Abstract
Ten commercially available rabbit polyclonal anti-NOS antibodies were tested for their immunohistological applicability in normal human, guinea pig, rat, and mouse organs. Most antibodies reacted as expected and described in the literature with various tissues of the investigated species. Several antibodies did not react with the expected cell populations in a certain species, or reacted in previously unknown patterns. In addition, different antibodies to the same isoform rarely detected identical cell populations, even within one species. Most of these unexpected immunoreactivities were observed in bronchial epithelial, glomerular epithelial, and vascular smooth muscle cells. These unexpected results usually occurred when the antibodies were tested in other organs or species than that to which they were originally raised. We therefore strongly recommend the use of anti-NOS antibodies only after careful immunohistological and biochemical analysis of their reactivity in the organ and species to be studied.
T
The goals of our study were to establish the immunohistological reactivity patterns of the commercially available isoform specific anti-NOS antibodies in these species; to explain potential differences in staining patterns among these species; and to determine the applicability of commercially available anti-NOS antibodies for animal models.
Investigated commercially available rabbit polyclonal anti-NOS antibodies
ABR, Affinity Bioreagents (Golden, CO); SC, Santa Cruz Biotechnology (Heidelberg, Germany); TL, Transduction Laboratories (Lexington, KY).
Materials and Methods
Sources of Human, Guinea Pig, Rat, and Mouse Organs and Tissues
Human tissues were obtained from normal donor organs that were disapproved for transplantation for technical reasons, or from autopsy material obtained within 6 hr of death from a total of six different individuals. The tissue donors were not suffering from any kind of systemic disease and the organs used in this study were considered histologically normal.
The most commonly used laboratory animals and strains were selected for this study. Two mature male individuals of about 3 months of age in two strains of outbred guinea pigs were tested (DH and Harlan outbred). The rats and mice investigated in this study were all sexually mature male animals at the age of about 10 weeks (generously provided by Harlan; Horst, The Netherlands). Four commonly used mouse (Balb/c, C57Bl/6, DBA2, SE) and five commonly used rat (BN, LEW, F344, SD, WU) strains were investigated. To exclude intrastrain variability, four animals were tested per strain.
The immunoreactivities of the NOS antibodies with the above-mentioned species were tested on the cerebrum, cerebellum, lung, spleen, kidney, and liver. These organs were chosen because they all should demonstrate specific patterns of expression of at least one of the NOS isoforms. In addition, this choice of organs covers main areas of research interest and allows comparison with the majority of published data on NOS expression (Robinson et al. 1996). Human tissues were snap-frozen in isopentane immediately after resection. The organs from the guinea pigs, rats, and mice were removed under halothane anesthesia after sacrificing the animals. Bleeding was performed to remove as much serum as possible from the organs to minimize unwanted immunological crossreactivities. The removed organs were snap-frozen in isopentane and stored at -80C for the preparation of cryostat sections.
Antibodies to NOS Isoforms
In this study, a selection of widely used antibodies from three different commercial sources was tested. To evaluate the crossreactivity of the antibodies in several species and to compare the antibodies with each other, we chose to investigate only rabbit polyclonal antibodies to NOS isoforms (Table 1). The antibodies were titrated in the species to which they were directed on organs in which immunoreactivity could be expected. For example, the immunoreactivity of anti-human eNOS from Transduction Laboratories (Lexington, KY) was tested on human lung which contained enough vessels for proper judgment. The proper antibody concentration was the one with clearly distinguishable positive reactivity with the expected cells and lowest nonspecific background reactivity. In general, this titration resulted in comparable IgG concentrations of 1-10 μg/ml, and these concentrations were used throughout the study in all species.
