Pitfalls and caveats in applying immunostaining to histopatho- logical diagnosis

Immunostaining is an essential histochemical technique for analyzing pathogenesis and making a histopathological diagnosis. The needs are prompted by technical development and refinement, commercial availability of a variety of antibodies, deepened knowledge of immunohistochemical markers, accelerated analysis of morphofunctional correlations, progress in molecular target therapy, and the expectation of advanced histopathological diagnosis. However, immunostaining does have various pitfalls and caveats. We should learn from mistakes and failures, as well as from false positivity and false negativity. The present review article describes various devices, technical hints and trouble-shooting guides to keep in mind in performing immunostaining.


1.
Introduction Techniques of immunohistochemistry or immunostaining have already been established. [1][2][3][4][5][6][7][8] We now need measures against technical artifacts and appropriate trouble-shooting tips. The present review article overviews technical aspects, knowhows, pitfalls and trouble-shooting guides in routinely performing immunostaining. Formaldehyde provokes a cross-linkage between tissue protein components, so that the epitope retrieval step is often needed. In contrast, dehydration-induced protein aggregation is the main mechanism of fixation in such organic solvents as ethanol and acetone. In general, the antigenicity and DNA/RNA structure are well preserved in organic solvent-fixed paraffin sections. Sato, et al. invented cold acetone-based tissue processing nicknamed as the AMeX method, that can preserve not only antigenicity but also high molecular-weight DNA sequence in paraffin blocks. 13) A disadvantage of this method includes that significant tissue shrinkage results in difficulty of cutting and mild alteration of H&E histology.
It is of note that antigenic substances packaged in small granules in the cytoplasm are lost after fixation in ethanol and acetone. This is particular true if the granules are ultrastructurally electron-dense. Presumably, the osmiophilic (lipid-rich) granule content is easily extracted by the organic solvent. Figure 6 represents false negativity of platelet factor 4 and insulin caused by fixation in acetone and ethanol. In contrast, somatostatin can be visualized in ethanol-fixed paraffin sections. It is well known that somatostatin granules show low electron density in ultrastructural appearance.

3)
Use of archival pathology material Some antigens and RNA signals are stably detectable in archival pathology specimens fixed in acidic formalin for a considerably long period of time. In other words, fixation in acidic formalin can keep the antigenic and genomic signals to the evaluable level of histochemical detection.
As shown in Figure 7, we detected HBs antigen of hepatitis B virus (HBV) and the genome of hepatitis C virus (HCV) in liver cirrhosis specimens soaked in the fixative for up to 110 years, by applying immunostaining using a monoclonal antibody against HBs antigen and nested RT-PCR against the core region of HCV, respectively. 14) Once the author visited the Gordon Museum of Pathology at Guy's Hospital in London, UK, he had a chance to get unstained glass slides of autopsy specimens of Hodgkin's lymphoma sampled by Dr. Thomas Hodgkin himself 170 years before. The specimens were fixed in ethanol for 80 years and then in formalin for 90 years. The unstained sections were then kept at room temperature for months. By applying the cell transfer technique described below, immunostaining for CD15 and in situ hybridization for Epstein-Barr virus-encoded small nuclear RNA (EBER) were performed. CD15 was immunoreactive on the plasma membrane of Reed Sternberg (RS) cells and Hodgkin's cells, and EBER signals were detected in the nuclei of RS cells and Hodgkin's cells (Figure 8). 15) Empirically, the antigenicity of amyloid A protein in secondary amyloidosis, Bacillus Calmette Guérin (BCG) antigens in tuberculous and leprosy mycobacteria, as well as bacterial antigens such as streptococcal and pneumococcal antigens are stable and reproducibly immunolocalized in specimens fixed in formalin for more than 70 years (Figure 9). HCV antigens are also immunohistochemically detectable with monoclonal antibodies in archival FFPE specimens (Figure 10).

6.
Artifacts by formalin fixation 1) Penetration of plasma proteins into the cytoplasm of certain cells: A diffusion artifact During formalin fixation, plasma proteins rich in the plasma and tissue fluid are penetrated into the cytoplasm of certain cells in FFPE sections. The nuclei remain completely negative. This represents a nonspecific diffusion artifact. 1) The phenomenon is often seen in giant cells. The plasma proteins such as albumin, fibrinogen, IgG, Ig A, IgM, and kappa/lambda chains of immunoglobulins permeate through a damaged plasma membrane into the cytoplasm of unfixed cells, and the cells are then fixed by formaldehyde. The immunoreactivity is specific, but the positivity represents a fixation artifact. Typical examples include RS cells and Hodgkin's cells in Hodgkin's lymphoma (Figure 11). 16) The cytoplasm of RS cells and Hodgkin's cells is polyclonally positive for both kappa and lambda chains. Such diffusion artifacts are observed in a variety of cells, including hepatocytes and hepatocellular carcinoma cells, normal and neoplastic thyroid epithelial cells, epidermal keratinocytes and non-Hodgkin's lymphoma cells (Figure 12). Since cells immunoreactive for IgG and albumin are surrounded by completely negative cells, it seems the positive signals mean something, but nothing is significant at all. Figure 13 illustrates albumin immunostaining in autopsied normal pancreas. The cytoplasm of ductular cells, acinar cells or islet cells/peripheral nerve appear positive in different part of the tissue. Accordingly, when we try to immunostain IgG or alpha-1 antitrypsin in FFPE sections, albumin should be chosen for an indifferent control.

