Abstract
It has been difficult to quantify protein production in small pathological specimens by conventional techniques. We describe a new method for semiquantification of immunohistochemical staining, which involves application of the enzyme-labeled avidin (LAB) technique, coupled with an ultra-sensitive and fast chemiluminescent substrate for alkaline phosphatase. The entire procedure can be completed in less than 3 hr. The final step involves X-ray film exposure for 30 min, and the optical density of the subsequent images is examined with a microcomputer imaging device. The optical densities are translated into relative protein concentrations by a reference standard curve, obtained via an immunoblot. To establish a model for semiquantification of endothelial constitutive nitric oxide synthase (eNOS) protein, we compared the coronary arteries of WKY rats fed a normal chow diet to the coronary arteries of WKY rats fed a cholesterol diet. Using this technique, we have found a relative 130-fold decrease in eNOS in the cholesterol-fed group compared to the normal chow-fed group.
Keywords
E
Quantification of protein in small pathological specimens is difficult with conventional methods. ELISA techniques or Western blotting often requires sample pooling to obtain a specimen sufficient for accurate analysis. The visual scoring system (H score) employs the ability of scientists to discriminate among immunohistochemical staining intensities (Adams et al. 1991; Schultz et al. 1992). This method is subjective and may be inaccurate.
We have developed a semiquantitative technique that can discriminate among antibody/antigen binding intensities. We have shown for the first time the difference between endothelial NOS binding intensities of hypercholesterolemic coronary arteries vs control coronary arteries. The method involves routine immunohistochemistry with a chemiluminescent final stage, which is detected by X-ray film. The resulting images on the X-ray film are processed with a microcomputer imaging device (MCID) and the optical densities of the images are recorded. The optical densities are then translated to relative protein concentration via a standard curve obtained with an immunoblot.
Materials and Methods
Tissue Collection and Preparation
Ten 8-week-old WKY (normotensive) rats were divided into two groups of five. One group was fed a normal rat chow diet. The other group received a normal rat chow diet supplemented with 2% cholesterol (Sigma Chemical; St Louis, MO) and 8% peanut oil (GFW; Port Melbourne, Australia). After dieting for 15 weeks, rats were sacrificed and their hearts were excised and snap-frozen in a dry ice/isopentane bath. Specimens were kept at −70C until processed. A control rat heart and a cholesterol-fed rat heart were mounted on a single chuck. Three sections were cut and fixed per slide. If possible, tissues to be compared should be mounted on a single glass slide, so as to keep incubation and wash times constant.
Immunohistochemistry
Immunohistochemical staining of tissues was performed using a Dako Universal LSAB+ kit (Dako; Carpinteria, CA). Briefly, 8-μm tissue sections were cut with a cryostat and allowed to dry for 2 hr. A normal diet heart and a cholesterol diet heart were mounted on a single chuck and three sections were fixed per slide. This was repeated for each successive pair of hearts. The sections were fixed/dehydrated in acetone (Sigma) at room temperature (RT) for 10 min and allowed to dry. Once dried, the sections were placed in a plastic container with desiccant crystals and stored at −20C until needed. Sections were removed from −20C and allowed to equilibrate to RT before immunhistochemistry. Sections were rehydrated in 0.05 M Tris-HCl (Tris[hydroxymethyl]aminomethane hydrochloride) (Sigma), pH 7.4, for 5 min. Constitutive NOS antibody (Transduction Laboratories; Lexington, KY) and alkaline phosphatase (Sigma) were diluted in 0.05 M Tris-HCl, pH 7.4, containing 1% BSA (Sigma). A 20-min incubation in 20% goat serum (Dako; Carpinteria, CA) was used to decrease nonspecific binding. Sections were incubated with the eNOS antibody (0.1 μg/ ml) or negative control antibody (Dako) (0.1 μg/ml) for 30 min and rinsed in a 0.05 M Tris-HCl, pH 7.4, bath with gentle rocking for 10 min. A biotinylated anti-mouse antibody supplied with the LSAB+ kit was used to incubate the tissue sections for 5 min and then the sections were rinsed in 0.05 M Tris-HCl, ph 7.4, in a gentle rocking manner for 10 min. The following 5-min incubation used a streptavidin molecule conjugated to alkaline phosphatase (Sigma) at a 1:100 dilution. Subsequently, a 10-min rinse in 0.05 M Tris-HCl, pH 7.4, in a gentle rocking manner was performed. The tissues were then equilibrated for 2 min in 100 mM Tris-HCl, pH 9.5, containing 100 mM NaCl. Incubation with disodium 4-chloro-3-(methoxyspiro{1,2-dioxetane-3,2′-(5′-chloro)tricyclo{3.3.1.13,7}decan}-4-yl) phenyl phosphate (CDP-Star substrate; Boehringer–Mannheim Biochemica, Mannheim, Germany) (diluted 1:100 in 100 mM Tris-HCl, pH 9.5, 100 mM NaCl) for 5 min was performed to provide chemiluminescence. The substrate included 10 mM levamisole (Sigma) to inhibit endogenous alkaline phosphatases. The sections were placed in an oven at 20C to dry. The sections were then exposed for 30 min under Mamoray MR5 X-ray film (AGFA; Mortsel, Belgium). The following image was analyzed using an MCID.
