Developmental and Anatomical Patterns Of IL-2 Gene Expression in Vivo in The Murine Thymus

Interleukin-2 (IL-2) is a potent growth factor that mature T lymphocytes synthesize and use as a proliferation signal. Much controversy has arisen concerning whether it is used to drive the extensive proliferation of immature pre-T cells in the thymus. Immature thymocytes acquire the competence to express IL-2 at an early stage, but it has remained uncertain whether they are activated to exercise this competence in vivo. Therefore, we have used in situ hybridization and immunohistochemistry on serial sections obtained from fetal and adult thymuses of normal C57BL/6 mice and of mice bearing the scid defect to determine where, when, and whether IL-2 is expressed in vivo. Our results show a striking spatial and temporal pattern of IL-2 expression in the normal fetal thymus. We detected a burst of IL-2 mRNA accumulation at day 14.5 of gestation, which rapidly decreased by day 15. At day 15, we observed maximal IL-2 protein production that subsequently decreased by day 16 of gestation. Both in situ hybridization and immunohistochemical staining revealed an unexpectedly strict localization of IL-2 expressing cells to patches around the periphery of the fetal thymus, creating a previously unrecognized compartment of high IL-2 protein content. IL-2 production in the day-15 fetal thymus appeared to be unaffected by the scid mutation, indicating that this response is likely to be T-cell receptor (TcR)-independent. Several features distinguish the IL-2 induction pattern in the adult thymus from that in the fetal thymus. In the normal adult thymus, IL-2-expressing cells are extremely rare (found at a frequency of 10-7), but they are reproducibly detectable as isolated cells in the outer cortex and subcapsular region of the thymus. Unlike the fetal thymic IL-2 producers, the IL-2 producers in the adult thymus are completely eliminated in mice homozygous for the scid mutation. This suggests that the IL-2-expressing cells in the normal adult thymus are of a more mature phenotype than the immature, TcR-negative cells that accumulate in the scid adult thymus. Thus, our work demonstrates that two developmentally distinct types of cell interactions induce IL-2 expression in vivo: one, a broadly localized interaction in day 14‑15 fetal thymus that is unaffected by the scid mutation; the other, a rare event that occurs asynchronously from late fetal through adult life, but which is completely eliminated by the scid defect. These results imply that significant differences exist between the physiological processing of thymocytes in the fetal and postnatal thymic microenvironments.

Interleukin-2 (IL-2) is a potent growth factor that mature T lymphocytes synthesize and use as a proliferation signal. Much controversy has arisen concerning whether it is used to drive the extensive proliferation of immature pre-T cells in the thymus. Immature thymocytes acquire the competence to express IL-2 at an early stage, but it has remained uncertain whether they are activated to exercise this competence in vivo. Therefore, we have used in situ hybridization and immunohistochemistry on serial sections obtained from fetal and adult thymuses of normal C57BL/6 mice and of mice bearing the scid defect to determine where, when, and whether IL-2 is expressed in vivo. Our results show a striking spatial and temporal pattern of IL-2 expression in the normal fetal thymus. We detected a burst of IL-2 mRNA accumulation at day 14.5 of gestation, which rapidly decreased by day 15. At day 15, we observed maximal IL-2 protein production that subsequently decreased by day 16 of gestation. Both in situ hybridization and immunohistochemical staining revealed an unexpectedly strict localization of IL-2 expressing cells to patches around the periphery of the fetal thymus, creating a previously unrecognized compartment of high IL-2 protein content. IL-2 production in the day-15 fetal thymus appeared to be unaffected by the scid mutation, indicating that this response is likely to be T-cell receptor (TcR)-independent. Several features distinguish the IL-2 induction pattern in the adult thymus from that in the fetal thymus. In the normal adult thymus, IL-2-expressing cells are extremely rare (found at a frequency of 10-7), but they are reproducibly detectable as isolated cells in the outer cortex and subcapsular region of the thymus. Unlike the fetal thymic IL-2 producers, the IL-2 producers in the adult thymus are completely eliminated in mice homozygous for the scid mutation. This suggests that the IL-2-expressing cells in the normal adult thymus are of a more mature phenotype than the immature, TcR-negative cells that accumulate in the scid adult thymus. Thus, our work demonstrates that two developmentally distinct types of cell interactions induce IL-2 expression in vivo: one, a broadly localized interaction in day 14-15 fetal thymus that is unaffected by the scid mutation; the other, a rare event that occurs asynchronously from late fetal through adult life, but which is completely eliminated by the scid defect. These results imply that significant differences exist between the physiological processing of thymocytes in the fetal and postnatal thymic microenvironments. INTRODUCTION T cells, unlike other cell types of the hematopoietic lineage, differen.tiate in the thymus, *Corresponding author. where differentiation is accompanied by extensive proliferation, apparently triggered by mitotic signals from the thymic stroma. Thymusspecific lymphostromal interactions are critical for driving T-cell development, yet both the nature of the signals and the molecular basis of their impact remain unknown. Among the 86 J.A. YANG-SNYDER AND E.V. ROTHENBERG properties acquired and modulated as thymocytes develop is the competence to express "response" genes that drive proliferation in mature T cells, such as the gene encoding the potent growth factor interleukin-2 (IL-2) (McGuire and Rothenberg, 1987;Howe and Mac-Donald, 1988;Rothenberg et al., 1988;Tentori et al., 1988b;Fischer et al., 1991;Chen and Rothenberg, 1993). Although the inducibility of these genes at certain stages of differentiation is easily demonstrable under artificial conditions in vitro, it is not known whether intrathymic signals provide inductive stimuli for these genes in vivo. Thus, it is unclear to what extent growth factors such as IL-2 are normally available to contribute to the hormonal microenvironment of the thymus. Similarly, it has remained uncertain whether the characteristic signaling pathways that activate IL-2 expression in vitro are utilized in normal thymocyte development in vivo.
