Murine Dendritic Cells Derived from Myeloid Progenitors of the Thymus Are Unable to Produce Bioactive IL-12p70¹

Dendritic cells (DC) are bone marrow (BM)-derived cells specialized in the capture, processing, transport, and presentation of Ags to T lymphocytes. In the secondary lymphoid organs, they are required for the initiation of immune responses (1, 2). However, in the thymus DC (3) are considered to have a different biological function; they display antigenic peptides to developing thymocytes undergoing repertoire selection and induce elimination of potentially autoreactive T cells by apoptosis (4). In the adult mice, thymic DC (TDC) represent ∼0.1% of all thymic cells and reside in the medulla and at the cortico-medullary junction.

DC comprise at least two distinct subsets according to their myeloid or lymphoid origin. DC of the myeloid lineage can be generated in vitro from BM (5, 6)-, spleen (7)-, or PBL (8)-derived progenitors using myeloid growth factors such as GM-CSF. In mice, the myeloid-related DC are CD11c⁺ CD11b⁺ CD8⁻ DEC 205⁻. The other lineage of DC appear to be closely related to lymphocytes and have been mainly characterized in the mouse. These lymphoid-related DC are found in the thymus and in the secondary lymphoid organs. They are CD11c⁺ CD11b⁻ CD8⁺ DEC 205⁺ and can be produced in recipient mice on transfer of CD4^low early thymocyte progenitors (9, 10).

Besides their distinct phenotypes, functional heterogeneity has been described between these two subsets of DC. CD8α⁺ DC from mouse spleen, unlike CD8α⁻ DC, have been considered to perform a tolerogenic function (11) due to their ability to induce apoptosis in responsive CD4⁺ T lymphocytes (12). Furthermore, recent reports proposed a differential role of CD8α⁻ and CD8α⁺ DC in the polarization of the immune response toward the Th-1 and Th-2 pathways. The orientation of the T helper development appeared to be mainly determined by the ability of DC to produce IL-12. Studies showed that IL-12 was produced by CD8α⁺ DC rather than CD8α⁻ DC (13, 14), but other authors demonstrated that CD8α⁻ DC were also able to secrete IL-12 (7, 15, 16). This discrepancy may result from the different methods of isolation or generation of DC and the origin of DC (human vs mouse). Finally, work performed with BM-derived DC (16) showed that, using different combinations of cytokines during generation in vitro, myeloid DC could direct the differentiation of Th-1 or Th-2 cells. This study supports the hypothesis that environmental factors may be more critical than lineage origin for DC functions (17).

To date, TDC were considered to be a homogeneous population of lymphoid-related DC. But their study has been hampered by the low number of cells that could be obtained either by purification (18) or by culture of CD4^low precursors (19).

We have recently described a highly effective culture system inducing proliferation and differentiation of DC from myeloid progenitors of the spleen (7). By adapting this procedure to the thymus, we questioned the presence of myeloid DC in the thymus. Using this original culture system, we were able to expand DC from thymic-resident myeloid-related progenitors. Cultured TDC (cTDC) exhibited a myeloid phenotype similar to a subset of freshly isolated DC from thymus and were obtained in sufficient number for further functional analysis. cTDC stimulate proliferation of allogenic lymphocytes but with less efficiency than myeloid splenic DC (SDC) obtained in the same culture conditions. Furthermore, unlike cultured SDC (cSDC) and purified CD8α⁺ TDC, they were unable to produce IL-12p70 under three different types of double-stimulation combining LPS, anti-CD40, and IFN-γ.

Four-week-old BALB/c and C57BL/6 mice were purchased from IFFA-CREDO Laboratories (L’Arbresle, France). Complete culture medium consisted of IMDM supplemented with 12.5% heat-inactivated FCS (both obtained from Life Technologies, Grand Island, NY), nonessential amino acids (0.1 mM), sodium pyruvate (1 mM), 50 μM 2-ME, 50 U/ml penicillin, 50 μg/ml streptomycin, and 0.5% fungizone. Murine-recombinant GM-CSF produced by a myeloma cell line transfected with the murine GM-CSF gene (provided by David Gray, Institute of Cell Animal and Population Biology, University of Edinburgh, Edinburgh, U.K.) was used at 1% in the culture medium. Purified murine recombinant Flt3-ligand, stem cell factor (SCF), and GM-CSF were purchased from R&D Systems (Abingdon, U.K.) and used at 50, 25, and 2 ng/ml, respectively.

