Thymic Medulla Epithelial Cells Acquire Specific Markers by Post-Mitotic Maturation

The development of thymocyte subsets and of the thymic epithelium in SCID and RAG-2-/– mice was monitored after normal bone-marrow-cell transfer. The kinetics of thymic reconstitution and their relationships with cell proliferation were investigated by using bromodeoxyuridine to detect DNA-synthesizing cells among lymphoid cells by 3-color flow cytometry, and in epithelial compartments by staining frozen sections. Thymocytes started to express CD8 and CD4 10 days after transfer, simultaneously with extensive proliferation. The first mature CD4+ single-positive cells were generated, from resting CD4+CD8+ cells after day 15. During this day 10–15 period, many epithelial cells positive for cortexspecific or panepithelial markers were labeled with BrdUrd after pulse-injection. Organized medullary epithelium also developed after day,15, that is, synchronously with the appearance of mature thymocytes, but medullary cells were never found BrdUrd+. These results suggest that, in these models, the reconstitution of the thymic epithelial network proceeds through expansion of preexisting cortical or undifferentiated cells and by later maturation (acquisition of specific markers) of medullary cells. This last process is dependent of the presence of mature thymocytes.


elopment.Medu
lary epithelial growth in these systems can be due to either proliferation of rare and dispersed cells present in the mutant thymus or to the acquisition of medulla-specific markers, without proliferation, by epithelial cells in contact with ma- ture thymocytes.To resolve this question, we em- ployed the BrdUrd method that we have established to study the kinetics of thymocyte differentiation (P6nit and Vasseur, 1988; Lucas et al., 1993), matura- tion (Lucas et al., 1993), and selection (Lucas et al.,  1994).We simultaneously determined the kinetics of thymocyte subset reconstitution, and the phenotype of cycling thymocytes in RAG-2-/-mice after i.v.injection of bone-marrow cells, by analyzing BrdUrd and surface staining of thymocytes at different time points after BM injection, by three-color cytometry.Sections from the same thymuses were stained with a series of epithelial or lymphoid markers and BrdUrd.The results show that extensive proliferation of cortical and/or nondifferentiated (putative precursors) epithelial cells is observed in parallel to lymphoid expansion, but that subsequent medullary growth is independent of cell proliferation.


RESULTS

Thymocyte Development in BM-Reconstituted RAG-2-/-Mice Mice deficient for the RAG-2 gene are unable to rearrange and express TCR genes.Consequently, the thymuses of these mice contain only precursor cell subsets, that is, CD3-CD4-CD8-cells, that stop de- velopment at the CD44-CD25/HSA stage (Fig. 1, top, and Mombaerts et al., 1992; Shinkai et al., 1992).

Normal bone-marrow cells injected i.v.into low-dose irradiated RAG-2-/mice develop normally and generate all normal thymocyte and peripheral T-cell subsets.Three transfer experiments were performed, and the kinetics of thymus reconstitution obtained are shown in Figs 1 and 2. Kinetics of precursor thymocytes were studied using CD44/CD25/ BrdUrd staining (Fig. 1).The loss of CD25 by CD3- CD4-CD8-(triple-negative, TN) thymocytes is de- pendent on productive rearrangement of TCR[3chain genes.This phenotypic change is accompanied by an intense cell proliferation and immediately precedes CD8, and then CD4 expression (Lucas et al.,  1993; P6nit et al., 1995).Both CD25 loss and DNA synthesis in CD44-CD251/-subsets were observed on day 11 after BM-cell transfer (Fig. 1), simultaneously with the first appearance of CD4/8 cells (Fig. 2).Transitions in the TN subset were then masked by the accumulation of CD44-CD25-cells, including all CD4/8 thymocytes.