Recent gene-disruption experiments have formally proven that IL-2 is not required for the generation of normal cell populations in the thymus (Schorle et al., 1991). However, severe perturbations can be induced in thymocyte development, in vivo or in organ culture, by anti-IL-2R antibody treatment (Jenkinson et al., 1987;Tentori et al., 1988a;Zuiga-Pfliicker and Kruisbeek, 1990;Zuiga-PfliJcker et al., 1990), by the addition of excess IL-2 (Skinner et al., 1987;Plum and de Smedt, 1988;Waanders and Boyd, 1990), or by the introduction of a transgenic IL-2Rc chain with inappropriate species specificity (Kroemer et al., 1991). Thus, the possibility remains that IL-2 production in vivo plays some developmental role. Furthermore, studies from our own laboratory and from others have shown that the intrinsic competence to express the IL-2 gene is shared by both mature and immature thymocytes (Howe and MacDonald, 1988;Rothenberg et al., 1990;Fischer et al., 1991). However, it is unknown whether the thymic microenvironment can actually provide signals to induce IL-2 expression in either or both populations of thymocytes. Therefore, IL-2 production (1) could potentially be induced at any of several stages in vivo and (2) might cause autocrine or paracrine effects, either early or late in T-cell development.
These issues make it instructive to determine the conditions under which developing thymocytes actually exercise their competence to express IL-2 in vivo.
Several laboratories have analyzed cytokine expression during fetal thymocyte development in vivo as an approach to determine which factors might be present in the thymus during T-cell development. Von Boehmer and coworkers initially reported that IL-2 is secreted as fetal thymocytes mature in organ culture (Kisielow et al., 1985). Subsequent studies have shown that murine embryonic thymocytes harvested at different times during gestation express a variety of cytokine mRNAs, using either in situ hybridization to cytocentrifuged preparations of freshly isolated cells (Carding et al., 1989(Carding et al., , 1990Zufiiga-Pfliacker et al., 1990) or reverse transcription/ polymerase chain reaction (rtPCR) analysis of total RNA samples obtained from fetal thymuses (Montgomery and Dallman, 1991). Most of these studies have focused exclusively on dissociated cell suspensions, making localization and identification of the IL-2-expressing cells within the thymus impossible. On a more general level, these studies have failed to distinguish between the developmental events involved in the fetal ontogeny of the thymus itself and those involved in the maturation of T-cell precursors processed therein. In adult animals, thymocyte development generates a different spectrum of T-cell subtypes than that generated during thymocyte development in the fetus (reviewed in Rothenberg, 1992). This could be attributable either to differences in the thymic environment or to the difference in hematopoietic origins between the precursors that seed the fetal and the postnatal thymus (Ikuta et al., 1990). Thus, it is not clear whether IL-2 production is a universal feature of T-cell development or a peculiarity of the cell types present in the fetal thymus.
