Reprints Available Directly from the Publisher Photocopying Permitted by License Only Epithelium-free Area in the Thymic Cortex of Rats

The histology of epithelium-free areas in the subcapsular region of the thymus was studied in Wistar rats. Lymphocytes in these areas were CD4/CD8 double-positive, TCR c/]/positive in low intensity, and in CD5 labeling either negative or positive in low intensity. There was a high proliferative activity as assessed by bromodeoxyuridine incorporation in vivo and detected by immunohistochemistry. Various macrophage types were observed. They were either large and round to slightly dendritic, or small and dendritic. Most large cells were positive for MHC Class II, and labeled by the antimacrophage antibodies ED1 and ED2. A few cells were strongly positive for Sudan black, Oil red O, nonspecific esterase, and acid phosphatase; they resembled the large rounded macrophages in the corticomedullary zone, although their MHC Class II and ED2 staining was more intense. A few cells showed features of tingible body macrophages, as they contained cellular debris. Serial sections showed that epithelium-free areas run from the subcapsular area to deep in the cortex, and often border the medulla. This opens the opportunity for immature lymphocytes to move into the medulla and corticomedullary zone without contacting and potential selection with cortical stromal elements other than macrophages in the epithelium-free areas. In this case, the epithelium-free areas may offer a separate intrathymic pathway for T lymphocytes.


INTRODUCTION
The thymus harbors various compartments or microenvironments, based on lymphoid-and nonlymphoid-cell characteristics. Among these are areas devoid of stromal elements. Adjacent to the capsule and septa of the thymus, areas can be discerned where there are no epithelial cells. These so-called epithelium-free areas (EFA) show an abundance of lymphocytes (rat: Duijvestijn et al. 1982;mouse: Van Ewijk, 1984;Godfrey et al., 1990; man: epithelium-free areas in the inner cortex; Von Gaudecker, 1986). The occurrence and extent of these EFA varies between strains of rats. In the thymus of WAG/Rij rats, such areas have not been observed, whereas in diabetes-prone (DP) and diabetes-resistant (DR) BB rats, they *Corresponding author. Present address: National Institute of Public Health and Environmental Protection, P.O. Box 1, 3720 BA Bilthoven, The Netherlands. make up 5% and 3% of the thymic volume, respectively (Rozing et al., 1989). The thymus of BB rats also showed EFA in the medulla and in the corticomedullary region (CMR). Medullary EFA were also found immediately after and during recovery from Cyclosporin treatment (Schuurman et al., 1990). It is questionable whether the areas in the cortex of untreated, healthy rats represent a similar histologic entity as EFA in the medulla and CMR of BB rats and rats after Cyclosporin treatment. This aside, EFA is evidently different from the perivascular space (PVS). PVS are connective tissue regions, containing collagen and matrix, and are as such considered as an extrathymic area (Christensen, 1952;Kendall, 1989 (Kuper et al., in press). Instead, perivascular spaces, especially in the CMR, were more prominent. A (transient) increase in PVS volume has also been demonstrated in the human thymus (Steinman, 1986).
The characteristics of EFA were not clearly defined, and their function, if there is any, is unknown. They may be reservoirs for lymphocytes (Van Ewijk, 1984) or proliferation sites of lymphocytes (Duijvestijn et al., 1982;Godfrey et al., 1990). We therefore performed an enzymeand immunohistochemical study in rats to investigate the lymphoid and nonlymphoid elements in the EFA of the thymus cortex, in order to elucidate possible functions of the compartment. To investigate the proliferative activity in the areas, the thymidine analog bromodeoxyuridine (BrdU) was injected in rats, and the presence of BrdU in thymus was detected immunohistochemically.

General Histology
Along the capsule and septa, areas were found that, in H&E-stained sections, were prominent due to their high number of lymphocytes but in which no epithelial cells are evident. The areas were negative for keratin (antibodies: see Table 1; Fig. 1). MHC Class II staining was also negative except for single cells (Figs. 2,and 8; see also what follows under "Macrophages'). Almost no laminin was found, either within the areas or between the areas and the epithelium-containing thymic tissue (Fig. 3). Vascularization was virtually absent. The subcapsular epithelial layer was found between these cortical epithelium-free areas (EFA) and the connective tissue of capsule and septa (Fig. 4). Some of the EFA ran from the capsule to the medulla. Serial sections from one thymus often showed medullary buds bordering the EFA (Fig. 2). The medullary epithelial network extended with cell processes into the areas.   cells with a rounded to dendritic morphology and small dendritic macrophage like cells (Figs. 2 and 8). The large cells showed a confluent Class-II reactivity, comparable to that of medullary IDCs and of single cells in the cortex. ED1 staining (panmacrophage marker, Table 1) showed predominantly large cells with a rounded to slightly dendritic morphology and a few small dendritic cells. There were some large cells with a rounded morphology, and a few small dendritic cells that were EDl-positive but negative or faintly immunoreactive for MHC Class II (Fig. 8). Most ED2-positive cells were large cells with a rounded to dendritic morphology (ED2, cortical macrophage marker, Table 1). In number and cell contour, these cells were comparable to cells identified in ED1 staining. ED2 staining of the large cells was more intense than in the rest of the cortex. In two-color mmunohistochemistry for ED1 and ED2, only a few large and small cells were ED1 positive/ED2 negative (Fig. 9).
A few large, rounded cells were strongly positive for Oil red O, Sudan Black, nonspecific esterase (NSE; Fig. 10) and acid phosphatase (AP). Other macrophages were weakly positive for NSE and AP.

