1, MD-2, Genzyme rabbit antibody, and RD Systems goat antibody were subjected to immunoenzyme double-staining experiments with cellular markers on the RA synovium and tonsil cryostat sections. It appeared that T-cells (CD2, CD3), natural killer cells (CD56), and macrophages (CD68) did not co-localize with the IFNγ-positive cell population (Table 3). However, a subpopulation of approximately 5% of all CD138 positive plasma cells co-localized with IFNγ-positive cells, observed as a purple mixed-color in the cytoplasm (Figures 4A-4D).

Discussion

In the present study we have compared the immunohistochemical applicability of 13 different anti-IFNγ antibodies on cryostat tissue sections. IFNγ immunohistochemical single- and double-staining analysis of the tissue sections showed a large variation in staining patterns, but positive staining of T-lymphocytes was never observed. In this context, it should be emphasized that not all 13 IFNγ-antibodies used in this study were originally developed and tested for an immunohistochemical application. This also could explain the highly variable immunohistochemical staining patterns (Figures 1A-1H), despite the fact that the antibodies were capable of detecting IFNγ in cultured and in vitro-stimulated intact T-cells (Table 3). For example, anti-IFNγ clone MD-2 was first described for ELISA application by Van der Meide et al. (1985) and only much later was applied for immunohistochemical purposes (Hoefakker et al. 1993). The latter application is subsequently cited by other investigators and has led to the use of this antibody on tissue sections (Dolhain et al. 1996; Thepen et al. 1996; Van Hoffen et al. 1996; Smeets et al. 1998b).

We also verified the immunoreactivity of the anti- IFNγ antibodies on intact PMA/ionomycin-stimulated T-cells, known to contain substantial amounts of IFNγ. Intact in vitro-stimulated T-cells showed a typical intracellular IFNγ staining pattern (Figure 3E) similar to the findings by others (Sander et al. 1991; Dolhain et al. 1993; Andersson and Andersson 1994; Krouwels et al. 1997). In addition, 12 of 13 anti-IFNγ antibodies subjected to FACS analysis showed positive staining with in vitro-stimulated T-cells, whereas unstimulated cells were negative. These experiments reveal that 12 of 13 anti-IFNγ antibodies have the potential to recognize IFNγ in in vitro-stimulated intact T-cells. In contrast to these intact in vitro-stimulated T-cells, IFNγ immunohistochemistry performed on cryosections prepared from these cells gave negative results (Figure 3F). Given the fact that these in vitro PMA/ionomycin-stimulated T-cells contain higher levels of IFNγ protein than activated T-cells in vivo (Del Prete et al. 1991), it may be not surprising that IFNγ immunostaining of T-cells in tissue sections of IAAA, RA synovium, and tonsil was negative.

The present experiments show that there is a clear discrepancy between the results obtained with cultured intact T-cells and those with cryosectioned T-cells. Apparently, immunocytochemical staining of intact T-cells is not a reliable model system for testing the immunohistochemical applicability of IFNγ antibodies on cryostat tissue sections, as has been suggested in the literature (Andersson et al. 1994; Dolhain et al. 1996).

Figure 1

Acetone-fixed cryostat serial sections from an inflammatory abdominal aorta aneurysm segment, showing a detail of the adventitia. A muscular vessel is surrounded by inflammatory cells and fibrotic tissue. (A) Overview of major cell types. Immunoenzyme triple staining was performed with anti-α-actin marking smooth muscle cells in turquoise (β-galactosidase), CD68 marking macrophages in brown (HRP), and anti-CD3 marking T-cells in red (AP). Nuclear counterstain in blue with hematoxylin. Anti-IFNγ clone/source: (B) 45-15, (IQProducts); (C) 25718.11 (RD Systems); (D) MMHG-1 (Accurate; Westbury, NY); (E) MD-2; (F) 25723.11 (Becton Dickinson; San Jose, CA); (G) H21 (Genzyme); (H) B-B1 (Serotec); (I) A section incubated with mouse IgG1 control antibody. (J) Staining pattern of anti-IFNγ receptor antibody, clone MMHGR-1. Media of a non-atherosclerotic normal aorta specimen. (K) Immunostaining with anti-IFNγ, clone MD-2. Note the intense staining of medial smooth muscle cells. (L) Immunostaining with anti-IFNγ receptor is completely negative. Bar = 100 μm.