In order to examine the spatial and temporal pattern of IL-2 gene expression as induced naturally during T-cell ontogeny in vivo, we have used a combination of in situ hybridization and immunohistochemical staining of mouse thymus sections isolated at different stages of development. Our findings show different and distinctive patterns of IL-2 gene expression in fetal versus adult thymuses. Additionally, IL-2 gene expression appears to be normal in the fetal thymus but perturbed in the adult thymus of animals that cannot generate mature T cells as a result of the severe combined immunodefficiency (scid) mutation.  Figure 1 shows the patterns of IL-2-specific hybridization in the murine fetal thymus at different stages of gestation as defined using 590 nt antisense or sense riboprobes. At day 14, a few highly positive cells were found, mainly in the outer region of the thymus (Fig. 1A). At day 14.5, the outer region of the thymus contained a dramatically increased number of highly positive cells, clustered in distinctive, large patches of hybridization one cell deep that "outlined" the periphery of the tissue (Fig. 1C). At this stage of gestation, no corticomedullary boundary was observed in the thymus; cells appeared equally distributed throughout the organ (Figs. 1C and 1D). Although the majority of sections of day 14.5 fetal thymus did not include any regions of strong hybridization, every thymus examined contained sections exhibiting this striking localization of highly positive cells. By day 15, the percentage of IL-2 mRNA positive cells decreased, and only an occasional highly positive cell was found (6-8 per half lobe equivalent; Fig. 1E). Isolated positive cells were also observed later, at day 18 of gestation, and distinct cortical and medullary regions w.ere identifiable by this time by DAPI staining (data not shown). The remaining IL-2 mRNA positive cells in the fetal thymus at later stages of gestation were still restricted to the cortex and not to the medulla. Thus, highly positive IL-2 mRNA-expressing cells were present at day 14 of gestation in the fetal thymus and some highly positive cells persisted through day 18 of gestation. However, a dramatic peak of highly localized IL-2 gene expression was observed at day 14.5, which decreased abruptly by day 15.
Pattern of IL-2 Expression in the Fetal Thymus at Different Stages of Development by Immunohistochemistry To confirm the pattern of IL-2 expression as determined from the in situ hybridization data, fetal thymuses isolated dt various stages of gestation were also stained with an anti-IL-2 mAb ( Fig. 2). At day 14-14.5, weak immunoreactivity was observed ( Fig. 2A; data not shown). At day 15, strong staining appeared in the outer region of the tissue (Fig. 2C). This staining pattern was not seen when normal rat IgG was substituted for the anti-IL-2 mAb (Fig. 2D); the IL-2 specificity of the staining reagent was further confirmed by control experiments with excess rmuIL-2 (see below). At day 16, immunoreactivity was still present around the periphery of the thymus, albeit at a reduced level (Fig. 2E), and at day 18, only isolated immunoreactive cells were found in the cortical region. A background of endogenous peroxidase-positive cells became detectable by day 18 (data not shown), but these cells could be distinguished from IL-2 immunoreactive cells by prior blocking and peroxidase quenching (for additional controls, see below). Nonetheless, based on these characteristics, the pattern of IL-2 protein expression, detected by immunohistochemistry, recapitulated the pattern of IL-2 mRNA accumulation as determined by in situ hybridization. Additionally, maximum IL-2 protein expression was observed 12 hr following maximum IL-2 mRNA accumulation.

Vy3 and CD16 Expression versus IL-2
Expression in Day 15 Fetal Thymus as Determined by Immunohistochemistry IL-2 has been implicated in the growth of Vy3 + T cells and natural killer (NK) cells from the fetal thymus in vitro (Skinner et al., 1987;LeClerq et al., 1990). Therefore, we were interested in determining whether intrathymic IL-2 was associated with the generation or expansion of particular subsets of T cells or NK cells in the fetal thymus in vivo. By day 15 of gestation, almost all of the rare thymocytes expressing surface TcR specifically utilize the Vy3 segment in TcR gene rearrangements (Havran and Allison, 1988 We used a V y3-specific (536) or CD16-specific (2.4G2) mAb along with the IL-2-specific mAb for immunohistochemical staining of adjacent serial sections of day 15 fetal thymus to determine the relative location of V,3 + and CD16 / cells with respect to IL-2-producing cells (Fig. 3). At this stage of gestation, the majority of the cells in the fetal thymus were cytoplasmic CD3e + (data not shown). CD16 / cells were detected in a heterogeneous distribution throughout the fetal thymus. V'3 + cells were also detectable in a similar distribution (Fig. 3), but were substantially less abundant than CD16 / cells. Neither cell type was concentrated in the peripheral region of the lobe. Thus, these results gave no clear evidence for colocalization of IL-2-producing cells with either V'3 or CD16 expression (see Discussion).
IL-2 Expression in the Scid Day 15 Fetal Thymus Mice homozygous for the scid mutation fail to rearrange their TcR genes normally, causing a developmental arrest that blocks the production of all classes of TcR / thymocytes (Bosma and Carroll, 1991). In order to determine whether the scid defect affected IL-2 expression in the fetal thymus, we used immunohistochemistry to detect IL-2-producing cells in the scid day 15 fetal thymus (Figs. 2I to 2L). Scid day 15 fetal thymuses appeared normal in terms of the intensity and localization of IL-2 immunoreactivity (Fig.   2I). This staining was also shown to be IL-2specific. When an excess of exogenous recombinant IL-2 was added, IL-2 immunoreactivity was abolished (Figs. 2K and 2L); as a control, addition of recombinant IL-4 had little effect (data not shown). Likewise, the gross architecture and size of the scid day 15 fetal thymus appeared to be normal in terms of cell density and number (data not shown). In situ hybridization of scid day 15 fetal thymus sections revealed that a similar number of highly labeled IL-2 mRNA positive cells were present as compared to the normal counterpart (data not shown). Hence, at this stage of gestation, TcR rearrangement and expression are not prerequisites for IL-2 expression in vivo.