2)
False positive and false negative results due to uneven formalin fixation Uneven formalin fixation may lead to both false positivity and false negativity. 1) Not only cells improperly fixed by formalin but also cells overfixed by formalin show false negativity. Since vimentin in lymphocytes, macrophages, fibroblasts and vascular endothelial cells often turns to be negative in poorly fixed tissues, vimentin serves as an internal control for judging whether the fixation is proper or not. 17) Lymphocyte surface markers such as CD20 may be positive at the peripheral part of B-cell lymphoma tissue, while lymphoma cells in the poorly fixed (vimentin-negative) central part are false-negative (Figure 14). Figure  15 illustrates glutathione peroxidase immunoreactivity seen at the outermost part of the paraformaldehyde-fixed frozen rat liver tissue. Occasionally, a reversed pattern of the antigen localization is also experienced. The peripheral area is overfixed to show false negativity for leukocyte common antigen (LCA, CD45), while the antigenicity is preserved in the central part. Vimentin is negative in CD45-positive central zone. In this case, vimentin does not function as an internal control. Figure 16 illustrates such paradoxical immunostained findings.

7.
Tips and pitfalls in immunostaining 1) To get low background staining The signal/noise ratio should be high. We must aim for immunostaining with distinct positive signals with minimal background staining. The recent development of epitope retrieval techniques and the availability of a wide variety of monoclonal antibodies have significantly contributed to the art of immunostaining: the staining must be artistic, the author believes.
When you encounter immunostaining with high background noise, the following three possibilities should be checked. a) The concentration of the primary or secondary antibodies is too high. b) The antibody titer is too low. c) Rinsing in PBS is insufficient. As a prerequisite for artistic immunostaining with a high signal/noise ratio, the appropriate antibody dilution should be predetermined (Figure 17). Generally speaking, monoclonal antibodies provide low background staining, when compared with polyclonal antibodies. However, monoclonal antibodies of IgM type may be aggregated during repeated freezing and thawing to result in high background staining.
In order to lower the background noise, the following tips should be tried. a) The primary antibody should further be diluted 10 times to incubate overnight, instead of short incubation time for 30-60 minutes. b) The PBS rinse should be prolonged to overnight. c) High concentration (1 M) sodium chloride should be added to PBS for rinsing. d) Non-ionic detergents such as Tween 20 or Triton X-100 (0.05-0.1%) should be added to PBS for rinsing. e) Skim milk (5-10%) or bovine serum albumin (1-5%) should be added to the diluted primary antibody.

2) The removal of endogenous confounding reactivities a)
Endogenous peroxidase activity Peroxidase activity in FFPE sections can be quenched by the following methods. Eosinophils and neutrophils have a strong peroxidase activity, and hemoglobin in red cells reveal pseudoperoxidase activity. Peroxidase activity in macrophages, platelets and epithelial cells (salivary gland, mammary gland, thyroid, and renal tubules) is completely inactivated during FFPE preparations. The most common method to inactivate endogenous peroxidase activity is dipping in 0.3% hydrogen peroxide-methanol solution or 3% hydrogen peroxide aqueous solution for 15-30 minutes. Oxidation in 0.5% periodic acid for 10 minutes is also effective, but never suitable for carbohydrate antigens. The hydrogen peroxide-mediated method is also harmful to the carbohydrate antigens when the treatment is too long up to 2 hours. Addition of 10 mM sodium azide to the DAB solution also assists at inhibiting endogenous peroxidase activity.
When a lymphocyte surface marker study is performed with ethanol-fixed fresh-frozen sections, the endogenous peroxidase quenching should be performed after the incubation of primary antibodies. The epitopes of lymphocytes surface markers frequently belong to sugar moieties, and the sugar molecules in fresh-frozen sections are easily destroyed by the oxidation step, as illustrated in Figure 18.

b)
Endogenous biotin activity Biotin (vitamin H), a coenzyme of the decarboxylation reaction, is mainly distributed in mitochondria. Cells rich in mitochondria, such as proximal renal tubules, hepatocytes and striated muscle cells, have a strong endogenous biotin activity, particularly in fresh frozen sections or cytology preparations. When one utilizes biotin-based detection systems such as ABC or LSAB method, quenching of endogenous biotin activity should be performed by preincubating with avidin solution. 18) Binding between biotin and avidin is quite strong, and once bound, they never dissociate any longer.
The endogenous biotin activity is inactivated by formalin fixation. Exceptions include "opaque nuclei" seen in endometrial gland cells, endometrioid carcinoma cells and embryonal-type lung adenocarcinoma cells. 19) The opaque nuclei reveal strong endogenous biotin activity in FFPE sections. It should be noted that the endogenous biotin activity in mitochondria-rich cells in FFPE sections is retrieved by the heat-induced epitope retrieval step. 9) This is especially evident when 10 mM citrate buffer, pH 7.0 or 1 mM ethylenediamine tetraacetic acid (EDTA), pH 8.0 is employed as a dipping aqueous solution ( Figure  19). 20) This is the reason why biotin-based detection methods should be avoided, and instead, the HRP-labeled polymer technique has become the mainstream.

c)
Endogenous protein A and protein G activity Staphylococcus aureus and Streptococcus pyogenes have protein A and protein G on the cell wall, respectively. These proteins strongly bind the Fc portion of IgG molecules. Their IgG-binding activity is lost by formalin fixation, but HIER to FFPE sections retrieves IgG Fc-binding activity of the cocci. This may cause a pitfall in immunohistochemical identification of the coccal pathogen seen in FFPE sections (Figure 20). 21) For the epitope retrieval of the coccal antigens, proteinase digestion should be employed, instead of the heating pretreatment. When the heating pretreatment is employed, preincubation with diluted normal human or animal serum is indispensable.