The following mAbs labeled with FITC, PE, Cy-Chrome, or biotin were purchased from PharMingen (San Diego, CA): anti-CD8α (53-6.7, PE-conjugated, biotin-conjugated and Cy-Chrome-conjugated); anti-CD11b (M1/70, FITC-conjugated and biotin-conjugated); anti-CD11c (HL3, PE-conjugated); anti-CD 40 (3/23, FITC-conjugated); anti-B220 (RA3-6B2, biotin-conjugated); anti-TER119 (TER119, biotin-conjugated); anti-Gr1 (RB6-8C5, biotin-conjugated); anti-I-A^b that cross-reacts with I-A^d of BALB/c mouse (25.9.17, FITC- and biotin-conjugated). Anti-CD80 (RMMP-2, PE-conjugated), and anti-CD86 (RMMP-1, PE-conjugated) were obtained from Caltag (Burlingame, CA). Anti-mannose receptor DEC 205 (NLDC-145) was purchased as neat culture supernatant from Serotec (Oxford, U.K.). Rat IgG2a (Dako, Glostrup, Denmark) was used as a negative control label for NLDC-145 unconjugated rat Ab. Purified 2.4G2 (rat anti-mouse FcγRII/III, CD32) was purchased from PharMingen. F(ab′)₂ goat anti-rat IgG (FITC-conjugated) was obtained from Jackson ImmunoResearch (West Grove, PA). Biotin-conjugated Ab was revealed using streptavidin-FITC, streptavidin-PE, or streptavidin-Cy-Chrome obtained from PharMingen.

Thymus was cut into fragments under aseptic conditions, and the entire tissue was digested for 30 min at 37°C with collagenase B (2 mg/ml)/DNase I (0.4 mg/ml) (Boehringer Mannheim, Mannheim, Germany) in PBS/10% FCS. Thymic cells were washed and counted. Cell concentration was adjusted to 4 × 10⁶/ml of complete culture medium supplemented with GM-CSF, Flt3-ligand, and SCF. This suspension (0.5 ml/well) was plated in multidish 4 wells (Nunclon, Glostrup, Denmark). At day 3 of culture, cells were boosted by addition of the three cytokines. At day 7, the supernatant was removed and replaced with fresh cytokine-supplemented medium. At day 14, medium was changed in the same manner unless the stroma had become confluent and acidified it. In this case, the aggregates were detached by PBS containing 3 mM EDTA (Sigma, St. Louis, MO). The removed cells were counted and split at a concentration of 1.5 × 10⁶ cells/ml. At different time points, cells were counted and cytospins prepared and stained with May-Grünwald-Giemsa (Sigma). cSDC were obtained as previously described (7).

Thymi and spleens from 4- to 6-wk-old BALB/c mice were digested for 30 min at 37°C with collagenase B (2 mg/ml)/DNase I (0.4 mg/ml; Boehringer Mannheim) in PBS/10% FCS and further dissociated in Ca²⁺-free HBSS/EDTA 10 mM. Thymic and splenic cells were separated on a Nycodenz (Nycomed, Oslo, Norway) gradient. Low-density cells were enriched for CD11c expression and further separated according to CD8α expression using a multisort anti-FITC kit (Miltenyi Biotec, Paris, France). For IL-12p70 detection purified DC (10⁶ cells/ml) were cultured for 24 h in regular medium containing GM-CSF, SCF, and Flt3-ligand without or with different combinations of LPS (1 μg/ml), anti-CD40 (1 μg/ml), and IFN-γ (20 ng/ml).

Cell surface phenotype was studied by flow cytometry. Two- or three-fluorescence color stainings were conducted in 96-well plates in 10 μl of mAb at optimal concentration for 10 min at room temperature with agitation. Cells were preincubated with 2.4G2 to prevent binding to FcγRII/III. Data of a minimum of 10,000 cells, collected using a FACSCalibur flow cytometer (Becton Dickinson), were analyzed using CellQuest software.

Intracellular distribution of MHC class II molecules was studied on cytospins. The slides were fixed in methanol/acetone (1:1) for 5 min at room temperature and washed in TBS before staining with mAb anti I-A. Ab fixation was revealed by alkaline phosphatase-conjugated avidin-biotin complex (ABC complex AP; Dako) according to the manufacturer’s recommendations.