Specimens were also processed using the LAB peroxidase method, according to the protocol provided in the LAB kit (Sally 1989), visualized microscopically, and compared to the image on X-ray film to positively identify coronary arteries. To obtain a reference standard curve to quantitate the relative concentration of eNOS, we used an immunoblotting technique. Halving concentrations of eNOS protein (75,000 pg-586 pg, triplicates) were absorbed onto Hybond N+ filter membrane (Amersham; Poole, UK) and dried for 10 minutes at 37C. The membrane was placed in a roller bottle and 5% low-fat skim milk solution was added for 30 min to inhibit nonspecific binding. The membrane was then rinsed thrice, 10 min each rinse, with Wash buffer [10 mM Tris-HCl, 150 mM NaCl, 0.05% Tween-20 (polyoxyethylenesorbitan monolaurate; Sigma)]. The membrane was then incubated with the eNOS antibody (0.1 μg/ml, 1% BSA) for 30 min and rinsed thrice, 10 min each rinse. Biotinylated goat anti-mouse antibody supplied with the LSAB+ kit was added to the membrane and incubated for 5 min. Again, the membrane was rinsed thrice, 10 min each rinse. The following 5-min incubation required a streptavidin molecule conjugated to alkaline phosphatase (Sigma) at 1:100 dilution. Subsequently, the membrane was rinsed thrice, 10 min each rinse. The membrane was placed in a bath containing 100 mM Tris-HCl, pH 9.5, 100 mM NaCl to equilibrate for 2 min. The membrane was then placed on clear, firm plastic film and incubated with CDP-Star substrate (Boehringer–Mannheim) diluted 1:100 in 100 mM Tris-HCl, pH 9.5, 100 mM NaCl, for 5 min and another clear, firm plastic film placed on top, to form a sandwich effect. This sandwich effect provides a uniform cover over the membrane. The membrane was then wrapped with clear plastic and exposed under X-ray film for 30 min. The resulting immunoblot was processed with an MCID and the optical density of each lane was recorded. The triplicate values per protein concentration were averaged, and the data (x axis) were plotted against the protein concentration (y axis). A computer graph program (Cricket Graph; Malvern, MacIntosh Computer, Malvern, PA) was used to calculate the standard curve formula
which was used to obtain the relative eNOS concentrations.
Microcomputer Imaging Device
The MCID was used to collect data in the following manner. The X-ray film was placed under the MCID camera and the corresponding microscope slide processed with the LAB-peroxidase method was placed under a microscope (low magnification, × 25). The camera height was adjusted to allow the image seen on the MCID monitor to be the same size as the image seen under the microscope. The relative light intensity on the MCID was adjusted to achieve a blue background and was increased until the negative control sections were not visible. The relative light intensity must not be changed during collection of data. The coronary arteries were then identified by scanning the microscope slide, and these arteries were assimilated to the coronary arteries seen with the MCID monitor. Once identified, the optical density of the endothelium was recorded by taking an outline of the endothelium.
Results
The typical eNOS immunoreactivity of coronary arteries, as seen with the MCID, is shown in Figure 1. The sections stained with negative control antibody were visible only under lower relative light intensity, because higher relative light intensities reduced negative control image. The higher relative light intensity was used in our experiment. Immunoblotting techniques were used to establish the linearity of the antigen/antibody binding intensity (Figure 2). Regression statistics of the data revealed an inverse log relationship between binding optical density and antibody concentration (r = 0.99; p < 0.0001, n = 3) (Figure 3). The formula
WKY rat coronary arteries showing ecNOS protein, using the chemiluminescent immunohistochemical method. (
Immunoblot obtained using the chemiluminescent detection system, showing the image gradient achieved by varying the cNOS protein concentration. The protein concentrations are shown (n = 3, p < 0.0001).