d)
Endogenous pigments Brown or black-colored endogenous pigments may hamper the observation of the DAB coloring reaction of target. Melanin shows metachromasia, so that Giemsa or methylgreen should be chosen for counterstaining, instead of hematoxylin. The greenish-colored metachromatic melanin is clearly distinguishable from the brown DAB deposit. Hemosiderin pigments can be distinguished from the DAB product by counterstaining with Berlin blue. Figure 21 illustrates representative examples. Black-colored formalin pigments are often dispersed in and around the lesion with hemorrhage. Formalin pigment is particularly troublesome in autopsy cases. For the removal of the formalin pigment, unstained slides should be treated with alcoholic solutions containing picric acid, sodium hydroxide or ammonium hydroxide. 22) In the liver, aggregated bilirubin pigments may be confused with the specific DAB products (Figure 22). A trouble-shooting tip is as follows. ALP-labeled probe should be chosen as the secondary reagent, instead of HRP-labeled probe. The specific reaction products of azo dyes can be visualized in red or blue.

3)
Staining artifacts: effects of dewatering, contamination of cytokeratin-positive squams and insufficient deparaffinization When sections happen to be dried during antibody incubation, DAB coloring reaction is seen in the area of drying, mainly located at the periphery of the sections. This is because the antibody molecules are dry-fixed onto the glass slide in the area of dewatering. Sections are occasionally contaminated with cytokeratin-positive squams of skin origin, which may cause false-positive judgment. Fine adjustment of the microscopic focus reveals that the positivity is seen on the section but not in the section. If deparaffinization is incomplete or when air bubbles are formed on the glass slide, round-shaped negative zones are observed because of the lack of antibody penetration. Representative features are demonstrated in Figure 23.

4)
Antigenic deterioration due to prolonged conservation of unstained sections As positive control sections, it is convenient for us to conserve unstained glass slides in diagnostic pathology divisions. Importantly, the detection of certain antigens is not suitable for the storage at room temperature for a long period of time. Particularly susceptible are nuclear antigens, such as Ki-67 (MIB-1), p53, ER and PgR (Figure 24). Positive control sections should thus be cut from the control paraffin block just before immunostaining. Another method of convenience is to keep unstained glass slides in a freezer at -20C or lower. 23)

5)
Effects of section-stretching temperature on a hot plate and drying period after cutting sections Some antigens are quite susceptible to high section-stretching temperature on a hot plate (Figure 25). In our experience, immunoreactivities of glutathione-S-transferase (GST)-, an isozyme of cytoplasmic peroxide-reducing enzyme, and orotate phosphoribosyltransferase (OPRT), a cytoplasmic enzyme phosphorylating 5-fluorouracil (5-FU), an anti-cancer drug, are markedly reduced after stretching on a hot plate at 70C for 2-3 seconds. 24) CD8 antigen is also susceptible to the high temperature stretching. The antigenicities of human epidermal growth factor receptor-2 (HER2), epidermal growth factor receptor (EGFR), Ki-67, cytokeratins and CD20 are mildly weakened. The temperature and period for the stretching should be set at 40C for 20 seconds or at 50C for 10 seconds for the safety purpose.
Regarding the drying period after cutting sections, HER2 immunoreactivity is significantly weakened after leaving unstained sections in an incubator at 40C for 3 days (Figure 26). 25) A milder effect for HER2 immunostaining is observed after incubation at 40C for 1 day. When sections are kept in the incubator at 60C for 3 days, immunoreactivities of cytokeratins and pepsinogen 2 are apparently weakened.

6)
Effect of decalcification Formalin-fixed bone or calcified tissue should be decalcified before preparing FFPE tissue. Four kinds of solution are commonly used for decalcification, including 10% formic acid in formalin, Plank-Rychlo solution (containing 8.5% hydrochloric acid, 5% formic acid and 7% aluminum chloride), 5% trichloroacetic acid, and 10% EDTA disodium solution, pH 7. EDTA-mediated decalcification, requiring a longer time to complete, has little effect on the expression of lymphocyte surface markers and Ki-67, while significant damages to the antigenicity are observed after decalcification by formic acid solution and particularly Plank-Rychlo solution and trichloroacetic acid. Therefore, Plank-Rychlo solution and trichloroacetic acid should be avoided for immunohistochemical analysis for lymphocyte surface markers. 26) In contrast, Mukai, et al. described that routinely decalcified tissues can be used for immunostaining without significant loss of immunoreactivity. 27) When necessary, the surface decalcification procedure is performed on paraffin blocks to prepare higher quality FFPE sections without chattering. The surface of paraffin blocks is exposed to the decalcification solution for 30 minutes to 24 hours. Ki-67 immunoreactivity is especially susceptible to the surface decalcification procedure (Figure 27). Surface decalcification with formic acid for less than 1 hour does show a negligible effect on the marker expression. 28)

