Characterization of an antigen expressed by human natural killer cells

A monoclonal antibody, anti-N901, was produced by fusing NS-1 myeloma cells with spleen cells of a mouse immunized with human CML cells. This antibody was reactive with a subpopulation of peripheral blood LGL, including the natural killer cells. Monocytes, granulocytes, B cells, T cells (T3+ cells), erythrocytes, and platelets were nonreactive. The N901-positive cells in the peripheral blood were heterogeneous with respect to expression of other cell surface antigens. The majority of N901+ cells co-expressed T11, Mo1, and HNK-1, whereas a smaller percentage expressed T8. Ia, T3, T4, Mo2, or B1 antigens were very uncommon on N901+ cells. The heterogeneity of the N901+ LGL was further investigated by examining the expression of N901 antigen on a series of cloned normal human NK cell lines. N901 antigen was expressed by each of the NK cell lines tested, and by a minority of cloned T cell lines without NK activity. Anti-N901 does not block NK activity and can be used to rapidly purify functional NK cells for further study.     

Characterization of natural killer cells induced in the peritoneal exudates of mice infected with Listeria monocytogenes: a study of their tumor target specificity and their expression of murine differentiation antigens and human NK-associated antigens

That M1/70, a monoclonal anti-murine macrophage antibody, recognizes murine natural killer cells (NK) and that there is an increase in NK following intraperitoneal infection with live Listeria monocytogenes (LM) was previously reported. Here, LM-induced NK cells were further characterized with respect to tumor target specificity and the expression of murine mast cell, mononuclear phagocyte, and lymphocyte differentiation antigens plus human NK-associated antigens. The M1/70-selected NK (Mac 1 NK) lysed Yac 1, RLmale 1, and WEHI 164.1, but not EL 4 or WTS cells. Immunoprecipitation with M1/70 demonstrated that Mac 1, the antigen recognized by M1/70, was present on NK and thioglycollate-elicited macrophages. Contaminating macrophages in the NK-enriched population did not account for the immunoprecipitated Mac 1. Mac 1 NK that lysed Yac 1 displayed Qa 5, LFA 1, asialo GM 1, Ly 5.1, and NK 1.2, but not Lyt 1, Lyt 2, Mac 2, Mac 3, or Mac 4. Thirty percent of these Mac 1 NK bore Thy 1.2. The presence of Thy 1.2 did not correlate with individual lytic efficiency or cell cycle. Antibodies to human NK antigens Leu 7, Leu 11a, and Leu 15 did not recognize LM-induced NK cells.     

Suppressor cell function of human granular lymphocytes identified by the HNK-1 (Leu 7) monoclonal antibody

The HNK-1 (Leu 7) differentiation antigen defines a subpopulation of human granular lymphocytes with natural killer (NK) and K cell function. In this study, we investigated whether HNK-1+ cells, identified with the monoclonal antibody and purified with a fluorescence-activated cell sorter (FACS), could function as suppressor cells. The results demonstrated that purified HNK-1+ cells efficiently suppressed both PWM-induced IgG production by B cells and T cell proliferation in mixed lymphocyte reactions (MLR). Manifestation of this suppressor cell activity required immune complex activation and was partially sensitive to 2000 rad irradiation. This suppressor cell activity was predominantly mediated by a subset of HNK-1+ cells that have previously been shown to have maximum NK function and lack expression of the E rosette (ER) receptor and T cell antigens (e.g., T3 and T8). Thus, HNK-1+ER- cells suppressed a MLR by an average 52%; HNK-1+ER+ were one-half as efficient, causing an average 23% suppression. For comparison, we also examined the characteristics of Leu 2a+ suppressor T lymphocytes. In contrast to HNK-1+ cells, unactivated Leu 2a+ cells suppressed both B and T cell responses. This suppressor activity was not augmented by immune complex activation and was absolutely radio-sensitive in PWM assays. HNK-1+ cells, especially the HNK+ER- subset, can therefore mediate suppressor cell function in addition to their spontaneous cytotoxic function. Furthermore, some of their suppressor cell properties are distinct from those attributed to other types of suppressor lymphocytes.     