Immunohistological Techniques
Frozen sections were cut on a cryostat microtome at 4 μm thickness. These sections were wrapped in aluminum foil and stored for approximately 2 weeks at -20C. Sections of several organs were collected on the same slide. After thawing, the sections were fixed in 100% acetone for 10 min at room temperature (RT), washed with PBS, and preincubated with 10% normal goat serum in PBS. After removal of the preincubation serum, the rabbit anti-NOS polyclonals were added in PBS containing 1% bovine serum albumin (BSA; Serva Feinbiochemica, Heidelberg, Germany) for 60 min at RT. After this, endogenous peroxidase was blocked by incubation for 30 min in 0.075% H2O2 in PBS. To detect bound antibodies, the slides were washed with PBS and then incubated for 30 min with affinity-purified, peroxidase-conjugated goat anti-rabbit antibodies (Dakopatts; Glostrup, Denmark) containing 1% (v/v) normal serum of the species from which the tissue was tested. Peroxidase reactivity was developed using 3-amino-9-ethylcarbazole (Aldrich Chemical; Milwaukee, WI) and H2O2, after which the slides were counterstained with hematoxylin. For identification of NOS reactivity in the human lung and brains from all species, double stainings with the anti-macrophage antibody CD68 (KP-1; Dakopatts) and vimentin (V9; Dakopatts) were performed, respectively. The latter antigens were identified with alkaline phosphatase-conjugated goat anti-mouse IgG antibodies (Southern Biotechnology Associates; Birmingham, AL) and the hematoxylin counterstain was omitted. The alkaline phosphatase label was developed with naphthol ASMX, Fast Blue BB, and levamisol (all from Sigma; St Louis, MO) in 0.1 M Tris-HCl (pH 8.2).
The sections were scored independently by two observers as strongly positive (++), weakly positive (+), or negative (blank) immunoreactivity.
Immunoblotting Technique
Immunoblotting using positive controls and tissue homogenates were used to test the specificity of the NOS antibodies. Lysates of IFN-γ- and LPS-stimulated mouse macrophages, human aortic endothelial cells, and a rat brain pituitary tumor (all from Transduction Laboratories) were used as positive controls for iNOS, eNOS, and nNOS, respectively.
Tissue homogenates were prepared as described previously for rat liver (Vos et al. 1997). Positive controls for NOS isoforms and tissue homogenates were fractionated on 10% SDS-PAGE and transferred to nitrocellulose (Amersham International; Poole, UK), using a semidry blotting system according to the manufacturer's instructions (Pharmacia; Uppsala, Sweden). Transfer of proteins was checked by Ponceau Red staining of the nitrocellulose filters after blotting. Molecular weight standards (BDH; Poole, UK) were used as marker proteins. The blots were incubated with polyclonal NOS antibodies in PBS containing 4% skim milk powder and 0.1% Tween-20, subsequently incubated with horseradish peroxidase-labeled swine anti-rabbit IgG (dilution 1:2000; Dakopatts), and finally developed using the ECL Western blotting system (Amersham).
Results
Immunohistology
Our results were generally in agreement with expected and previously published staining patterns in different species (Springall et al. 1992; Kobzick et al. 1993; Bredt and Snyder 1994; Dinerman et al. 1994; Tracey et al. 1994; Bachmann et al. 1995; Sancesario et al. 1996; Xue et al. 1996a, b), but important exceptions were found (see below). The Affinity Bioreagents antibodies all gave high background reactivity, making it difficult to titrate these antibodies. The following antibodies did not allow proper titration because of absence of specific immunoreactivity with the expected positive control cell population: anti-rat nNOS from Affinity Bioreagents; anti-human nNOS from Santa Cruz; anti-human eNOS from Santa Cruz; anti-mouse iNOS from Affinity Bioreagents; and anti-mouse iNOS from Santa Cruz. In the cases where there was no expected reactivity, we titrated the antibodies on cell types displaying specific positive immunoreactivity. For example, anti-human nNOS from Santa Cruz did not react with human neuronal cells but did react with smooth muscle cells in the vessel wall in all tissues of all species and with rat kidney glomeruli. The optimal concentration in the latter cell types was then titrated and chosen for further study.
The anti-rat nNOS antibody from ABR showed some reactivity in neuronal cells in the cerebrum and cerebellum of all tested species except the rat. This antibody did show reactivity with rat bronchial epithelium and diffusely with glomeruli in the rat kidney. The anti-human nNOS from SC reacted with the smooth muscle cell layer of all vessels in all species, in the human lung diffusely with the alveolar walls, in the rat lung with bronchial epithelial cells, and in the human kidney with parietal epithelial cells of the glomerulus. The anti-human nNOS from TL reacted only with neuronal cells in the human brain, and reacted diffusely with mouse kidney glomeruli and strongly with podocytes in guinea pig glomeruli (Figure 1A). Contrary to several reports, none of the antinNOS antibodies reacted with the renal macula densa (Tojo et al. 1994; Bachmann et al. 1995).
