Three Populations of Mouse Lymph Node Dendritic Cells with Different Origins and Dynamics¹

Dendritic cells (DCs)³ have been shown to be very heterogeneous with regard to their origin, maturation state, and rate of turnover. There is general agreement that DCs originate from a hemopoietic progenitor (1). However, they apparently have multiple pathways of differentiation. There are numerous arguments for a myeloid origin of some DCs (2, 3). Culture of bone marrow cell progenitors in semisolid medium has allowed the characterization of mixed colonies that can differentiate into either DCs or myeloid cells (4). Human monocytes can also further differentiate into typical DCs when cultured in the presence of IL-4 (5, 6, 7). The differentiation of these myeloid DCs from either immature precursors or more differentiated monocytes always requires granulocyte-macrophage CSF. On the other hand, there is now substantial evidence for the existence of a pool of DCs with a lymphoid-related origin (8). A precursor that can differentiate into either T cells or DCs has been described in the mouse thymus. This precursor has lost the ability to differentiate into either myeloid cells, B cells, or NK cells. DCs that develop from these thymic precursors express the CD8α molecule (9). Granulocyte-macrophage CSF is not required for the development of these lymphoid-related DCs (10). A common precursor for T cells, B cells, NK cells, and DCs has also recently been identified in human bone marrow (11). Langerhans cells may belong to a third DC lineage, and TGF-β₁ is required for their development (3, 12, 13, 14).

In addition to these distinct lineages, DCs differ in their maturation stage and functional properties. “Immature” DCs appear very efficient for macromolecule capture and processing. Inflammatory mediators induce their differentiation into more mature or activated cells that have acquired the capacity to efficiently stimulate T lymphocytes (5, 15, 16, 17). Finally, DCs also appear heterogeneous in terms of turnover (18). Some of them, such as the Langerhans cells (LCs), appear to be renewed very slowly in the absence of inflammation (19, 20). Other DCs have a rapid turnover with a t_1/2 < 1 wk (21, 22, 23, 24).

DC heterogeneity has already been described in some lymphoid tissues. In the spleen, two DC populations with distinct phenotypes and localizations have been described: the DCs of the marginal zone and the DCs localized in the periarteriolar T cell region of the white pulp (25, 26, 27, 28), that seem to be of myeloid- and lymphoid-related origin, respectively (9, 29). Similarly, DC heterogeneity has also been described in the Peyer’s patches (30) and tonsils (31).

The characterization of these distinct populations of DCs in various tissues or organs is important because they most probably have different functions in the immune system. Indeed, it has already been suggested that in the spleen the myeloid-related DCs a play major role in triggering immune responses, while the lymphoid-related DCs participate in the regulation of these responses (32, 33). Surprisingly little is known about lymph node (LN) DCs, although these secondary lymphoid organs are major sites of immune response initiation. Until now, a single DC type has been described, which is located in the paracortical zone and referred to as an interdigitating cell (34). It is believed that these cells are derived from immature DCs localized in regional tissues, which migrate to the draining LN via afferent lymph (Refs. 35–37; and reviewed in 38 .

In this work, we present a detailed analysis of LN DCs. We have identified three distinct populations that can be distinguished on the basis of their morphology and phenotype. Interestingly, these cells also differ in their capacity to uptake Ag delivered by different routes and in their dynamics.

Inbred mouse strains DBA/2, C57BL/6, CBA/J, and FVB/N were obtained from IFFA-Credo (L’Arbresle, France). Except where indicated, all experiments were performed with DBA/2 mice. Chimeric mice used to analyze DC turnover were derived from LTR-TK transgenic mice expressing the HSV1-thymidine kinase (TK) in DCs as previously described (24, 39). They were generated by transplantation of 3 to 5 × 10⁶ bone marrow cells from LTR-TK transgenic mice into lethally irradiated (13 Gy) normal recipients with the same DBA/2 genetic background.

LNs were cut into small fragments and incubated in RPMI 1640 supplemented with 1.6 mg/ml (500 IU/mg) collagenase (type IV, Sigma Chemical Co., Saint Quentin Fallavier, France) and 200 μg/ml DNase I (Boehringer Mannheim, Mannheim, Germany) at 37°C for 30 min. Cells were dissociated by repeated pipetting, reincubated at 37°C for 10 min, and washed. Cell suspensions were then incubated with 200 μg/ml DNase I for 15 min at room temperature and resuspended in staining buffer (PBS, 3% FCS, 0.02% azide) for further flow cytometric analyzes.

