Ligation of E-cadherin on in vitro–generated immature Langerhans-type dendritic cells inhibits their maturation

Dendritic cells (DCs) represent a developmentally heterogeneous class of leukocytes that are highly specialized in antigen uptake, processing, and presentation. DCs differ in migration pathways, tissue location, and functional abilities. Mature DCs in T-cell areas of lymph nodes are known as interdigitating DCs and are in part derived from a peripheral DC migration pathway that involves immature DCs in peripheral organs such as epithelial Langerhans cells (LCs).1-3 After homing to T-cell areas of secondary lymphoid organs, DCs rapidly undergo apoptosis unless they receive a survival signal from antigen-specific T cells.4 

Immature epidermal LCs fulfill a sentinel role by filtering the surrounding tissue for foreign antigens and pathogens. They form a 3-dimensional network in suprabasal epidermal layers and spend long time periods at an immature or a nonactivated differentiation stage in the epidermis before they migrate to the lymph nodes.5 LCs express high levels of the homophilic adhesion molecule E-cadherin and undergo E-cadherin–dependent adhesion with epidermal keratinocytes.6 E-cadherin expression is markedly down-regulated upon the migration and maturation of epidermal LCs, and lower expression levels of E-cadherin on the surfaces of cultured LCs correlate with decreased cell adhesiveness.6-8 LC migration can be induced in vivo by the topical application of allergens or the intradermal or systemic injection of tumor necrosis factor-α (TNFα), interleukin-1, or lipopolysaccharide (LPS) (reviewed in Austyn9). Migration of LCs involves the adhesion of LCs to the epidermal basement membrane by α6 integrin,10 and it can be inhibited by monoclonal antibodies (mAbs) to CD44-variant isoforms.11 Passage of migrated DCs from lymphatic vessels into T-cell areas of draining lymph nodes at least partially requires chemokine SLC interaction with the mature DC-associated receptor CCR7.12,13 

LC migration from the epidermis to the draining lymph nodes is preceded by the activation and maturation of immature LCs locally in the epidermis. Because immature epidermal LCs spontaneously undergo maturation on in vitro culture,14 inhibitory signals that are provided by the epidermal microenvironment might likely counteract the maturation of LCs in vivo.

We recently established defined serum-free culture conditions for the in vitro generation of immature LCs from human CD34⁺ cord blood progenitor cells. We demonstrated that LC differentiation from CD34⁺ cells is dependent on the addition of transforming growth factor β1 (TGF-β1) to defined serum-free cultures supplemented with granulocyte macrophage–colony-stimulating factor (GM-CSF) plus TNFα and stem cell factor (SCF).15TGF-β1 induces LC growth from shared clonogenic precursor cells with myelomonocytic cells, and this effect on LC growth is strongly enhanced by a synergistic activity of SCF plus flt3 ligand (FL).16LCs selectively accumulate in this differentiation model at an immature stage resembling epidermal LCs in vivo (CD86^dim/−, CD83⁻, Birbeck granule⁺⁺, Lag⁺⁺, CD1a⁺⁺, E-cadherin⁺⁺).

The availability of a differentiation model that allows the maintenance of LCs in vitro in defined serum-free cultures at an immature stage provided the opportunity to study signaling mechanisms leading to the maturation of immature LCs. Our data suggest that functional activation of the TGF-β1–induced cell surface molecule E-cadherin actively suppresses the maturation of LCs in this differentiation model.

Murine monoclonal antibodies of the following specificities were used in flow cytometry analyses. CD34 (clone HPCA2) and CD11c (clone S-HCL-3) were obtained from Becton Dickinson Immunocytometry Systems (San Jose, CA). CD83 (clone HB15a) was obtained from Immunotech (Marseilles, France). CD86 (clone IT2.2) was obtained from PharMingen (San Diego, CA). Lag antibody (clone Lag), specific for a 40-kd glycoprotein associated with Birbeck granules,17was kindly provided by Drs Imamura and Yoneda (Kyoto, Japan). CD1a (clone VIT6b) and a control mAb (clone VIAP, IgG1) were produced in our laboratory. Antibodies used in functional analyses are listed in Table2. Monoclonal antibody preparations showing functional effects were controlled for endotoxin (LPS) and were found to contain less than 0.1 ng/mL endotoxin, with the exception of the commercially obtained mAb HEC/75 (CD31), which contained low amounts (3.5 ng/mL) LPS. It is, however, highly unlikely that the observed functional effects of CD31 mAb resulted from LPS contamination because we performed all the experiments in serum-free medium that did not contain significant amounts of LPS-binding protein and soluble CD14. Furthermore, we detected similar amounts of LPS in 2 additional mAb preparations that did not show functional effects on DC maturation when added to parallel cultures (clones 4B4, CD29 and HP2.1, CD49d).

For membrane staining, 50 μL cells (10⁷/mL) were incubated for 15 minutes at 0°C to 4°C with 20 μL conjugated mAb. Cells were washed and analyzed by flow cytometry or submitted for intracellular staining.