Semisolid assays for clonogenic progenitors present in the initial thymic cell suspensions and PBL were prepared in complete collagen culture medium, as previously described (7). Briefly, cell suspensions were cultured in triplicate wells (multidish 4 wells; Nunclon) containing 0.3 ml IMDM supplemented with 25% FCS, 50 μM of 2-ME, antibiotics, and collagen (I + III from Hemeris Laboratories, Sassenage, France) at a concentration of 1 mg/ml of culture. Thymic cells were plated at 2 × 10⁶ cells/ml, and PBL were plated at 2 × 10⁵ cells/ml. The growth of colonies was stimulated by an optimal mixture of growth factors containing SCF (25 ng/ml), Flt3-ligand (50 ng/ml), and GM-CSF (10 μl/ml). The cells were incubated for 13 days at 37°C in a fully humidified incubator containing 5% CO₂, and the colonies were counted under an inverted microscope. Hence, the total number of colonies per initial suspension could be calculated from this colony count and the initial total number of cells. To characterize the proportion of MHC class II-positive cells present in each colony, the gels were then desiccated in situ and used for the immunocytochemical detection of MHC class II molecules.

CD4⁺ T lymphocytes were negatively selected from C57BL/6 splenocytes by magnetic cell sorting using anti-CD8α, anti-B220, anti-CD11b, anti-Gr-1, and anti-TER 119 biotin-conjugated Abs and streptavidin magnetic microbeads (Miltenyi Biotec). Purity of CD4⁺ T cells was between 90 and 95%. Triplicates of 1 × 10⁵ enriched CD4⁺ T cells were seeded into a 96-well round-bottom plate (Falcon; Elvetec, Venissieux, France) together with titrated numbers of mitomycin C-treated (50 μg/ml, 20 min, 37°C; Sigma) splenic or thymic-derived DC in 0.2 ml RPMI 1640 supplemented with 10% FCS and 2-ME (50 μM). Three days later, cells were pulsed with 1 μCi/well of [³H]TdR for an additional 12 h before harvesting and scintillation counting.

IL-12p70 production was determined by the OptEIA set for mouse IL-12p70 from PharMingen, according to the procedure recommended by the manufacturer. The Ab used in this test specifically recognizes the p70 heterodimer but not free p40 chains. The detection limit was 30 pg/ml of IL-12. Days 13–14 cSDC or cTDC (5 × 10⁵ cells/ml) were cultured for 24 h in regular medium containing SCF (25 ng/ml), Flt3-ligand (50 ng/ml), and GM-CSF (10 μl/ml). Cells were stimulated or not by three types of double stimulation combining LPS (1 μg/ml) + anti-CD40 (1 μg/ml), LPS (1 μg/ml) + IFN-γ (20 ng/ml), and anti-CD40 (1 μg/ml) + IFN-γ (20 ng/ml), respectively referred to as DS1, DS2, and DS3. IL-12 measurements were performed in duplicate on each culture supernatant. To exclude the contribution of a contaminating cell type to the secretion of IL-12p70, positive sorting of cSDC and cTDC was performed after labeling with biotinylated anti-CD11c and anti-class II Abs followed by streptavidin magnetic microbeads (Miltenyi Biotec). Total and purified CD11c⁺ and class II⁺ cell suspensions were stimulated by LPS+anti-CD40 (DS1) and LPS+IFN-γ (DS2), and IL-12p70 was measured 24 h later in the supernatants as previously indicated.

Unfractionated thymic cell suspensions were cultured in the presence of Flt3-ligand, SCF, and GM-CSF. A low number of cells with mature DC morphology appeared in the supernatant within 7 days of culture, mixed with dead thymocytes and some granulocytes and macrophages. At this time, these nonadherent cells were completely removed with the supernatant allowing the observation of adherent fibroblasts in contact of which small aggregates of round cells were growing (Fig. 1,A). After 12–14 days, culture cells grew in characteristic clusters (Fig. 1,B) strongly adherent to a stroma made up of fibroblasts and macrophages. At the same time, a low number of DC were released but most of them remained in large adherent clusters that persisted until days 19–21. Cytological examination of cells from clusters at day 14 of culture indicated that >90% of them had typical bean-like nuclei and irregular membranes (Fig. 1,C). Very few (0.1%) blastic cells with round nuclei and reduced basophilic cytoplasm were found. Immunostaining for MHC class II showed that cells produced from day 14 possessed extensive, fine, and often beaded dendritic processes expressing high level of MHC class II molecules (Fig. 1 D).

