Abstract
Eighteen commercially available antibodies reactive against different cytokeratin proteins were tested on classic examples of lobular intraepithelial neoplasia (LIN) and of ductal intraepithelial neoplasia (DIN) of the breast. About 90% of higher-grade DIN (AIDH and DCIS) show no or substantially diminished reaction with clone 34βE12 (specified as reactive against keratins 1, 5, 10, and 14 as determined by the manufacturer), while the cells of LIN were found to express the antigen reactive with this antibody. To determine which of these four keratins are present in the cells of LIN, antibodies reactive against these individual four keratins were tested. None of the four antibodies to keratins 1, 5, 10, or 14 reacted with the cells of LIN. To investigate this further, 13 additional monoclonal antibodies to various other keratin proteins were tested on the cells of LIN. Those that successfully reacted with the cells of LIN were further tested on the cells of DIN. All of the individual antibodies reactive with the cells of LIN were also reactive with the cells of DIN to a degree, with clone RCK108 (reactive against keratin 19) coming the closest to demonstrating the reactivity seen with 34βE12. We conclude that the reactivity seen in the cells of LIN with 34βE12 is due to either (a) a crossreaction with keratin 19 that is slightly less prominent than the reaction of the individual clone RCK108, (b) a crossreaction with a keratin protein that was not tested (3, 11, 12), (c) a crossreaction with a protein closely resembling keratin in formalin-fixed, paraffin-embedded tissue, or (d) the detection of a mutated or truncated form of keratin 1, 5, 10, or 14 that cannot be detected by the individual monoclonal antibody.
Keywords
L
This study was designed to determine which of the four keratin proteins identified by 34βE12 were detected in the cells of LIN. To that end, we tested individual antibodies reactive against keratins 1, 5, 10, and 14 on classic LIN. To our surprise, none of these reacted with the cells of LIN. We then obtained monoclonal antibodies (MAbs) to an additional 13 keratin proteins to determine if there was a previously unrecognized or undetected crossreaction of 34βE12 with another keratin protein in formalin-fixed, paraffin-embedded tissue. We tested these antibodies against not only the cells of LIN but also the cells of ductal intraepithelial neoplasia (DIN), variants that were shown to be nonreactive with the clone 34βE12 in our previous studies (Moinfar et al. 1999). If 34βE12 is crossreacting with another keratin protein, it should be one that is present in the cells of LIN but absent in the cells of higher-grade DIN (ductal carcinoma in situ).
Monoclonal antibodies used: dilutions, pretreatments, and sources
Materials and Methods
Of the sixty LIN specimens reactive with 34βE12, five were selected for testing with individual antibodies to keratins 1, 5, 10, and 14. Of these, two were selected for examination with the 13 other keratin MAbs. Sections were selected for the quantity of lesion and the classic reactivities noted with 34βE12.
The tissue preparation and IHC method were the same as described in our earlier study (Bratthauer et al. 2002). Briefly, sections were pretreated with heat-induced epitope retrieval using a pressure cooker and Reveal epitope retrieval buffer (Biocare Medical; Walnut Creek, CA) for 3 min. Antibody detection was with peroxidase ABC (Vector Laboratories; Burlingame, CA), followed by diaminobenzidine (DAB)/H2O2 (Sigma Chemical; St Louis, MO). For some of the antibodies, an enzyme digestion method of pre-treatment was employed. Sections were deparaffinized in xylenes and rehydrated through ethanol. Once in buffer, they were subjected to protease VIII digestion for 2 min. The oxidative quenching and antibody applications were the same as before. In addition, some antibodies were tested after heat pretreatments in an environment containing EDTA using Trilogy epitope retrieval buffer (Cell Marque; Hot Springs, AR) in the pressure cooker. Sections were assayed for both 34βE12 and all of the individual MAbs tested (Table 1). Results of the reactivities were assessed in the cells of both LIN and DIN lesions. Negative controls were performed as above on like sections with normal mouse IgG in substitution for primary antibody. Positive controls included sections of skin when individual keratin proteins were not identified in normal breast tissue.
Results of keratin clone applications to the cells of LIN, DIN, and normal TDLU a
LIN, lobular intraepithelial neoplasia; DIN, ductal intraepithelial neoplasia; TDLU, terminal duct lobular unit.