CD11c expression was analyzed with the unlabeled N418 mAb (hamster IgG, HB224; American Type Culture Collection (ATCC), Rockville, MD) revealed by a PE-conjugated F(ab′)₂ goat anti-hamster IgG (Caltag Laboratories, San Francisco, CA). Depending on the analysis performed, we used different mAbs against MHC class II molecules: the M5/114 mAb (ATCC TIB120) was revealed by a TriColor-conjugated F(ab′)₂ goat anti-rat IgG (Caltag Laboratories) (see Figs. 1, 5, A and B, and 6); the 14.4.4S mAb (PharMingen, San Diego, CA), either FITC-conjugated or biotin-conjugated, was revealed by streptavidin TriColor (Caltag Laboratories) (see Figs. 3, A and B, and 7); the biotin-conjugated 2G9 mAb (PharMingen) was revealed by streptavidin TriColor (see Fig. 4); the 28-16-8S mAb (Caltag) was revealed by an allophycocyanin-conjugated goat anti-mouse IgM (Caltag) (see Figs. 3 C and 5C).

Panels of mAbs were selected to study the phenotype of LN DCs by three-color flow cytometry analysis. We used a panel of FITC-conjugated mAbs against B220 (RA3-6B2, Caltag Laboratories), mThy-1.2, mCD4 (CT-CD4, Caltag Laboratories), mCD8α (CT-CD8a, Caltag Laboratories), Gr-1 (RB6-8C5, Caltag Laboratories), F4/80 (F4/80, Caltag Laboratories), Mac-1 (M1-70.15, Caltag Laboratories), heat stable Ag (HSA) (M1/69, PharMingen), MHC class II (M5/114, Boehringer Mannheim), CD44 (IM7, PharMingen), and CD62L (Mel-14, PharMingen), a panel of biotin-conjugated mAbs against B7-1 (16.10. A1), B7-2 (GL-1), H2-K_d (SFI-1.1, PharMingen), and CD54 (KAT-1, Caltag Laboratories) and a panel of uncoupled rat mAbs against CD2 (AT37, Serotec, Oxford, U.K.), spleen DC (33D1, ATCC TIB 227), DEC 205 (NLDC145 (25)), CD40 (3/23, Serotec), and FcγII/IIIR (2.4G2, PharMingen).

To minimize nonspecific binding, cells were preincubated either with 2.4G2 mAb and then analyzed using FITC- or biotin-conjugated third mAbs or with 10% mouse serum and analyzed using mAbs from the panels of uncoupled mAbs. In all stainings, the N418 mAb was revealed by a PE-conjugated F(ab′)₂ goat anti-hamster IgG (Caltag Laboratories). For MHC class II, the staining differed according to the third mAb used. For FITC-conjugated mAbs, we used a biotin-conjugated anti I-E mAb (14.4.4S) revealed by streptavidin TriColor. For mAbs from the panels of biotin-conjugated or uncoupled rat mAbs, revealed respectively by streptavidin TriColor (Caltag Laboratories) and a TriColor-conjugated F(ab′)₂ goat anti-rat IgG (Caltag Laboratories), the anti I-E staining was performed with the FITC-conjugated 14.4.4S mAb.

The following isotypic Ig controls were used: FITC-conjugated rat IgG2a (LODDNP-16, Immunotech, Marseille, France), FITC-conjugated rat IgG2b (Cedarlane, Hornby, Ontario, Canada), and biotin-conjugated rat IgG2a, biotin-conjugated hamster IgG, unconjugated rat IgG2a, and unconjugated rat IgG2b (Caltag Laboratories).

Cells were fixed in 1% formaldehyde, and analyses were performed on a FACScan (Becton Dickinson Co., Mountain View, CA).

After collagenase digestion of inguinal, brachial and axillary LNs from 35 DBA/2 mice, cells were fractionated on a discontinuous gradient of BSA (density = 1.082) as previously described (40). The recovered low-density cells, preincubated with the 2.4G2 mAb to reduce nonspecific binding, were stained with N418 revealed by PE-labeled anti-hamster Ig and FITC-labeled anti-I-E (14-4-4S) mAbs. Cells were sorted at a rate of 3000 events/s on a FACStar^Plus (Becton Dickinson) and were kept at 4°C throughout the procedure.

