Lymphocyte depletion in thymic nurse cells: a tool to identify in situ lympho-epithelial complexes having thymic nurse cell characteristics

Summary In situ pre-existing complexes of epithelial cells and thymocytes having thymic nurse cell characteristics were visualized in the murine thymus cortex using dexamethasone as a potent killer of cortisone-sensitive thymocytes. The degradation and subsequent depletion of cortisone-sensitive thymocytes enclosed within cortical epithelial cells appeared to be paralleled by thymocyte degradation and depletion in thymic nurse cells isolated from thymic tissue fragments from dexamethasone-treated animals. This suggests that thymic nurse cells are derived from pre-existing sealed complexes of cortical epithelial cells and thymocytes. Not all thymocytes situated within in situ epithelial or thymic nurse cells complexes appear to be cortisone-sensitive: a minority of 1–2 thymocytes per complex survives the dexamethasone-treatment, thus constituting a minor subset of cortical cortisone-resistant thymocytes predominantly localized within cortical epithelial cells in situ and within thymic nurse cells derived from such structures. Cortisone resistance in thymocytes thus seems to be acquired within the cortical epithelial cell microenvironment. Cortisone-resistant thymocytes in thymic nurse cells express the phenotype of mature precursors of the T helper lineage, indicating that the in situ correlates of thymic nurse cells may play an important role in T cell maturation and selection.     

Identification and in situ localization of the "thymic nurse cell" in man

The observation of the "thymic nurse cell" (TNC), a reticuloepithelial cell with intracytoplasmic lymphocytes, in suspension of murine thymic tissue prompted us to investigate the existence of this cell in cell suspension, as well as in tissue sections of the human thymus. TNC-like cells were enriched in suspension by enzymatic disintegration of thymic tissue and 1 X G sedimentation over 50% fetal calf serum gradients. TNC-like cells were negative for lysosomal enzymes: in this respect, as well as in light microscopic morphology, the cells were different from tissue macrophages with intracytoplasmic lymphocytes. In electron microscopy, TNC-like cells showed reticuloepithelial characteristics. In 1-micron tissue sections, clusters of lymphocytes with a possible reticuloepithelial nucleus were observed close to blood capillaries in the cortical area. Ultrastructural analysis confirmed the epithelial nature of this cell, as well as its location adjacent to blood capillaries. We concluded that there is in situ existence of TNC in man. This observation enables studies on the role of TNC in intrathymic T cell maturation.     

The identification of thymic nurse cells in vivo and the role of cytoskeletal proteins in thymocyte internalization

Much debate has been generated about the existence of thymic nurse cells within the thymus. Until now, the authenticity of an epithelial cell capable of internalizing developing thymocytes within the thymic cortex has been in question. Here, we use the thymic nurse cell-specific monoclonal antibody, ph91, to define the in vivo location of thymic nurse cells. For the first time, thymic nurse cells enclosing several thymocytes were detected in the subcapsular region of the thymic cortex in a “honeycomb-like” configuration. In vitro studies show the internalization process using digitalized time-lapse microscopy. Internalized thymocytes have also been reported to interact with macrophages within the TNC complex. The cytoplasmic interaction between thymocytes and macrophages was detected using time-lapse microscopy. Using fluorescence microscopy, we show polymerization of actin within macrophages at the contact point with thymocytes, which is indicative of an immunological synapse. Microfilaments and microtubules within TNCs were shown to be associated with thymocyte binding and internalization, but neither interacted with macrophages. Also, we provide data to show that thymocytes are actively involved in the internalization process. These experiments show for the first time the existence of thymic nurse cells within the thymic microenvironment. They provide a visual documentation of thymocyte uptake by thymic nurse cells, and define an interaction between thymocytes and macrophages within the TNC complex.     

Avian thymic accessory cells

On the basis of morphologic criteria and ingestion of latex particles, two basic types of accessory cells can be identified from quail and chick thymuses, dendritic cells, and macrophages. By using embryonic grafting techniques, we show that cells of this lineage enter the thymus during the initial colonization of the epithelial thymic rudiment by hemopoietic cells, and within a few days differentiate into cells exhibiting properties of glass adherence, Ia expression, and formation of rosettes with thymocytes. It appears that the precursors of this lineage undergo extensive, but finite, proliferation and are eventually replaced by further influx of the accessory cell lineage. In chimeric grafts, quail thymocytes were seen forming rosettes with chick accessory cells, and vice versa, indicating, as in the interaction between the epithelial cells and thymocytes, that the molecules involved in thymocyte-accessory cell association can interact across species barriers in our system.     