It is remarkable that in the majority of papers the observed IFNγ positivity simply was assumed to represent T-cells. Confirmation by double staining was performed only by Dolhain et al. (1996), who showed CD3/IFNγ co-localization in RA synovium samples. In this study we could not reproduce those results, even in the same type of tissue. Attempting to demonstrate the co-localization of IFNγ in T-cells, we performed double staining with both CD2 and CD3. Double staining with CD2 was also performed because CD2 is not downregulated after T-cell activation (Moingeon et al. 1991), in contrast to CD3 (Van Lier et al. 1987). Furthermore, we have been able to detect IFNγ/CD3 co-localization in in vitro-stimulated T-cells (not shown). We have no explanation for the different results in our study compared to that of Dolhain et al. (1996).

The obvious difference between in vitro-stimulated intact T-cells and cryostat tissue sections prepared from these cells is the presence of an outer membrane. Therefore, it appears attractive to hypothesize that intact T-cells retain IFNγ more effectively during the fixation procedure, either with acetone or PFA, than a sectioned cell. Possibly IFNγ protein is lost from the cryosectioned T-cells during the fixation procedure due to extraction of the antigen (Larsson 1993) or due to poor fixation. Arguments in favor of this leakage-hypothesis are the observations that (a) IFNγ is detectable by ELISA in the buffer covering the unfixed in vitro-stimulated T-cell cryostat sections. However, after fixation of the T-cell cryostat sections with either acetone or PFA, almost no IFNγ is detectable in the overlying buffer, whereas the immunohistochemical visualization of IFNγ in cryosectioned stimulated T-cells was completely negative also (Figure 3F). (b) Acetone fixation of intact T-cells, which completely dissolves the fatty membrane structures, shows a near-negative IFNγ staining result (Figure 3D), whereas the same cells were intensely positive after PFA fixation (Figure 3E).

IFNγ immunohistochemistry on cryostat tissue sections from tonsil, RA synovium, and IAAA revealed highly variable staining patterns with different antibodies. A variety of cellular constituents were positive, with a high preference for smooth muscle cells. Apart from staining of various cellular constituents, 12 of 13 anti-IFNγ antibodies revealed staining of a CD138-positive plasma cell subset (Figure 4) but not of T-cells. Non-T-cell positivity with anti-IFNγ antibodies has been reported previously and is considered a result of receptor-bound IFNγ (Dolhain et al. 1996). Indeed, some of the IFNγ staining patterns resemble the IFNγ receptor staining pattern (Figure 1J). However, there are indirect arguments that certainly not all non-T-cell IFNγ positivity can be explained as receptor bound. (a) In the non-atherosclerotic aortas, without any morphological signs of inflammation, the medial smooth muscle cells showed massive IFNγ positivity, whereas the IFNγ receptor was negative (Figures 1L and 1K). (b) When non-T-cell IFNγ staining in non-atherosclerotic aortas was caused indeed by receptor-bound IFNγ, this would have induced major histocompatibility complex Class II expression (HLA-DR, DP, DQ). However, this was not observed. This absence of HLA-DR, DP, and DQ positivity in non-atherosclerotic aortas also excludes the possibility that IFNγ receptor staining is perhaps missed because IFNγ blocks the antibody binding site.

Considering the IFNγ positivity of plasma cells, some investigators claimed expression of trace amounts of IFNγ by B-lymphocytes under in vitro conditions (Pang et al. 1992; Jelinek and Braaten 1995). In our opinion, these findings cannot be matched with the abundant and strong IFNγ staining of the plasma cell subpopulation as observed in all three tissues studied. Therefore, apart from the non-T-cell staining of various cellular constituents, also the plasma cell IFNγ-positivity was considered as a staining artifact. This artifactual staining might be caused by the formation of conditional epitopes (Willingham 1999) or redistribution of small molecular weight antigens (Larsson 1993), both of which may occur during the fixation procedure.

