U.S. flag

An official website of the United States government

Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Immunohistochemistry in Investigative and Toxicologic Pathology

Kyathanahalli S. Janardhan, Heather Jensen, Natasha P. Clayton, and Ronald A. Herbert
Toxicologic Pathology (2018) DOI: http://dx.doi.org/10.1177/0192623318776907 PMID: 29966501

Publication

Abstract

Immunohistochemistry (IHC) is a valuable tool in pathology. This review provides a brief description of the technical aspects of IHC and a detailed discussion on the variables that affect the results, interpretation, and reproducibility of IHC results. Lists of antibodies that have and have not worked in IHC on various mouse and rat tissues in our laboratory are provided as a guidance for selection of antibodies. An approach to IHC method optimization is presented. Finally, the critical information that should be included as a part of peer-reviewed manuscript is also discussed.

Figures

Figure 1. Basics of immunohistochemistry.

(A) Illustration of the direct method of detecting a protein by immunohistochemistry where a polyclonal (upper) or monoclonal (lower) antibodies tagged to an enzyme such as horseradish peroxidase (HRP) are used.
(B–F) Illustration of various indirect methods of detecting a protein by immunohistochemistry. A simple indirect method is shown in (B). In this method, primary antibody (polyclonal—top and monoclonal—bottom) is unconjugated and secondary antibody is conjugated to HRP or alkaline phosphatase.
(C) Illustration of the peroxidase–antiperoxidase (PAP) method. First, unconjugated primary and secondary antibodies are added to the reaction, followed by addition of PAP complex.
(D) Illustration of the avidin-biotin complex (ABC) method. After the addition of unconjugated primary antibody, biotinylated secondary antibody is added followed by addition of HRP-conjugated ABC.
(E) Illustration of the labeled streptavidin-biotin method is shown. Unconjugated primary antibody, biotinylated secondary antibody, and HRP-conjugated streptavidin are added sequentially.
(F) Polymer-based method is demonstrated. After adding unconjugated primary antibody, secondary antibody and enzyme conjugated polymer is added to the reaction.
(G and H) Illustration of the catalyzed signal amplification method which is also known as tyramide signal amplification. After addition of unconjugated primary antibody and HRP-conjugated secondary antibody, biotinylated tyramide and hydrogen peroxide are added to the reaction. In the presence of hydrogen peroxide, HRP converts biotinylated tyramide to a reactive molecule which binds to the tyrosine residues present in the proximity of the protein of interest. This is followed by addition of HRP-conjugated streptavidin.

Figure 2. Illustration showing tissue lifting in collagen rich tissues such as skin.

This is one of the disadvantages of using pressure cooker for heat-induced antigen retrieval in immunohistochemistry (IHC) of collagen rich tissues.
(A) IHC for S-100 protein performed without antigen retrieval. Notice the lack of any artifactual changes in the collagen and adipose tissues (asterisks).
(B) IHC for S-100 protein performed using pressure cooker and citrate buffer for heat-induced antigen retrieval. Notice the excessive lifting of collagen (asterisks) which interferes with staining and interpretation.

Figure 3. Immunohistochemistry for peroxisomal membrane protein 70 on rat liver demonstrates that certain proteins do not require antigen retrieval (AR).

Figure 3. Immunohistochemistry for peroxisomal membrane protein 70 on rat liver demonstrates that certain proteins do not require antigen retrieval (AR). (A) No AR. (B–D) Pepsin, pressure cooker/citrate buffer, and pressure cooker/EDTA buffer were used for AR, respectively. Presence of immunostaining in (A) indicates that AR is not required. However, using the method which produces specific, higher intensity staining (in this case D) will allow the user to increase the specificity by increasing the antibody dilution.

Figure 4. Influence of antigen retrieval (AR) on the results of immunohistochemistry (IHC).

(A) and (B) show IHC using paired box protein 5 antibody, a B-cell marker, on rat spleen without and with AR, respectively. The staining is very weak when AR is not preformed (A) compared to the staining after AR (B). Use of an incorrect AR method can result in misinterpretation.
(C) and (D) demonstrate TUNEL method in mouse lung sections in which pressure cooker/citrate buffer and proteinase K were used for AR, respectively. Use of pressure cooker in TUNEL method induces nonspecific nuclear staining in all the cells as seen in (C). Based on the protein biology and histomorphology, this should be interpreted as nonspecific staining. When proteinase K is used for AR, no nonspecific staining is observed in the same lung tissue (D).
Use of AR can induce nonspecific staining for certain proteins. Specific staining for smooth muscle actin in the myometrium (star) and blood vessels (arrowheads) of mouse uterus is present in (E). Note that the epithelium (arrow) and stromal cells (asterisk) are not staining.
Use of AR, when not required, can induce nonspecific staining as seen in (F). Here, in addition to the staining in myometrium and blood vessels, nonspecific staining is present in epithelium (arrow) and stromal cells (asterisk).