The anti-bovine eNOS antibodies from ABR showed, apart from high backgrounds, strong luminal reactivity in all vessels in all species, which was not restricted to the endothelium. The anti-human eNOS from SC did not show endothelial immunoreactivity with any of the tested tissues, but this antibody did show strong reactivity with mouse lung bronchial epithelial cells (Figure 1C). The anti-human eNOS antibodies from TL demonstrated endothelial reactivity only in human tissues and in mouse spleen and liver. This antibody also reacted strongly in the glial cell in the cerebrum and cerebellum of all species (Figures 1E and 1F) and sometimes with the smooth muscle cell layer of vessels.
The anti-mouse iNOS antibodies from ABR showed dot-like reactivity in all cells in the mouse lung and reactivity with some isolated cells in human organs. This antibody also showed weak reactivity with vessels in human tissues and was reactive only with the WU strain of rats, specifically with smooth muscle cells in the vessels and with endothelial cells in the liver. The anti-human iNOS from SC (649) showed reactivity with the glomerulus in human and with tubuli in the rat kidney. This antibody was also reactive with guinea pig endothelium and with human hepatocytes and macrophages in the spleen. The anti-mouse iNOS from SC (650) was strongly reactive with parietal epithelial cells of the glomerulus in the kidney of the SE mouse and in all mouse lungs with bronchial epithelial cells and nerves. This antibody was also strongly reactive with rat kidney tubuli, the red pulp in the spleen, and with hepatocytes. Weak reactivity was found with vessel walls in the human kidney and with guinea pig macrophages. The anti-human iNOS from TL was strongly reactive with human lung macrophages (Figure 1B), with guinea pig and mouse podocytes in the kidney, with bronchial epithelial cells of the guinea pig and rat (Figure 1D), and with mouse liver dendritic cells.
There were only small differences in intensity of the staining in the individual human specimens, and no differences could be detected among the individual animals of a particular strain. However, the aberrant staining patterns with several antibodies in the WU rat and the SE mouse, in comparison with other rat and mouse strains, demonstrates that interstrain differences in immunoreactivites may exist.
On the basis of our own experience as well as suggestions in the literature (Gonzalez-Hernandez et al. 1996), the antibodies were also tested on their positive control tissues using fixation with 2% and 4% buffered paraformaldehyde, but this generally resulted in a decrease or loss of immunoreactivity (not shown). In addition, amplification of the signals by third-step antibody incubations or the avidin-biotin method resulted in stronger but not other or more specific signals.
Selection of immunohistological staining results with NOS antibodies. Frozen sections of tissue were incubated with NOS antibodies and processed for peroxidase staining only (
Immunoblotting
Most antibodies detected the proper protein band in Western blots of the positive control preparations (Figure 2). The high background in immunohistology demonstrated by the Affinity Bioreagents antibodies against eNOS and nNOS therefore probably results from the lack of affinity purification.
Immunoblot showing reactivity of NOS antibodies against corresponding commercially available positive control preparations. SDS-PAGE and immunoblotting were performed as described in Materials and Methods. Antibodies used are as described in Table 1. Molecular weights were verified by running molecular weight markers in lanes adjacent to the positive controls.
Immunoblot showing reactivity of selected NOS antibodies against a selection of tissue homogenates. SDS-PAGE and immunoblotting were performed as described in Materials and Methods. Antibodies used are eNOS antibodies (upper panel) and iNOS antibodies (lower panel) from ABR (left) and TL (right). The position of the 56-kD molecular weight marker is shown to indicate the molecular weights of the proteins recognized by the antibodies. Lane 1, human lung; Lane 2, rat lung; Lane 3, rat kidney; Lane 4, mouse lung.
All antibodies against the iNOS isoform recognized a single band of approximately 130 kD, with the exception of the Santa Cruz antibody against human iNOS (#649). This antibody is raised against a C-terminal peptide of the human sequence and may not recognize the murine iNOS present in the positive control owing to the low homology between the murine and human iNOS in this region. The antibody from Transduction Laboratories against eNOS recognized a single band of approximately 140 kD in the endothelial cell lysate. In contrast, the anti-eNOS antibody from Santa Cruz did not react at all with the endothelial cell lysate, whereas the anti-eNOS antibody from ABR recognized several low molecular weight proteins but not a protein at the proper molecular weight. Finally, the antibodies against nNOS from Transduction Laboratories and Santa Cruz reacted with a single protein of approximately 155 kD in the rat pituitary lysate. In contrast, the anti-nNOS from ABR predominantly recognized lower molecular weight proteins.