The nonspecific positive staining of the plasma cell subpopulation was shown not to be unique for anti- IFNγ antibodies. Using, for example, anti-IL-2 or anti-IL-4 antibodies, known to be secreted by T-cells upon activation, we also observed a strong positivity in RA synovium of the same plasma cell population that was found positive with anti-IFNγ antibodies, whereas positive T-cells were not observed. Non-T-cell IFNγ staining may be the basis for conflicting immunohis tochemical staining results when different anti-IFNγ antibodies were applied to similar tissue specimens. For example, Andersson et al. (1994) reported in chronic recurrent tonsillitis a few IFNγ positive cells using clone DIK-1, whereas Hoefakker et al. (1993) reported massive IFNγ positivity with clone MD-2 in the same tissue. Moreover, the application of extremely high concentrations of IFNγ antibody (MD-2, 45–100 μg/ml) (Thepen et al. 1996; Van Hoffen et al. 1996) for immunostaining of cryostat sections may introduce staining artifacts. Not only may the IFNγ antibodies themselves may play a role but also the application of different tissue fixatives before the IFNγ staining procedure may have an effect on the final staining result. For example, acetone fixation was employed by Hoefakker et al. (1993), Dolhain et al. (1996), Smeets et al. (1998b), and Thepen et al. (1996), whereas PFA fixation and saponin treatment were used by Ulfgren et al. (1995), Smeets et al. (1998a), and Andersson et al. (1994). In our hands, a comparison of these two popular fixatives showed a dramatic loss of IFNγ staining after acetone fixation of intact T-cells (Figures 3D and 3E), confirming the use of PFA fixation as indicated by Andersson et al. (1994). However, in considering the non-T-cell IFNγ staining of tissue sections after either acetone or PFA fixation, there was no difference regarding localization and intensity. This observation is another indication that the non-T-cell staining in tissue sections after acetone fixation has a nonspecific basis. The present description of nonspecific IFNγ immunohistochemical staining shows analogy with the false-positive staining of anti-RAP-5 antibody detecting the ras oncogene product p21 as reported by Samowitz et al. (1988) and Gutheil et al. (1989). Although these investigators employed immunoprecipitation and Western blotting as contra-evidence, they finally proved that the initial RAP-5 immunohistochemical positivity could be recognized as completely nonspecific. Furthermore, Kaino et al. (2000) described the crossreactivity of anti-rat CD45RA antibody with glucagon-producing islet α-cells in formalin-fixed and paraffin-embedded rat pancreatic tissue.

Figure 2

Acetone-fixed cryostat serial sections from a synovium of a rheumatoid arthritis patient. (A) Section incubated with mouse IgG₁ control antibody. Anti-IFNγ clone/source. (B) B-B1, (Serotec). (C) MMHG-1 (Accurate). (D) 35B10G6 (Biosource); nuclear counterstain in blue with hematoxylin. Bar = 100 μm.

Figure 3

PMA/ionomycin-stimulated (D–F) and unstimulated (A–C) cultured T-cells immunostained with anti-IFNγ, clone MMGH-1 (Accurate). (A,D) IFNγ immunocytochemistry on acetone-fixed intact T-cells. (B,E) IFNγ immunocytochemistry on PFA-fixed and saponin-treated intact T-cells. (C,F) IFNγ immunohistochemistry on PFA-fixed and saponin-treated cryostat sections from T-cells. Note the weak IFNγ immunostaining of stimulated cultured T-cells after acetone-fixation in D compared with PFA/saponin immunocytochemistry in E, and completely negative immunostaining after cryostat sectioning of stimulated cultured T-cells in F. Bar = 20 μm.

Figure 4

Detail of a synovium with rheumatoid arthritis in four adjacent acetone-fixed cryostat sections, showing a focal area with inflammatory cells. (A) IFNγ immuno-single staining using antibody MMHG-1 (Accurate), nuclear counterstaining in blue with hematoxylin. Immunohistochemical double stainings marking IFNγ, clone MMHG-1 in red (HRP) and either CD3 marking T-cells (B), CD68 marking macrophages (C), or anti-syndecan-1, CD138 marking plasma cells (D) in blue (AP). (D) Red and blue reaction products merged to purple indicate co-localization, whereas in B and C no co-localization is observed. (A) Nuclear counterstaining with hematoxylin. (B–D) Weak nuclear counterstaining with methyl green. Bar = 50 μm.

Figure 5

PMA/ionomycin in vitro-stimulated (A) and unstimulated (B) PFA prefixed, frozen, and cryosectioned cultured T-cells, immunostained with anti-IFNγ, clone MMGH-1 (Accurate). Cells are counterstained in blue with hematoxylin. Note the characteristic IFNγ staining as a perinuclear Golgi spot in the stimulated T-cells (A), whereas unstimulated T-cells in B are completely negative. Bar = 20 μm.