7)
Pitfalls and caveats in high-sensitivity immunostaining HRP-labeled polymer techniques such as Envision Flex, Simple Stain Max and Novolink have enabled us to visualize various tissue antigens in FFPE sections. The CSA employing fluorescein isothiocyanate (FITC)-labeled tyramide (CSA-II, available from Agilent Co./Dako) is an ultrasensitive detection system to visualize hidden antigens, which have been never detectable in FFPE sections. However, the higher the sensitivity of detection, the higher the background staining. Intracellular localization of antigens becomes blurred: i.e., the membrane antigen may diffuse into the cytoplasm.
PharmDx TM is an immunostaining kit available from Agilent/Dako for the detection of EGFR in colonic adenocarcinoma. The problem is that positive signals are commonly weak not only in cancer cells but also in the normal colonic mucosa. The kit employs the EnVision as the secondary reagent. However, the use of the CSA-II system available from the same company significantly enhances the positive signals ( Figure 28). 29) When the concentration of the primary antibody is too high, false-negative results may happen paradoxically when one uses the CSA-II system. 30) We evaluated the localization of CD4, interleukin-6 (IL-6) and interferon-gamma (IFN-) in heatretrieved FFPE sections of the pharyngeal tonsil with the CSA-II system. When the antibodies were diluted at 1:5,000, clear positivity was observed in immunocytes in FFPE sections. False negative findings were seen with anti-IFN- polyclonal antibody at a 1:500 dilution. In case of CD4 (monoclonal) and IL-6 (polyclonal), the antibodies at 1:50 dilution gave false-negative results (Figure 29). The background staining was increased. False negativity was partially recovered by 10-times dilution of the HRP-labeled secondary reagent and the FITC-labeled tyramide.
The autostainers are usually programmed for detecting the antigen with ultrahigh sensitivity. For example, HER2 immunostaining for breast and gastric cancer using a Ventana's Benchmark ULTRA (Roche Diagnostics, Rotkreuz, Switzerland) gives very strong positivity in HER2-overexpressed lesions. and normal mammary ducts and normal gastric foveolar cells also tend to be stained fairly strongly. When pathologists are unfamiliar with the result of this system, false positive judgment may happen in cancerous lesions not overexpressing HER2.

8)
Pitfalls and caveats in antigen retrieval sequences FFPE sections often require an antigen retrieval step before immunostaining. Cross-linkage of cellular proteins formed during formalin fixation may mask antigenic sites (epitopes). Epitope retrieval sequences, pretreatments before antibody incubation, can expose antigenic sites to allow antibodies to bind by breaking or loosening the formaldehyde-mediated methylene bridges. Two methods are mainly utilized for the epitope retrieval: enzymatic digestion and heat-induced epitope retrieval (HIER). To prevent detachment of sections during the pretreatments, the use of coated glass slides is inevitable: (3-aminopropyl)trimethoxysilane-coated glass slides are most commonly utilized.

a)
Proteinase pretreatment A variety of proteinases have been employed for the digestion of FFPE sections for the epitope retrieval. These include proteinase K, protease-1, pronase, actinase, trypsin, pepsin and ficin. The proteinase pretreatment is effective for retrieving the antigenicity of type 4 collagen (Figure 30) and laminin. Cytokeratins and lymphocyte surface markers may be retrieved with this pretreatment.
Glomerular immune deposits consisting of immunoglobulins and complements can be detected reproducibly by thoroughly digesting sections (Figure 31). 31) Prolongation of the digesting period is required. In case of trypsin digestion, the immune deposits are visible after 2 hour treatment (4 times longer than the usual use). We detected IgM deposits in glomeruli of IgM nephropathy in the FFPE autopsy kidney. 32) IgG deposition on the plasma membrane was proven in the skin, gut and bronchus in lethal paraneoplastic pemphigus. 33) Of note is that the antigenicity may be lost by the proteinase pretreatment. Representative examples include peptide hormones (substance P and gastrin-releasing peptide), IgD and J-chain of IgA and IgM (Figure 32). The epitopes detected by certain monoclonal antibodies, such as anti-vimentin V9 and anti-cytokeratin KL-1, are lost by the pretreatment. Of note is that cytokeratin immunoreactivity detected with monoclonal antibody KL-1 is lost after trypsin treatment but enhanced by pepsin digestion. The protease pretreatment should also be avoided for localizing CD20 (