In contrast to the satisfactory Western blot results with the commercially available positive control lysates, we rarely found the expected molecular weights using tissue homogenates. This was unexpected because the tissue homogenates comprised human, rat, and mouse lung and kidney, which were selected on the basis of positive immunohistological results with eNOS and iNOS antibodies from ABR and TL. As shown in Figure 3, the NOS antibodies, particularly the eNOS antibodies, recognized a variety of proteins in different tissue homogenates. Most of the recognized proteins were around or below the 56-kD molecular weight marker. The pattern of recognized proteins in, e.g., tissue homogenates, varied among species and among the sources of antibodies. This is illustrated in Figure 3 for lung homogenates and eNOS antibodies. This inconsistency was also seen, albeit to a lesser extent, with the iNOS antibodies.
Discussion
This report describes for the first time a thorough comparison of the immunoreactivity of several commercially available antibodies to NOS in human, guinea pig, rat, and mouse tissues. Most antibodies reacted as expected in the species from which the antigen used for immunization was originally derived, in a pattern characteristic for the cell types known to express the specific isoform. In addition, most antibodies identified a protein with the expected molecular weight in the Western blots with the commercially available positive control extracts. The demonstration of specificity of antibodies using immunohistological and immunoblotting techniques is normally considered sufficient for further research applications. In our study, however, we also observed unexpected crossreactivities and recurring patterns of aberrant immunoreactivities that are discussed below.
First, anti-NOS antibodies detect different structures in different species. For example, we found, as expected, that anti-human nNOS from Transduction Laboratories recognized human neuronal cells. However, the reactivity of this antibody with guinea pig glomerular epithelial but not neuronal cells is striking and is difficult to explain. If the anti-human nNOS antibody from Transduction Laboratories would be species- specific, causing absence of neuronal recognition in the guinea pig, then it remains to be explained why the antibody reacts so strongly with guinea pig glomeruli. This explanation goes beyond the scope of this study and will be the subject of further study (manuscript in preparation). It could imply that a humanlike nNOS is expressed in glomerular epithelial cells in the guinea pig or that this antibody recognizes a NOS different from nNOS in guinea pig tissues. Interestingly, similar glomerular staining patterns can be obtained with the anti-mouse iNOS from Transduction Laboratories in the mouse kidney. Another example of species-specific differences in staining is the antihuman iNOS from Santa Cruz which showed weak renal apical tubular reactivity in the rat, whereas in the guinea pig these antibodies reacted with endothelial cells.
Second, unrelated antibodies can recognize identical structures in different species. Bronchial epithelial cells in the rat are recognized by the iNOS antibody from Transduction Laboratories. The same cell type in the rat is also recognized by the nNOS antibody from Santa Cruz. However, the latter antibody does not recognize neuronal cells in the rat brain. The iNOS antibody from Transduction Laboratories also recognizes bronchial epithelial cells in the guinea pig, but only in the apical region of the cells. Other investigators have described the apical expression of eNOS in rat (Xue et al. 1996a) and human (Shaul et al. 1994) bronchial epithelial cells, but it is unlikely that the anti-mouse iNOS from Transduction Laboratories would recognize eNOS in the guinea pig becauses no endothelial reactivity was identified in guinea pigs with this antibody. In contrast, the anti-mouse iNOS from Santa Cruz does react with endothelial cells in the guinea pig, suggesting that eNOS recognition by iNOS antibodies may occur.
Third, different antibodies raised against the same isoform of the same species rarely result in similar staining patterns. Specifically, the anti-human nNOS from Transduction Laboratories identifies neuronal cells in human neuronal tissues but the anti-human nNOS antibody from Santa Cruz fails to detect these cells, even in high concentrations. However, immunoblotting of the latter antibody with rat pituitary extract demonstrated its ability to recognize nNOS. Whether or not the specific vascular immunoreactivity of this Santa Cruz nNOS antibody should be interpreted as recognizing neuronal or another NOS isoform in the vasculature remains to be established.