Figure 5. Choice of detection method can influence the immunohistochemistry (IHC) results.

In some instances, using a polymer-based detection method is better than other methods such as avidin-biotin complex (ABC) method to avoid nonspecific staining.
(A) Mouse brain stained with ionized calcium-binding adapter molecule 1 antibody using ABC method. Note nonspecific staining in the nuclei of neurons (arrows) along with specific staining of microglial cells (arrowheads).
When a polymer-based method is used (B), only specific staining in the microglial cells (arrowheads) is present. However, polymer-based method is not the best for all proteins.
In (C), IHC for Ki67, a marker of proliferating cells, by ABC method demonstrates intense staining in crypt epithelium (arrow) and lymphocytes (asterisk).
However, only small proportion of cells are stained in (D), when the section is stained using a polymer-based method. This comparison allows the user to decide that the method used in (C) is optimal for Ki67 staining. This example highlights the need to determine optimal detection system at the method optimization stage.

Figure 6. (A)–(C) demonstrate how choice of an antibody can influence the outcome of immunohistochemistry.

In (A), an antibody claimed by the company (Abcam; catalog # ab11197) to detect mucin 2 (MUC2) in mouse produced nonspecific staining and did not recognize goblet cells which contain MUC2 (arrows). Although the negative control stained with an isotype matched nonspecific immunoglobulin produced no staining (B), staining in (A) can’t be interpreted as positive as the staining is present everywhere except goblet cells. Section of intestine in (C), stained by another MUC2 antibody (Novus Biologicals, catalog # NBP1-31231) shows specific staining in goblet cells (arrows).

Figure 7. In addition to proper choice of an antibody, comprehensive knowledge on the protein of interest is important to avoid misinterpretations.

(A) is a section of testis stained with an antibody claimed to be specific for SRY Box 9 protein (SOX-9; Santa Cruz Biotechnology, catalog # sc-20095), a Sertoli cell marker. Based on the negative control section stained with an isotype matched nonspecific immunoglobulin (B), staining in (A; arrows) can be interpreted as specific. However, applying the knowledge of Sertoli cell distribution and morphology (arrows) shown in (C) using another Sertoli cell marker, GATA binding protein 4 (GATA4; Abcam, catalog # ab84593), staining in (A) doesn’t seem to be in Sertoli cells. Further, SOX-9 is a transcription factor and is known to be localized to nucleus. However, in (A), the staining appears to be perinuclear. Considering all these together, staining in (A) must be interpreted as nonspecific.

Figure 8. Importance of choice of antibody, antigen retrieval (AR) method and importance of protein biology to avoid misinterpretation in immunohistochemistry are highlighted in (A)–(D).

Peroxisome proliferator-activated receptor-alpha (PPARα) is a nuclear receptor expressed in various tissues including liver, and the protein is localized to nucleus. (A) When a section of rat liver is stained with PPARα antibody (Abcam; catalog #ab8934) using pepsin for AR, intense cytoplasmic staining can be observed in centrilobular hepatocytes (arrows). The corresponding negative control section shown in (B) has no staining. However, when the liver section is stained with the same antibody but with a different AR (pressure cooker and EDTA buffer), the staining pattern changes. Now the staining is present only in bile duct cytoplasm (C; arrow) and not in hepatocytes. The corresponding negative control shown in (D) has no staining. The results vary based on the type of AR used. Further, staining observed in both (A) and (C), although appear specific, is not true because the localization seen here is cytoplasmic and not nuclear as reported for PPARα.

Figure 9. (A)–(F) demonstrate how tissue other than the region of interest can help in determining the specificity of an antibody.

(A) Section of rat testis stained with octamer-binding transcription factor 4 (OCT4) antibody (Santa Cruz Biotechnology, catalog # sc-5279) showing no staining in a seminoma. (B) However, some cells in the surrounding seminiferous tubules (arrow) show nuclear staining. (C) Section of a uterine tumor stained with OCT4. The neoplastic cells appear to be positive based on the absence of any staining in the negative control (D). However, when the surrounding normal tissue (E) and corresponding area in the negative control (F) are evaluated, staining appears to be also present in the normal endometrial epithelium (arrow) and stroma (asterisk). Since these cells are terminally differentiated, and OCT4 expression is associated with stemness, this antibody can be interpreted to be not appropriate for rat tissues.

Figure 10. (A)–(F) demonstrate how using antibodies from different lots can affect staining results.