Analysis of IL-2Rc expression confirmed the lack of gross perturbation of thymocyte development in scid mice at this age. When sections were stained with a rat anti-mouse IL-2Ro mAb, normal and scid day 15 fetal thymuses were similar with regard to the number and distribution of IL-2Ro + cells in the thymus (Figs. 4A and 4B). By contrast, differences were readily apparent when comparing the IL-2Rc staining pattern of the normal versus scid adult thymus. The overwhelming majority of the cells in all regions of the scid adult thymus are IL-2Rc+, as compared with the irregular clusters of IL-2Rc + cells in the cortex of the normal tissue (Figs. 4C and 4D). This aberrant IL-2Rc staining pattern of the scid adult thymus convincingly showed the developmental arrestblocking progression beyond the IL-2Rc + immature stage. These staining results indicated that the scid mutation did not severely affect the control of IL-2 or IL-2Rc expression in the fetal thymus, in spite of its severe developmental effects observed in the adult thymus.
In Situ Hybridization of Normal Adult Thymus Sections In previous analyses of cytocentrifuged preparations of adult thymus cell suspensions, we had failed to detect any cells expressing IL-2 mRNA (McGuire and Rothenberg, 1987;Rothenberg et al., 1990). Ribonuclease protection assays failed to detect any IL-2 transcripts in samples of up to 100/g of adult thymus total RNA, roughly equivalent to the amount of RNA isolated from half a thymic lobe (400tg total RNA/thymus; data not shown). However, the results of four independent experiments confirmed the presence of extremely rare, highly labeled IL-2 mRNApositive cells in the thymus of 4-week-old mice. Approximately 160 sections hybridized alternately with each strand of the 430-nt probe were analyzed per experiment (approximately 320 sections examined overall, corresponding to a thymus lobe; i.e., a half lobe equivalent for each probe). Highly labeled cells were found only in sections that had been hybridized with the antisense riboprobe. These IL-2-positive cells were reproducibly found in the outer cortex and subcapsular region of the thymus (Fig. 5A); the cortical versus medullary regions of the thymus were distinguished by higher or lower cell density within the section, as shown by DAPI staining (Figs. 4A and 4B). Surprisingly, of all the IL-2-FIGURE 1. (See Colour Plate VI at the back of this publication). In situ hybridization to sections of normal thymuses from days 14 to 18 of gestation for IL-2 expression, using either a 35S-labeled, 590-nt IL-2 antisense I.NA probe (IL-2; Figs. 1A, 1C, 1E, 1G) or the complementary sense strand (NC; Figs. 1B, 1D, 1F, 1H). Samples were counterstained with DAPI and observed under dark field microscopy. 14-day exposure; magnification x200. The average number of copies IL-2 per IL-2 mRNA ce11=224+18 aEL4 cells induced for hr with calcium ionophore and phorbol ester mixed with uninduced EL4 cells at the percentages indicated (a total of 5x106 viable cell equivalents per sample).
Total RNA extracted according to the method described by Chomczynski and Sacchi (1987). One-fifth of each sample (or RNA from lx106 cell equivalents) was used per protection. No protection was observed above background using total RNA isolated from uninduced EL4 cells.
bIdentical samples of cells processed described in Materials and Methods and analyzed by in situ hybridization. Signals from at least 5000 cells analyzed per sample.
'No. copies protected conversion factor phosphor imaser value for protected product cpm undigested probe used per protection ng probe used per protection 6 1023 molecules per mole molec, wt. protected product where the conversion factor (liquid scintillation cpm determined for known amounts of undigested probe. phosphor imager value)' No. copies IL-2 protected 106 No. copies IL-2 per avg. cell % ind. EL4 10 No. copies IL-2 per ind. cell % induced EL4 %IL-2 mRNA by in situ least 100-fold for immature thymocytes at all stages leading up to the CD3-CD4-CD8-(TN) IL-2Rc / stage of development (see Fig. 4) (Bosma and Carroll, 1991). If the IL-2 mRNApositive/immunoreactive cells in the thymus of normal animals were of an immature phenotype, they should be present in greater numbers in the scid adult thymus than in the normal counterpart.