CD4/CD8 subset reconstitution was monitored from day 8 to day 21 post-BM tran

oth CD4/CD8 /TCR[3 and CD4/CD8/BrdU staini
g.Intermediate cells in the DN-DP transition comprised CD4-CD81, CD4-CD8 and Cd4CD8 cells that ex- press no or very low levels of c[3-TCR.They were gated in a single region (Fig. 2) and will be referred as pre-DP subsets.They appeared on days 11-13, and reached a normal percentage (Fig. 2) and absolute number (n t shown) by day 19.CD4+CD8 thymo- cytes showed similar reconstitution kinetics.Finally, CD4+CD8 and CD4/CD8 cells were detectable after day 15, and displayed their normal repre- sentation on day 21.The maturation status of single-positive and intermediate subsets was as- sessed by CD4/CD8/TCR[3 staining (Fig. 3).As soon as their first appearance (day 15), CD4+CD8 and CD4/CD8 cells were TCR hi in majority (60% on day 15, 80% on day 19), but still expressed a high HSA density on day 19 (data not shown).By contrast, mature TCRHiCD4-CD8 cells were completely absent on day 15, were first detected in significant percent- ages on day 19, and were still underrepresented on day 21.These reconstitution kinetics are in accord with the normal sequence of thymocyte-subset gen- eration, which shows a 2-day delay between CD4/8 and CD4-8 cell generation (Lucas et al., 1993), and are very close to those observed in normal BM- radiation chimeras (P6nit and Ezin , 1989).Indeed, in those animals, thymocyte-subset generation pro- ceeds through proliferation of pre-DP precursors, cell proliferation stops at the DP stage, and the subse- quent generation of mature SP cells is essentially proliferation-independent.This process also oc- curred in reconstituted RAG-2-/-mice, as shown by measuring the DNA-synthesis rate after BrdUrd pulse labeling (Fig. 4).Thymocyte DNA-synthesis frequency peaked on day 13 and returned to normal values in all subsets by day 19.The proliferation rate of CD4/CD8 and CD4/CD8 cells was always found to be low (less than 2%).The very high proliferation rate of pre-DP intermediates (70-80% on days 13- 15) only concerned cells with null or very low TCR expression, in chimeras and normal mice, as recently confirmed by four-color labeling (not shown).

The distribution of cycling thymocytes was also studied on thymus frozen sections.However, under our labeling conditions, only one thymocyte marker can be detected, besides BrdUrd, on sections.These staining experiments were also done with reconsti- tuted SCID mice.We have previously shown (P6nit, 1986, 1988) that normal DNA-synthesizing thymocytes are almost exclusively located in the cortex, with the highest frequency in the subcapsular region (SC).IL2-R0-positive cycling cells are located in the SC cortex.Some rare IL2-R cells are also scattered in the inner cortex, but most of them are not BrdUrd- labeled.

In the present study, the same double-color stainings were performed on bone-marrow (BM)- injected SCID or RAG-2-/-mouse thymuses.Development of thymocytes in BM-transferred RAG-2-/-mice: CD44/CD25 subsets.Thymuses of untreated (day 0) and BM-transferred (days 8, 11, 14, and 21) mice were taken h after BrdUrd pulse and thymocyte suspensions were stained with anti-CD44-PE and biotinylated anti-CD25 revealed with Tricolor-streptavidin.BrdUrd was then detected with the 76-7 antibody and FITC-conjugated anti-mouse IgG1.Dot plots and numbers indicated in defined regions represent the subset distribution in total (left) and DNA-synthesizing (right) thymocytes.Downregulation of CD25 is clearly apparent on day 11, and the corresponding CD251 and CD25-subsets  Development of hymocytes in BM-transferred RAG-2-/-mice: CD4/CD8 subsets.Same protocol as in Fig. 1.Surface staining was done with PE-anti CD4 and biotinylated anti-CD8 plus TC-streptavidin.CD8 and CD4 expression starts to be detect- able on day 11, and corresponding intermediate subsets are enriched in cycling cells at this time and thereafter.

CD4+CD8"


CD4"CD8+

Day 15

Day 19 BTCR FIGURE 3. TCR chain expression in thymocytes from BM-reconstituted RAG-2-/-mice.Thymocytes harvested on day 15 and day 19 after BM-cell transfer were stained with anti-CD4-PE, anti-CD8-FITC, and biotinylated H57-597 anti-[TCR antibody re- veale with Tricolor-streptavidin.The [TCR histogram obtained for unseparated (left), gated CD4/8 (middle), and CD4-8 (right) thymocytes are shown with the percentages of TCR hi cells.Time (days) FIGURE 4. DNA-synthesis rate of thymocyte subsets in BM-transferred RAG-2-/-mice.Percentages of BrdUrd cells were meas- ured in CD4/CD8 subsets between day 7 and day 21 after transfer.Cells expressing CD8 and CD4 were first detected on day 11.DNA-synthesizing cell frequency in the total thymus was maximum on day 13.systems and were in accord with the cytometry data.They are only briefly described here.On day 7 after BM injection, cycling cells were detected as large, irregularly shaped nuclei scattered inside the gland, without any preferential location.Interestingly, they were absent in the SC region, which then appears as the exclusive site of coreceptor induction.The sec- tions were essentially negative for CD8 expression, and the very rare CD8 cells always were BrdUrd .