Graph showing the method's linearity. A computer graph program was used to calculate the standard curve (y = 4.2 × −6.5). The y axis shows the protein concentration and the × axis the binding intensity. Note the inverse log relationship between the actual values plotted and the real values that were assessed. Each value is the mean ± SE; n = 3.
where y = amount of protein in pg and × = optical density units was calculated via the computer graph program and can be used to quantify the relative amount of eNOS protein present in a particular area. For example: The optical density of a specific area chosen is 0.455, as given by the MCID. The inverse of this number is 2.2, and the log of 2.2 is 0.342. This number can then be used in the formula
The formula now reads
which computes to 1/y = 8.64 × 10-6. Finally, the inverse of 8.64 × 10-6 is 115,784, which equals the relative amount of eNOS protein in the specific area chosen, in picograms. Therefore, various areas of immunostaining can be semiquantified using an MCID. Coronary artery endothelial NOS binding density decreased by approximately 60% in the WKY rats fed a high-cholesterol diet compared to coronary arteries of rats fed a control diet (0.09 ± 0.01 compared to 0.30 ± 0.08). This corresponds to a relative 130-fold decrease in NOS protein in the coronary endothelium of the cholesterol diet group (0.15 ± 0.01 vs 20 ± 0.08); p < 0.001, n = 5.
Discussion
This article reports a novel method for immunohistochemical semiquantification. The method is simple, efficient, and rapid. Results are attainable in less than 3 hr. This method has vast potential applications, e.g., semiquantification of a protein in pathological tissues compared to normal tissues. We have also shown, for the first time, semiquantitative analysis of eNOS in coronary arteries of hypercholesterolemic rats.
Recently, Habib et al. (1996) employed a DAB/peroxide immunohistochemical method with nickel enhancement, in which two unbiased scientists randomly selected microscopic fields for quantification and the chosen areas were measured using gray-scaled computer-assisted image analysis. Nickel enhancement of DAB adds a blue tinge to the specimen, which is visible by microscopy. However, any spectral colors present in the specimen will affect the discriminative capabilities of the MCID, because the instrument utilizes a black-and-white camera to obtain an image from film or glass slide. The blue color present in nickel enhancement of DAB may decrease the contrasting representation of the actual antibody binding optical density, and this may lead to quantitative inaccuracies. For example, a blue tinge in a specimen on a glass slide may “seem” dark gray to an MCID, and therefore the MCID will process this as higher binding optical density, producing darker optical density in this area on the image visualized on the computer monitor screen. This will lead to inaccurate quantification. Our method provides an exposed film with a full gray-scale image, with no spectral color interference. The MCID can more accurately differentiate between image optical densities on film, and therefore this method is more acceptable for quantitative analysis. More importantly, we use an immunoblot to provide a reference standard curve to translate the antigen/antibody binding optical density to final relative protein concentrations.
Time-course studies for primary antibody incubations should be set for each laboratory, because various antibodies require different incubation times. Previous experiments in our laboratory show a 30-min primary antibody incubation time to be optimal. Antibody-antigen saturation must be avoided if suitable comparison between samples is required (McBride 1995). The main advantage of using a phosphatase/ chemiluminescent reaction over peroxidase/DAB color reaction is that the image can be intensified by the former, simply by extending exposure times. Increasing the link (biotinylated goat anti-mouse IgG) incubation time can also increase image intensity, but also increases nonspecific binding (results not shown). A 5-min link incubation step was optimal for our experiments. The benefit of using CDP-Star substrate over CSPD substrate (disodium 3-(4-(methoxyspiro{1,2-dioxetane-3,2′-(5′-chloro)tricyclo {3.3.1.13,7} decan}-4-yl) phenyl phosphate (Boehringer–Mannheim) is the increased sensitivity of the former. The manufacturers state a tenfold increase in sensitivity. We have shortened the exposure time to 30 min when we use CDP-Star substrate, whereas an overnight exposure time was used with CSPD substrate. Choice of X-ray film is also significant. Enhanced sensitivity is obtained by using Mamoray MR5 X-ray film, over Ektoscan B (Kodak; Coburg, Australia).