The kinetics of cTDC production (Fig. 2) showed that the highest number of MHC class II⁺ cells was obtained at day 13 of culture with a mean of 123,000 ± 34,000 DC produced per 10⁶ initial thymic cells. The variations observed in the number of cTDC produced seemed to result from the relative composition for macrophages and fibroblasts of the stroma that developed in parallel with DC. The presence of fibroblasts helped DC production, whereas the invasion of the stroma by macrophages resulted in poor DC recovery. Regardless of the efficiency of the stroma, the number of DC decreased and production stopped at the end of the third week of culture. Splitting the culture after day 14 did not improve the TDC production.

Similar DC yield was obtained using purified rGM-CSF instead of the conditioned medium, excluding the role of other cytokines potentially produced by the myeloma cell line. Furthermore, in the absence of GM-CSF, no DC were observed in culture (data not shown).

cTDC obtained between days 12 and 21 could be subdivided into two populations on the basis of their intermediate or high surface expression of MHC class II molecules (Fig. 3, A and B). All MHC class II⁺ cells were CD11c⁺ CD11b⁺ but CD11b expression was reduced on MHC class II^high DC. All MHC class II^high DC coexpressed B7.2 and CD40, whereas DEC 205 was only detectable on a small proportion of them. MHC class II^int DC were negative for these three markers. According to the B7.2 and CD40 expression, MHC class II^high DC exhibited a more mature or activated phenotype than MHC class II^int DC. A small proportion of each MHC class II⁺ population expressed F4.80, and both MHC class II⁺ populations were negative for B7.1 and CD8α. The surface phenotype of TDC remained roughly similar between days 12 and 21 with a progressive increase in the proportion of MHC class II^high. Nevertheless, the rapidity of this maturation differed from one culture to another, presumably as a consequence of the variability of the stroma. The cTDC phenotype was compared with the corresponding cSDC (Fig. 3, C and D). The presence of CD11b and the absence of DEC 205 and CD8α confirmed that both DC types are related to myeloid DC lineage. Conversely, the percentages of MHC class II, B7-2, and CD40 positive cells exhibit some variations probably due to differences in the kinetics of maturation from one cSDC culture to another. However, the percentage of CD40⁺ cells was consistently higher in cSDC than in cTDC.

A few cells (<3%) among the MHC class II^high exhibited a discrete positivity for B7-1, whereas the percentage of B7-2 positive cells correspond to the MHC class II^high population, which increased with the maturation between days 14 and 20 of the incubation. To determine whether DC with a similar myeloid phenotype were present in the thymus, we analyzed the expression of CD11c, CD8α, and CD11b in TDC purified according to CD11c expression (Fig. 3 E). TDC could be separated into CD11c^high CD8^high and CD11c^int CD8^−/low populations. Further analysis showed that CD11c^high CD8^high cells were CD11b^low, whereas CD11c^int CD8^−/low cells were CD11b^high. Due to the small number of CD11c positive cells recovered from thymi, we observed in one experiment that CD11c⁺ CD8α⁻ cells were B7-1 negative, whereas 58% of CD11c⁺ CD8α⁺ cells were also B7-1 positive. The CD11c^int CD8^−/low CD11b^high cells represented between 16 and 28% of the TDC in four separate experiments. So, our results clearly indicated that DC with myeloid phenotype could be found in significant numbers in the adult mouse thymus.

Comparative phenotypic studies were performed on CD11c⁺CD8α⁺ and CD11c⁺CD8α⁻ DC populations purified from the spleen of BALB/c mice. As previously reported by others (13, 14) the CD11c⁺ CD8α⁺ SDC represented between 19 and 26% of the total CD11c⁺ cells that were predominantly CD8α⁻ (data not shown).

To approach the question of the origin of the cTDC obtained in our cultures, we look for the presence of DC progenitors in the total thymic cell suspensions. Semisolid cultures were performed in the presence of Flt3-ligand, SCF, and GM-CSF, the cytokines present in the liquid cultures. The use of collagen gel allowed the in situ identification of DC in the colonies by immunostaining of MHC class II cells after desiccation. Total thymic cell suspensions formed 2.7 ± 0.6 colonies per 10⁶ cells (Table I). The mean of colonies obtained at day 13 per total thymus was 586 ± 271. One third of the colonies were of large size, relatively compact, and contained between 1000 and 3000 cells. May-Grünwald-Giemsa staining and MHC class II immunocytochemistry revealed that these colonies contained between 60 and 90% of MHC class II⁺ DC mixed with class II⁻ macrophages (Fig. 4, A–C). The other colonies were of small size and contained 50–200 loosely arranged cells, which were all positive for MHC class II (Fig. 4, D and E). Thus, the majority of DC found in semisolid culture came from progenitors common with macrophages. It is likely that cTDC obtained in liquid culture with the same cytokines mainly originated from these myeloid progenitors.