(A) Section of spleen stained with a CD3 antibody shows specific staining in T-cells. (B) Section of the spleen stained with CD3 antibody from a different lot has significant nonspecific staining (asterisks). (C) The CD3 antibody used in (B) was diluted 10 times more to obtain specific staining in the spleen. (D) A section of uterus stained with a cytokeratin 18 antibody shows specific staining in the glandular epithelium (arrows). (E) Section of the uterus stained with the cytokeratin antibody from a different lot produces substantial nonspecific staining in the stroma and myometrium (asterisks). (F) The antibody used in (E) was diluted 10 times more to obtain specific staining in uterus.

Figure 11. Nonspecific staining in a variety of cell types.

(A) Nonspecific staining of neutrophils (arrows) by Ki67 antibody in a section of mouse lung.
(B) Nonspecific staining of mast cells in a section of rat uterus.
(C) Section of mouse intestine stained with a bromodeoxyuridine (BrdU) mouse monoclonal antibody. The nuclear staining in the crypt epithelial cells is specific (arrows). However, cytoplasmic staining in the plasma cells (arrowheads) in the lamina propria is nonspecific.
(D) A section of the same intestine in (C) stained by omitting primary antibody. Notice cytoplasmic staining in plasma cells (arrowheads) even in the absence of primary BrdU antibody. This results from using mouse monoclonal antibody on mouse tissues. Various methods have been described to overcome this issue.
(E) and (F) provide two such examples. In (E) and (F), sections of intestine are stained for BrdU using a mouse-on-mouse horseradish peroxidase (HRP) polymer and a mouse on mouse HRP kit, respectively. In both, there is reduction in the nonspecific cytoplasmic staining of plasma cells (arrowheads). However, the staining for BrdU is also substantially reduced (arrows). Such a reduction can be an issue when the antigen of interest is not abundant in tissue.

Figure 12. Technical error as a source of nonspecific staining.

(A) Section of uterus allowed to dry during immunohistochemical staining with a CD31 antibody. As a result, instead of staining only the endothelium, nonspecific staining was present throughout the section.
(B) Tissue not allowed to dry stained with CD31 antibody resulted in a specific staining of endothelial cells (arrows). Epithelium, stroma, and myometrium (asterisks) that serve as internal negative controls are not stained.

Figure 13. Endogenous peroxidase is not a major source of nonspecific staining in tissues.

Sections of a kidney incubated with chromogen without (A) and with (B) preincubation in hydrogen peroxide. Only weak nonspecific staining is present in red blood cells (arrow) in (A) compared to no staining in (B). Similarly, staining sections of spleen (C) and liver (D) without preincubation in hydrogen peroxide produces weak nonspecific staining in red blood cells (arrows).

Figure 14. Blocking with serum or other agents is not critical in immunohistochemistry on formalin fixed paraffin embedded tissues.

Sections of mouse spleen (A and B), intestine (C and D), and brain (E and F) are stained with paired box protein 5, Ki67, and glial fibrillary acidic protein, respectively. Sections in (A), (C), and (E) are stained after blocking with 10% normal serum, and sections in (B), (D), and (F) are stained without blocking with serum, before incubating with the primary antibody. Staining is similar in sections stained with or without blocking. There is no nonspecific staining when blocking with serum is omitted.

Figure 15. Examples of internal controls in immunohistochemistry.

(A) Section of a rat uterus stained with smooth muscle actin. Since smooth muscle actin is known to be expressed by smooth muscle cells (in myometrium and around blood vessels) and not by epithelium or endometrial stroma, they serve as internal positive and negative controls, respectively. In this section, smooth muscle cells in the myometrium (asterisks) and around the blood vessels (arrowhead) are stained. The epithelium (arrow) and endometrial stroma (star) are not stained. Therefore, the observed staining can be interpreted with confidence as specific.
(B) Similarly, in a section of mouse lung stained with surfactant protein C (SPC; B), type II alveolar epithelium serves as an internal positive control and type I alveolar epithelium and bronchiolar epithelium serve as internal negative controls. In this section, only type II cells (arrows) are stained; type I cells (arrowheads) and bronchiolar epithelium (short arrow) are not stained. Based on these, staining for SPC can be interpreted as specific.

Figure 16. Method optimization for immunohistochemical staining with a new antibody.

Tables

Table 1. List of Antibodies that Did Not Work Optimally in Immunohistochemistry on Rat or Mouse Tissue in our Laboratory.

Table 2. List of Antibodies Used in Immunohistochemistry (IHC) on Mouse Tissues.

Table 3. List of Antibodies Used in Immunohistochemistry (IHC) on Rat Tissues.

Table 4. List of Antibodies that Worked in Immunohistochemistry (IHC) on both Rat and Mouse Tissues.

Table 5. Recommendation on the Details to be Included in the Methods in Published Manuscripts with Immunohistochemistry Results.