We consider further testing of the anti-IFNγ antibody specificity using absorption with IFNγ antigen not to be useful. In our opinion and that of others (Burry 2000), an absorption control does not prove the specificity of an antibody for its antigen in a fixed tissue section. For example, anti-IFNγ clone MD-2 was subjected successfully to such an absorption experiment (Hoefakker et al. 1993), but our present data clearly show this antibody as not to be specific for tissue IFNγ staining (Table 3; Figure 1L).

Future Perspectives for IFNγ Immunostaining?

At the request of one of the reviewers and after discussions at the latest Congress of Histochemistry and Cytochemistry, we tested the option of prefixation by perfusion or immersion in PFA before freezing. This method has been applied, for example, by Van Noorden and Polak (1985), Van Noorden (1986), and Larsson (1988), and is especially recommended for detection of small peptides and hormones. However, among the vast majority of reports applying acetone or PFA post-fixation of cryosections, we could find only one report using prefixation for the detection of IFNγ (Kakazu et al. 1999). Our pilot experiment showed that PFA prefixation of in vitro-stimulated T-cells indeed retains and preserves IFNγ after freezing and sectioning (Figures 5A and 5B), in contrast to PFA postfixation (Figure 3F). Apparently, PFA postfixation of a cryosection is too slow (Fox et al. 1985) and cannot prevent leaking of IFNγ from cryosections. However, PFA prefixation of a closed cell allows a firm fixation of IFNγ, thus preventing leakage even from tissue sections. Thus far, however, we have not been able to demonstrate IFNγ immunohistochemically in a few PFA-prefixed and 20% sucrose freeze-protected tissue blocks tested. Nevertheless, it could be that PFA prefixation is the key to successful immunohistochemical visualization of IFNγ and perhaps of other cytokines. Moreover, we suspect that missing the IFNγ staining of activated T-cells in cryosections from PFA prefixed tissue blocks, in contrast to prefixed in vitro-stimulated T-cells, may be a sensitivity problem of current detection systems even when tyramide amplification is used.

Conclusion

Using the present panel of anti-IFNγ antibodies, specific staining of activated T-cells (Th-1 type) in tissue sections is not observed with generally applied staining procedures, a phenomenon most likely due to either the lack of sensitivity and/or to leakage of IFNγ from the sections. On the basis of the present findings, we conclude that IFNγ immunostaining of cryostat tissue sections may result in significant staining artifacts. As a consequence, IFNγ immunostaining results on cryostat tissue sections should be interpreted with great caution. The performance of double staining is highly recommended, at least for the distinction between “true” IFNγ positivity and nonspecific staining.

Footnotes

Acknowledgements

We wish to express our gratitude to Prof Dr P.P. Tak and Mr T. Smeets (Academic Medical Center, Dept. of Rheumatology) for providing us with synovial tissue from patients with rheumatoid arthritis.

References

Andersson

Abrams

Bjork

Funa

Litton

Agren

Andersson

(1994) Concomitant in vivo production of 19 different cytokines in human tonsils. Immunology 83:16–24

Andersson

(1994) Immunolabeling of cytokine-producing cells in tissues and suspension. In Fradelizi

, ed. Cytokine Producing Cells. Paris, INSERM, 32

Burry

(2000) Specific controls for immuncytochemical methods. J Histochem Cytochem 48:163–165

Camoglio

Te Velde

Tigges

Das

Van Deventer

SJH

(1998) Altered expression of interferon-gamma and interleukin-4 in inflammatory bowel disease. Inflam Bowel Dis 4:285–290

De Boer

Van der Wal

Verhagen

Becker

(1999) Cytokine secretion profiles of cloned T-cells from human aortic atherosclerotic plaques. J Pathol 188:174–179

Del Prete

De Carli

Mastromauro

Biagiotti

Macchia

Falagiani

Ricci

Romagnani

(1991) Purified protein derivative of Myocbacterium tuberculosis and excretory-secretory antigen(s) of Toxocara canis expand in vitro human T cells with stable and opposite (type 1 T-helper or type 2 T-helper) profile of cytokine production. J Clin Invest 88:346–350

Dolhain

RJEM

Andersson

Ter Haar

Brinkman

BMN

Verweij

Daha

Breedveld

Miltenburg

AMM

(1993) Detection of intracellular interferon-gamma by light microscopy using an immunoperoxidase technique: correlation with the corresponding mRNA and protein product. J Leukocyte Biol 54:545–551

Dolhain

RJEM

Ter Haar

Hoefakker

Tak

De Ley

Claassen

Breedveld

Miltenburg

AMM

(1996) Increased expression of IFN-gamma together with IFN-gamma receptor in the rheumatoid synovial membrane compared with synovium of patients with osteoarthritis. Br J Rheumatol 35:24–32

Farrar

Schreiber

(1993) The molecular cell biology of interferon-gamma and its receptor. Annu Rev Immunol 11:571–611

10.