Zonal unit-gravity elutriation. A new technique for separating large cells and multicellular complexes from cell suspensions

A new and simple technique, zonal unit-gravity elutriation, has been devised for separating very large cells, multicellular complexes, or small organisms from suspensions consisting mainly of small cells. The separation vessel is a conical chamber with an entrance at the lower, narrower part of the cone and an exit at the upper, wider part of the cone via a dome-shaped lid. A baffle at the entrance prevents turbulence from incoming fluid. Chambers of differing widths and wall slopes are chosen depending on the sedimentation rate of the particles to be separated. A small volume of the cell suspension is placed in the chamber on the bench in a cold-room. Medium stabilized by a shallow density gradient is pumped into the base of the chamber and ascends, creating a decreasing velocity gradient. Cells sediment at unit-gravity against this ascending counterstream, and are separated into bands according to sedimentation velocity. By adjusting the flow rate of the medium, different sizes of cells can be separated. Tumor cells can be enriched, and larger blast cells can be separated from small cells in lymphoid cell suspensions. The procedure produces complete separation of thymic nurse cells (epithelial-lymphoid complexes) from free thymocytes in digested thymus suspensions and produces substantial enrichment of thymic rosettes (macrophage-lymphoid complexes). A very favorable situation for applying this technique is the isolation of Taenia taeniaformis larvae, which can be completely purified from infected liver suspensions, representing a 4 X 10(5)-fold enrichment of the parasites, with high recovery, in a single 30 min operation.     

Zonal unit-gravity elutriation

A new and simple technique, zonal unit-gravity elutriation, has been devised for separating very large cells, multicellular complexes, or small organisms from suspensions consisting mainly of small cells. The separation vessel is a conical chamber with an entrance at the lower, narrower part of the cone and an exit at the upper, wider part of the cone via a dome-shaped lid. A baffle at the entrance prevents turbulence from incoming fluid. Chambers of differing widths and wall slopes are chosen depending on the sedimentation rate of the particles to be separated. A small volume of the cell suspension is placed in the chamber on the bench in a cold-room. Medium stabilized by a shallow density gradient is pumped into the base of the chamber and ascends, creating a decreasing velocity gradient. Cells sediment at unit-gravity against this ascending counterstream, and are separated into bands according to sedimentation velocity. By adjusting the flow rate of the medium, different sizes of cells can be separated. Tumor cells can be enriched, and larger blast cells can be separated from small cells in lymphoid cell suspensions. The procedure produces complete separation of thymic nurse cells (epithelial-lymphoid complexes) from free thymocytes in digested thymus suspensions and produces substantial enrichment of thymic rosettes (macrophage-lymphoid complexes). A very favorable situation for applying this technique is the isolation ofTaenia taeniaformis larvae, which can be completely purified from infected liver suspensions, representing a 4×10⁵-fold enrichment of the parasites, with high recovery, in a single 30 min operation.     

Lympho-epithelial interactions in rat thymus during pregnancy

Alterations in the thymic epithelial cell activity were analysed during pregnancy and lactation in Wistar rats by examining the presence and in situ distribution of lymphoepithelial complexes formed by thymic nurse cells (TNC). TNC were identified in paraffin sections by their expression of MHC class II antigens, CD54 molecule and a neuromarker, protein gene product 9.5 (PGP9.5). On the first days of pregnancy (gestational days, GD) the number of PGP9.5+ TNC was found to decrease abruptly. On GD 14, a transient increase was noted in the number of PGP9.5+, MHC+, CD54+ TNC. Another increase was observed in the course of lactation, when the weight of the thymus reached the lowest values. While the increase in TNC numbers during lactation may be linked to the process of reconstruction of the thymic lymphoid population, the augmented activity of lymphoepithelial interactions on GD14 may be associated with thymic engagement in pregnancy-induced immune processes.     