To evaluate the possibility that these progenitors were contaminants from blood, we determined the colony formation capacity of PBL and the intrathymic blood volume (Table I). Under our cloning conditions, PBL formed 70 ± 63 colonies per 10⁶ cells. Seventy percent of them were mixed macrophage/DC colonies containing 200-1000 cells, and 10% of them also contained granulocytes (data not shown). These mixed macrophage/DC colonies were predominantly macrophagic and contained only 10–30% MHC class II⁺ DC. Thirty percent of the PBL-derived colonies were macrophagic, and very occasional pure DC colonies (<1%) containing 50–100 loosely arranged MHC class II⁺ cells were observed. Thymic blood volume was 0.4 ± 0.1 μl, as estimated by red cell counting, leading to a total contamination by blood of ∼3400 PBL, which would give a mean of 0.2 ± 0.07 colonies per thymus. These results showed that contamination of the total thymic cell suspensions by blood-derived precursors could be excluded and that the cTDC derived from true thymus-resident progenitors.

To determine their functionality, cTDC were compared with cSDC obtained in the same culture conditions for their ability to stimulate allogenic MLR. In the experiment shown in Fig. 5,A, unstimulated cTDC were as competent as unstimulated cSDC in the induction of proliferation of allogenic CD4⁺ lymphocytes even though the proportion of MHC class II^high B7.2⁺ cells was higher in cTDC (57%) than in cSDC (34%) (Fig. 5, B–D). Double stimulation by anti-CD40 and LPS induced maturation of both cTDC and cSDC as shown in Fig. 5, C–E. Most of the cSDC (74%) and cTDC (84%) became MHC class II^high B7.2⁺ and expressed these molecules at similar levels (identical mean of fluorescence intensity). Both mature cSDC and cTDC induced a significantly stronger MLR than the corresponding unstimulated cells. Proliferation induced by mature cTDC was greater than by unstimulated cSDC. Nevertheless, MLR induced by mature cTDC was lower than those induced by mature cSDC even though the proportion of MHC class II^high B7.2⁺ cells was higher in cTDC than in cSDC.

Thus, cTDC are able to stimulate allogenic MLR, and this capacity was increased by stimulation with LPS (data not shown) or LPS + anti-CD40. However, activated cTDC appeared to be less potent stimulators than activated cSDC despite a more mature phenotype.

We have previously described that cSDC secrete large amounts of bioactive IL-12p70 when double-stimulated by anti-CD40 and LPS and that each stimulus alone was unable to induce IL12p70 production (7). We extended this observation by studying the IL-12p70 production of cSDC and cTDC in one part and ex vivo purified SDC and TDC on the other part. All types of DC were subjected to three different double stimulations, namely, LPS + anti-CD40, LPS + IFN-γ, and anti-CD40 + IFN-γ, respectively referred to as DS1, DS2, and DS3. As shown in Fig. 6,B, cSDC produced high levels of IL-12p70 when stimulated by DS1 or DS2, whereas DS3 was reproducibly inefficient. We also verified that PBL-derived DC cultured in the same conditions produced significant amounts of IL-12p70 when stimulated by DS1. As previously reported (13, 14), we confirmed that only the CD11c⁺ CD8α⁺ SDC subset purified ex vivo was able to secrete IL-12p70 under DS1, DS2, and DS3, whereas the CD11c⁺ CD8α⁻ subset was not, whatever the type of stimulation used (Fig. 6,A). Then, we investigated the capacity of the cTDC to secrete IL-12p70 in response to the same stimuli. As shown in Fig. 6,B, cTDC were unable to produce detectable amounts of IL-12p70 under the three types of double stimulation used. To exclude that a contaminating cell type may be responsible for the effect obtained, we verified that similar IL-12p70 responses are observed by using DS1 and DS2 on purified CD11c and class II-positive cells from cSDC and cTDC suspensions. We performed double labeling of the cells with biotinylated anti-CD11c and anti-class II Abs followed by magnetic sorting with streptavidin microbeads. We used double labeling of cultured DC because we observed that a single staining with anti-CD11c Ab results in the sorting of the most mature DC, which secreted very low amounts of IL-12p70 as previously shown (20). The addition of class II Ab allowed the positive sorting of immature DC, which are responsible for the main part of IL-12p70 production. As shown in Fig. 6 C, the levels of IL-12p70 secreted under DS1 and DS2 stimulation by the unseparated cell suspensions and by the purified CD11c class II-positive cells are similar. Thus, the incapacity of cTDC to produce IL-12p70 after stimulation indicated that they were functionally different from cSDC and PBL-derived DC obtained in the same culture conditions.