For suspension staining of intracellular antigens, we used the commercially available reagent combination Fix&Perm (An der Grub, Kaumberg, Austria) according to the manufacturer's procedure. Briefly, cells were first fixed for 15 minutes at room temperature (50 μL cells plus 100 μL formaldehyde-based fixation medium). After one washing with phosphate-buffered saline, pH 7.2, cells were resuspended in 50 μL phosphate-buffered saline and mixed with 100 μL permeabilization medium plus 20 μL fluorochrome-labeled antibody. After further incubation for 15 minutes at room temperature, cells were washed again and analyzed by flow cytometry. Indirect suspension staining for the intracellular Birbeck granule marker molecule Lag was performed as described previously.15 

Flow cytometric analyses were performed using a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems) equipped with a single laser emitting at 488 nm. Cell sorting was performed using a FACS Vantage flow cytometer (Becton Dickinson). The purity of the CD1a⁺ cell fractions obtained by sorting was determined by re-analysis and was found to exceed 95%.

Cord blood (CB) samples were collected during healthy full-term deliveries. Mononuclear cells were isolated within 10 hours of collection by discontinuous Ficoll/Hypaque (Pharmacia, Uppsala, Sweden) density gradient centrifugation. CD34⁺ cells were isolated from CB mononuclear cells using the MACS CD34 Progenitor Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) as described previously,15 according to the instructions of the manufacturer. The purity of the CD34⁺ population exceeded 90%.

Primary (1°) cultures of purified CD34⁺ CB cells were grown in 24-well plates (Costar, Cambridge, MA) (5 × 10³ cells in 1 mL/well) at 37°C in a humidified atmosphere and in the presence of 5% CO₂, as previously described by us.16 The serum-free medium X-VIVO 15 (BioWhittaker, Walkersville, MD) contained L-glutamine (2.5 mmol/L), penicillin (125 U/mL), and streptomycin (125 μg/mL). Cultures were supplemented with optimized concentrations of the following human cytokines: FL (100 ng/mL; kindly provided by Immunex, Seattle, WA), TGF-β1 (0.5 ng/mL; purified from platelets; British Biotechnology, Abington, UK), rhTNFα (50 U/mL; Bender, Vienna, Austria), rhGM-CSF (100 ng/mL; Novartis, Basel, Switzerland), and rhSCF (20 ng/mL; Amgen, Thousand Oaks, CA). A 50% medium exchange was regularly performed at culture day 7.

Secondary (2°) culture experiments were set up using cells generated in 1° LC generation cultures as described above. Cells from 1° cultures were harvested at days 10 to 14. Single-cell suspensions were prepared by pipetting, and resuspended cells were plated in the above-described growth medium X-VIVO 15 in 24-well plates at a cell density of 1 × 10⁵ cells per well (1 mL). Cultures were supplemented with GM-CSF, TNFα, or TGF-β1, or any combination of them, at concentrations described above. CD40 ligand trimer (CD40L; 200 ng/mL; kindly provided by Immunex) was added when indicated.

Cells from 1° cultures (see above) were harvested, carefully resuspended by pipetting, and submitted to 2° cultures (ie, aggregation cultures) of 5 × 10⁴ cells/well in flat-bottomed 96-well plates (Costar) under exactly the same growth conditions as for 1° cultures. These reaggregation cultures were supplemented with 20 μg/mL mAb when indicated, preincubated with mAb for 20 minutes at 4°C, and transferred to 37°C. After additional culture for 24 hours, cell cluster formation was analyzed by phase-contrast microscopy using a scoring system previously established in our laboratory18 with slight modifications. Scores ranged from 0 to 4, as follows: 0, less than 10% of the cells were in aggregates; 1, 10% to 50% of the cells were in aggregates; 2, approximately 50% to 75% of the cells were in aggregates; 3, 75% to 90% of the cells were in aggregates; 4, 90% to 100% of the cells were in aggregates. In the first series of experiments, these cultures were initiated with total cells generated in primary cultures at day 8. To further study direct homotypic LC-to-LC interaction, we initiated identical cultures (see above) with flow sorted CD1a⁺ cells generated after 10 to 12 days in 1° cultures. To assess the differentiation stage of LC, cells from these reaggregation cultures were stained for CD1a or were double-stained for CD1a versus CD86 or CD83 expression and analyzed by flow cytometry as indicated in “Results.”