b) Heat-induced epitope retrieval (HIER)
Hydrated heating is quite effective for retrieving hidden antigenicities in FFPE sections. 3,6) Typically, deparaffinized sections are heated at 60-121C in 10 mM citrate buffer, pH 6.0 or pH 7.0 or in 1 mM EDTA solution, pH 8.0. The devices used for heating include the incubator (at 60C, overnight), water bath (at 95C), microwave oven (at 100C), steamer (at >100C), pressure pan (at 121C) and autoclave oven (at 121C). The author strongly recommends pressure pan cooking, because of its stability, reproducibility and ease of handling (Figure 33).
A variety of antigens (epitopes) can be retrieved, such as intranuclear antigens (p53, Ki-67, ER, PgR, thyroid transcription factor-1 [TTF1], caudal-type homeobox-2 [CDX2], etc.), plasma membrane antigens (lymphocyte surface markers, epithelial membrane antigen [EMA], etc.), cytoskeletal proteins (cytokeratins, vimentin, desmin, actin, etc.), and secretory proteins (parathyroid hormone, -fetoprotein, fibrinogen, etc.). Typical examples are displayed in Figures 34 and 35. Peptide hormone immunoreactivity may also be retrieved by the heating pretreatment. Choice of the solution is important for appropriate HIER of the respective antigens (Figure 36). Supposedly, chelation of calcium iron by citric acid and EDTA is a key factor for HIER. It is of note that nuclear antigens such as ER and p53 show an HIER effect in ethanol-fixed cytology preparations (Figure 37). 34) The heating treatment changes double-stranded DNA into single-stranded to allow nuclear antigens hidden within the DNA stretches to bind antibodies. Occasionally, we happen to experience false-negativity of nuclear antigens such as p53, ER and PgR when monoclonal antibodies are incubated overnight at 4C or at room temperature. Nuclear reactivity is evident when the same antibodies at the same dilution are incubated for 60 minutes (Figure 38). This paradoxical phenomenon can be explained as follows. During the prolonged incubation period, heat-provoked single-stranded DNA naturally returns to doublestranded, so that the nuclear antigens are hidden again to become inaccessible to the antibodies.
Hematoxylin is suitable for nuclear counterstaining in heat-treated sections. Methylgreen stains the nuclei only faintly in the heat-treated sections. This is explained by the fact that the methylgreen dye has an affinity to double-stranded DNA. Nuclear affinity to hematoxylin is evidently decreased after heating in 1 mM EDTA solution, pH 8.0, so that prolongation of hematoxylin staining time is needed (Figure 39). 35) It should also be noted that, during heating in 10 mM citrate buffer, pH 7.0, sections tend to be detached off.
Ironically, the antigenicity of some antigens is markedly weakened or lost after the heating pretreatment ( Figure  40). These include BM-1 (a marker of the myeloid precursors), neutrophil elastase (NP57), von Willebrand factor, NSE and GST (,  and ).
One must know that the nuclear localization of Ki-67 is dependent upon how to cool down sections after heating. 36) When sections are rapidly cooled down in tap water after heating, Ki-67 often reveals a false negative finding. In contrast, clear nuclear positivity is observed in sections cooled down gradually by leaving sections in the heating solution at room temperature for more than 30 minutes (Figure 41). When one employs a high molecular weight polymer reagent (EnVision, Agilent/Dako Co) as the secondary probe, the cytoplasm of the mitotic cells is stained for Ki-67 while the nuclei of proliferative cells remain unstained (Figure 42). Such a strange phenomenon is not experienced when the secondary polymer reagent of smaller molecular size such as EnVision Plus or EnVision Flex is employed.

c) Other methods for epitope retrieval
Immunoreactivity of -amyloid protein in the brain is retrieved by soaking in 100% formic acid solution for 5 minutes (Figure  43), 37) while protease treatment and HIER are ineffective. Bromodeoxyuridine (BrdU) experimentally incorporated in the nuclei of proliferative cells can be detected in FFPE sections after the exposure in 2-4 N hydrochloric acid solution for 20-90 minutes. 38) DNase I from the bovine pancreas can also be utilized for the same purpose. 39) Actin immunoreactivity of intracytoplasmic eosinophilic inclusion bodies in infantile digital fibromatosis is retrieved by treating sections in 1 N potassium hydroxide in 70% alcohol for 60 minutes followed by trypsin digestion. 40) 8. How to judge the immunostained results 1) False positivity or equivocal negativity When the target cells appear to be immunostained faintly, it is critically important how to judge the result: whether they are weakly positive or equivocally negative. The author once experienced a regrettable misjudgment for a mediastinal small round tumor of a middle-aged man. Placental alkaline phosphatase (PALP) was judged as weakly positive, so that the author's final diagnosis was germinoma (seminoma). However, it was actually small cell carcinoma of the lung with false-positive (or equivocally negative) PALP immunoreactivity. In general, immunoreactivity of secretory proteins in neoplastic cells tends to be weaker than the normal cells. Examples include chromogranin A, insulin, gastrin, von Willebrand factor and immunoglobulins.

2)
Intracellular localization pattern of antigens The specificity of immunostaining can be judged by the intracellular localization pattern of the antigenic substances, as schematically illustrated in Figure 44. 1) Representative immunohistochemical markers are displayed in one panel (Figure 45).
Peptide hormones and chromogranin A are localized as intracytoplasmic fine granules. Lysosomal and mitochondrial proteins are visualized as intracytoplasmic coarse granules. Secretory proteins such as alpha-fetoprotein, human chorionic gonadotropin and immunoglobulins are seen as small vesicles in the cytoplasm, representing the ultrastructural localization in the rough endoplasmic reticulum and Golgi apparatus. Particularly in neoplastic cells, immunoreactivity of secretory proteins may be accentuated and clustered in the Golgi area. Diffuse cytoplasmic reactivity is seen for cytosolic proteins such as myoglobin, NSE and GST. S-100 protein, heat-shock proteins, ubiquitin and -catenin are distributed diffusely in both the cytoplasm and nucleus. The nuclei are diffusely positive for Ki-67 (MIB-1), p53, CDX2, ER and PgR, but nucleolar staining is frequently accentuated for Ki-67. Plasma membrane proteins reveal distinct membrane staining, and the Golgi area is also frequently positive. Carcinoembryonic antigen (CEA) is expressed along the apical plasma membrane in normal gastric and colonic mucosa, while in adenocarcinomas of the stomach and colon, the plasma membranes are circumferentially positive. 41) Such intracellular antigen distribution patterns are clearly recognized in PPFE sections, when compared with fresh-frozen sections.
When Ki-67 is localized along the plasma membrane instead of the nucleus, one can easily recognize they are nonspecific (Figure 46). Ki-67 antigen is positive along the plasma membrane in certain tumors, and the finding has been utilized as a diagnostic marker. 42) PgR may be expressed along the plasma membrane of breast cancer cells, and the lymphocyte surface markers happen to be localized in the nucleus (Figure 47). When an ultra-sensitive method is employed for immunostaining together with HIER, plasma membrane proteins may appear to be localized diffusely in the cytoplasm. Periodic acid treatment after HIER may suppress such non-physiological (artificial) localization.