To determine whether the inability of cTDC to produce IL-12p70 is characteristic of DC normally found in the thymus, we purified TDC to test their IL-12p70 secretion capacity. In the first experiment shown in Table II, TDC sorted according to CD11c expression produced detectable levels of IL-12p70 when stimulated by LPS and anti-CD40. In the next two experiments, TDC were further separated according to CD8α expression to determine which TDC subset produced IL-12p70. Due to the small number of TDC obtained after the CD11c⁺ sorting, we only performed one round of separation for CD8α yielding fractions enriched in either CD8α⁺ or CD8α^−/low cells. The highest levels of IL-12p70 were found in the CD8α⁺ TDC-enriched fractions, and low levels of IL-12p70 were detected in the CD8α^−/low TDC-enriched fractions. For each fraction, the amount of IL-12p70 exactly correlated with the ratio of CD8α⁺ TDC to CD8α^−/low TDC and the absolute number of CD8α⁺ TDC found in the cell suspension (correlation coefficient: R = 1 and p < 0.01). These results strongly suggested that the low levels of IL-12p70 detected in the CD8α^−/low TDC-enriched fractions were produced by the few contaminant CD8α⁺ TDC. DS2 stimulation was tested in two other experiments using purified ex vivo TDC showing that CD11c⁺CD8α⁺ TDC also respond to the combination of LPS + IFN-γ by an increased secretion of IL-12p70, whereas CD11c⁺ CD8α⁻ cells exhibit a low level of IL-12p70 (data not shown). Thus, purified CD8α^−/low TDC did not secrete significant amounts of IL-12p70 under these conditions of stimulation. In conclusion, the incapacity of cTDC to produce IL-12p70 was shared with CD8α⁻ TDC purified from thymus.

For years, TDC were considered as a homogeneous population of lymphoid-related DC characterized by the expression of CD8α and DEC 205 and the lack of CD11b (21, 22). TDC and T cells were shown to be closely related as they are produced in the thymus from a common CD4^low precursor population (9, 10). However, several recent papers challenged this consensus. Radtke et al. (23) used mice in which the Notch 1 gene had been conditionally inactivated as BM donors for lethally irradiated recipients. These mice exhibit a very early block in T cell development. BM chimeras has normal macrophage, granulocyte, NK, and B cell development but a total T cell deficiency. This contrasts with an entirely normal thymic and peripheral DC development showing that thymic DC and T cells are derived from distinct precursors. Furthermore, Martin et al. (24) reported that thymic CD4 low precursors isolated from C57BL/6 mice are able to reconstitute both populations of CD8α⁻ and CD8α⁺ DC of the spleen of an irradiated recipient. Merad et al. (25) showed that epidermal Langerhans cells, obtained from murine BM cultures exhibiting a typical myeloid phenotype (CD11c⁺, CD11b⁺, CD8α⁻) are able to express CD8α when they migrate to the draining lymph node after injection in the footpads of mice and became highly immunostimulatory. Hence, CD8α expression on these DC appears to reflect a state of activation, mobilization, or both rather than lineage specificity. Rodewald et al. (26) identified putative lymphoid and myeloid DC in the thymus of wild-type and c-kit⁻ γ_c⁻ 5-day-old mice with regard to their respective CD11c^high CD11b^−/low and CD11c^int CD11b⁺ phenotype. In these mice, the putative myeloid DC represented ∼37% of the TDC. Finally, Traver et al. (27) recently showed that CD8α⁺ and CD8α⁻ DC can arise from clonogenic common myeloid progenitors in both thymus and spleen of adult mice. Vremec, in his analysis of DC subtypes in mouse (28), evoked the existence of a few CD8α^low CD11b⁺ TDC. In this study, using an original bulk culture system allowing differentiation of DC from myeloid progenitors, we demonstrated the presence of a distinct subset of myeloid-related DC exhibiting functional specificities in the thymus of normal adult mice. Furthermore, we clearly observed that in adult mice ∼20% of the DC purified from thymus were CD11c⁺ CD8α^−/low CD11b^high. This result was in agreement with the observations of Rodewald. The smaller proportion of CD8α⁻ TDC observed in our study may be due to the strain or to the age of the mice analyzed (adult vs postnatal) as it has been shown that the myeloid activity of the thymus decreases with age (see below). The discrepancy with the observations of Vremec et al. (21) may be due to the use of anti-CD11b in the procedure for purification of TDC resulting in the elimination of most of CD8α⁻ DC.