Fox

Johnson

Whiting

Roller

(1985) Formaldehyde fixation. J Histochem Cytochem 33:845–853

11.

Frostegard

Ulfgren

Nyberg

Hedin

Swedenborg

Andersson

Hansson

(1999) Cytokine expression in advanced human atherosclerotic plaques: dominance of pro-inflammatory (Th1) and macrophage-stimulating cytokines. Atherosclerosis 145:33–43

12.

Gutheil

Mane

Kapil

Needleman

(1989) Immunoprecipitation of cell lysates with RAP-5 does not specifically detect ras oncogene product p21. Hum Pathol 20:1176–1180

13.

Hansson

Holm

Jonasson

(1989) Detection of activated T-lymphocytes in the human atherosclerotic plaque. Am J Pathol 135:169–175

14.

Harabuchi

Wakashima

Murakata

Yoshioka

Yokoyama

Kataura

(1996) Cytokine expression and production by tonsillar lymphocytes. Acta Otolaryngol Suppl 523:75–77

15.

Haynes

Jackson

Tuan

Shepley

Smith

(1993) Cytokine production in first trimester chorionic villi: detection of mRNAs and protein products in situ. Cell Immunol 151:300–308

16.

Hoefakker

Van 't Erve

Deen

Van den Eertwegh

Boersma

Notten

Claassen

(1993) Immunohistochemical detection of co-localizing cytokine and antibody producing cells in the extrafollicular area of human palatine tonsils. Clin Exp Immunol 93:223–228

17.

Jelinek

Braaten

(1995) Role of IL-12 in human B lymphocyte proliferation and differentiation. J Immunol 154:1606–1613

18.

Kaino

Ito

Hirai

Kida

(2000) Anti-rat CD45RA monoclonal antibodies cross-react with glucagon. Acta Histochem 102:151–157

19.

Kakazu

Hara

Matsumoto

Nakamura

Oshitani

Arakawa

Kitano

Nakatani

Kinjo

Kuroki

(1999) Type 1 T-helper cell predominance in granulomas of Crohn's disease. Am J Gastroenterol 94:2149–2155

20.

Kotake

Schumacher

Jr Yarboro

Arayssi

Pando

Kanik

Gourley

Klippel

Wilder

(1997) In vivo gene expression of type 1 and type 2 cytokines in synovial tissues from patients in early stages of rheumatoid, reactive, and undifferentiated arthritis. Proc Assoc Am Phys 109:286–301

21.

Krouwels

Nocker

RET

Snoek

Lutter

van der Zee

Weller

Jansen

Out

(1997) Immunocytochemical and flow cytofluorimetric detection of intracellular IL-4, IL-5 and IFNgamma: application in blood- and airway-derived cells. J Immunol Methods 203:89–101

22.

Larsson

L-I

(1988) Immunocytochemistry: Theory and Practice. Boca Raton FL, CRC Press

23.

Larsson

L-I

(1993) Tissue preparation methods for light microscopic immunohistochemistry. Appl Immunohistochem 1:2–16

24.

C-Y

Ziesmer

Lazcana—Villareal

(1987) Use of azide and hydrogen peroxide as inhibitor for endogenous peroxidase in the immunoperoxidase method. J Histochem Cytochem 35:1457–1460

25.

Moingeon

Lucich

Stebbins

Recny

Wallner

Koyaso

Reinherz

(1991) Complementary roles for CD2 and LFA-1 adhesion pathway during T cell activation. Eur J Immunol 21:605–610

26.

Pang

Norihisa

Benjamin

Kantor

RRS

Young

(1992) Interferon-gamma gene expression in human B-cell lines: induction by interleukin-2, protein kinase C activators, and possible effect of hypomethylation on gene regulation. Blood 80:724–732

27.