3) Specificity of antibodies
The specificity of the antibody should be confirmed before practical use. Information supplied by the manufacturer is valuable, but one must not believe it in blindness. It is of note that rabbit antiserum occasionally contains natural antibodies against intermediate filament proteins. 43) For example, anti-myoglobin antiserum stains the epidermal keratinocytes, vascular endothelial cells and pancreatic islet cells (Figure 48). Monoclonal antibodies supplied as a form of the ascitic fluid may show unexpected cross-reactivity due to mouse ascitic components. 44) Small molecules (haptens) such as peptide hormones and thyroxine are immunized with carrier proteins, so that antibodies designated against peptide hormones also contain anti-carrier protein antibodies. 45) When bovine thyroglobulin is used as a carrier protein, the antiserum reacts with colloid and epithelial cells of the human thyroid gland. Anti-cholecystokinin (CCK) antiserum immunized with Ascaris proteins as a carrier protein reveals cross-reactivity to the smooth muscle, cartilage and some epithelial cells in human tissues, as well as to neuroendocrine tumors (Figure 49). Since the extract of Mycobacterium tuberculosis is commonly used as an adjuvant for immunization, 46) many antisera may stain tuberculous bacilli and other mycobacteria in FFPE sections (Figure 50).

4)
Specificity of the specific markers The poorer the degree of cancer differentiation, the more infrequent the expression of specific markers. The lack of marker expression does not necessarily exclude the possibility of specific differentiation of neoplastic cells. The epithelial cells may express vimentin, while a variety of non-epithelial tumor cells are immunoreactive for cytokeratins. Spindle cell carcinoma of the skin co-expresses cytokeratin and vimentin. 47) Anaplastic large cell lymphoma may express cytokeratins. 48) Usefulness and pitfalls of cytokeratin immunostaining are displayed in Figure 51.
The lack of basal cells is an excellent indicator of prostatic adenocarcinoma. Postatrophic hyperplasia of the prostate histologically resembles adenocarcinoma, causing diagnostic confusion. 49) The preservation of the cytoleratin 5/6-immunoreactive basal cells around the respective acini is of significant diagnostic value (Figure 52).
Neuron-specific enolase (NSE) or gamma-enolase is not necessarily specific to neurons and neuroendocrine cells. 50) A variety of non-neuroendocrine cells and their tumors may express NSE. Glial fibrillary acidic protein (GFAP) is known to be expressed in some myoepithelial cells and their tumors of the salivary gland. 51) Antiserum against prostate-specific antigen (PSA) may be cross-reactive to salivary gland duct cells. The reason is that PSA or human kallikrein-3 shares epitopes common to other kallikreins expressed in the salivary gland ducts. 52) Prostatic acid phosphatase (PAcP) is frequently expressed in neuroendocrine cells and tumors of the rectum. 53) Representative examples are illustrated in Figure 53. Diagnostic pathologists must study and know such unexpected expression of "specific" markers in unrelated cells.  Figure 54. Stromal collagen fibers also tend to show nonspecific binding with antibodies. 57)

6)
Positive and negative controls Negative controls employing normal animal serum or PBS are important to check the specificity of the immune reaction. When multiple antibodies are evaluated simultaneously, the antibodies mutually function as indifferent antibodies, so that negative control sections may be unnecessary.
Appropriate positive controls are inevitably needed in each run of immunostaining. They function as the evidence to exclude the possibility of inactivation of the antibodies. When positive cells are consistently included in the same sections, for example, cells positive for vimentin, vascular markers and epithelial markers, they act as internal positive controls. 17) For the diagnostic practice for breast cancer, immunostaining for ER, PgR, HER2 and Ki-67 is consistently requested. When the author looked at borrowed slides sent from an outside hospital as a second opinion consultation for "ER-negative" breast cancer, not only breast cancer cells but also non-neoplastic mammary ductal cells were negative for ER. ER was then immunostained again in our laboratory, and clear nuclear positivity of ER was confirmed in both the normal and cancerous cells. Appropriate judgment for such technical pitfalls is critically important to give precious information for the treatment of the patient.

9.
When unexpected results are obtained It is a chance of particular significance if immunohistochemical results are different from H&E-based expectation. However, pathologists must make a careful judgment by considering the specificity of the markers they immunostained. We may encounter cases of cytokeratin-positive malignant lymphoma, glioma, sarcoma or melanoma. Vimentin is expressed in certain normal and neoplastic epithelial cells. We should know the expression of EMA (as a MUC-1-related molecule) on normal and neoplastic perineurial cells and plasma cells (Figure 55). 58) CD138, a marker of normal plasma cells, may not be expressed in neoplastic plasma cells. 59) Cautions are needed when the clinical and histopathological diagnoses are discrepant. First of all, we must exclude the possibility of tissue contamination or mistake from other patient. The container of the formalin-fixed specimen and the paraffin block should be checked first just to be sure. We can immunostain blood group substances, A, B and H or Lewis a and b, for blood typing. The blood group substances are expressed consistently in vascular endothelial cells and occasionally in epithelial cells, so that we can apply immunohistochemical blood typing to proving the tissue contamination or mistake of FFPE sections, when the blood group of the cases of subject is different. 60) Figure 56 displays successful immunohistochemical blood typing for contaminated needle biopsy specimens of the prostate, where one specimen contained cancer but not in another.