Finally, we cannot explain the recurrent patterns of the aberrant immunoreactivities. Bronchial epithelial, glomerular epithelial, and vascular smooth muscle cells were remarkably often the cell types that reacted unexpectedly with the antibodies. The significance of this finding remains to be established.
These inconsistencies in immunohistological results were paralleled by the immunoblotting results. Although we performed our blots on a far from complete selection of tissues, it was nevertheless unexpected that in none of the lysates proper molecular weights were detected, despite their selection on the basis of positive immunohistology. The main reason for this may be that in normal, nondiseased tissue homogenates the dilution of NOS proteins is such that it remains below the detection level of the Western blots. We have increased the sensitivity of the Western blots to be able to see bands at all, but the observed bands are not necessarily proof of lack of specificity. It could very well be possible that specific NOS proteins can be detected with Western blots only in homogenates of purified cell populations or under induced conditions (Vos et al. 1997).
In general, the unexpected results occurred when the antibodies were used to detect their corresponding antigens in other species or organs than those to which they were originally raised. Until recently, the expression patterns of NOS isoforms appeared rather simple. Neuronal and endothelial NOS were constitutively expressed in neuronal and endothelial cells, respectively, and inducible NOS was expressed in activated macrophages and was therefore also called macrophage NOS (Stuehr et al. 1991; Bredt and Snyder 1994). The expression patterns of the NOS isoforms were subsequently found to be less exclusive (Dinerman et al. 1994; Shaul et al. 1994; Tojo et al. 1994; Gonzalez-Hernandez et al. 1996; Sancesario et al. 1996). Thus, iNOS was identified in and cloned from hepatocytes (Geller et al. 1993), vascular smooth muscle cells (Geng et al. 1994), and chondrocytes (Charles et al. 1993). In addition, co-expression of different NOS isoforms in a single cell type was described (Dinerman et al. 1994). The significance of our unexpected antibody reactivities in other species therefore remains to be established.
The potential relevance of NOS expression in many (patho)physiological situations has led to an enormous number of publications in the past few years. Several of these studies demonstrated the expression of NOS isoforms by immunohistochemical techniques, using the same antibodies as those we tested in this study. Frequently, the specificities of these antibodies were demonstrated in Western blots on extracts of other cell types and/or other species than were studied in immunohistochemistry. Our results demonstrate that the detection of a specific band on purified positive control lysates does not justify immunohistochemical application on tissues without further study on tissue homogenates.
In conclusion, our findings demonstrate that (commercially available) NOS antibodies should be used carefully and critically. We discourage the use of anti- NOS antibodies for research applications on species and organs other than those to which these antibodies were raised. We recommend the following strategy to verify the immunohistological applicability of anti-NOS antibodies. Only if the immunohistological results on a certain tissue corresponds with the expression as well as the proper molecular weight of the protein in immunoblotting on its homogenate, do we believe that it is safe to use the antibody. If immunoblotting does not confirm the immunohistological result, we recommend the use of preferably antibody-independent techniques such as mRNA detection to verify the results. Alternatively, if an antibody-independent technique cannot be used, the tissue homogenates under study should be enriched for the studied cell type, e.g., for iNOS studies after or followed by induction of enhanced expression with LPS and cytokines. Again, only when the bands in these blots have the proper molecular weight and confirm the expression can the antibodies be applied in immunohistochemistry.
Footnotes
Acknowledgments
Supported in part by grant #95.24 from the J.K. de Cock Foundation (Groningen, The Netherlands). The rats and mice used in this study were generously provided by Harlan B.V. (Horst, The Netherlands). We appreciate the generous gift of the anti-NOS antibodies from Santa Cruz Biotechnology (Heidelberg, Germany). The guinea pigs used in this study were kindly provided by Martin Schuiling (Department of Molecular Pharmacology, Groningen, The Netherlands).
We would like to thank Nathalie Nauta, Msc, and Annemiek van Oosten, Msc, from the Department of Pathology for their contributions. Sippie Huitema and Hans Vos from the Department of Pathology are greatly appreciated for generating important results in pilot experiments and their continuous technical advice. We are grateful to all members of the CARA Research Group Groningen, especially Nick ten Hacken, MD, PhD, for helpful and stimulating discussions.