Reprints Available Directly from the Publisher Photocopying Permitted by License Only Induction of Apoptosis in Thymocytes by Prostaglandin E2 in Vivo

In vivo administration in mice of a synthetic analog of prostaglandin E2 (PGE2) caused a selective and dramatic decrease of CD4+CD8 double-positive, CD3/T-cell-receptor-ccb 1 cells in the thymus. This loss was corticosteroid-independent and not affected by Cyclosporin A. The disappearance of CD4+CD8 thymocytes was strictly correlated with the induction of apoptosis inside the thymus as shown by morphological studies and by the induction of intracellular transglutaminase expression. Considering that PGE2 has been found to be produced by different cell populations inside the thymus, these results indicate that PGE2 may act as endogenous signals for apoptosis during T-cell differentiation.


INTRODUCTION
It is generally acknowledged that prostaglandin E2 (PGE2) exerts a powerful modulatory action on mature T cells (Goodwin, 1985;Vercammen and Ceuppens, 1987;Betz and Fox, 1991). Conversely, little information exists on its effects in the course of T-cell differentiation (Rinaldi-Garaci et al., 1983). The programmed cell death (PCD)of thymocytes is considered to be a crucial event that occurs during the intrathymic phase of T-cell differentiation. This death of physiological significance occurs in tissues by an active cellular phenomenon of self-destruction, called "apoptosis." It requires coordinate expression of regulatory proteins, as bc12, and enzymes, as Ca++/Mg+/-dependent endonuclease and tissue transglutaminase (tTG), causing morphological modifications of the cell and leading to a final irreversible damage of DNA, characterized by molecule fragmentation Arends and Wyllie, 1991;Piacentini et al., 1991aPiacentini et al., , 1991b. Apoptosis would lead to the controlled removal of thymocytes, during both positive and *Corresponding author.
negative selection of functionally mature T cells (Rothemberg, 1990;Boyd and Hugo, 1991;von Boehmer, 1991;Zugic, 1991). The intrinsic mechanisms and the biochemical mediators of PCD in thymocytes in vivo are not yet clear (McConkey et al., 1990b). Moreover it is difficult to understand how T-cell-receptor (TCR)-mediated signals could result in two distinct thymocyte fates, that is, protection from PCD in the case of positive selection or induction of PCD in the case of negative selection. Recently, it has been demonstrated that an intracellular increase of cAMP stimulates PCD in rat and mouse thymocytes in vitro (McConkey et al., 1990a;Suzuki et al., 1991).
Here we report the evidence that administration in mice of a synthetic analog of PGE2, 16,16dimethyl-PGE2 methyl ester (DI-M-PGE2), induces apoptosis of thymocytes in vivo, and that thymocytes at phenotypically different stages of differentiation show variable sensitivity to PGE2.  (Fesus et al., 1987(Fesus et al., , 1989Arends and Wyllie, 1991;Piacentini et al., 1991aPiacentini et al., , 1991b. By contrast, the enzyme expression is not enhanced during necrosis (Fesus et al., 1987). tTG, by catalizing covalent cross links between polypeptide chains, leads to the assembly of a stable protein scaffold (insoluble in detergents and chaotropic agents) that prevents the release of harmful molecules from the dying cell before its final degradation by phagocitosis (Fesus et al., 1989). The induction of tTG could so be considered as an early event during apoptosis. On the basis of these findings, we monitored the expression of tTG, in parallel with the morphology of thymus cells, in order to characterize apoptosis after DI-M-PGE2 in vivo administration. Upon a single DI-M-PGE2 injection, tTG activity was increased over the controls as early as 3 hr, reaching a twofold increase at 24 hr. Repeated treatments had additive effects in enhancing the enzyme activity ( Table 1). The effect of PGE2 was limited to the thymus, as indicated by the absence of induction of tTG in other organs, such as spleen (control= 0.75+0.04nmol/hr/mg protein, mean+SD, n=3; DI-M-PGE2=0.35+0.06 nmol/hr/mg protein, n=3) or liver (control=0.62+0.05 nmol/hr/mg protein, n=3; DI-M-PGE2=0.45+0.08 nmol/hr/mg protein, n=3). Immunohistochemical analysis of thymus  after DI-M-PGE2 administration showed a large induction of tTG protein in several cells mainly localized in the thymus cortex (Figs. 2A to 2C). The morphology of positive cells showed the distinctive features of apoptosis (condensed chromatin and nuclear fragmentation) that appeared unequivocal when the immunostaining was performed in single-cell suspensions of thymocytes freshly isolated from DI-M-PGE2 treated mice (Fig. 2D).