Cells were cytocentrifuged for 5 min at 300 rpm on slides. Cells were fixed in 1% formaldehyde, 0.2% glutaraldehyde for 5 min at 20°C and conserved at 4°C in PBS until use. Slides were incubated for several min in TBS (50 mM Tris, pH 7.6, 0.9% NaCl), then in TBS supplemented with 0.5% BSA and 0.02% Nonidet P-40 (staining buffer; Sigma Chemical Co.). Slides were immunostained in staining buffer supplemented with saturating levels of M5/114 hybridoma supernatant for 45 min at room temperature, washed twice in TBS, incubated in the staining buffer with peroxidase-conjugated rabbit anti-rat Ig (P 0450, Dako SA, Trappes, France), washed in 50 mM Tris, colored by 3,3′-diaminobenzidine hydrochloride (DAB tablets, Sigma Fast, Sigma Chemical Co.) for 2 to 3 min, rinsed 5 min in H₂O, counterstained with hematoxylin, rinsed 5 min in H₂O, fixed in 70% ethanol then 100% ethanol, and mounted.

LN cells were lysed in an RNA extraction solution (RNA-BTM; Bioprobe, Montreuil, France). Cellular RNA was then reverse transcribed at 42°C for 1 h in a 20-μl reaction containing: 1 mM of all four deoxynucleotide triphosphates, 0.04 U of random primer P(dN)6, 40 U of RNase inhibitor (Pharmacia LKB Biotechnology, Uppsala, Sweden), and 200 U of MMLV reverse transcriptase (Life Technologies, Gaithersburg, MD). Five microliters of the reverse transcription product were used for amplification in a 50-μl reaction containing: 1 μM of murine CD4 or β-actin primers, 200 μM of all four deoxynucleotide triphosphates, 1.5 mM MgCl₂, and 1 U Taq DNA polymerase (Goldstar DNA polymerase; Eurogentec, Seraing, Belgium). The murine CD4 primers were: sense primer, 5′-TGTGGCAGTGTCTGCTGAGTGA-3′, in the D4 domain; antisense primer, 5′-TGGCAGGTCTTCTTCTCACTGA-3′, in the cytoplasmic region. There are two introns present between the genomic position of these primers, and thus they can only amplify CD4 cDNA. Reactions were performed in a DNA thermal cycler (Hybaid Ltd., Teddington, U.K.) as follow: an initial denaturation cycle lasting 10 min at 94°C, followed by 35 cycles of amplification each comprising denaturation for 30 s at 94°C, annealing for 30 s at 62°C, and extension for 30 s at 72°C. The last cycle was followed by an extension cycle lasting 10 min at 70°C.

Nylon wool-passed T cells from mesenteric, brachial, axillary, and inguinal LN cells and graded numbers of FACS-sorted stimulator cells irradiated at 20 Gy were cocultured in RPMI 1640 supplemented with 10% FCS (Flobio, Asnieres, France), 50 μM 2-ME, and antibiotics in U-bottom 96-well microplates. After 108 h of culture, cells were pulsed with 1 μCi/well of [³H]TdR for an additional 12 h before harvesting and scintillation counting.

Mice were anesthetized with avertin (2.5% tribromoethanol) and the skin painted at the level of the triceps muscle or on the abdomen at the level of inguinal LN with 25 μl of 0.8% FITC (isomer 1, Sigma Chemical Co.) dissolved in a 1:1 mixture of acetone:dibutylphthalate just before application.

Mice were anesthetized with avertin and injected i.v. in the retro-orbital sinus with 100 μl of 50 to 100 mg/ml FITC-dextran, m.w. 40,000 (FD-40, Sigma Chemical Co.) or 150,000 (FD-150, Sigma Chemical Co.). Potential small FITC contaminants were removed by dialysis through a dialysis membrane (membra-cell 24037, Polylabo France) with a m.w. cut-off of under 12,000 to 16,000. Dialysis was performed for 24 to 48 h at 4°C in PBS in the dark.

Ganciclovir (GCV) was continuously administered at a dose of 50 to 55 mg/kg/day for 7 days by using miniosmotic pumps (model 2001, Alza Corp., Palo Alto, CA) implanted s.c. as previously described (24). Flt3 ligand (Immunex, Seattle, WA) was injected i.p. once a day for 7 days at a dose of 10 μg/day.