On the other hand, if these cells had matured past the stage of TcR gene rearrangement, they should be depleted in the scid adult thymus. In situ hybridization revealed that IL-2 mRNA-positive cells were conspicuously absent from the scid adult thymus; both antisense and sense probes gave identical results in all sections analyzed (Figs. 5D and 5E). The absence of hybridization in the scid adult thymus samples was not due to artifactual degradation of probes hybridized to the scid thymus sections, because alternate sections hybridized with IL-2Rc antisense probes yielded strong signals (Fig. 5F), consistent with the results obtained from direct immunofluorescent staining (Fig. 4D).
For direct quantitative comparison, hybridiz-ation of scid thymus sections were carried out in parallel with hybridization of normal thymus sections (Fig. 6). On average, fewer sections per scid thymus were analyzed than per normal thymus because scid adult thymuses were substantially smaller than normal adult thymuses (scid thymuses, which lack all TcR / cell types, contain approximately 1/100 the number of viable cells found in normal thymuses; data not shown). Due to the reduced cell numbers in the scid adult thymus, the frequency of IL-2-expressing cells could not be proven to be lower in the scid adult thymus than in the normal adult thymus on a p_ cell basis. Nonetheless, the depletion of IL-2 mRNA-positive cells per lobe in the scid thymus was highly significant (Fig. 6). The absence of cells producing IL-2 protein in the scid adult thymus was difficult to confirm by immunohistochemistry due to a high background of endogenous, peroxidase-positive cells that were "reactive" in the absence of specific antibody, even after extensive H202/peroxidase quenching (data not shown). Nevertheless, based on the unambiguous in situ hybridization data, it 95 OOD oo Normal scid FIGURE 6. Compiled data of in situ hybridization for IL-2 expression in adult thymus from normal and scid mice. Adult thymus sections from normal and scid animals (open and shaded points, respectively) were analyzed by in situ hybridization with the 430-nt antisense or sense IL-2 RNA probe. Approximately half a thymic lobe was hybridized with each probe. The results using the antisense probe were compiled from four independent experiments. No signals were detected with the sense probe (data not shown). Normal and scid sections hybridized and analyzed in parallel are denoted with identical shapes (circles, diamonds). Positive cells were considered as having greater than 10 silver grains. The average, total numbers of sections analyzed per normal and scid thymus were 320 and 80, respectively, corresponding to complete thymic lobes. Slides were exposed for 21 days. appeared that IL-2 mRNA-positive cells in the normal adult thymus were very likely to be of a more mature phenotype than the developmentally arrested thymocytes found in the scid adult thymus.

DISCUSSION
Our results show that IL-2 expression is induced in thymocytes by intrathymic signals in vivo. This expression is subject to complex developmental regulation that constrains the possible roles of IL-2 as a mediator of signals. On the other hand, the induction of IL-2 itself serves as a probe for specific cell-cell interactions, which reveal surprising differences between the fetal and adult thymus.
Restricted Activation of IL-2 Expression in the Fetal Thymus Our results show that the day 15-16 fetal thymus includes distinct, substantial zones of low and high regional IL-2 protein concentration, in good agreement with the results of Waanders (1990). Twelve hours prior to maximal protein production, a burst of IL-2 message accumulation is observed in cells restricted to the identical area where IL-2 protein will be found, namely, the periphery of the fetal thymus. The combined in situ hybridization and immunohistochemical staining data reported here for day 14.5-15 fetal thymus show that only a minority of the cells in the fetal thymus were expressing IL-2 message or protein at the given time points. This interpretation is not in complete agreement with that of Zufiiga-Pflficker et al. (1990), who reported that 50-60% of fetal thymocytes expressed IL-2 mRNA at day 15 gestation, based on in situ hybridization to cytocentrifuged samples. The disparity in percentage of IL-2 mRNA-positive cells could be due to any of several factors. Because the kinetics of IL-2 expression are clearly transient, our time points might have missed the true peak. Alternatively, the timing of the actual impregnation of the mice could be affected by the setting of the 12-hr light-dark cycle in the respective mouse colonies, leading to apparent shifts in kinetics. Also, in principle, there could be a lower threshold of detection for in situ hybridization to cytocentrifuged cell suspensions than to tissue sections. Finally, as Zufiiga-Pfliicker et al. (1990) did not subdivide categories of IL-2 mRNA-positive cells based on silver grain count, they might have included cells that we would not consider to be unequivocally IL-2positive. However, our immunohistochemical staining data support our in situ hybridization results, indicating that only a highly localized minority population express IL-2 protein that is detectable at days 15-16 of gestation. Thus, according to these staining results, if 50-60% of cells do express an IL-2 message at the time of  (Rodewald et al., 1992). Furthermore, transgenic mice constitutively expressing IL-2 and IL-2Ro chain show great preferential expansion of NK cells (Ishida et al., 1989). Based on these data, the generation of either of these cell types in the fetal thymus in vivo could be positively regulated by the local concentration of IL-2. However, when we examined V,3, CD16, and IL-2 expression in the day 15 fetal thymus by immunohistochemical staining, we were unable to show a relationship between the localization of V,3 + or CD16 / cells and IL-2 producers (Fig. 4). Although we cannot formally disprove the possibility that V,3or CD16precursors of either cell type respond to IL-2, these results make it unlikely that IL-2 is limiting for the generation of Vy3 / T cells or CD16 / cells (presumptive T/NK precursors) in the fetal thymus in vivo. The targets for the burst of fetal IL-2 production therefore remain undefined.