Luminal epithelial cells.
Myoepithelial cells.
Results
The results are shown in Table 2. None of the MAbs tested matched the exact reactivity seen with 34βE12 (Figure 1). Antibodies reactive against keratins 4, 7, 8, 18, and 19 all demonstrated antigen in the cells of LIN. Antibodies reactive against keratins 1, 2e, 5, 6, 9, 10, 13, 14, 15, 16, 17, and 20 did not react with the cells of LIN. Antibodies reactive against keratins 4, 7, 8, and 18 all reacted strongly with the cells of higher-grade DIN as well as the cells of LIN.
Among the 17 antibodies tested in addition to 34βE12, the one reactive against cytokeratin 19 (RCK108) came the closest to duplicating the reactiveity seen with 34βE12 for LIN and DIN cells (Figure 2). The reaction product seen in the cells of LIN was the typical intense perinuclear pattern previously described. Unlike clone 34βE12, there was some reactivity seen in the cells of higher-grade DIN with RCK108, but this reactivity was significantly weaker than that seen in the epithelial cells of the normal terminal duct lobular unit. In some samples the intensity of this reactivity in DIN did vary but was always significantly less intense than that for cytokeratin 8.
(
(
All antibodies used in this study reacted with epithelial antigens in formalin-fixed, paraffin-embedded tissue control sections and, in some cases where antigen was present, reacted with breast epithelial cells in sections containing LIN. This was especially true for the MAbs reactive against keratins 5 and 14. All negative control sections were nonreactive.
Discussion
Cytokeratin antibody clone 34βE12, reactive against the high molecular weight cytokeratins 1, 5, 10, and 14, has been widely used in pathology laboratories worldwide since its development in 1982 (Gown and Vogel 1982). The proteins recognized by this antibody are absent or downregulated in a majority of atypical ductal epithelial cells of the breast. However, the cells of LIN, usually characterized by lack of the E-cadherin protein, consistently react with 34βE12 with the use of a very sensitive heat-induced epitope retrieval technique. The cells of LIN failed to react with individual antibodies against the high molecular weight cytokeratins recognized by 34βE12.
The individual MAbs to keratin proteins 1, 5, 10, and 14 all recognized an antigen present in normal epithelial cells in formalin-fixed, paraffin-embedded tissue control sections. In addition, the antibodies reactive against keratins 5 and 14 both reacted with the myoepithelial cell layer present in sections containing LIN, yet failed to react with the cells of LIN. Antibody 34βE12, reactive with a common epitope on keratins 1, 5, 10, and 14, reacted strongly in the myoepithelial cell layer as well as the cells of LIN. If a given antibody reactive with keratin 5, for example, is able to detect that protein in one cell type in a section, it should also be able to detect that protein wherever it might exist in other cell types in the section.
Because antibody 34βE12 reacts with a shared epitope, it is possible that this epitope is also present to a degree on some of the other keratin proteins. To assess this possibility, individual antibodies reactive in paraffin-embedded, formalin-fixed tissue against most of the keratin proteins were tried. None of these duplicated the reaction in either quality or extent observed with the 34βE12 antibody in the cells of LIN and DIN. The antibody clone RCK108, specific to cytokeratin 19, most closely approached the reaction pattern of 34βE12, showing intense reactivity in the cells of LIN but slightly more reactivity than anticipated in the cells of DIN. It is therefore difficult to state unequivocally that the reactivity seen with 34βE12 in LIN is due to a crossreaction with keratin 19. This is because a slight reaction with RCK108 (anti-cytokeratin 19) was noted in higher-grade DIN and the protein detected in LIN using 34βE12 went undetected in higher-grade DIN with that antibody. It is remotely possible, however, that the difference in the intensity of the reactivities observed may be due to the difference between a stronger immunoreaction occurring with a perfect idiotype-epitope match (RCK108 and keratin 19), and a weaker immunoreaction occurring with a crossreacting epitope (34βE12 and keratin 19).
In theory, it is possible, although unlikely, that the crossreacting keratin protein identified in the cells of LIN is one to which a usable antibody has not been developed. The only proteins we did not assess directly were keratins 3, 11, and 12. Of these, keratins 3 and 12 are cornea-specific and keratin 11 is believed to be restricted to keratinizing epidermal squamous cells. None of these keratins is likely to be present in LIN of the breast.