Analysis of hemopoietic lineage of accessory cells in the developing thymus of Xenopus laevis

The developmental history of accessory cells in the thymus was studied by grafting hemopoietic stem cells into cytogenetically distinct frog embryos (diploid-2N or triploid-3N) before the establishment of circulation and overt differentiation and colonization of the thymus. The DNA content of cortical thymocytes and circulating erythrocytes was quantified by staining with propidium iodide and measuring the amount of red fluorescence emitted by individual nuclei with the use of flow cytometry. Accessory cells from thymic medulla were separated by incubating for 2 hr on glass slides. For comparison, the developmental history of peritoneal macrophages was examined as representative, myeloid-derived phagocytic cells. DNA content of adherent cells was quantified by staining with the DNA-specific Feulgen reaction and measuring light absorption of individual nuclei by microdensitometry. Thymic accessory cells were subdivided into phagocytic and nonphagocytic phenotypes on the basis of latex bead ingestion. Phagocytic cells in the thymus were usually nonspecific esterase positive and phenotypically resembled peritoneal macrophages. Nonphagocytic cells from the thymus were usually esterase negative and had a dendritic morphology characterized by branched cytoplasmic extensions. Nonphagocytic cells were positive for cytoplasmic RNA based on staining with methyl green-pyronin Y. Phagocytic cells from both the thymus and the peritoneal cavity had no levels of cytoplasmic RNA detectable by this method. Analysis of the embryonic derivation of thymic accessory cells, based on the proportion of cells carrying the cytogenetic marker, demonstrated that thymic lymphocytes and thymic accessory cells were a concordant pair of cells, distinct from myeloid-derived erythrocytes and possibly macrophages. These experiments provide circumstantial evidence suggesting thymocytes and thymic accessory cells could arise from a bipotential precursor that diverges into these separate lineages after colonization of the epithelial thymic rudiment during early development.     

Thymic epithelium in vitro. V. Binding of thymocytes to cultured thymic epithelial cells

Direct contact between thymocytes and thymic stromal elements may be one of the mechanisms involved in thymocyte differentiation. Thymic lymphoepithelial complexes have been isolated in which thymocytes appear to be in direct association with cortical epithelial cells. We have previously reported the isolation and successful culture of two morphologically distinct types of murine thymic epithelial cells. We have utilized these to study the interactions of lymphoid and epithelial cells by means of an in vitro assay of the binding of radiolabeled thymocytes to monolayers of these cultured thymic epithelial cells. The percentage of bound cells increased rapidly during the first hour of incubation, reaching approximately 40% binding. Binding continued to increase slowly until plateau levels were reached at approximately 5 hr. Thymocyte binding to thymic epithelium, but not fibroblast monolayers, was trypsin-sensitive, suggesting that specific protein interactions may be involved. Binding of thymocytes to epithelium was temperature-dependent, involved formation of cytoplasmic projections, and was inhibited by cytochalasin B. We also found that cortical thymocytes (peanut agglutinin-positive (PNA+)cells) bound to cultured epithelium to a greater degree than medullary thymocytes (PNA- cells). This correlates with in vivo studies by others in which thymocytes associated with lymphoepithelial complexes have been found to have immature phenotypes. This system provides a means for a quantitative study of the role of cell to cell contact in the process of thymocyte selection and differentiation.     

Differences in turnover between thymic medullary dendritic cells and a subset of cortical macrophages

To investigate the turnover of thymic accessory cells, we performed vascular thymus transplantation in RT7 congenic rats. mAb specific for one of the two allelic variants of the RT7 molecule, as well as mAb specific for either medullary interdigitating cells or a subset of cortical macrophages (M phi), were used on cryostat sections and cell suspensions prepared from grafted thymuses to monitor the turnover of these two cell types. In contrast to the complete turnover of interdigitating cells within 3 wk after transplantation, ED2-labeled cortical M phi showed a very slow turnover. Seventy-six days after transplantation, more than 30% of these M phi were found to be still of donor origin. The different turnover rates of these thymic accessory cells could reflect their function in T cell development.     