In contrast to the results obtained from day 14.5-15 fetal thymuses, IL-2-expressing cells are extremely rare in the adult thymus with a frequency on the order of 10 -7 as determined by both in situ hybridization and immunohistochemical staining. Any cells responding to IL-2 in the adult thymus in vivo would most likely need to be in close proximity to these IL-2 producers, based on the minimal diffusion of IL-2 protein observed in the cortex. Unlike IL-2 production in the fetal thymus, no continuous zones of high IL-2 concentration are found in the adult organ. Furthermore, IL-2 may be available in these local sites for only a limited time. Our data suggest that the steady-state frequency of IL-2-expressing cells reflects the asynchronous activation of different rare cells. In the adult thymus, cells producing IL-2 protein were detected at an approximately threefold higher frequency than cells expressing IL-2 mRNA at any one time. The IL-2 mRNA is known to be labile (Shaw et al., 1988;Lindsten et al., 1989), and our results with the fetal thymus indicate that IL-2 protein can persist in the thymus at least 1 day longer than IL-2 mRNA. Thus, we interpret the excess of IL-2 immunoreactive cells over IL-2 mRNA-positive cells as evidence that IL-2 gene induction in the adult thymus is also transient but asynchronous; that is, that most cells synthesizing IL-2 protein have already shut off synthesis of the short-lived IL-2 mRNA. When the resulting limited avail-ability of IL-2 is compared with the 10-20% of thymocytes that are actively in cycle (Penit, 1986;Boyer et al., 1989), it is clear that thymocytederived IL-2 is grossly inadequate to drive the extensive proliferation in this organ.
Our results support the finding that transgenic mice homozygous for an IL-2 gene disruption did not make IL-2 and yet generated all major thymocyte populations in normal numbers (Schorle et al., 1991). However, the extreme rarity of IL-2-rich "niches" in the normal adult thymus raises a caveat about the interpretation of the gene disruption results. If IL-2 were normally required as a growth factor for a particular subset of cells in the thymus, then its targets must represent a very minor fraction of cells in the cortex, the deletion of which might not drastically affect thymocyte population dynamics as a whole. Thus, an effect of IL-2 on some minor or transient branch of the T-cell lineage cannot be ruled out.
Developmental Significance of IL-2 Induction In Vivo: IL-2 as an Indicator of a Rare Activation Event What may be the most telling aspect of our data is the access they provide into the intrathymic activation events that are detected as the induction of IL-2 gene expression in the murine thymus in vivo. These activation events depend both on the delivery of the appropriate signal from the thymic microenvironment and on the developmental status of the responding thymocyte (Howe and MacDonald, 1988;Rothenberg et al., 1990;Fischer et al., 1991). Little is known about the biochemical basis of thymic lymphostromal interactions, and key events in which activation signals are delivered have been more commonly inferred than demonstrated. On the other hand, the IL-2 induction response in vitro is relatively well characterized, both in terms of cytoplasmic signaling cascade and the transcription factors that are mobilized in the activated cells. Thus, our results provide strong evidence that this particular cascade of responses is indeed triggered in vivo in particular cells in the fetal and postnatal thymus.
As both immature (TcR-) and mature (TcRhigh) thymocytes can in principle make IL-2, the intrathymic IL-2 production observed could represent a response either of an immature cell, presum-98 ably to differentiating signals, or of a mature cell (possibly to antigen). We have utilized two complementary kinds of evidence to reach a provisional conclusion. One is the intrathymic location of IL-2-expressing cells: In postnatal mice, the cortex is a zone of differentiation and precursor expansion, and the medulla is a domain where mature thymocytes accumulate and encounter circulating cells and antigen. The other relates to the effect of the scid mutation on intrathymic IL-2 expression. By preventing correct TcR gene rearrangement, this mutation blocks all TcR-expression-dependent events in Tcell development while permitting those that precede TcR gene rearrangement. The results of our analysis suggest that both fetal and postnatal cells that make IL-2 in the thymus are responding to developmental signals, rather than environmental antigen, but that distinct, nonequivalent cell types are responding to these signals in the fetal versus postnatal thymus.