Double staining of LN cells, performed using the N418 mAb, which recognizes the murine CD11c (41), and a mAb against MHC class II, allowed us to distinguish four cell populations. Besides double-negative and MHC class II⁺/CD11c⁻ cells, which are mainly T and B cells, respectively, two distinct populations of double-positive cells were detected (Fig. 1,A). One expressed high CD11c levels and cell surface MHC class II at a level comparable to that of B cells (s-cells, Fig. 1,A). The other expressed 5 to 10 times more cell surface MHC class II and lower levels of CD11c (l-cells, Fig. 1,A). These cells represented approximately 0.5 and 1.5% of all LN cells, respectively. Size analysis of gated cells showed that l-cells were significantly larger than s-cells, hence the label s (small) and l (large) (Fig. 1 B). Two MHC class II⁺/CD11c⁺ populations were identified in all of the LNs (brachial, inguinal, axillary, para-aortic, or mesenteric) and mouse strains (DBA/2, C57Bl/6, CBA/J, FVB) analyzed (not shown). However, in mesenteric LNs, l-cells expressed lower levels of cell surface MHC class II compared with peripheral LNs.

To analyze their T cell stimulatory capacity, s- and l-cells were sorted by flow cytometry and used as stimulators in mixed lymphoid reaction. Both s- and l-cells had a strong T cell MLR-stimulating activity, similar to that of splenic DCs used as control (data not shown).

To observe the morphology of s- and l-cells, flow cytometry cell sorting was performed. Cytospins of sorted cells revealed for both types of cells the typical dendritic cellular processes that characterize DCs (1). However, some differences could be observed: l-cells were larger and had longer and more numerous dendritic processes; s-cells contained intracellular clusters of MHC class II-rich compartments (Fig. 2).

To further characterize s- and l-cells, we performed three- and four-color flow cytometry analysis (Fig. 3). The two populations did not express the classical markers of B cells (B220), T cells (Thy-1), or granulocytes (Gr-1). They did express a set of molecules associated with the Ag-presenting function of DCs such as MHC molecules and costimulatory and adhesion molecules ICAM-1, B7-1, B7-2, and CD40. l-Cells expressed intermediate to high levels of these molecules, while expression levels were lower in s-cells.

l-Cells did not express CD4, CD8α, the myeloid marker F4/80, and the 33D1 marker of splenic DC; they expressed low levels of the interdigitating cell marker DEC-205 and heterogeneous levels of HSA and Mac-1. In contrast, these molecules clearly defined two distinct populations of s-cells. One expressed CD4, F4/80, Mac-1, 33D1, and low levels of HSA and did not express DEC-205, while the other population expressed the lymphoid DC marker CD8α and DEC-205, high levels of HSA, and low to zero levels of CD4, F4/80, Mac-1, and 33D1 (Fig. 3 and data not shown). To verify that the detected CD4 expression was due to endogenous production, we analyzed CD4 transcription by RT-PCR on sorted cells. CD4 transcripts were indeed detected in s-DC using primers that encompass an intron and thus cannot amplify genomic DNA (data not shown). Altogether, based on their phenotype, sm- and sl-DCs appear to be myeloid or lymphoid related, respectively (9, 29).

Therefore, based on morphology, phenotype, and T cell-stimulatory capacity, s- and l-cells can be considered as typical DCs (1) and will subsequently be referred to as s-DC and l-DC, according to their small and large size, respectively. Based on their putative lymphoid- or myeloid-related origin, the two s-DC populations will subsequently be referred to as sl-DC and sm-DC, respectively.

The migration of DCs from extravascular compartments of nonlymphoid tissues to draining LNs via afferent lymph has been well documented (36, 37, 38). In this respect, 24 h after skin painting with FITC diluted in organic solvents, this tracer was detected in LN DCs, suggesting that epidermal LCs migrate to the draining LN (42, 43). We analyzed FITC staining of s-DC and l-DC at different time points after FITC skin painting. After 24 h, up to 60% of l-DCs showed high levels of FITC fluorescence in the draining brachial LN, whereas no staining was observed in the contralateral one. FITC staining of l-DCs became detectable 12 h after application and peaked at 24 h (Fig. 4). Interestingly, in preliminary experiments, when we applied a 10-fold higher volume of FITC, we observed a significant increase in l-DCs proportion, suggesting that new cells immigrated into the LN (data not shown). However, this procedure was not suitable for additional experiments due to staining of the contralateral LN DCs. The proportion of FITC-stained cells then decreased, with only a few stained cells remaining at day 5 (Fig. 4); this could be due to either their migration outside the LN, their death in situ, or FITC degradation. During the 5-day follow-up, no significant FITC staining could be detected in CD11c-negative cells, nor in either s-DC population (Fig. 4).