Ia-positive macrophages bind and internalize viable lymphocytes in murine thymus

Macrophages have been shown to be present in thymus throughout its development. In the present study monoclonal and polyclonal antimacrophage reagents were used to identify, quantitate, and determine the distribution of thymic macrophages. Those studies demonstrated that significant numbers of macrophages were evenly distributed throughout the cortex and medulla, and that macrophages account for most, if not all, Ia positivity in murine thymus. Suspensions of thymic cells prepared by enzyme digestion contained 2-4% macrophage antigen-positive cells, over 95% of which were I-Ak positive in double-labeling studies. Removal of lymphocytes and macrophages left only epithelial cells and those failed to label for Ia. Subsequent to mild enzymatic digestion, up to 80% of the thymic macrophages were in the form of lymphocyte/macrophage rosettes. Morphologic evaluation of the thymocyte rosettes revealed that some of the macrophages contained internalized lymphocytes. The proportion of macrophages with internalized lymphocytes generally was less than 10%, but during the first 4 weeks of life values often approached 30%. Nurse cells, which were shown through double labeling to express both Ia and macrophage-associated antigen, were included in the population of rosetted cells which had internalized lymphocytes. The results demonstrated that there is a high level of interaction between lymphocytes and Ia-positive macrophages in the thymus which is greatest during the immediate postnatal period.     

Immunologic characterization of hairy cell leukemias in continuous culture.

The immunobiologic characteristics of three continuous cell lines established from hairy cell leukemia cells were investigated. All three cell lines continued to produce tartrate-resistant acid phosphatase, the enzymatic marker of hairy cells. Two of these cell lines were B lymphoid in nature. They carried Fc and C receptors, had surface and internal immunoglobulin, and did not form spontaneous sheep red blood cell rosettes. Experiments employing biosynthetic radiolabeling of immunoglobulin demonstrated distinctive immunoglobulin kinetics for each of these two hairy cell lines. One cell line remained quite similar to the original hairy cells from which it was derived whereas the other B lymphoid hairy cell line had undergone a switch in the immunoglobulin isotype produced. The third hairy cell leukemia line was shown to be of thymic derivation. These cells formed spontaneous sheep red blood cell rosettes and did not carry Fc or C receptors. The spontaneous sheep red blood cell rosette-forming cells contained tartrate-resistant acid phosphatase. They did not possess surface on internal immunoglobulin and did not synthesize immunoglobulin in vitro. Hairy cell leukemia cells maintained in permanent cell culture retain their immunobiologic properties and offer the opportunity for indepth study of these unusual cells.     

Immunohistology of thymic nurse cells

The demonstration of thymic nurse cells (TNC), complexes between stromal cells and thymocytes, in cell suspensions of murine thymuses, prompted us to investigate (1) the relationship of TNC to other thymic stromal cell types defined in situ, and (2) the maturation stage of the enclosed thymocytes. To this purpose we incubated frozen sections of TNC suspensions with various monoclonal antisera directed to T cells and stromal cell types, using immunohistology. This approach enabled us to study antigen expression on the "nursing" cell itself and to analyze the phenotype of the enclosed lymphocytes in cross sections of TNC. The results show that lymphocytes enveloped by TNC express high levels of Thy-1, moderate levels of T200, and variable amounts of Lyt-1. Due to enzymatic degradation Lyt-2 expression could not be studied. The enveloped cells also bear PNA receptors, but no detectable I-A/E antigens. Expression of H-2K antigens on enclosed thymocytes varied from weak to absent. The "nursing" cells react with ER-TR4, a monoclonal antibody which detects cortical epithelial-reticular cells. In addition TNC express I-A/E and H-2K antigens. In contrast, TNC do not react with ER-TR 5 and 7, monoclonal antibodies, which detect medullary epithelial cells and reticular fibroblasts, respectively. TNC do not express the macrophage antigens Mac-1 and Mac-2. We conclude that TNC in vitro represent the in vivo association of epithelial-reticular cells with cortical thymocytes. However, the enclosed thymocytes do not constitute a phenotypically distinct subset of subcapsular or outer cortical cells.     

Role of accessory cells in B cell activation. I. Macrophage presentation of TNP-Ficoll: evidence for macrophage-B cell interaction

The importance of cell interaction for thymic independent antigen responses has not been widely appreciated. The present report demonstrates, however, that macrophage-B cell interaction may be an important feature of B ce-l activation for the response to at least one polysaccharide thymic independent antigen, TNP-Ficoll. Experiments were performed demonstrating that a strict accessory cell requirement exists for the thymic independent response to soluble TNP-Ficoll, and that such accessory cells are both adherent and phagocytic, that is, macrophages. It was further demonstrated that macrophages could be pulsed with TNP-Ficoll and that these pulsed macrophages could activate B cells to respond, but only if the pulsed macrophages were viable. Thus, one function that macrophages can fulfill in responses to TNP-Ficoll is the specific function of antigen presentation. Such presentation of TNP-Ficoll by macrophages to B cells suggests that the antigen may not be activating B cells directly, and raises the possibility that the interaction of B cells and macrophages might be genetically restricted.     