Human melanoma-associated antigen expression on human neuroblastoma cells: Effects of differentiation inducers

Summary We have described two human melanoma-associated antigens (HMAA), recognized by the murine monoclonal antibodies LS62 and LS109. LS62 recognizes the neuroglandular antigen (NGA), which is overexpressed in neoplastic melanocytes as well as in several tissues of neuroectodermal origin. These antibodies were used to screen six neuroblastoma cell lines and one neuroepithelioma cell line. A melanoma cell line, G361, known to express the two antigens, was used as the positive control. Variable expression of the two antigens was detected in neuroblastoma cells. The surface expression of NGA and of the LS109 antigen was modulated in parallel with the morphological differentiation induced by retinoic acid, 5-bromodeoxyuridine, or cyclic AMP analog/activators. The modulation of the expression of the two HMAA was detected in G361 melanoma cells and in one of the neuroblastoma cell lines, SK-N-SH. These results suggest altered expression of both antigens during melanoma and neuroblastoma cell differentiation in culture.     

Expression of the HNK-1/NC-1 epitope in early vertebrate neurogenesis

Summary A family of glycoconjugates has recently been shown to share a common carbohydrate epitope recognized by the mouse monoclonal antibody HNK-1. The specificity of HNK-1 was found to be similar to that of another monoclonal antibody, NC-1. These two IgM monoclonal antibodies were raised after immunization of mice with a human T-cell line and avian neural crest-derived ganglia, respectively. The antigens recognized by these antibodies include the myelin-associated glycoprotein, MAG, a glycolipid of defined structure, and a set of molecules involved in cell adhesion. The timing and pattern of appearance of these antigens are distinct. Moreover, the epitope may be absent on an antigen at a given stage or in a given tissue. Therefore, although the molecules able to carry the NC-1/ HNK-1 epitope are numerous and expressed in various tissues, the use of the monoclonal antibodies on tissue sections has proven adequate for following the migration of avian neural crest cells, the major cell lineage recognized by NC-1 and HNK-1 during early embryogenesis. Analogies in several other species have been found on the basis of HNK-1 reactivity. In this study we show that NC-1 and HNK-1 can be used successfully to label migrating neural crest cells in dog, pig and human. On the other hand, the NC-l/HNK-1 epitope was not present on migrating crest cells in amphibians or mice and was found only transiently on the neural crest of rats.     

Human Neuroectoderm-Derived Cell Line Secretes Fibronectin that Shares the HNK-1/10C5 Carbohydrate Epitope with Neural Cell Adhesion Molecules

A human malignant melanoma cell line, Melur, secretes several glycoproteins that contain a unique carbohydrate epitope shared by neural cell adhesion molecules and recognized by the monoclonal antibodies HNK-1, L2, and 10C5. In this report, we present evidence that one of the major melanoma glycoproteins containing the HNK-1/10C5 epitope is the cell adhesion molecule, fibronectin, or a fibronectin-like molecule. Melanoma-derived fibronectin was isolated from serum-free conditioned medium by gelatin-Sepharose affinity adsorption and shown to react with monoclonal antibodies HNK-1 and 10C5 in Western blot analysis. HNK-1-containing fibronectin was purified on a gelatin-Sepharose column followed by an affinity column using a monoclonal antibody against the HNK-1 carbohydrate. The purified HNK-1-fibronectin then could be incorporated into the extracellular matrix of hamster fibroblasts in vitro, and such a matrix was detectable using the HNK-1 monoclonal antibody in an immunofluorescence assay. Of the seven neuroectoderm-derived tumor cell lines tested, only the Melur melanoma cell secreted fibronectin containing the HNK-1 carbohydrate. Identification of human neuroectoderm-derived fibronectin as a potential carrier of the HNK-1 carbohydrate suggests a new role for fibronectin in neural development and regeneration, and represents a new model for studying the function of this carbohydrate domain in neural cell adhesion.     

Characterization of a unique glycoprotein antigen expressed on the surface of human neuroblastoma cells

