Reprints Available Directly from the Publisher Photocopying Permitted by License Only Modulation of Cytokeratin Expression in the Hamster Thymus" Evidence for a Plasticity of the Thymic Epithelium

Cytokeratin (CK) expression was investigated, by means of immunocytochemistry, in the hamster thymic epithelium during ontogeny, as well as in primary cultures and upon glucocorticoid hormone treatment in vivo. As compared to the distribution pattern of distinct monoclonal antibody-defined cytokeratins in the normal adult thymus, CK modulation was evidenced in the three situations studied. During thymus ontogeny, both cytokeratins of simple lining epithelia, as CK8 and CK18, as well as the CK1/CK10 pair (typical marker of terminal stage of keratinization), were expressed since early stages of thymus development. They were located in the central region of thymic lobules preceding the cortical-medullary distinctions. This differed from what had been previously shown for mouse thymus ontogeny, revealing that the interspecific diversity in the distribution pattern of thymic cytokeratins occurred early in fetal life. A modulation of CK expression was also detected when hamster thymic epithelial cells (TEC) were led to grow in culture, with a down-regulation of CK19 contrasting with an enhancement of CK18 expression. This diverged from the maintenance of the in situ pattern when human TEC were cultured. Last, in vivo hydrocortisone treatment, known to increase the numbers of KL1 cells in the mouse thymus medulla, promoted a cortical expression of the CK1/CK10 pair in the hamster thymus. Taken together, our findings demonstrate a continuous plasticity of the thymic epithelium, at least regarding cytokeratin expression, and enlarge the concept of interspecific diversity of intrathymic CK distribution in conditions as morphogenesis, in vitro system, and responsiveness to glucocorticoid hormone treatment.


INTRODUCTION
Studies on thymocyte differentiation are often carried out on fetal specimens or in vitro, providing important information for the role of the thymic microenvironment in this process. Nonetheless, to evaluate the function of each microenvironmental component, it is essential that the markers known to recognize it in the adult thymus in situ could also be used in other experimental systems. A large number of monoclonal antibodies (MAbs) was postulated to be markers of thymic epithelial cell (TEC) subsets (Haynes, 1984;van Vliet et al., 1984.;de Maagd et al., 1985;Kaneshima et al., 1987;Takacs et al., 1987;Izon *Corresponding author. and Boyd, 1990). Yet, no specific functional subset was isolated so far, despite the several TEC lines produced (Itoh, 1979;Nieburgs et al., 1985;Potworowski et al., 1986;Mizutani et al., 1987;Naquet et al., 1989). Interestingly, these markers reveal a close antigenic similarity between the thymic epithelium and epidermis (Haynes, 1984;Lobach et al., 1985;Schmitt et al., 1987), suggesting that TEC heterogeneity could also be a reflex of their type of differentiation, as demonstrated for the epidermis. In this regard, van de Wijngaert et al. (1984) proposed a TEC differentiation process occurring essentially in the cortical region, yet participating in Hassall's corpuscle formation. This hypothesis contrasted to that previously raised by Mandel (1968) postulating that Hassall's corpuscle rose from some medullary TEC. In this respect, many anti-TEC MAb reacting to Hassall's corpuscles and few surrounding medullary TEC recognize the stratum corneum (Haynes, 1984;Lobach et al., 1985); other MAb, pan-specific to the thymic epithelium (including Hassall's corpuscles), label all epidermal sheets, whereas MAb recognizing the whole thymic epithelium network, but Hassall's corpuscles restrictly stain epidermal basal layers (Schmitt et al., 1987). It was then postulated that Hassall's corpuscles might correspond to a final degree of TEC differentiation, as the stratum corneum keratinocytes. In this context, anticytokeratin (anti-CK) MAb, previously defined as epithelial differentiation markers might be regarded as relevant tools in the study of TEC subsets, as evidenced in rodents and humans (Nicolas et al., 1985(Nicolas et al., , 1986(Nicolas et al., , 1989Laster and Haynes, 1986;Colic et al., 1988aColic et al., , 1988bDardenne, 1988a, 1988b;Farr and Brady, 1989).