Repopulation of the mouse thymus after sublethal fission neutron irradiation. II. Sequential changes in the thymic microenvironment

The stromal cells of the thymus of sham-irradiated and sublethal fission neutron-irradiated CBA/H mice were analyzed with immunohistology, using monoclonal antibodies directed to I-A and H-2K antigens as well as specific determinants for cortical and medullary stromal elements. In the control thymuses, I-A expression in the thymus shows a reticular staining pattern in the cortex and a confluent staining pattern in the medulla. In contrast, H-2K expression is mainly confluently located in the medulla. Whole body irradiation with 2.5 Gy fission neutrons reduces within 24 hr the cortex to a rim of vacuolized "nurse cell-like" epithelial cells, largely depleted of lymphoid cells. The localization of I-A antigens changes in the cortex and I-A determinants are no longer associated with or localized on epithelial reticular cells. Medullary stromal cells, however, are more or less unaffected. A high rate of phagocytosis is observed during the first 3 days after irradiation. About 5 days after irradiation, the thymus becomes highly vascularized and lymphoid cells repopulate the cortex. The repopulation of the thymic cortex coincides with the appearance of a bright H-2K expression in the cortex which is associated with both stromal cells as well as lymphoid blasts. During the regeneration of the thymus, the thymic stromal architecture is restored before the expression of cell surface-associated reticular MHC staining patterns. The observed sequential changes in the thymic microenvironment are related to the lymphoid repopulation of the thymus.     

Primary stage of feline immunodeficiency virus infection: viral dissemination and cellular targets.

The objective of this study was to identify cellular and organ targets of acute feline immunodeficiency virus (FIV) infection in vivo. Tissues of FIV-infected cats were studied at eight time points during the first 3 months after experimental infection. FIV nucleic acids were first detected by in situ hybridization 21 days after infection, approximately 1.5 weeks after lymph node enlargement was first observed and 3 weeks before the primary acute flu-like illness. The majority of FIV-infected cells were present in lymphoid organs, though low numbers of infected cells were noted in nonlymphoid organs as well. Germinal centers harbored many of the FIV-infected cells within lymphoid tissues. The thymic cortex was also a major site of early infection. Combined in situ hybridization and immunohistochemistry revealed that T lymphocytes were the primary target of early FIV infection in tissues of cats before the onset of clinical signs of acute illness. An unidentified population of mononuclear cells and a few macrophages were also infected. During the ensuing acute flu-like illness, the proportion of FIV-infected macrophages in tissues increased dramatically. This early shift in the predominant cellular localization of FIV from T lymphocytes to macrophages may be important for establishing viral persistence.     

Effect of cytokines on thymic hematopoietic precursors

Summary We have previously shown that the interaction of thymocytes with thymic accessory cells (macrophages and/or interdigitating cells) is one of the factors required for thymocyte activation. Precursors of both thymic accessory cell and thymocytes are included in the CD4^- CD8^- Mac-1^- Ia^- subpopulation, and their respective maturation and/or activation may be modulated by granulocyte-macrophage colony-stimulating factor, interleukin 1 and interleukin 2. When CD4^- CD8^- thymic cells are activated with granulocyte-macrophage colony-stimulating factor plus interleukin 2, both macrophages and interdigitating-like cells are present, as shown by electron microscopy. When activated with interleukin 1 plus interleukin 2, the interdigitating-like cells is the only accessory cell present. In both culture conditions, large clusters are formed between interdigitating cells and lymphoid cells. These results have led us to propose two-step signals for thymocyte proliferation: first, the maturation of macrophages under granulocyte-macrophage colony-stimulating factor control and the production of interleukin 1, and secondly, the maturation of interdigitating cells under interleukin 1 control, their clustering with thymocytes which are then activated.Abbreviations CFU-S colony-forming units in the spleen - CSF colony-stimulating factor - DC dendritic cells - DN double negative cells (CD4^- CD8^-) - EC epithelial cells - GM-CFC granulocyte/macrophage colony-forming cells - GM-CSF granulocytemacrophage CSF - IDC interdigitating cell - IL-1 interleukin 1 - IL-2 interleukin 2 - MØ macrophage - P-TR phagocytic cell of the thymic reticulum     