Our study showed that DC with a myeloid phenotype, similar to those observed in freshly isolated TDC, could be produced in vitro in significant numbers from mouse thymus suspension. TDC represent ∼0.1% of all thymic cells (18, 29) and to date, their study has been hampered by the low number of cells that could be obtained either by purification or by culture of early T lymphocyte progenitors CD4^low. Recoveries of DC using the purification procedure described by Vremec et al. (18) ranged from 1 to 5 × 10⁵ per thymus. An equivalent production was obtained by Saunders et al. (19) who succeeded in generating TDC from CD4^low precursors after 6 days of culture. Our culture system allowed the production of a mean of 20 × 10⁶ DC per thymus. Such a procedure was useful for generating myeloid TDC in numbers allowing functional characterization.

The efficiency of our culture system was due to the use of the early cytokine, SCF, combined with GM-CSF and Flt3-ligand. This mixture previously has been described to be highly effective in inducing proliferation of murine SDC progenitors in vitro (7). Secondly, preliminary results indicated that the stromal cells in the culture provide interactions necessary for the survival and subsequent proliferation and differentiation of TDC progenitors. Similarly, thymic epithelial cell lines have been shown to induce growth and differentiation of CD4^low precursors and allowed a limited production of DC (30).

The phenotype (CD11c⁺ CD8α⁻ CD11b⁺) and GM-CSF requirement of the cTDC suggested the presence in the thymus of myeloid-related progenitors giving rise to DC. Semisolid cultures performed with the same mixture of cytokines as in liquid cultures clearly showed that DC could arise from a colony-forming progenitor common to macrophages and present in the adult murine thymus. The estimated contamination by PBL of the thymic suspensions was close to the value found by Antica et al. (31). This result coupled with the colony-forming capacity of PBL provided evidence that the macrophage/DC progenitors found in the cell suspensions were not blood contaminants but thymus-resident cells. Myeloid activity has been described in the fetal murine thymus, but this myeloid potential declined between days 12 and 14 of gestation (32). Nevertheless, in the adult mice, a erythroid-myeloid activity was described within the population of CD4⁻ CD8⁻ thymocytes (33). More recently Wu et al. (34) using in vitro colony assay showed that the CD4^low precursors contained macrophage colony-forming cell at a frequency of 1 per 200 cells. Considering that CD4^low precursors represent ∼0.05% of all thymocytes, the frequency of myeloid activity in total thymocytes can be estimated at ∼2.5 per 10⁶ thymocytes, which is close to those obtained in our semisolid cultures. Furthermore, preliminary sorting experiments performed in our laboratory indicated that the macrophage/DC progenitor activity was found within the CD3⁻ CD8⁻ B220⁻ Gr1⁻ Ter119⁻ thymocytes and lost when anti-CD4 was added to the depletion mixture. So, it is likely that the thymic macrophage/DC progenitor belongs to the CD4^low population. The absence of mixed macrophage/DC colonies in the colony assay realized by Wu et al. (34) with CD4^low thymocytes could be due to the lack of Flt3-ligand in the culture. Actually, it remains to be determined whether the macrophage/DC progenitor identified in our cultures is committed to the myeloid lineage or whether it can also give rise to lymphoid cells. But the cytological composition of colonies containing DC clearly indicated that the thymic macrophage/DC progenitor differs from the classical DC/myeloid progenitor found in PBL, spleen (7), or BM (35). Furthermore, the thymic macrophage/DC progenitor had a very limited self-renewing capacity when compared with the spleen and BM-derived CFU-GM as indicated by the incapacity to increase the number of colonies in the first days of culture (data not shown, Ref. 7) and the absence of remaining blastic MHC class II⁻ CD11c⁻ for the second week of culture. Finally, contrary to spleen- and BM-derived CFU-GM, the macrophage/DC progenitor appeared to be very dependant on stromal cells that developed in parallel with them. The small pure DC colonies obtained in the semisolid cultures may derive from more committed precursor than the macrophage/DC progenitor. This committed precursor would be only capable of differentiation into DC and would have a limited proliferative capacity. Pure DC colonies were very rare in the spleen (7) and never detected in the BM by Inaba, who showed that DC arise as a component of mixed granulocytic and macrophagic colonies (35). However, Young (36) reported the generation of a number of small, loosely arranged pure DC colonies in semisolid agarose cultures seeded with purified CD34⁺ cells from human BM in the presence of GM-CSF and TNF-α.