Many IL-2R cells were also found on day 8, some of which were BrdUrd/, and most BrdUrd-, but, anyway, the majority of BrdUrd cells were IL-2R0/.

On day 10, thymus sections were still not stained by CD8 antibodies, and the number of BrdUrd cells strongly increased, simultaneously with IL-2R cells (not show

.

On day 15, cycling cells became very numero
s and were distributed all over the section.CD8 ex- pression was now extensive, and many CD8 cells were cycling.IL-2R0 cell frequency had already dropped and positive cells were located mainly in CD8-region .


Thymic Epithelium Proliferation

In normal mouse thymus, cycling cortical thymocytes are surrounded by epithelial-cell processes de- tected by MTS 5 antibody.BrdUrd nuclei are small, and always clearly separated from MTS 5 epi- thelial cells.If BrdUrd epithelial cells exist, they appear very infrequently.Similar pictures were ob- tained with MTS 39 (pan-epithelium) and MTS 44 (cortical epithelium-specific) antibodies.MTS 10 only stains medullary epithelium, and BrdUrd cells were absent in MTS 10 regions (data not s own).

On day 8, post-BM injection, SCID thymuses showed strong MTS 5 (Fig. 5a), MTS 39, and MTS 44 staining (not shown).Negative zones were also found, however.Among the relatively rare BrdUrd cells found, some, but not all, were labeled with epithelial markers.On days 11 (Fig. 5b, MTS 44), the thymus was still hyperepithelial, and many cycling MTS 5 or MTS 44+, but also MTS 39 cells were observable.During the day 8-15 period, MTS 10 cells were very scarce, and appeared as isolated scat- tered cells; they were always BrdUrd-(Figs 5d and  5e).On day 19 (Fig. 5f), small but typical MTS 10 medullary zones were found, which contained only very few BrdUrd nuclei.At this time, association of BrdUrd nuclei with cortical epithelial cells resem- bled the normal situation (Fig. 5c) with absence of contact of labeled nuclei with epithelial MTS 5 processes.MTS 44 staining showed negative areas, cor- responding to MTS 10 cells (data not s own).

Hence, the rapid growth of the thymic medulla was observed without significant proliferation of medullary-type epithelial ells.

The relationships between the nuclear BrdUrd green staining and the cytoplasmic epithelial red staining were more precisely analyzed using con- focal microscopy.The analysis of a representative field of a section made on day 11 post-BM injection is shown in Fig. 6a.In the large photograph (left), three large labeled nuclei are observed, irregularly shaped and containing yellow patches corres- ponding to the superposition of red and green fluorescences.The small pictures correspond to 4 successive plans of increasing depth, and we can see that when visible, all three labeled nuclei retain the same aspect: close contact with red staining, with yellow patches.Other green nuclei of smaller size are also observed, which show only partial contacts with cytoplasmic epithelial staining, without yellow patches, and are likely to correspond to cycling thymo ytes.

On day 19 (Fig. 6b), many labeled thymocyte nuclei were observed, uniformly green, with only occasional contacts with the epithelial MTS 5 stain- ing.No cycling epithelial cells could be detected.Confocal microscope analysis of MTS 10/BrdUrd staining was done on day-17 sections (Fig. 6c), that is, at the same time as maximum medullary g owth.

The rare BrdUrd nuclei found in MTS 10 foci did not show any association with the cytoplasmic stain- ing, suggesting that, like in normal thymus, medul- lary epithelial cells are resting during formation of th

medulla.
ISCUSSION

Using the BrdUrd labeling method, we determined the kinetics of thymocyte-subset reconstitution in BM transferred RAG-2-/mice.The release of the de- velopmental block at the CD44-CD25 hi stage of TN precursors was observed on days 10-11, as shown by the successive appearance of CD44-CD251 and then CD44-CD25-CD4-CD81// cells.IL-2R0 expression then dropped sharply, in parallel with the emer- gence of the CD4+8 subset and the decrease of total mitotic activity.CD4-8 cells showed a high proliferation rate, and were TCR -/1 until at least day 15.The first CD4/8 cells emerged on day 15 and were also gathered in the subcapsular cortex, but positive cells scattered in the inner cortex (and often resting) were present at all late time points, as in the norm l thymus.