In order to develop a molecular probe to delineate chemical and biological characteristics of human neuroblastoma cells, a murine monoclonal antibody (Mab 5G3) was produced that is directed to a glycoprotein, preferentially expressed on the surface of such cells. This antibody is of IgG2a isotype, has an association constant of 8 X 10(9) M-1, and reacts preferentially with human neuroblastoma cell lines and fresh frozen tissue sections in enzyme-linked immunosorbent assay and immunoperoxidase assays, respectively. Minimal reactivity is observed with a variety of lymphoblastoid cell lines and normal fetal and adult tissues. Mab 5G3 specifically recognizes a neuroblastoma target glycoprotein antigen of 215 kDa that is derived from a 200-kDa precursor, as evident from pulse-chase biosynthetic studies. Treatment with tunicamycin revealed that both molecules contain N-asparagine-linked oligosaccharides; however, only the 215-kDa species is resistant to treatment with endo-beta-N-acetylglucosaminidase H and sensitive to neuraminidase, indicating that it contains trimmed and terminally sialylated oligosaccharides of the "complex" type. In contrast, the 200-kDa precursor is sensitive to endo-beta-N-acetylglucosaminidase H and resistant to neuraminidase treatment indicating that it contains high-mannose non-processed oligosaccharides. The 215-kDa molecule is sulfated, phosphorylated at serine residues, and expressed on the cell surface. A molecule of 200 kDa is detected by Mab 5G3 in spent culture medium of human neuroblastoma cells which is neither sulfated nor phosphorylated.     

Phenotypic characterization of Ewing sarcoma cell lines with monoclonal antibodies

The histogenesis of Ewing sarcoma, the second most frequent bone tumor in humans, remains controversial. Four Ewing cell lines were analyzed by immunological methods. A panel of antibodies directed to T, B, and myelomonocytic markers gave negative results. Surface antigens recognized on Ewing cells were found to be related to the neuroectoderm lineage. Ganglioside GD2, a marker of neuroectodermal tissues and tumors, was present on all lines. These were also stained by the mouse monoclonal antibody HNK-1, which detects a carbohydrate epitope present on several glycoconjugates of the nervous system, including two glycoproteins, the myelin-associated glycoprotein and the neural cell-adhesion molecule (N-CAM), and an acidic glycolipid of the peripheral nervous system. The P61 monoclonal antibody, which reacts with a peptide moiety of N-CAM, and a rabbit antiserum, raised to purified mouse N-CAM and not recognizing the HNK-1-defined epitope, were also reactive. By contrast, all antibodies specific for hematopoietic cell surface antigens were totally negative. Besides these antigenic features, Ewing sarcoma cells are characterized by a specific t(11;22)(q24;q12) translocation also observed in neuroepithelioma, a neuroectodermal tumor, suggesting a possible evolutionary related origin. The recent finding that the human N-CAM gene is located at the vicinity of the breakpoint on chromosome 11 indicates that it might be involved in genetic rearrangements occurring in this region.     

A phenyl-beta-galactoside-specific monoclonal antibody reactive with murine and rat NK cells

The phenyl-beta-galactoside (phi-beta-gal)-specific monoclonal antibody (mAb) 49H.8 cross-reacts with the terminal disaccharide structure of the asialo GM1 (AGM1) molecule. It was found to react with phi-beta-gal determinants on murine and rat splenic natural killer (NK) cells, as measured by complement depletion studies. Flow cytometric analysis identified the antigen on two IL 2-dependent cloned murine NK cell lines and the rat large granular lymphocyte leukemia RNK. We have compared the 49H.8 reactivity to that of anti-AGM1 antisera (alpha-AGM1) on NK cells and a panel of NK related killer cells, including bone marrow-derived killer cells, lymphokine-activated killer cells (LAK), and anomalous killer cells (AK). We found that the 49H.8 specificity closely paralleled that of alpha-AGM1. When tested against Con A-reactive T cells, the 49H.8 mAb was less reactive than the alpha-AGM1, indicating that it may be a more specific marker for splenic NK populations than the alpha-AGM1.     

Systemic administration of human leukocyte interferon to melanoma patients. III. Increased helper:suppressor cell ratios in melanoma patients during interferon treatment

Malignant melanoma patients treated with human leukocyte interferon (IFN-alpha) displayed increased natural killer (NK) cytotoxicity to K562 targets within the first 2 weeks of therapy. This study explored the possibility of T-cell regulation of this NK response, as evidenced by a variation in T-cell subpopulations. T-cell subset levels were studied in 9 patients who received daily doses of IFN-alpha over a period of 42 days. Five monoclonal antibodies to T-cell surface markers were used: Leu 1 (pan-T), Leu 2a (T suppressor/cytotoxic), Leu 3a (T helper/inducer), HNK-1 (Leu 7, a marker for NK cells), and B73.1 (an antibody against the Fc receptor on NK/K cells). Percentages of these markers were measured on days 0, 3, 7, and 21 of treatment. Percentages of Leu 1-positive cells and Fc-receptor-positive cells remained relatively constant throughout treatment in all patients. A trend toward a decrease in suppressor cells and an increase in helper cells peaking on day 7 and returning to earlier values by day 21 was found in 5 patients. The increase in NK cytotoxicity was not reflected consistently in proportions of HNK-1-positive cells or Fc receptor-bearing cells within the first week of therapy. The most significant finding was an increase in the helper:suppressor cell ratio peaking on day 7 and returning to pretreatment values by day 21. This increase was seen in every patient studied. The average pretreatment Leu 3a:Leu 2a ratio was 0.67 increasing to an average value of 1.47 on day 7 (p less than 0.005). Leu 3a:Leu 2a ratios returned to pretreatment values, in parallel to NK activity, despite continuation of interferon therapy.     