It is also possible that antibody 34βE12 recognizes an epitope on one of the four keratin proteins it defines (or others), that is conserved in the protein existing in LIN, but that the epitope to which the individual MAb reacts is somehow missing in these cells. This could be the case if one of those keratin proteins were truncated or mutated at the epitope to which the individual MAb reacts. The clone LL002 recognizing keratin 14, for example, was developed against only the last 18 amino acids. There are many reports in the literature concerning the truncation and/or mutation of keratin proteins in genetic or acquired diseases of the skin (Livingston et al. 2001; Ku and Omary, 2001; Reichelt et al. 1997). It is possible that a similar problem may affect patients with LIN, resulting in the detection of a protein at one but not another epitope.
Another possibility is the existence of a related but undescribed keratin or a non-keratin protein that exists in LIN. However, clone 34βE12 did not react unexpectedly in normal or tumor tissue from over 45 different cell types and morphologies. Moreover, no literature reports of non-keratin crossreactivity exist in 20 years of steady use. This suggests that this putative non-keratin protein might exist only in the cells of LIN; although not impossible, this is unlikely.
The possibility of technical artifact could not be excluded because methodology can enhance or hinder expression, especially when formalin-fixed tissue is used. In fact, examination of the cells of LIN with 34βE12 using an enzymatic digestion retrieval technique failed to demonstrate these proteins, as was mentioned in the earlier study (Bratthauer et al. 2002). However, when various tissue controls other than breast containing epithelial as well as other cell types were used, the pattern of reactivity seen with 34βE12 was consistent regardless of which antigen enhancement procedure was used. Therefore, we conclude that the clone recognizes proteins in epithelial-type cells (presumably keratins) and does not recognize protein found in any other cell type (hematological, neural, mesenchymal, myoid, or glial) regardless of the technique used. A difference in sensitivity is seen with various techniques, and the ability to uncover an epitope otherwise hidden is the purpose of applying new tissue preparation methods.
The existence of unrecognized reactivity is always a concern in IHC. A crossreactive epitope can be easily identified when the reactivity is found in an unexpected cell type or location. For example, reaction with antibody to IgG is anticipated in plasma cells but would be unexpected in mammary epithelial cells. As far as we can tell, the clone 34βE12 reacts solely with cells of epithelial/myoepithelial origin. If there are proteins other than the keratins with which this antibody reacts that exist only in epithelial cells, they would have to be identified through absorption or Western blotting experiments on fresh tissue. The protein that is identified through the use of antibody 34βE12 in the cells of LIN does not exist in the cells of DIN and does not react with individual MAbs raised against the component keratin proteins reactive in paraffin-embedded, formalin-fixed tissue.
The use of antibody 34βE12 with an antigen recovery system using heat rather than enzymes provides information of significant value in distinguishing LIN from DIN. To refer to the reaction product as high molecular weight cytokeratins is not incorrect because the manufacturer and the literature support those proteins as being identified through the use of this antibody. However, as with any IHC procedure, one can only factually state that a given reactivity is one that is antigenically similar to the immunogenic antigen. Certainly, with an antibody that is by definition nonspecific in that it recognizes common epitopes on more than one antigen, the specter of crossreactivity is present. However, the common epitope in this case is one that exists on very, very similar proteins (the keratin protein family) and has not been detected on any other protein as examined by cell type and location.
We conclude that the protein being detected in the cells of LIN with clone 34βE12 is due to either (a) a crossreaction with keratin 19 that is slightly less prominent than that seen using antibody clone RCK108, (b) a crossreaction with a keratin protein that was not tested (3, 11, 12), (c) a crossreaction with a protein closely resembling keratin in formalin-fixed, paraffin-embedded tissue, or (d) the detection of a mutated or truncated form of keratin 1, 5, 10, or 14 that cannot be detected by these individual MAbs. Further studies will be needed to determine the 34βE12 reactive material in LIN. The use of fresh tissue or frozen sections containing LIN will have to be used to conduct immunoprecipitation studies or Western blotting experiments to better identify the keratin subtype (if any) present in these cells.