The set of cytokeratins expressed by an epithelium is typical of this tissue and related to its differentiation (Moll et al., 1982a;Quinlan et al., 1985;Sun et al., 1985). Nonetheless, concerning the thymic epithelium, we evidenced the expression of cytokeratins both of simple and stratified epithelia, as well as an interspecific diversity in the distribution of these intermediate filament proteins. These results suggested that the thymic epithelium is complex and unique, and that the general classification of epithelial tissues based on their CK expression patterns cannot be applied to the thymus (Meireles de Souza et al., 1993).
In the present work, we used the hamster as a model to study if in a given species, the in situ pattern could be reproduced in other experimental situations, which seems to be essential for a further evaluation on the precise role of distinct thymic microenvironmental components.
We addressed the question concerning TEC differentiation and CK expression, using the hamster model during thymic ontogeny, primary TEC cultures, and following hydrocortisone treatment, the latter known to augment KL1defined CK expression in the mouse thymus (Savino et al., 1988). It is important to note that the hamster thymus presents a late cortical-medullary distinction (1 week after birth), although the appearance of Hassall's corpuscles precedes this distinction (Adner et al., 1965) situations analyzed, thus evidencing a continuous plasticity of thymic epithelial cells. These data suggest that TEC heterogeneity depends on the developmental stage, species of origin, hormonal status, and in situ/in vitro systems. Lastly, our findings indicate that the thymic epithelium may not be divided in definitive subsets, and that phenotypic markers of TEC heterogeneity should be considered as markers to trace TEC plasticity.
CK expression in hamster TEC appeared to occur, we asked the question whether such modulation could also take place in vitro. The

Cytokeratin Expression During Hamster
Thymus Ontogeny As reported in Meireles de Souza et al. (1993), the adult hamster thymus presented the coexpression of CK typical of simple and stratified epithelia ( Fig. 1). Interestingly, this was already observed since day 13 of fetal life (Fig. 2). In fact, at early stages of thymus ontogeny, CK8/, CK18/, and KL1 / cells were predominantly located in the central region of the thymic lobes. Cortical and medullary TEC were distinguished by anti-CK MAb solely at the time of cortical-medullary distinction. It is noteworthy that the predominance of CK18 expression observed in the adult hamster, as compared to CK8 distribution, was stressed during ontogeny. Regarding the expression of CK19, we noticed that the whole thymic epithelium was already stained by the anti-CK19 MAb on the 13th day of fetal age ( Fig. 2b). Actually, the anti-CK19 MAb was the only reagent that reacted similarly during fetal or adult age. In contrast, CK13, which is expressed by medullary/subcapsullary TEC in the adult thymus, was not detected during ontogeny, even in animals of age 1 week of postnatal life.
On the 15th fetal day, thymus lobulation was initiated, and we could detect subseptal TEC intensely labeled by anti-CK19 MAb. Lastly, the appearance of Hassall's corpuscles at birth was clearly defined by KL1 staining (Fig. 3).
The acquisition and compartmentalization of cytokeratin expression during the hamster thymus ontogeny are summarized in Fig. 4.

Cytokeratin Expression in Hamster Thymus Primary Cultures
Because an ontogenetic-dependent plasticity of FIGURE 2. Coexpression of cytokeratins typical of simple and stratified epithelia in the 13-day fetal thymus of hamster. Panel (a) depicts CK8 TEC being predominantly found in the central region of thymic lobules (x160). In panel (b), the whole epithelial network was labeled by the anti-CK19 MAb (x250), whereas in panel (c) (showing the same thymic lobule seen in (b)), less cells in the central region are stained with KL1 (x250). rationale for that was based on our previous findings showing that human primary TEC cultures revealed a CK pattern similar to that found in situ (Savino and Dardenne, 1988b). Nonetheless, the hamster thymic epithelium presented a CK pat-tern largely different from that found in vivo (Fig. 5). Although anti-CK19 was a pan-TEC marker in situ, a rather low percentage of cells (20-40%) in the hamster thymus primary cultures was labeled. Actually, the cytokeratin predominantly FIGURE 3. Cytokeratin expression pattern in the neonatal hamster thymus. In panel (a), the whole TEC network is seen labeled by the anti-CK19 MAb (xl00). Panel (b) shows the same thymic region stained with the anti-CK18 MAb, revealing that CK18 TEC are predominantly found in the central region (xl00). CK19-expressing TEC network is seen in panel (c) (x160), whereas in the same region, KL1 TEC are restricted to the central region (panel (d); x160). Additionally, panel (e) reveals KL1 subseptal TEC (arrowheads; x400), and panel (f) shows KL1 Hassall's corpuscles (large arrow) and subseptal TEC (short arrow; x400). expressed in vitro was CK18 (>80%), with CK8 expression being consistently lower, in parallel to what was observed during ontogeny. Regarding CK1/CK10 expression, immunocytochemically defined by KL1 MAb, only a minor percentage of TEC (1-5% was KL1-reactive. Interestingly, the previously described in vitro pattern was maintained in cultures of 1, 2, or 3 weeks. Moreover, it was not altered by hydrocortisone addition at the doses of 10-3, 10-5, or 10 -7 M, previously known to increase the numbers of KL1 + cells in a mouse TEC line (Savino et al., 1988). As summarized in Table 1, the in vitro pattern of CK expression by hamster TEC differs from that described for human TEC cultures (Savino and Dardenne, 1988b).