The distribution of cells expressing a natural killer cell marker (HNK-1) in normal human lymphoid organs and malignant lymphomas

A murine monoclonal antibody, HNK-1, is known to react with some human leukocytes including all natural killer (NK) cells in peripheral blood. The distribution of cells reacting with this antibody (HNK-1+ cell) was studied in human peripheral lymphoid organs, consisting of five lymph nodes, two specimens of gastric mucosa with lymphoid tissue, two tonsils, one appendix, and two thymuses. Fourteen cases of malignant lymphoma (ML) were also examined. For the demonstration of HNK-1+ cells, the peroxidase-antiperoxidase (PAP) bridge method was applied to cryostat sections of these specimens. It was found that in normal lymphoid organs most HNK-1+ cells were located in lymph follicles, especially in germinal centers, and some were found in 'mixed' regions which indicate outsides of both the follicles and T-zones. Amongst the ML, large clusters of HNK-1+ cells were observed only in two cases of follicular lymphoma, although a few scattered HNK-1+ cells were noted in other ML, including five diffuse B-cell lymphomas, six T-cell lymphomas and one null cell lymphoma. The possible significance of these findings is discussed.     

The epitope of mouse embryonic antigen(s) recognized by monoclonal antibody TEC-02 is a carbohydrate carried by high-molecular-weight glycoconjugates

The expression and properties of mouse embryonic antigens, recognized by monoclonal antibody TEC-02, were analyzed in teratocarcinoma-derived cell lines. TEC-2 antigens were found in the majority of the parietal endoderm cells PYS-2 and in a fraction of cultured embryonal carcinoma cells but not in other cell lines tested. During the course of retinoic acid-induced differentiation of embryonal carcinoma cells F9, the expression of TEC-2 was transiently increased. Immunolabeling of extracts from F9 and PYS-2 cells separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed that TEC-2 antigens are polydisperse glycoconjugates of high molecular weight (mostly greater than 100,000). The TEC-2 epitope was shown to be carbohydrate which in F9 cells might be attached to the same carrier as another developmentally regulated carbohydrate epitope TEC-1. The TEC-2 antigens, isolated by indirect immunoprecipitation, were degraded by extensive pronase digestion or mild alkaline treatment to mostly large products. Immunostaining of glycolipid standards suggested that TEC-2 epitope involves the GalNAc beta 1----4Gal beta 1----4R sequence. Combined data indicate that TEC-2 is a new developmentally regulated carbohydrate epitope carried in embryonal carcinoma cells predominantly on glycoprotein-bound large carbohydrates.     

Differentiation, proliferation and adhesion of human neuroblastoma cells after treatment with retinoic acid

Because of the known property of spontaneous regression in stage IVS of neuroblastoma all attempts are made to elucidate whether differentiation inducers possibly could be applied for neuroblastoma therapy. Here we examined the influence of retinoic acid (RA) in vitro on differentiation, proliferation and adhesion of 10 permanent and 4 primary cell lines as well as of several SCID-mouse tumour transplants. In general, after RA treatment morphologically different cell types which are characteristic for neuroblastoma cells have changed. N (neuronal)-type cells prolonged their neuronal processes, whereas S (epithelial, substrate-adherent, Schwann cell-like)-type cells lost their adherence to substratum and became apoptotic. Additionally, the reactions of all neuroblastoma cell lines with monoclonal antibodies against beta-tubulin (for neuronal cells) and glial fibrillary acidic protein (for epithelial cells) were determined. The anti-proliferative effect of all-trans-RA as well as 13-cis-RA was more profound in S-type cells (up to 40% in primary cell lines). To elucidate the role of adhesion molecules during neuronal cell differentiation, we have analysed the adhesion of neuroblastoma cells on poly-D-lysin-precoated plates under RA influence. While N-type cells displayed an increased adhesion, all S-type cell lines as well as all primary cell lines exhibited a reduced adhesion (IMR-5 and IMR-32: p < 0.001; JW, SR and PM: p < 0.05). RA treatment increased predominantly the tested antigens (HCAM, ICAM-1, NCAM, PECAM-1, VCAM-1, cadherin, FGF-R, IGF-R, NGF-R, TGF-beta/1, NF200, NF160, NF68, NSE, HLA-ABC) in all cell lines independently of their phenotypes (TGF-beta/1: p < 0.001; NF68: p < 0.01; PECAM-1 and NGF-R: p < 0.05). In recultured SCID-mouse-passaged tumour cells antigens were down-regulated (FGF-R: p < 0.01), but increased again after RA influence (TGF-beta/1: p < 0.05). In summary, the RA differentiation model demonstrates the possibility to interfere in cell adhesion and to diminish growth potential both in N-type as well as S-type neuroblastoma cells.     