Ramshaw

Roskell

Parums

(1994) Cytokine gene expression in aortic adventitial inflammation associated with advanced atherosclerosis (chronic periaortitis). J Clin Pathol 47:721–727

28.

Samowitz

Paull

Hamilton

(1988) Reported binding of monoclonal antibody RAP-5 to formalin-fixed tissue sections is not indicative of ras p21 expression. Hum Pathol 19:127–132

29.

Sander

Andersson

(1991) Assessment of cytokines by immunofluorescence and the paraformaldehyde-saponin procedure. Immunol Rev 119:65–93

30.

Smeets

Dolhain

RJEM

Breedveld

Tak

(1998a) Analysis of the cellular infiltrates and expression of cytokines in synovial tissue from patients with rheumatoid arthritis and reactive arthritis. J Pathol 186:75–81

31.

Smeets

Dolhain

RJEM

Miltenburg

De Kuiper

Breedveld

Tak

(1998b) Poor expression of T-cell-derived cytokines and activation and proliferation markers in early rheumatoid synovial tissue. Clin Immunol Immunopathol 88:84–90

32.

Szekanecz

Shah

Pearce

Koch

(1994) Human atherosclerotic abdominal aortic aneurysms produce interleukin (IL)-6 and interferon-gamma but not IL-2 and IL-4: the possible role for IL-6 and interferon-gamma in vascular inflammation. Agents Actions 42:159–162

33.

Thepen

Langeveld—Wildschut

Bihari

Wichen

Van Reijsen

Mudde

Bruijnzeel—Koomen

CAFM

(1996) Biphasic response against aeroallergen in atopic dermatitis showing a switch from an initial Th2 response to a Th1 response in situ: an immunocytochemical study. J Allergy Clin Immunol 97:828–837

34.

Ulfgren

Lindblad

Klareskog

Andersson

(1995) Detection of cytokine producing cells in the synovial membrane from patients with rheumatoid arthritis. Ann Rheum Dis 54:654–661

35.

van der Loos

(1999) Immunoenzyme Multiple Staining Methods. Oxford, BIOS Scientific Publishers

36.

van der Loos

Göbel

(2000) The animal research kit (ARK) can be used in a multistep double staining method for human tissue specimens. J Histochem Cytochem 48:1431–1437

37.

Van der Meide

Dubbeld

Schellekens

(1985) Monoclonal antibodies to human interferon and their use in a sensitive solid-phase ELISA. J Immunol Methods 79:293–305

38.

Van Hoffen

Van Wichen

Stuij

De Jonge

Klopping

Lahpor

Van den Tweel

Gmelig—Meyling

De Weger

(1996) In situ expression of cytokines in human heart allografts. Am J Pathol 149:1991–2001

39.

Van Lier

RAW

Boot

Verhoeven

De Groot

Brouwer

Aarden

(1987) Functional studies with anti-CD3 heavy chain isotype switch-variant monoclonal antibodies. Accessory cell-independent induction of interleukin 2 responsiveness in T cells by epsilon-anti-CD3. J Immunol 139:2873–2879

40.

Van Noorden

(1986) Tissue preparation and immunostaining techniques for light microscopy. In Polak

Van Noorden

, eds. Immunocytochemistry. Modern Methods and Applications. 2nd ed. Bristol, Wright, 26–53

41.

Van Noorden

Polak

(1985) Immunocytochemistry of regulatory peptides. In Bullock

Petrusz

, eds. Techniques in Immunocytochemistry. 3. London, Academic Press, 115–154

42.

Willingham

(1999) Conditional epitopes: is your antibody always specific? J Histochem Cytochem 47:1233–1235

Immunohistochemical Detection of Interferon-γ: Fake or Fact?

Abstract

Keywords

Materials and Methods

Human Tissue Samples

Cell Specimens

Immunostaining Reagents

Immunocytochemistry and Immunohistochemistry

Immunoenzyme Multiple Staining of Tissue Sections

FACS Analysis

Leakage of IFNγ from Tissue Sections

Results

IFNγ Staining Characteristics with Cultured T-cells

IFNγ Staining Characteristics with Tissue Sections

Cellular Specificity of Anti-IFNγ Antibodies

Discussion

Future Perspectives for IFNγ Immunostaining?

Conclusion

Footnotes

Acknowledgements

References