Medullary foci, defined by MTS 10 expression de- veloped in parallel'with the appearance of CD4+8 cells, that is, after day 15.In the absence of cell pro- liferation, the medullary growth could be due either to gathering of already present and dispersed med- ullary cells or to maturation of cortexlike or undif- ferentiated cells acquiring medulla-specific markers such as MTS 10.Simple reorganization and concen- tration appear unlikely because the rare MTS 10 cells observed at early time points could not constitute the large medullary zones seen at days 19-21.The most probable mechanism, therefore, is the mat- uration of epithelial cells that are already present, expressing panepithelial, cortical, or unknown mark- ers into medulla-specific cells, under the influence of recently matured thymocytes.This hypothesis is also supported by the occurrence of rare MTS 10-MTS 44 double-positive cells in the medulla (Boyd et al., unpublished data).Hence, the devel- opment of the thymic medulla is primarily due to maturation of epithelial cells that have previously expanded.These results are summarized in Fig. 7.This process appears to be dependent on the pres- ence of mature cells, recently produced in situ in the present study or artificially transferred like in the work of Surh et al. (1992).The mechanism by which mature thymocytes induce maturation of the med- ullary epithelium remains unknown, but the present data show that the putative T-cell factors involved in this process are not simply mitogenic for preexisting epithelial cells, but are true maturation factors.


EXPERIMENTAL PROCEDURE

Mice C57BL/6 mice were obtained from Iffa Credo (UArbresle, France), and used as controls and source of bone-marrow cells.CB1

in University of Inns
ruck facilities.One hundred twenty-nine RAG-2-/mice (Shinkai et al., 1992), obtained by courtesy of EW.Alt, were bred at CDTA- CNRS (Orl6ans, France).


Bone-Marrow Transfer

The work was started with SCID mice that were whole-body irradiated (3 Gy) and 107 bone-marrow cells were injected i.v. 3 h later.To avoid any bias due to the leakiness of the scid mutation, we also used RAG-2-/mice.RAG-2-/mice were irradi- ated (3 Gy) and received 2 x 107 bone-marro cells in a first series and 5 x 10 in a second one.In all cases, BM cells were thoroughly depleted of T cells with antibodies against CD3, CD4, and CD8 and magnetic beads.

Depending on the donor-cell number and on the recipie t mouse strain (SCID or RAG-2-/-), small quantitative differences in thymic cellularity were observed, but thymic reconstitution was qualitatively and kinetically the same.

BrdUrd Labeling and Thymus Section Seven to 21 days after BM transfer, SCID and RAG- 2-/-mice received two intraperitoneal injections of BrdUrd (1 mg ea h in 100 tl of PBS) at a 4-hr inter- val.One hour after the second injection, SCID thymuses were dissected, immediately frozen in liquid nitrogen, and stored at-80C until use.

Frozen thymuses were embedded in Tissue Tek (Miles, Elkhart, Indiana) and Cryostat (Leica) cut sec- tions (4 tm) were immediately fixed in 100% acetone and stored at-20C.Normal C57B1/6 mice were submitted to the same protocol and used as controls.

For each BM reconstituted RAG-2-/-mouse, one thymus lobe was frozen for immunochemistry and the other prepared as a cell suspension for flow cytometry.

Immunofluorescence on Frozen Sections Fixed sections were stained with anti-CD8 (clone 53- 67), anti-Thyl.1 + 1.2, anti-Thyl.1 (19XE5), anti-IL2- R0 (PC61), MTS 5, MTS 10, MTS 12, MTS 39, MTS  44, or anti-Mac-1 (M1-70) revealed with biotinylated goat anti-rat Ig followed by Texas Red-streptavidin (Caltag).

Stained sections were fixed again in 70% ethanol and treated for BrdUrd detection, as described (P6nit,   1988).Briefly, semidried sectio