To elaborate, the fetal thymocytes that make IL-2 are unaffected by the scid mutation, even though the scid mutation effectively blocks the generation of fetal TcR / thymocytes. Therefore, the activation signal that induces IL-2 expression in fetal thymocytes is likely to be delivered to cells at a stage upstream of the scid arrest point, that is, prior to TcR expression. This interpretation is supported by the failure of IL-2-expressing cells to colocalize with TcR-expressing cells in the day 15 fetal thymus (i.e., V,3 + cells). By contrast, since the scid mutation does block the appearance of all IL-2-expressing cells in postnatal thymus, it is probable that postnatal thymocytes normally receive an IL-2-inducing signal downstream of the scid arrest point, as discussed further below. Thus, in the context of their own developmental lineage, the postnatal IL-2 expressors are likely to be more advanced than the fetal IL-2 expressors. However, the cortical location of the postnatal IL-2 producers argues against any possibility that they have completed their intrathymic processing. Instead, the signal that induces IL-2 expression is apparently delivered while cells are still in the differentiative domain of the thymus.

Successful induction of IL-2 expression in vivo
indicates not only that. thymocytes have the capacity to express IL-2 and that the thymic stroma has the capacity to deliver triggering signals, but also that the signal is matched to satisfy the activation requirements for the particular classes of thymocytes present. The failure of adult scid thymocytes to activate IL-2 expression could be caused by defects in any of these aspects. However, our in vitro studies (Chen and Rothenberg, 1993;Rothenberg et al., submitted) clearly show that the scid thymus is significantly enriched for cells with the competence to express IL-2, provided that a certain set of triggering conditions are used in vitro. Therefore, either a lack of inductive signal or a mismatch between the available signals and the requirements of the developmentally arrested cells seems to be implicated. One possible explanation for the failure of cells in the scid adult thymus to express IL-2 in vivo would be that the scid thymic microenvironment (i.e., the nonlymphoid component of the thymus) does not develop properly, due to the lack of feedback interactions from mature thymocytes. Indeed, there is some evidence for architectural abnormality of the scid thymus (Shores et al., 1990(Shores et al., , 1991van Ewijk, 1991). However, the main components reported to be reduced or absent in the scid thymus are medullary epithelium, and as previously noted, the induction of IL-2 expression that we detect in the normal postnatal thymus is never associated with the medulla. Thus, another explanation for the inability of cells in the scid adult thymus to express IL-2 in vivo seems more plausible, namely, that the scid mutation directly prevents the maturation of thymocytes themselves to the stage where they can respond to those IL-2 inductive signals generated by the postnatal thymic cortex. In this case, responsiveness to IL-2inducing signals in the adult thymus in vivo would be at least indirectly dependent on successful TcR gene rearrangement. Whether the TcR itself is implicated in the triggering event remains to be determined. We are currently in the process of defining the signals that induce IL-2 gene expression in the postnatal thymus in vivo.

Conclusions
To summarize, our work shows that IL-2 is expressed in the fetal and adult thymus of normal mice in vivo. Although the pattern of IL-2 gene expression differs in the fetal versus adult thymus, IL-2 expression is highly localized during both times in development. In the fetal thymus, high expression of IL-2 is synchronous and 99 transient. Surprisingly, fetal thymic expression of IL-2 in vivo is not disrupted by the scid mutation, suggesting that normal TcR expression is not a requisite for IL-2 gene expression in the cell types present in the fetal thymus. On the other hand, IL-2 gene expression appears to be dependent upon TcR gene expression in the adult thymus, because the scid mutation abrogates the appearance of IL-2-expressing cells in the adult thymus. Finally, our results indicate that IL-2 is available to cells in highly restricted areas of the thymus during development. Taken together, this work implies that TcR-independent and TcR-dependent interactions are made within the thymus at different stages of development that, in turn, provide the proper inducing signal(s) in resident cells for IL-2 expression in vivo. Likewise, differences between the fetal and postnatal thymic microenvironments were clearly evident, both in terms of IL-2 protein content and in terms of the subsets of thymocytes induced to express IL-2 therein. The nature of the responsive and responding cells, and the criteria by which they are distinguished from others in the same stage of differentiation are intriguing issues for further investigation.

MATERIALS AND METHODS
Mice C57BL/6 Tla / and C.B-17-scid animals were bred and maintained in our own facility. Thymuses from embryos at various stages of gestation were obtained from females using timed matings. The appearance of a vaginal plug was designated as day 0 and on this day the females were separated from the males. Young "adult" animals were used at 4 weeks of age. All scid animals were maintained in an Isotec flexible film isolater (Indianapolis, Indiana) without antibiotic treatment, except those used for timed matings that were kept outside the isolator.