10.
When only one preparation is available It happens that we have only one unstained or H&E-or Papanicolaou-stained specimen using non-coated glass slide. Particularly, cytology specimens often have a chance of this situation. Can we use the one slide for immunostaining analysis? Yes, two methods can be applied. One is re-staining method and another is the cell transfer technique. See the reference 61 for details.

1)
Re-staining method Suppose you have only one glass slide already stained with H&E or Papanicolaou sequences. We can re-use the specimen. 62) The target cells or areas should be photographed beforehand. The first step is the removal of the cover glass in warm xylene. The stained dyes are bleached in acid alcohol solution (0.5% hydrochloric acid in 70% ethanol) for more than 1 hour. Or, you can leave the specimen in tap water overnight to bleach the dyes. Then the glass slide is ready for immunostaining. One typical example of cervical cytology smear re-stained for a Chlamydia trachomatis antigen is illustrated in Figure 57.
When the silane-coated glass slide has been used, you can apply HIER to immunostaining. If not, the sections or cells before the bleaching step should be transferred to silane-coated glass slide so as to apply to HIER-assisted immunostaining, as mentioned below. Finally, the same cells or areas of target should be re-photographed. If multiple markers are needed to be immunostained, the antigenic substances should be visualized repeatedly with an ALP-labeled secondary reagent. The azo dyes used for the color reaction for ALP, colored red or blue, can easily be bleached by soaking specimens in ethanol.
When the immunostained (DAB-colored) preparations are re-stained, negative control sections with negative results can be utilized for new immunostaining. If there is focal positivity of the DAB reaction, one can choose the second immunostaining with the ALP-labeled method to have a double-immunostained section. The cover glasses should be removed by soaking in warm xylene. The sections or cell preparations are covered with mounting resin solution, and, after leaving them for 1-2 days in an incubator at 40C, the solidified resin forming a thick membrane can be peeled off after soaking in warm water for 1 hour. The sections or cells have been detached and transferred to the membrane side. In warm water, the resin membrane is then placed onto silane-coated glass slides. The glass slides with transferred resin membrane should be fully dried in the incubator. The resin can easily be removed by dipping the specimens in xylene. By scissors, you can cut the resin membrane into several pieces to obtain plural silane-coated glass slides for immunostaining ( Figure  58).
Itoh, et al. invented a modified method for rapid cell transfer. 64) The mounting resin solution should be diluted double with xylene, and the resin is solidified on a hot plate at 70-80C. With this modification, the cell transfer can be achieved in one hour.
It should be noted that the cell transfer technique is neither applicable to dry-fixed Giemsa-stained preparations nor to the preparations mounted on silane-coated glass slides, such as cytology specimens of the urine, cerebrospinal fluid or effusions. For these specimens, the re-staining method should be tried. We found miracle coated glass slides supplied for cytology practice termed TACAS (Thinlayer Advanced Cytology Assay System supplied from Medical & Biological Laboratories, Nagoya, Japan), which allow both immunostaining with HIER and cell transfer. The TACAS glass slides prevent detaching off sections during immunostaining with HIER and allow the cell transfer after immunostaining employing the inverted beam capsule method. 65) Both the antigen and genome of severe fever with thrombocytopenia syndrome (SFTS) virus were successfully localized at the ultrastructural level with the pre-embedding technique (Figure 59).

11.
Immunostaining with low specificity and high sensitivity Pathogens infected in tissues are proven with antibodies with low or unknown specificity. Two examples are described here. Please refer to the author's previous article (reference 66).

1)
Use of antisera against pathogens showing wide cross-reactivity The antigenicities of pathogens are distinct from those of human cells. Therefore, pathogens can suitably be immunolocalized in FFPE sections. Bacteria or bacterial antigens are visualized with antisera against pathogens showing wide cross-reactivity. These include commercially available rabbit antisera against Bacillus cereus, BCG, Treponema pallidum and Escherichia coli. The immunostaining is useful for screening of bacterial infection in FFPE sections. 66) In other words, the identification of bacteria or bacterial antigens in the infectious lesions is of diagnostic value, even if the specificity is unknown.
A representative example includes Corynebacterium kroppenstedtii infected in granulomatous mastitis. The lipophilic bacteria are visualized with the antisera described above mainly in fat vacuoles within the inflammatory breast lesion. 12) Flesheating bacteria, Vibrio vulnificus, provoking gangrene of the extremity can be localized in the affected skin with antisera against Bacillus cereus, BCG and Treponema pallidum. 67) Leptospira interrogans infected in the liver is immunoreactive with E. coli antiserum. The causative bacteria in Haemophilus pertussis-provoked pneumonia are clearly shown in FFPE sections with all of the antisera. In colon biopsy specimens of intestinal spirochetosis, the long basophilic spiral-shaped bacteria, Brachyspira aalborgi, adhered on the surface of the colonic epithelial cells are strongly immunostained with all of the antisera described above. 66) Representative features are illustrated in Figures 60 and 61.