Effects
Effect of PGE2 Administration on CD4-and CD8-Identified Thy, mocytes in

Adrenalectomized Mice
Considering that a cell loss of CD4/CD8 / thymocytes, similar to that observed by us after DI-M-PGE2 treatment, is also induced in vivo by glucocorticoids (Screpanti et al., 1989), we have investigated if the elimination of CD4/CD8 + thymocytes caused by PGE2 could be mediated by endogenous steroid production. In fact, it is known that glucocorticoids induce apoptosis of thymocytes Wyllie and Morris, 1982;Ojeda et al., 1990). Therefore, we have repeated the experiments in adrenalectomized mice. After 4 days of treatment at the dose of 0.25 mg/kg/day (the maximal tolerated dose of DI-M-PGE2 by adrenalectomized mice), DI-M-PGE2 administration resulted in a cell loss of CD4/CD8 + thymocytes, with respect to untreated controls, that was similar in intact as well as in adrenalectomized mice (Table 2). Thus, it is reasonable to hypothesize that PGE2 action on thymocytes is not mediated by endogenous corticosteroids.
Effect of PGE2 Administration on CD4-and CD8-Identified Thymocytes in Cyclosporin A-Treated Mice It has been previously reported that apoptosis of thymocytes induced in vivo by anti-CD3 antibodies can be inhibited by Cyclosporin A (CsA) (Shi et al., 1989). In order to verify an eventual relationship between apoptosis induced in vivo FIGURE 2 Immunolocalization of tTG protein in mouse thymus and freshly isolated thymocytes. Immunostaining both on paraffin included thymus sections and freshly isolated thymocytes was performed 3 hr after the last DI-M-PGE2 injection using an affinity-purified monospecific rabbit IgG against soluble "tissue transglutaminase." Biotinylated goat antirabbit IgG was used as second antibody followed by a preformed avidin-horseradish peroxidase complex. Cells were counterstained in Mayer's hemalum. In thymus from control animals, the positivity to the anti-tTG antibody is limited to the endothelial cells lining the vessels (A). On the contrary, note the presence in the thymus cortex of many immunopositive (arrows) presumptive preapoptotic thymocyte after PGE2 administration (B); Bar=60/ria. Higher magnification of the thymus cortex from PGE2-treated animals (C); Bar=12/m. Intense positivity to anti-tTG antibody is localized in cells showing a picknotic nucleus (arrows). Isolated thymocytes. After extensive washing in PBS, cells were smeared on slides, fixed in 2.5% paraformaldehyde, and, after immunostaining, counterstained with Mayer's henaalum. In thymocytes from control mice, no positive reaction was observed (data not shown); in cells isolated from PGE2-treated animals, an increase in tTG protein expression is observed (D); Bar=60/m. Note the intense staining of shrinked cells showing the condensed chromatin marginated at the cell periphery typical of apoptotic bodies (inset, Bar=12/m). High concordant results were obtained in three independent experiments. (See Colour Plate VII at the back of this publication). by anti-CD3 antibodies and that induced by DI-M-PGE2, we have investigated the effects of a simultaneous administration of PGE2 and CsA. The flow cytometry analysis performed 24hr after the last treatment did not show any difference between the cell loss of CD4/CD8 + thymocytes in mice treated with DI-M-PGE2 alone and those treated with DI-M-PGE2 plus CsA (Table 3).
In addition, no changes in tTG activity were observed (data not shown).

DISCUSSION
A restricted number of experimental models of induction of thymocyte apoptosis has so far been described. In fact, apoptosis of thymocytes has been unequivocally proved to be induced in vitro by glucocorticoids Wyllie and Morris, 1982;Ojeda et al., 1990), radiations (Yamada and Ohyama, 1988), calcium ionophores and phorbol ester (Kizaki et al., 1989), or in vitro as well as in vivo by anti-CD3 antibodies (Shi et al., 1989;Smith et al., 1989) and some peptide antigens Murphy et al., 1990). Our results provide the first direct evidence for a pharmacological induction of thymocyte apoptosis in vivo. In fact, it is described by a simple and highly reproducible pharmacological experimental model of thymocyte apoptosis in vivo. Moreover, thd recently obtained evidence for the induction of thymocyte apoptosis in vitro by PGE2 and other intracellular cAMP elevating agents (McConkey et al., 1990a;Suzuki et al., 1991) furnishes a clear in vitro experimental support to the observation we have obtained in vivo. The in vivo PGE2-induced PCD includes pecularities that extend and clarify the previously described thymocyte apoptosis features. In fact, it is not mediated by endogenous glucocorticoids, nor inhibited by CsA as in vivo CD3induced PCD. Indeed, glucocorticoids seem not to be produced inside the thymus, nor could radiations or anti-CD3 antibodies constitute physiological triggers for thymocyte apoptosis. On the contrary, PGE2 are produced in large amounts inside the thymus by a wide variety of cell types as thymic epithelial cells, dendritic cells, macrophages (Gallily et al., 1985;Homo-Delarche, 1985;Nieburgs et al., 1985), and also nurse cells as recently reported (McCormack et al., 1991). Interestingly, a selective elimination of double-positive immature thymocytes by a thymic epithelial cell line has been reported in vitro (Nakashima et al., 1990); on the other hand, it has been recently demonstrated that thymic macrophages or dendritic cells are associated to different subsets of developing thymocytes, suggesting a role for thymic rosettes, consisting of thymocytes attached to a central stromal cell, in the maturation steps in the thymic cortex (Shortman and Vremec, 1991). Thus, it is possible that PGE2, endogenously produced by stromal cells inside the thymus during the intercellular contacts, which seem to play an important role during Tcell development, could locally reach concentrations active to induce cell death of thymocytes by apoptosis. The different sensitivity of CD3/TCR-c]/lo, CD3/TCR-c]/int, or CD3/TCRc]/hi, thymocytes to PGE2, could be involved in the mechanism that leads to opposite final events (selection or removal by PCD) in different phases of T-cell differentiation. Our results could contribute to explain the mechanism involved in the physiological elimination of the majority of thymocytes at the CD4+CD8 / stage. A very rapid clearance by macrophages of apoptotic thymocytes could result in the difficulty to identify well the physiologically occurring phenomenon. On the other hand, the effects caused byPGE2, at the pharmacological doses we have used, could overcome the capacity of phagocytic cells to eliminate dying cells, thus rendering the phenomenon as evident when using an early marker of apoptosis as tTG.