Regulation by retinoic acid of ICAM-1 expression on human tumor cell lines.

In a group of four human tumor cell lines comprising one melanoma, one glioma, one teratocarcinoma and one neuroblastoma, the expression of the intercellular adhesion molecule-1 (ICAM-1) was found to be significantly increased following treatment with 10 microM of all-trans retinoic acid. In the melanoma and glioma cell lines HS 294T and HS 683, greater than 90% of the cells reacted with the anti-ICAM-1 monoclonal antibody (mAb) CL203.4 in the absence of treatment. Retinoic acid increased the cell surface expression of the molecule by 2-fold. In the teratocarcinoma and neuroblastoma cell lines, TERA-2 and SK-N-SH, the constitutive expression of ICAM-1 was weak, the percentage of cells stained above the background being less than 25%. Retinoic acid induced ICAM-1 expression in greater than 80% of the cells and increased the levels of expression by 2.5 to 3-fold. Immunoprecipitation studies in biosynthetically labeled cells as well as RNase protection analysis confirmed that retinoic acid treatment increased the amount of ICAM-1 at both the protein and mRNA level. The induction or stimulation occurred within 24 h, was maximal after 4 days and reversible.     

Tumor-activated NK cells trigger monocyte oxidative metabolism

We have examined the hypothesis that tumor cells can stimulate a respiratory burst by human natural killer (NK) cells in vitro as measured by luminol-dependent chemiluminescence (CL). Percoll-purified NK cells, containing 40% HNK-1+ cells and less than 1 to 4% esterase-positive contaminating monocytes, can generate a strong CL response after stimulation with the NK-susceptible K562 tumor but not with the NK-resistant P815 tumor cells. Although the response was NK dependent, as shown by depletion with NK-directed monoclonal antibodies (HNK-1, OKT-11, and OKM-1), the cell generating the CL response was not the NK cell. On the basis of several independent experimental approaches the CL response always required the presence of monocytes in the NK preparation. a) Treatment with a monocyte-specific monoclonal antibody (MO2) and complement completely abolished CL. b) The cells producing the CL response were strongly adherent to nylon wool columns (NWC), and large granular lymphocyte preparations containing less than 0.1% esterase-positive cells were inactive. c) NK cells cultured in IL 2-containing medium and tested over several days did not generate CL. d) Optimal numbers of monocytes (less than 1 to 2%) added to a non-CL NWC-purified NK population restored CL, whereas larger or smaller amounts were ineffective. Neither these procedures nor the addition of superoxide dismutase (which completely blocked CL) had any effect on NK lytic activity. We subsequently demonstrated that a factor present in supernatants obtained from NK/K562 incubations, but not from NK or tumor cells alone, could stimulate monocyte CL. We therefore propose that the CL response measured in NK-enriched Percoll fractions originated from contaminating monocytes that were triggered by factor(s) released from tumor-activated NK cells, and that superoxide anion was not required for NK lysis.     

Glycan array analysis of the antigen repertoire targeted by tumor-binding antibodies

Immunization with whole cells has been used extensively to generate monoclonal antibodies, produce protective immune responses, and discover new disease antigens. While glycans are abundant on cell surfaces, anti-glycan immune responses have not been well-characterized. We used glycan microarrays to profile 49 tumor-binding monoclonal antibodies generated by immunizing mice with whole cancer cells. A substantial proportion (41%) of the tumor binding antibodies bound carbohydrate antigens. The antibodies primarily recognize a group of 5 glycan antigens: Sialyl Lewis A (SLeA), Lewis A (LeA), Lewis X (LeX), blood group A (BG-A), and blood group H on a type 2 chain (BG-H2). The results have important implications for monoclonal antibody production and cancer vaccine development.     