Removal and Sectioning of Tissues
Fetal thymuses were removed from embryos every 12 hr on and following day 14 of gestation (pregnant females were sacrificed at 10:00 A.M. and 10 P.M.) and were immediately embedded in Tissue-Tek O.C.T. compound (Miles Inc., Kankakee, Illinois). Thymuses removed from adult animals were frozen in n-pentane/dry ice and stored at-70C and embedded in O.C.T. prior to sectioning. For in situ hybridization, 6-8-/m serial cryosections were collected onto polycolysine-coated glass slides, dried briefly, fixed in freshly made 4% paraformaldehyde in PBS for 1-4 min, rinsed in PBS and stored in 70% ethanol at 4C. Alternatively, rinsed sections were stored with desiccant at-70C and transferred to ethanol at 4C, 3-7 days prior to hybridization. For sensitivity controls/hybridization standards, the murine thymoma EL4 was induced to express IL-2 by culturing the cells in the presence of the calcium ionophore A23187 (70nM) and phorbol myristate acetate (PMA; 17 nM, 10 ng/ml) for 5 hr in RPMI 1640 (Irvine Scientific, Santa Ana, California) based complete medium supplemented with 5% fetal bovine serum (v/v) as described elsewhere (Chen and Rothenberg, 1993). Following induction, cells were harvested, mixed with uninduced cells at a known ratio of induced-to-uninduced cells, embedded in O.C.T., sectioned, fixed, and stored accordingly. For immunofluroescence and immunohistology, 6-8-/m cryosections were collected onto 3-aminopropyltriethoxysilane (Fluka, Ronkonkoma, New York) coated slides, air dried for approximately 1 hr, fixed in cold acetone for 10 s and stored with desiccant at-20C. For activated lymph node samples, adult mice (8-12 weeks) were immunized by hind foot pad injection using Freund's complete adjuvant. Popliteal, inguinal, and pericardial lymph nodes were removed 8 days after immunization and tissues were processed as described before.

Hybridization Probes
All IL-2 probes used for in situ hybridization were IL-2 cDNA fragments cloned into the polylinker of the pGEM-1 and -2 plasmids (Promega, Madison, Wisconsin) using standard procedures. To ensure IL-2-specific hybridization, RNA probes did not include the CAG-repeat located in the 5'-region of the IL-2 cDNA. Two IL-2 RNA probes were used; a 590-nucleotide (nt) probe gave a strong signal, and a 430-nt AT-rich probe (contained within the 590-nt probe) gave extremely low backgrounds. For the 590 nt probe, a 750-base pair (bp) PstI-AccI fragment derived from the full-length IL-2 cDNA (DNAX, Palo Alto, California) was subcloned into pGEM-1 and -2 (pCmlL2-1, -2). For the 430-nt probe, a 430-bp HindIII-AccI fragment derived from the same full-length cDNA clone was also subcloned into pGEM-1 and -2 (pmIL2.3-1,-2; McGuire and Rothenberg, 1987). The pCmIL-2 plasmids were linearized with PvuII and the pmIL2.3 plasmids were linearized with HindIII. The IL-2Rc probes were derived from two IL-2Rc cDNA fragments (Miller et al., 1985), a 5' 410-bp PstI fragment and an internal 460-bp TaqI-PvuII fragment. Both were subcloned into pGEM-1 and were linearized with HindIII. 35S-labeled sense or antisense transcripts were synthesized using T7 RNA polymerase as described previously (McGuire and Rothenberg, 1987;McGuire et al., 1988). Transcription products were resuspended in 200/fl of 100 mM DTT and 0.1% SDS in 10 mM Tris, 1 mM EDTA, pH 7.4, and stored at-70C prior to use. All steps in the procedure were carried out using solutions made with diethylpyrocarbonate-treated water and RNase-free glassand plasticware.
In Situ Hybridization For each in situ hybridization experiment, alternate serial sections were hybridized with antisense and sense (negative control, "NC" in figures) riboprobes. Cells were only considered positive where the signal with the antisense probe clearly exceeded the sense-strand background. Hybridization was carried out with modifications to a protocol described previously (Rothenberg et al., 1990). The hybridization buffer described previously was modified by the addition of 4 mM vanadyl ribonucleoside complexes, and hybridization was carried out for 8-12hr. Following hybridization, cover slips were soaked off in 4xSSC. Unhybridized probe was removed with three washes in 50% formamide, 2xSSC, 10 mM DTT, and 1% 2-ME at 50C (1 hr each). Slides were briefly rinsed in 2xSSC, transferred to digestion buffer (0.5M NaC1,