Drugs and Treatment
Prostaglandin Mice were injected i.p. with 16,16-dimethyl prostaglandin E2 (Di-M-PGE2) (Cayman Chem. Co., Ann Arbor, MI) at the doses of 0.25 mg/kg body weight, 0.5 mg/kg, and 1 mg/kg once a day for a time ranging from I to 4 consecutive days.
Immunofluorescence staining and flow cytometry analysis were then performed. The following antibodies were utilized: phycoerythrin conjugate antimouse L3T4 and fluorescein conjugate antimouse Lyt-2 (Becton Dickinson, Mountain View, CA) for a two-colors analysis of CD4-and CD8-positive cells, respectively; fluorescein conjugate anti-CD3-e (clone 145-2Cll) (Boehringer Mannheim Bioch., Mannheim, Germany) for a single-color analysis; fluorescein conjugate anti-aft TCR (H57-597 mAb), kindly provided by L. Jones (NCI, NIH, Bethesda), for a single-color analysis. Phycoerythrin conjugate antihuman CD4 and fluorescein conjugated antihuman CD8 (Becton Dickinson) were used as unrelated controls for background detecting. Staining was performed at 4 C for 30 min. After treatment, cells were washed twice in PBS containing 0.02% sodium azide and flow cytometry analysis was performed using a FACScan (Becton Dickinson). In two-color analysis, marks were set to indicate quadrant boundaries limiting 99.8% of the background events in the lower left quadrant. In single-color analysis, markers were set to indicate the upper and lower boundaries of CD3/TCR-ofl lo, int, or hi populations for comparison among treatment groups. The first boundary was obtained by limiting 99.7% of the background events, and the others were set arbitrarily on the basis of the curve profile obtained in control samples and maintained in the experimental samples. Data collection was gated on live thymocytes by forward and side angle scatter, utilized to exclude dead cells, debris, very large nonlymphoid cells, and cell aggregates. Data represent 5000 events.
Cyclosporin A CsA (Sandoz, Basel, Switzerland) was administered at the dose of 50 mg/kg of body weight, daily for 4 consecutive days, alone or immediately after Di-M-PGE2 administration.

Immunofluorescence Staining and Flow
Cytometry Analysis Thymuses were individually processed 12hr after the last injection by gentle teasing in RPMI 1640. The resultant cell suspension was filtered through a nytex mesh, washed twice with RPMI 1640, and resuspended in PBS at 2x107/ml cells.

Immunocytohistochemistry
Immunostaining on paraffin included thymus sections and freshly isolated thymocytes was performed 3 hr after the last DI-M-PGE2 injection using as primary antibody (diluted 1:100) an affinity-purified monospecific rabbit IgG raised against soluble "tissue transglutaminase" of human red blood cells (kindly furnished by L. Fesus, University Medical School of Decebren, Hungary) in a wet chamber overnight at 4 C. Biotinylated goat antirabbit IgG was used as a second antibody followed by a preformed avidin-horseradish peroxidase complex (Immunon, Detroit, MI). The reaction was devel-oped using aminoethylcarbazole as chromogen substrate and 0.01% H202. Cells were counterstained in Mayer's hemalum. Endogenous peroxidase activity was blocked by methanol-H202.
Isolated thymocytes were obtained as previously described for flow cytometry analysis. After extensive washing in PBS, cells were smeared on slides, then fixed in 2.5% paraformaldehyde and, after immunostaining, counterstained with Mayer's hemalum.