a)
Use of patients sera The sera of patients have high-titer antibodies against the causative pathogen, particularly in cases at the recovery stage or with chronic persistent infection. 66,68) When the host reactions such as abscess or granuloma are observed microscopically, the serum, diluted at 1:500 to 1:1,000, is applicable as a highly sensitive probe to detecting causative pathogens in the infected lesion embedded in paraffin. 66,68) What you need is to make a brief phone call to a laboratorian to save a small aliquot of patient's serum. Indirect immunoperoxidase method using HRP-labeled anti-human IgG or immunoglobulins is suitable in order to obtain low-background immunostaining. The methods with high sensitivity of detection such as LSAB or the polymer method give high background staining due to diffuse distribution of endogenous IgG in tissues. Usually, HIER is unnecessary for the detection, but HIER may occasionally assist at increasing the sensitivity of detection. 66) This low-cost sequence is particularly useful in case of protozoan and helminthic infections, including cryptosporidiosis, isosporiasis, toxoplasmosis, amebiasis, schistosomiasis, ascariasis and gnathostomiasis. 69) When the clinical diagnosis has been confirmed, we can visualize the causative pathogen in FFPE sections of the infected lesion. Even if the clinical diagnosis is totally unsettled, candidate causative pathogens can be visualized within the lesion. In some cases, the specificity of the patient's serum is totally unknown. The size, shape and localization pattern of la-belling may suggest certain causative microbes, when considered together with clinical features. Once the specificity of the patient's serum is known, the serum can be used thereafter as a specific probe for immunostaining. In order to avoid biohazard, the serum of the carriers of hepatitis B and C or acquired immunodeficiency virus must not be utilized.
As a representative example, a liver biopsy of visceral leishmaniasis (kala azar) is displayed in Figure 62. The patient was a middle-aged Japanese businessman who stayed long in Australia, Thailand and India for business purpose. He complained of fever and malaise, and accompanied anemia, thrombocytopenia and liver dysfunction. The patient's general condition was poor. FFPE liver biopsy preparation showed multifocal small non-caseous granulomas. No positive findings were obtained by immunostaining with a routine panel of anti-microbial antibodies. Immunostaining using 1:500 diluted patient's own serum demonstrated red cell-sized and rounded positive signals in the cytoplasm of epithelioid cells in the granuloma or in Kupffer cells. The size, shape and localization pattern strongly suggested the possibility of visceral leishmaniasis caused by Leishmania donovanii, which is endemic in India. Visceral leishmaniasis, never endemic in Japan, was serologically confirmed thereafter, and antimony-based therapy saved patient's life. 66,68) Another example is a case of Balamuthia encephalitis with a skin nodule. Here, the 1:500 diluted serum of another patient who suffered lethal chronic meningoencephalitis caused by Balamuthia mandrillaris was used. 66) In biopsy samples of both the skin and brain, brown-colored amebic bodies (trophozoites and some cysts) are clearly identified, as shown in Figure 63.
The same strategy is applicable to autoimmune disorders. The patient's serum of autoimmune (type A) gastritis can be utilized to detect proton pump molecules on intracytoplasmic secretory canaliculi of parietal cells in the normal oxyntic gastric mucosa in FFPE sections (Figure 64). 70) The dilution at 1:20 was employed in this situation. Autoantibodies against amphiphysin, a 128 kDa presynaptic protein, are seen in a patient of breast cancer who manifested progressive rigidity and painful spasms (an autoimmune encephalitis called Stiff-Man syndrome). 71) The patient serum at a 1:20 dilution stained the plasma membrane of the breast cancer cells of her own. The serum of another patient with autoimmune limbic encephalitis of non-paraneoplastic type 72) stains astroglial cells and glial fibers in FFPE basal ganglia of autopsied normal brain. Immunostaining using diluted patient's serum may contribute to clarifying the pathogenesis of autoimmune encephalitis, as illustrated in Figure 65.

12.
Concluding remarks Nowadays, it makes sense to have beautiful and specific immunostaining. Biomedical engineers (medical technicians) should know how to select markers and how to judge the result. Pathologists and investigators must understand the pitfalls and caveats of immunostaining procedures. For applying immunostaining to diagnostic pathology services, it is critically important for us how to stain, how to select markers and how to judge the result. The author wants to emphasize the importance of close teamwork and cooperation between the biomedical engineers and pathologists.                            . Immunohistochemical blood typing in needle biopsy specimens of the prostate. Specimens from two patients were mixed up in one tissue container. The blood group was type A in one patient, and type O in another. Immunostaining for type A substance is positive in endothelial cells in the left panel, but negative in the right panel. It is evident that the patient with type A blood group contains prostatic adenocarcinoma (asterisk).  After removal of the cover glass, the cell preparation is covered with mounting resin solution to be solidified as a thick membrane sheet. Cells are transferred to the solidified resin (left). After cutting into several pieces, the resin membrane is placed onto silane-coated glass slides, which should be fully dried in the incubator (right).

Preprints
The resin component can easily be removed by dipping the specimens in xylene. Now, the specimens are ready for immunostaining using HIER.

Figure 59.
Pre-embedding immunoelectron microscopy for SFTS virus using a TACAS glass slide. Left: light microscopic immunostaining, Right: immunoreactive round-shaped viral particles, around 100 nm in size, at the ultrastructural level. The FFPE splenic red pulp from an autopsy case was immunostained with a monoclonal antibody to SFTS virus after HIER. For pre-embedding immunoelectron microscopy, the immunostained section was peeled off to transfer to an Epon-embedded block by the inverted beam capsule method. The miraculous TACAS glass slide prevents detaching off sections during immunostaining and allows the cell transfer after immunostaining. Bar indicates 500 nm.