Human neuroblastoma cell lines are susceptible to lysis by natural killer cells but not by cytotoxic T lymphocytes

The susceptibility of human neuroblastoma cells to direct cellular cytotoxicity has not been previously established. This is of particular interest because of their aggressive growth and low HLA expression. Neuroblastoma lines CHP 100 and CHP 126 were found to be excellent targets in 4-hr CML assays. Natural killer (NK) cells from fresh PBL and from an NK clone, 3.3, have high lytic activity against both cell lines. We also studied mixed lymphocyte culture-generated cytotoxic lines containing allo-specific cytotoxic T lymphocytes (CTL) directed against HLA antigens present on the neuroblastoma target cell lines. These lines did show excellent lytic activity, but cold target competition studies indicated that all of the lysis resulted from NK activity. This was verified by using inhibition studies with the use of monoclonal antibodies. OKT 3 and anti-HLA antibodies that block CTL function caused no reduction in kill. In contrast, anti-lymphocyte function antigen-1 (anti-LFA-1), which blocks both NK and CTL function, significantly inhibited lysis. These results serve as a functional confirmation of earlier findings of a very weak expression of HLA-A,B,C and beta 2-microglobulin on neuroblastoma cells.     

Expression of a drug resistance gene in human neuroblastoma cell lines: modulation by retinoic acid-induced differentiation.

Expression of a multidrug resistance gene (mdr1) and its protein product, P-glycoprotein (Pgp), has been correlated with the onset of multidrug resistance in vitro in human cell lines selected for resistance to chemotherapeutic agents derived from natural products. Expression of this gene has also been observed in normal tissues and human tumors, including neuroblastoma. We therefore examined total RNA prepared from human neuroblastoma cell lines before and after differentiation with retinoic acid or sodium butyrate. An increase in the level of mdr1 mRNA was observed after retinoic acid treatment of four neuroblastoma cell lines, including the SK-N-SH cell line. Western blot (immunoblot) analysis demonstrated concomitant increases in Pgp. However, studies of 3H-vinblastine uptake failed to show a concomitant Pgp-mediated decrease in cytotoxic drug accumulation. To provide evidence that Pgp was localized on the cell surface, an immunotoxin conjugate directed against Pgp was added to cells before and after treatment with retinoic acid. Incorporation of [3H]leucine was decreased by the immunotoxin in the retinoic acid-treated cells compared with the undifferentiated cells. These results demonstrate that whereas expression of the mdr1 gene can be modulated by differentiating agents, increased levels of expression are not necessarily associated with increased cytotoxic drug accumulation.     

Expression of class III β-tubulin in normal and neoplastic human tissues

 The class III β-tubulin isotype is widely used as a neuronal marker in normal and neoplastic tissues. This isotype was, however, also immunodetected in certain tumours of non-neuronal origin such as squamous cell carcinoma. Using a newly described monoclonal antibody we compared the distribution of class III β-tubulin in normal and neoplastic tissues. The TU-20 mouse monoclonal antibody was prepared against a conserved synthetic peptide from the C-terminus of the human class III β-tubulin isotype, and its specificity was confirmed by immunoblotting, by competitive enzyme-linked immunosorbent assay and by immunofluorescence microscopy on cultured cells. In different cell lines of various origins the antibody reacted only with neuroblastoma Neuro-2a cells and with embryonal carcinoma P19 cells stimulated to neuronal differentiation by retinoic acid. Immunohistochemistry on formaldehyde-fixed paraffin-embedded normal human tissues revealed the presence of the class III β-tubulin isotype in cell bodies and processes of neuronal cells in the peripheral and central nervous systems. In other tissues, this β-tubulin isotype was not immunodetected. Class III β-tubulin was found in all cases of ganglioneuroblastoma, ganglioneuroma, medulloblastoma, neuroblastoma, sympathoblastoma and in one case of teratoma. In contrast, no reactivity was detected in tumours of non-neuronal origin, including 32 cases of squamous cell carcinoma. The results indicate a specific TU-20 epitope expression exclusively in neuronal tissues. The antibody could thus be a useful tool for the probing of class III β-tubulin functions in neurons as well as for immunohistochemical characterisation of tumours of neuronal origin. Accepted: 29 July 1997