In Vivo Modulation of CK Expression by Hydrocortisone In a third group of experiments, we probed the responsiveness of hamster thymic microenvironment following hydrocortisone treatment in vivo. This strategy promoted an increase in the numbers of KL1 + cells in mice (Savino et al., 1988), as well as in the expression of extracellular matrix proteins (Lannes-Vieira et al., 1991).
Regarding KL1 reactivity, besides the normal subcapsullary-medullary staining, cortical TEC were also recognized by this MAb in hydrocorti-   sone-treated animals (Fig. 6a). Additionally, CK8 and CK18 expression (normally restricted to the cortex in young adult hamster thymus) was also detected in peripheral TEC of Hassall's corpuscles and some scattered medullary TEC (Fig.   6b).
Interestingly, concerning the other microenvironmental compartment, namely, the extracellular matrix component (whose distribution was demonstrated to be rather conserved in mammals; Meireles de Souza et al., 1993), the expression of fibronectin, laminin, and type-IV collagen was extended to the cortical region of hamster thymus (Fig. 7), similarly to that described in hydrocortisone-treated mice (Lannes-Vieira et al., 1991).

DISCUSSION
In the present work, we bring consistent data showing the modulation of cytokeratin expression by hamster thymic epithelial cells, following three distinct biological situations. As discussed in detail in what follows, the findings presented herein indicate a continuous plasticity of the thymic epithelium, at least in terms of intermediate filament proteins.
The first aspect to be considered regards CK expression during hamster thymus ontogeny. In relatively early fetal stages at day 13, it was noteworthy that both CK8/CK18 and CK1/CK10 pairs, respectively, markers of simple and stratified highly keratinized lining epithelia, were expressed in the central region of the thymic lobes. This simultaneous CK expression largely preceded the cortical-medullary distinction, which occurs 1 week after birth (Adner et al., 1965).
The results on KL1 staining contrasted with the data previously reported for the mouse thymus ontogeny (Savino and Dardenne, 1988a) in which KL1 / cells appeared later. In this regard, the concept for an interspecific diversity in the thymus FIGURE 6. Immunofluorescence of hamster thymus sections 24 hr after in vivo hydrocortisone treatment labeled with KL1 and anti-CK18 MAb, respectively, in panels (a) and (b). KL1 TEC can be seen in the cortical region, whereas CK18 staining is found in a Hassall's corpuscle (arrow) and cortical TEC (x 320): C: cortex. CK distribution should also comprise ontogeny. Moreover, because in the adult thymus, the localization of these CK pairs is different, it is likely to occur as downand up-regulation of CK expression along with the definition of the adult pattern. This is further supported because CK13 is acquired much later during hamster development, whereas some weak CK13 reactivity was detected from day 17 of fetal life to 1 week after birth in the rat thymus, and then was absent in the adult organ (Colic et al., 1990).
The second point deserving discussion concerns the modulation of CK expression in hamster TC cultures. This was clearly evidenced by the analysis of CK19 labeling. Although the anti-CK19 MAb stained the whole TEC network in situ, less than half of cultured cells were reactive. In contrast, there was a predominance of CK18 / TEC, with TEC expressing CK18 but not CK19. This may not be the result of cortical TEC selection in vitro (at least as they are phenotyped in situ), because the hamster thymic cortex is CK18/, CK8/, and CK19/. Nevertheless, such decrease in CK19 expression may be due rather to culture conditions used herein, which may be suitable for human but not for hamster primary cultures in terms of keeping CK19 expression.