We similarly tested the capacity of the three populations of LN DCs to acquire blood Ag using i.v. injection of FITC-dextran. High m.w. dextrans are known to remain in the plasmatic compartment (44). They have also been used as a marker to quantify the endocytic capacity of DCs (17). Cells of peripheral LN were analyzed at different time points after i.v. injection of FITC-dextran. At 30 min, 17% of s-DCs were highly fluorescent in both brachial and inguinal LNs. In contrast, only 3% of l-DCs were FITC⁺, and moreover, the fluorescence of these stained cells was weak (Fig. 5). The proportion of FITC-stained s-DCs increased rapidly during the first 30 min after injection, and then more slowly during the next 12 h to reach 30% of the cells. This percentage then remained stable for 5 days. The percentage of fluorescent l-DCs increased slowly, reaching the same proportion after 12 h as that observed at 30 min for s-DCs. At day 5, the percentages of fluorescent s- and l-DCs were similar. During this time period, only traces of CD11c⁻ cells were FITC⁺ (Fig. 5). Similar results were obtained using FITC-dextran of 40,000 or 150,000 m.w. or after dialysis to remove potential small m.w. FITC contaminants (data not shown).

To analyze which of the s-DC acquire FITC-dextran, we performed four-color flow cytometry analysis. 30 min after FITC-dextran i.v. injection, 17% of the sl-DCs were stained, while >50% of the sm-DCs were FITC⁺ (Fig. 5,C). However, sl- and sm-DCs populations overlap based on CD4 staining (Fig. 3 A), and it should be noticed that the FITC-stained sl-DCs are positioned close to the bar separating sm- and sl-DC on the CD4 staining axis. Therefore, it can be assumed that most if not all stained cells are sm-DCs. Similar results were obtained at 12 h after FITC-dextran injection (data not shown).

Careful analysis of FITC staining in s- and l-DCs revealed two interesting features. First, there was a progressive increase in the cell surface expression of MHC class II in most FITC⁺ cells, which resulted in a shift from the s- to the l-DC population (Fig. 6,A). Furthermore, while CD11c expression was heterogeneous among l-DCs, the l-DCs that became FITC⁺ had high CD11c expression levels, comparable to that of s-DCs. Second, the intensity of FITC staining increased with time among s-DCs and also, but in a delayed manner, in l-DCs (Fig. 6 B).

We next analyzed the in vivo turnover of l- and s-DCs. We used a model of transgenic mice allowing the conditional ablation of dividing DC precursors expressing an HSV1-TK gene upon GCV treatment (24, 39). This enzyme allows the conversion of the nontoxic GCV into GCV triphosphate, which can be incorporated into elongating DNA, inducing elongation termination and ultimately cell death. Because only those cells that express HSV1-TK and that are dividing can be killed, size variation of a cell population during a GCV treatment reflects the turnover of these cells if they divide or have a dividing precursor. However, this system will not discriminate between the rapid turnover of a resident cell population that divides in situ or of a circulating cell population with a dividing precursor. Using this model, we previously showed the rapid turnover of spleen DCs, amounting to 10 to 15% renewal per day, in agreement with previous measurements obtained with other techniques (21). Here, we found that a 7-day GCV treatment led to the almost complete disappearance of both populations of s-DCs, while the l-DC population was only slightly reduced (Table I). These results reveal further differences between the three populations of LN DCs, sm- and sl-DC having a rapid turnover, l-DC a slow turnover.

The in vivo administration of Flt3 ligand, a stimulator of hemopoietic progenitor cells, has recently been described to induce hypertrophy of lymphoid tissues and to dramatically increase the proportion of spleen DC (45, 46). The effects of Flt3 ligand on the distinct subpopulations of LN DCs were analyzed by three-color flow cytometry after a 7-day treatment. The l-DCs did not appear to be affected by Flt3 ligand. In contrast, the proportion of s-DCs reached 30% of total LN cells vs 0.5% in controls (Fig. 7,A). The expanded population did not express CD4, F4/80, or Mac-1, but expressed low levels of CD8α and high levels of HSA, indicating that Flt3 ligand treatment affected mostly if not exclusively the sl-DC subpopulation (Fig. 7 B). A moderate increase in the other DC populations could be masked by the dramatic increase of the sl-DCs.