Acetone was the fixative of choice because of its gentle fixative properties. Usually, paraformaldehyde is used in immunohistochemistry, although it is a harsher fixative. Although chemiluminescent Western blotting techniques are used routinely, localization of protein of interest is lost and ample specimen is needed for protein extraction and electrophoresis. Our method provides full localization of protein and the ability to use small amounts of specimen.
An increasingly recognized feature of vascular disease is an early and significant imbalance between endothelium-derived vasoactive agents. An equilibrium among endothelium-derived vasodilators and vasoconstrictors is essential for vascular homeostasis. Endothelium-dependent vascular relaxation is impaired in humans and animals with hypercholesterolemia, and this is believed to be an initial step in atherogenesis. Previously, Osborne et al. (1989) suggested that the lack of endothelium-dependent relaxation in coronary arteries of cholesterol-fed rabbits may be due to a dysfunction in the synthesis or release of endothelium-derived relaxing factor (NO) from the endothelium. A decrease in NO production has also been shown to accelerate neointimal formation (Cayatte et al. 1994), which might contribute to fatty streak and atherosclerotic plaque formation. We have shown for the first time, immunohistochemically, a relative 130-fold decrease of NOS protein in the coronary artery endothelium of cholesterol-fed rats compared to coronary arteries of rats fed the control diet, suggesting a mechanism for the inhibition of endothelial NOS enzyme production in hypercholesterolemia. Whether or not hypercholesterolemia affects eNOS transcription and/or translation warrants further study.
It is important to note that this method can report only the protein concentration relative to the standard curve obtained using the immunoblot. Purified eNOS protein used in the immunoblot might be structurally different to the eNOS protein found in tissue, and this could affect the antigen/antibody binding affinity. In addition, although incubation and wash times are consistent in both immunohistochemistry and immunoblotting, an immunoblot provides optimal conditions for antigen/antibody binding, whereas, in immunohistochemistry, tissue fixation may interfere with the permeability of the antibody or may directly affect the eNOS protein motif. Therefore, we stress the importance of reporting results with the use of the term “relative concentration” according to the immunoblot data. Unfortunately, not all proteins are available commercially, and if extraction and purification of a commercially unavailable protein cannot be considered, the antibody itself can be used instead of the protein in an immunoblot to create a standard curve. This is not as accurate, although the basic characteristics of the procedure can be established. Furthermore, it is not necessary to repeat the immunoblot on a daily basis as part of the protocol, once the characteristics of the system are established. Immunoblots in our laboratory are consistent on a day-to-day basis. However, we advise repetition of the immunoblot if the experiments are undertaken over an extended period of time, because phosphatase activity will decrease.
In summary, the protocol used is as follows:
Quantitative immunohistochemistry
Rehydrate specimens in Tris buffer
Incubate 20% goat serum
Incubate 30 min primary antibody
Rinse 10 min in Tris bath
Incubate 5 min secondary antibody
Rinse 10 min in Tris bath
Incubate 5 min strep./alk.phos.
Rinse 10 min in Tris bath
Equilibrate in CPD-Star buffer 2 min
Incubate 5 min CPD-Star + levamisole
Dry and expose under X-ray film
Immunoblot
Prepare protein dilutions
Absorb protein onto membrane, 10 min
Dry membrane 37C/5 min
Incubate 5% skim milk 30 min
Rinse 3 × 10 min in wash buffer
Incubate 30 min primary antibody
Rinse 3 × 10 min in wash buffer
Incubate 5 min secondary antibody
Rinse 3 × 10 min in wash buffer
Incubate 5 min strep./alk.phos
Rinse 3 × 10 min in wash buffer
Equilibrate in CPD-Star buffer 2 min
Incubate 5 min CPD-Star
Dry and expose under X-ray film
In conclusion, we describe a novel semiquantitative immunohistochemical method for quantitation of protein in fresh frozen cryostat sections. The method involves the use of chemiluminesence and detection via X-ray film. The total procedure is completed in less than 3 hr, including X-ray film exposure time. An immunoblot is employed to create a standard curve, which is then used to translate optical density to relative protein concentration. The method is very sensitive, accurate, and rapid. Its potential applications are wide-ranging, because it is now possible to semiquantify protein production in small pathological specimens. This method provides a useful and convenient tool for quantitative studies in immunohistochemistry.