DISCUSSION
Epithelial-free areas (EFA) are found in the outer cortex of the thymus, mainly immediately bordering the subcapsular epithelial-cell layer. They can run deep into the cortex and even reach the medulla. At other sites, medullary buds contact the EFA. Immediate contact between lymphocytes in EFA and cortical and medullary epithelium is feasible, because no basal lamina and connective tissue are found between the epithelium and the EFA. Moreover, in keratin and MHC Class-II labeling, the medullary epithelial lining and, at some places, also the cortical epi-  thelial lining with the EFA suggest that there is free cell movement of lymphocytes between the EFA and the thymic epithelial network. In addition, free cell movement between EFA and CMR appears possible (Figs. 2 and 7). Serial staining with CD4 and CD8 suggests that the predominant lymphocyte is CD4/CD8 double-positive. This is in accordance with findings of Godfrey et al. (1990) and Rozing et al. (1989). EFA also contain TCR c/]/positive cells. This implies that lymphocytes in the EFA already passed the first intrathymic development, including TCR gene rearrangement. Godfrey et al. (1990) have suggested that the EFA are isolated "bags" of CD4+/CD8+, proliferating lymphocytes before they contact the thymic stroma. Proliferation in EFA is evident from BrdU labeling, performed in double staining with keratin or MHC Class II and BrdU. Moreover, EFA contain several lymphocytes that are either strongly positive or faintly positive for HIS44. Kampinga (1990) has suggested that lymphocytes lose this marker during intrathymic proliferation. Following his suggestion, the presence of this marker indicates that lymphocytes in the EFA stay a while before proliferation, or stay sufficiently long after proliferation to regain the marker. Boyd and Hugo (1991) have hypothesized that cells in EFA are accumulations of double-positive lymphocytes, which are not under the influence of positive selection, and subsequently die by apoptosis.
In EFA, macrophages with features of TBM are not frequent. However, under "stressfull" conditions, for example, after dexamethason administration, TBM may accumulate in EFA ( Fig. 11; unpublished results). Moreover, the large rounded macrophages in EFA, which are strongly positive for Oil red O, NSE, AP and Sudan Black and resemble CMR macrophages (Milicevic et al., 1987;Milicevic and Milicevic, 1989), might be precursors of tingible body macrophages (TBM), because they sometimes contain nuclear debris. Aggregates of the CMR macro-120 J.P. BRUIJNTJES et al. between 8 to 11 weeks old were used. They were kept under conventional laboratory conditions. Tissue Sampling and Preparation Part of the animals were intraperitoneally injected with 15 mg/kg body weight Bromodeoxyuridine, 2 hr before sacrifice. All animals were anesthetized with ether, and bled to death via the abdominal aortia. The thymus was removed, fixed in neutral, phosphate-buffered 4% solution of formaldehyde or snap-frozen in isopentance in liquid nitrogen, and stored at -80C. Formaldehyde-fixed tissues were embedded in paraffin, sectioned at 5/m, and stained with H&E. Cryostat sections (5-7/m) were stained with oil red O and Sudan Black (Pearse, 1968) for lipids. Immunohistochemistry Cryostat sections (5-7/tm) from the thymus were air dried on glass slides, and fixed for 10 min in acetone. Thereafter, they were rinsed in phosphate-buffered saline (PBS, 0.01 M, pH 7.4) and preincubated with 10% normal rabbit serum for 20 min. Serial sectiohs were incubated for 60 min with one of the monoclonal or polyclonal antibodies listed in Table 1. The sections were then rinsed in PBS and layered for 30 min with a peroxidase-conjugated rabbit antimouse Ig (RAMPO, Dakopatts, Denmark), which was diluted in PBS with 4% normal rat serum. The sections were subsequently rinsed in PBS and Tris/HC1 (0.05 M, pH 7.6) and finally incubated with the chromogen 3'3'-diaminobenzidine-tetrahydrochloride (Sigma) in a concentration of 0.5 mg/ml in Tris/HC1 containing 0.01% H202 for 10 min. The whole procedure was carried out at room temperature. Most sections were slightly counterstained with haematoxylin. Control

Enzyme-Histochemistry
Acid phosphatase activity was demonstrated according to Burstone (Pearse, 1968) with naphthol AS-BI phosphate (Sigma) as the substrate.
The incubation time was 30-60 min at 37C. The substrate for the demonstration of nonspecific esterase was alpha-naphthyl acetate (Sigma; Pearse, 1972). Cryostat sections were used. Incubation time was 10-20 min at room temperature.
For both reactions, hexazotized pararosaniline was used as the diazonium salt.

ACKNOWLEDGMENTS
We acknowledge greatly the gifts of the following monoclonal antibodies: Monoclonal antibodies in the ED series from C.D. Dijkstra, Free University of Amsterdam, The Netherlands; R73 from Th. Hunig, University of Wurzburg, Germany; monoclonal antibodies in the HIS series from J. Kampinga and F.G.M. Kroese, Uni-versity of Groningen, The Netherlands; and antikeratins from F.C.S. Ramaekers, University Hospital Maastricht, The Netherlands. We thank A. Stenus-van Basten and S. de Vlugt-van den Koedijk for their help in preparation of the manuscript. (Received August 11,1992) (Accepted October 21, 1992)