In mouse lymphoid tissues, N418 is the only mAb that strongly reacts with DCs but not with freshly isolated macrophages or lymphoid cells (29, 41). We looked for a possible heterogeneity of DCs in LN within CD11c⁺, as already described in the spleen (27, 29, 32, 47) and in Peyer’s patches (30). We used double-staining analysis of whole LN cells without any purification procedure or culture, which could have modified the cell characteristics or resulted in the loss of a DC subpopulation. Three distinct CD11c⁺/MHC class II⁺ populations were clearly disclosed by four-color flow cytometry analysis. Given their overall heterogeneity, it is admitted that identification of DCs is based on a combination of characteristics including: 1) their morphology, with the existence of dendrites; 2) their phenotype, showing the presence of molecules involved in T cell activation; and 3) their T cell stimulatory capacity (1). The three CD11c⁺/MHC class II⁺ cell populations fulfill all of these criteria and can thus be considered as typical DCs.

Due to their high level cell surface expression of MHC class II and costimulatory molecules, their size, and their dendritic morphology with numerous and long cell processes, l-DCs resemble mature DCs. On the other hand, due to the mainly intracellular location of MHC class II molecules, their smaller size, and their smaller dendrites, both populations of s-DCs resemble more immature DCs (15, 17, 48). Despite differences in MHC class II expression, all of these cells strongly stimulate in vitro allogeneic responses to similar levels. This is probably due to a rapid in vitro maturation of s-DCs during the culture, as suggested by preliminary experiments (data not shown).

Expression of the Mac-1 and F4/80 myeloid markers on the sm-DC population supports a myeloid origin of these cells. CD4 expression does not argue against such a hypothesis, since CD4 is not considered a marker of lymphoid-related DCs (49). Nonetheless, sm-DC are the only identified DCs that clearly express CD4 at levels similar to that of CD4 T cells. In addition, based on 33D1 expression and low HSA expression levels, sm-DCs resemble marginal zone splenic DCs (26, 29), a subpopulation considered to be myeloid related (9, 50). In contrast, sl-DCs expressed CD8α, a marker considered to define lymphoid-related DCs (9). sl-DCs also express DEC-205 and high levels of HSA, thus resembling splenic DCs of the T cell zone that have been considered lymphoid related (9, 26). Finally, the observation that Flt3 ligand treatment induced the expansion of only sl-DCs reinforces the hypothesis that the two s-DCs populations have distinct lineages.

In addition to these phenotypic differences, we showed that these DC subpopulations have a different turnover by using an animal model of conditional ablation of HSV1-TK-expressing DC upon GCV treatment. Using this model, we previously showed that a 7-day GCV treatment led to complete disappearance of splenic DCs of the white pulp marginal zone, while LCs that also expressed HSV1-TK were minimally affected (24, 39). Since GCV kills only HSV1-TK-expressing cells that are dividing (51), these experiments indicate that the splenic DCs have a very rapid turnover, whereas LC have a slow turnover, in agreement with previous observations (19, 20, 21). Using the same experimental system, we show here that sm- and sl-DC populations have a rapid turnover, since a 7-day GCV treatment resulted in the disappearance of almost all s-DCs. This rapid turnover could be due to either a high division rate of the s-DC themselves or to a rapid transit into LN of cells that have a rapidly dividing precursor. Additional experiments will be needed to address this issue. On an other hand, GCV treatment induced only a slight depletion of l-DCs (1.2 vs 1.6% in control mice). This result is compatible with the persistence of the LC-derived l-DC and the disappearance of the sm-DC-derived l-DC (see below).