In any case, this in vitro pattern did not significantly change even in 3-week-old cultures. In this respect, KL1 / cells, expected to represent the terminal stage of TEC differentiation, actually remained as a minor population, thus differing from the in vitro differentiation process reported for a rat TEC line (Itoh et al., 1982;Lobach and Haynes, 1987). Furthermore, the hydrocortisoneinduced increase in KL1 / cells that occurs in mouse TEC line (Savino et al., 1988) was not detected in hamster cultures. Thus, acquisition of high-molecular-weight CKs, typical of epithelial differentiation, did not occur in the hamster model in vitro. In addition, the results obtained with cultured hamster TEC, when compared to those reported for human TEC cultures, again point out an interspecific diversity of CK expression even when in vitro conditions are considered.
Besides the modulation of CK expression by hamster TEC during thymus ontogeny and in vitro conditions, the in vivo treatment with hydrocortisone significantly changed the distribution pattern of distinct cytokeratins. CK18 / cells (normally restricted to the cortex) also appeared in the medulla (adjacent to Hassall's corpuscles), whereas KLl-defined CK expression was enhanced, being evidenced throughout the cortex. A modulation of KL1 / cells upon a single dose of hydrocortisone was previously observed in the mouse thymus (Savino et al., 1988). Nonetheless, in this species, the medullary restriction of these cells was maintained. Thus, although a plasticity in the expression of a KLl-defined CK pair occurred in both species following this experimental condition, an interspecific diversity was again evidenced. In contrast, the responsiveness concerning the expression of extracellular matrix proteins (whose normal distribution is conserved in mammals) was similar to that observed in mice (Lannes-Vieira et al., 1991). In conclusion, the data presented and dis- cussed herein clearly demonstrate the continuous plasticity of the thymic epithelium, at least regarding expression of intermediate filament proteins. In this context, anti-CK monospecific antibodies can be ascribed as useful tools to define such plasticity, even when rather subtle modulations are involved, as we previously demonstrated in human thymomas (Savino and Dardenne, 1988b), murine Chaga's disease (Savino et al., 1989), and in the nonobese diabetic mouse . In this respect, our findings suggest that the thymic epithelium may not be divided in definitive subsets, and that phenotypic markers of TEC heterogeneity should be rather considered as markers to trace TEC plasticity. Briefly, we applied pan-CK markers, respectively, HTK and KL4, as well as monospecific anti-CK reagents recognizing CK8, CK18, CK19 (found in simple epithelia), and CK13 and the CK1/CK10 pair (typical of stratified epithelia). With regard to ECM reagents, we used rabbit polyclonal antisera specifically recognizing laminin, fibronectin, or type-IV collagen (Institut Pasteur, Lyon). The distribution of these proteins in the normal thymus have been described previously for humans (Berrih et al., 1985), mice (Lannes-Vieira et al., 1991), and a variety of mammals (Meireles de Souza et al., 1993). Sec-ondary antibodies used herein were described in Meireles de Souza et al. (1993). also Immunohistochemistry Acetone-fixed, 4-Bm thymus sections or methanol-fixed cell cultures were subjected to indirect immunofluorescence or immunoperoxidase assays. Description of these techniques can be found in Meireles de Souza et al. (1993).
We chose the MEM DVal medium because it favors the epithelial cell proliferation in comparison to fibroblast growth (Gilbert and Migeon, 1975). Culture medium was changed twice a week, and cells were analyzed on days 7, 14, and 21 of culture.

Hydrocortisone Treatment
Primary cultures of hamster TEC were treated with sodic hemisuccinate of hydrocortisone (Sigma Co., St. Louis, MO), with doses ranging from 10 -3 to 10 -7 M. Cultures were fixed in absolute methanol on days 7 and 14, and evaluated for CK expression by immunofluorescence assay. When in vivo experiments were carried out, young adults were injected i.p. with a single dose of hydrocortisone sodic hemisuccinate (10mg/20g of body weight), being sacrificed 24 hr later. Thymuses were excised, snap frozen into liquid nitrogen, and processed for immunohistochemistry.