The functional analysis of myeloid cTDC showed that they are able to stimulate allogenic MLR but with less efficiency than cSDC despite a more mature phenotype. Different allostimulatory capacity has previously been described between purified CD8⁺ and CD8⁻ SDC. CD8⁺ DC induced a lower response, associated with marked T cell apoptosis due to interaction of Fas on activated T cells with Fas-ligand on CD8⁺ DC (12). With regard to this finding it would be interesting to evaluate the ability of cTDC to kill lymphocytes. Several authors have emphasized the pivotal role of IL-12 secreted by DC in the polarization of Th responses (37, 38, 39). We have previously reported (7) that cSDC secreted large amounts of IL-12p70 when double-stimulated with anti-CD40 and LPS. This study extended this observation by using LPS + IFN-γ, which also induced active secretion of IL-12p70 by cSDC, whereas IFN-γ + anti-CD40 combination was inefficient. The capacity of myeloid-related cultured DC to secrete IL-12p70 in response to appropriate stimulus has been previously reported by using murine BM (16, 37) and human monocyte-derived DC (40, 41). However, in agreement with others (13, 14, 20), we showed that IL-12p70 is produced by the lymphoid-related CD8α⁺ cell subset purified ex vivo from the spleen of mice. Reis e Sousa et al. (42) reported that CD11c⁺CD8α⁺ (and not CD11c⁺CD8α⁻) SDC secreted IL-12p70 following in vivo systemic triggering with Toxoplasma gondii tachyzoites but that they can no longer be restimulated during 1 wk in vivo, whereas CD8⁺ as well as CD8⁻ DC continue to respond in vitro. This study demonstrate that CD8α⁻ SDC are not intrinsically unable to secrete IL-12 but they are susceptible to inhibitory activity of the splenic microenvironment. Concerning TDC, we showed for the first time that in the thymus also, IL-12p70 is produced by the CD11c⁺ CD8α⁺ cells and not by the CD11c⁺ CD8α⁻ subset. This finding is in agreement with the assigned biological function of TDC: to induce the elimination of potential autoreactive developing T cell by apoptosis (4, 43, 44) rather than to initiate immune responses. Recent data indicated that intrathymic IL-12 production was essential for the negative selection of early CD4^low CD8^low thymocytes but did not play a significant role in the negative selection of CD4⁺ CD8⁺ thymocytes and single-positive thymocytes. Nevertheless, to date, thymic IL-12-producing cells were not characterized (45). Although we cannot rule out that purified CD8α⁻ TDC produced small if any amounts of IL-12p70, our results clearly demonstrated that IL-12p70 was mainly produced by CD8α⁺ TDC. To our knowledge, it is the first evidence of a functional difference between subsets of purified TDC. This finding suggests that CD8α⁻ and CD8α⁺ TDC may play distinct roles in thymocyte development. They may be involved in negative selection of distinct subsets of thymocytes. CD8⁺ TDC would induce negative selection of CD4^low CD8^low thymocytes by an IL-12-dependent mechanism, whereas CD8⁻ TDC induce negative selection of more mature thymocytes by a mechanism remaining to be elucidated. As we are able to culture myeloid-related TDC, we investigate their ability to secrete IL-12p70. We found that in contrast with their myeloid cSDC counterparts, they were reproducibly unable to secrete IL-12p70 in response to the three types of double stimulation used. This could be due to an intrinsic defect of myeloid-related TDC or/and to an inhibitory factor present in TDC culture. If we compared the SDC and TDC culture systems, the kinetics of DC production is very different. cSDC precursors proliferate quickly and vigorously in the liquid phase of the culture during the first 10 days of incubation resulting in frequent medium changes. In contrast, a few cTDC precursors tightly bound to numerous foci of fibroblastoid cells appeared after 7 days of culture and give birth to semiadherent typical DC aggregates developing after 10–12 days of culture. The prolonged interactions of cTDC with the adherent stromal layer could induce inhibitory signals for IL-12p70 production. This hypothesis is currently under investigation.

We are grateful to Muriel Moser for her helpful discussions and criticism. We thank Maighréad Gallagher for correcting the English text.