The majority of l-DCs come from the peripheral territories drained by the regional LN. Indeed, at 3, 12, 24, and 120 h after FITC skin painting, l-DCs but not s-DCs were FITC stained. Because there was a progressive increase in the proportion of l-DC-stained cells from 12 to 24 h after skin painting, without any detectable stained s-DCs throughout this time period, it is unlikely that we could have missed a transition from stained s-DCs to l-DCs. This suggests that neither sm- nor sl-DCs represent an intermediate differentiation stage from LCs to l-DCs and that LCs have completed their phenotypic maturation when they arrive in the draining LN. These l-DCs were thus derived from LCs, which themselves are believed to derive from the CFU-DC progenitors yielding pure DC colonies in semisolid medium (3, 12). A maximum of 60% of all l-DCs was stained after skin painting. This may indicate that not all the territory drained by the analyzed LN was painted and/or that some l-DCs have a different origin. In this respect, there is some evidence that a fraction of the l-DC population is derived from sm-DCs. First, sm-DCs were the first to be stained after i.v. injection of FITC-dextran, followed by a delayed staining in l-DCs, which appeared with kinetics compatible with a maturation of sm-DC to l-DC. Second, although CD11c expression in l-DCs was quite heterogeneous, those l-DCs stained by the plasmatic tracer had a CD11c expression that was more homogeneous and was similar to that of s-DCs. Finally, preliminary experiments suggest that some s-DCs can evolve toward an l-DC-like phenotype upon culture (data not shown).

Altogether, these results suggest that LNs contain DCs from the three DC lineages that have been proposed so far, originating from: 1) a pure DC progenitor (for most of l-DC) (3, 12); 2) a common myeloid/DC precursor (for sm-DC) (3, 4); and 3) a common lymphoid/DC precursor (for sl-DC) (10).

Intravenous injection of FITC-dextran led to a remarkably rapid staining of sm-DCs. This staining was observed with both 40,000 and 150,000 m.w. FITC-dextran. These molecules are known to remain in the blood circulation and are actually used in humans to replenish the plasma compartment. Therefore, it is unlikely that they diffused directly from the blood to the LN parenchyma. On the contrary, the staining was specific in many ways. First, only sm-DCs, representing 0.2% of all LN cells, were stained by FITC-dextran 30 min after i.v. injection, whereas all of the LN cells were stained when incubated with FITC-dextran in vitro (data not shown). Second, contamination with plasma molecules present in the LN blood vessels and occurring during preparation of LN cells was ruled out; no staining could be observed when LN cells from an untreated mouse were prepared in the presence of the plasma of a FITC-dextran treated mouse (data not shown). Finally, stained s-DCs were only detected in peripheral but not in mesenteric LN. These observations also indicate that the staining of sm-DCs was not due to passive diffusion of FITC-dextran by blood vessel leakage. Therefore, these results suggest that FITC-dextran was acquired within the blood compartment by sm-DCs; this could be due to either the existence of DCs within the blood vessel endothelium layer of peripheral LNs or to a translocation of blood DCs to LNs. In this respect, we observed that the intensity of FITC staining in sm-DCs increased with time, suggesting that some sm-DCs were exposed several times to the plasmatic tracer. These cells may therefore be extremely mobile, migrating from the blood to LNs and back to the blood. Such a recirculation has not been observed for spleen or lymph-borne DCs (52, 53, 54). Altogether, our results suggest a possible circulation of DC from blood to LN. Further careful analysis will be required to clarify this important point, which conflicts with the current paradigm for DC circulation (38).

Until now, Ags delivered through the blood supply have been assumed to be trapped mainly by APCs in the spleen, as well as in the liver for particulate Ags (38, 55). Our results demonstrate that regardless of the mechanism, previously unidentified LN DCs can specifically uptake a plasmatic molecule. These cells, therefore, might play an important role in the control of blood pathogens. In this line, this cell population seems to play a role in the transport of HIV from the blood to the LNs (our manuscript in preparation).

In conclusion, we show for the first time the existence of a DC heterogeneity in the LN. The three DC populations identified appear to belong to distinct lineages and to differ in their capacity to uptake Ags administrated by different routes. These results, together with further analysis of the functional properties of these DCs, may have important implications for a better understanding of the relationship between the route of Ag introduction and the nature of the subsequent immune response.

We thank Drs. Geneviève Milon, Polly Matzinger, Olivier Boyer, and Pieter Leenen for helpful discussions, Drs. Jean Claude Gluckman and Michelle Rosenzwajg for critical reading of the manuscript, Sylvie Bruel and Catherine Pioche for their contribution to some of the experiments, Dr. Eugene Maraskovsky, Immunex Corp., for providing us with the flt3 ligand, and Geneviève Milon and Jeffrey Bluestone for providing some of the mAbs.