Thymus-derived lymphocyte differentiation and lymphocytic leukemias. I. Evidence for the existence of functionally different subpopulations of thymus-derived cells in leukemic AKR mice

The effects of transplanting thymic (LTC), splenic (LSC), and lymph node (LLNC) lymphocytes derived from overtly leukemic AKR mice into preleukemic syngeneic animals were studied. Each of these thymus-derived (T cell) populations produced a different and distinct pathology in recipient mice. Animals receiving LTC exhibited thymoma and enlargement of peripheral lymphoid tissues. Gross organomegaly was also noted in mice given LSC, but thymic atrophy was uniformly observed. The thymus appeared normal in mice receiving LLNC, but marked enlargement of peripheral lymphoid tissues again were observed. The differences noted in disease pathologies correlated with the “homing” patterns of the subpopulations investigated. These findings suggest that subpopulations of T cells exist in mice with a thymus-derived neoplastic disorder.     

Circulating macrophages as well as developing thymocytes are enclosed within thymic nurse cells.

Both thymic nurse cells (TNCs) and macrophages have been reported to function as antigen-presenting cells during the process of MHC restriction. Negative selection, which results in the apoptosis of potentially autoreactive thymocytes, is believed to be associated with both macrophages and TNCs in the cortex. Both cell types have also been reported to ingest thymocytes undergoing positive and negative selection. However, macrophages ingest apoptotic thymocytes, while TNCs have been shown to internalize viable cells. A subset of the TNC-engulfed population is allowed to mature and is released, while the remaining fraction becomes apoptotic and is absorbed within the TNC cytoplasm through lysosomal activity. A recent report described a subset of rat TNCs that contain macrophages as well as thymocytes within their cytoplasm. We examined freshly isolated TNCs from C57BL/6 mice and found that, of the TNC population recovered, 1.7% contained macrophages within its cytoplasm. There also were macrophages tightly bound but not internalized into the multicellular structure at a rate of 2.9%. The total association of macrophages with TNCs was approximately 4.6%. This unique association of macrophages with TNCs was also observed in vitro when freshly isolated thymocytes (containing macrophages) were added to cultures of cells from the TNC cell line tsTNC-1. The macrophage-TNC interaction was found to be dynamic, with macrophages moving rapidly into and out of TNCs containing cytoplasmic thymocytes. Macrophages within TNCs showed a close association with cytoplasmic thymocytes. We then labeled peritoneal macrophages with CFDA SE, a cell tracking dye, and returned them to the mouse peritoneum. Within 1 h, labeled macrophages were detectable in the thymus. This is the first investigation to show a direct interaction between peripheral macrophages and TNCs. These results suggest that TNCs and macrophages work together as antigen-presenting cells.     

Thymic nurse cells in culture: morphological and antigenic characterization

Epithelial monolayers were derived from thymic nurse cells (TNC), and were seeded onto collagen-coated dishes immediately after their isolation from young adult C3H-murine thymuses. Different media and supplements were tested in order to obtain cultures that were as pure as possible. Primary cultures were enriched in epithelial cells but always contained non-epithelial components among which fibroblasts predominated. Immunodetection of keratins, and repeated light- and electron-microscopic observations established the epithelial nature of the elongated cells derived from TNC; these elongated cells were cortical reticular cells, and were different from medullary globular cells that immediately adopted a mosaic pattern in vitro. At the beginning of the culture, the necrosis of cortical lymphocytes appeared to be toxic for epithelial cells; when epithelial cells survived, they showed a temporary lipid accumulation. After a 5-day culture, they still synthesized DNA but lost this capacity thereafter and dedifferentiated. The lympho-epithelial symbiosis appeared to be necessary to maintain some epithelial characteristics of the cultured cells, such as the clear vesicles and the expression of la antigens. In sub-cultures, the monolayers were almost purely epithelial in nature but growth was no longer observed. The cells remained reticular in shape, as they were in vivo, but their cytoplasm and their nucleus became larger and numerous cells were multinucleated. Confluence was not obtained with classical media even after mitogenic stimulation. The frequent observation of strongly keratinized areas suggested a process of terminal differentiation; this could not be avoided by using low serum concentration.