Graded numbers of irradiated (30 Gy; cesium Cs 137 source) stimulator cells (generated LC) were added to constant numbers (5 × 10⁴/well) of purified (greater than 98%) allogeneic T cells in round-bottom 96-well tissue culture plates (Costar). Triplicate analyses were performed. Stimulation of responding T cells was monitored by measuring ³H-thymidine (Amersham Life Science, Buckingham, UK) incorporation on day 5 of culture. Incorporated radioactivity was measured using a Top-Count microscintillation counter (Packard Instrument, Meriden, CT). Allogeneic T cells used in these experiments were purified from peripheral blood mononuclear cells by negative immunomagnetic depletion (MACS beads; Miltenyi Biotec) using mAbs specific for CD14 (clone VIM13; generated in our laboratory), CD16 (clone 3G8; Caltag), CD19 (clone HD37; kindly provided by Dr G. Moldenhauer, Heidelberg, Germany), HLA-DR (clone L243; ATCC, Rockville, MD), and CD33 (4D3; generated in our laboratory) as previously described by us.16 

We recently described that the addition of TGF-β1 to serum-free cultures of CD34⁺ cells, supplemented with the cytokines GM-CSF plus TNFα, SCF, and FL, results in the generation, after a culture period of 7 to 10 days, of a large cell fraction that strongly expresses CD1a (approximately 60%).16 Indistinguishable from epidermal LC in vivo, most CD1a⁺ cells generated in these cultures are predominantly CD86 (B7.2)^dim/− and lack expression of the mature DC marker molecule CD83.16 This immature DC phenotype of most of the generated cells remains preserved on extension of the culture period to 14 to 16 days (Figure1A,B). Furthermore, most CD1a⁺ cells generated in these TGF-β1–supplemented cultures express high levels of the homophilic epithelial adhesion molecule E-cadherin (Figure 1A, upper right diagram), and a large proportion of cells expresses the Birbeck granule–associated molecule Lag (Figure 2, upper panel; mean, 52% at day 14 to 16; n = 7). Cells generated in identical culture medium in the absence of TGF-β1 lack expression of Lag (Figure 2, lower panel), and most of these cells remains CD1a⁻/E-cadherin⁻ (Figure 1A).

The most striking morphologic feature of TGF-β1–supplemented cultures is the occurrence of large cell clusters from culture day 7 on (Figure 2, upper panel). In comparison, little or no cell clustering occurs in TGF-β1–nonsupplemented cultures (Figure 2, lower panel). TGF-β1–dependent LC generation and cell cluster formation are accompanied by dramatic cell multiplication. At day 0, 5 × 10³ CD34⁺ cells/mL plated in 24-well plates give rise to 2.1 × 10⁵ to 5.4 × 10⁵ (mean, 3.5 × 10⁵; n = 7) cells after day 12. Thus, cells generated in these cultures in response to TGF-β1 stimulation resemble immature LCs. They fail to undergo further DC maturation despite the presence of TNFα in culture, which has been previously demonstrated to induce maturation (CD86^bright, CD83⁺) of in vitro–generated immature DCs in cultures of CD34⁺ hematopoietic progenitor cells.19-25 We investigated whether this LC maturation arrest was caused by a lack of stimulatory factors not present in our defined serum-free cultures or by an inhibitory effect of TGF-β1.

LCs generated in the above-described primary cultures of CD34⁺ cells for 10 to 14 days were harvested, resuspended to obtain single-cell suspensions, and plated in 2° serum-free cultures. Stimulation of LC in these 2° cultures in the presence of GM-CSF plus TNFα for 48 hours resulted in the acquisition of mature DC features by most generated CD1a⁺ cells. As shown in Figure 3, panel A (top diagrams), most immature CD1a^bright cells from primary (1°) cultures acquired increased expression of CD86 and were found to express the mature DC marker molecule CD83 on stimulation with GM-CSF plus TNFα. These phenotypic changes were correlated with a marked reduction of mean CD1a expression density of cultured cells (Figure 3A). Morphologic examination of GM-CSF plus TNFα-supplemented 2° cultures revealed numerous loosely plastic adherent and floating single cells with highly dendritic processes (Figure 4). GM-CSF plus CD40L-supplemented parallel cultures showed a similar maturation pattern, though these cells were found to be consistently brighter CD86⁺ and CD83⁺ than those cultured in GM-CSF plus TNFα, and they showed a more mature DC morphology (Figures 3A,4). Interestingly, even in the presence of GM-CSF alone (Figure 3A, third panel), 2° cultures contained higher percentages of CD86⁺ cells than 1° control cultures in which cell clusters remained undisturbed (Figure 3A, bottom). Parallel staining of cultured cells for additional marker molecules revealed a marked up-regulation of cell surface expression of HLA-DR, CD80, and CD54 molecules in parallel with the above-described DC maturation-linked phenotypic changes (CD83 induction, CD86 up-regulation, and decreased CD1a expression). Conversely, most CD1a⁺ cells expressed CD11c, and CD11c expression levels did not clearly change in 2° cultures (data not shown).

Figure 3, panel B shows a phenotypic analysis of cells stimulated in the presence of fresh medium supplemented with the initial cytokine combination (TGF-β1 plus GM-CSF, TNFα, SCF, and FL). As can be seen, cells that were resuspended acquired mature DC characteristics (Figure 3B, open histograms), whereas cells that remained as cell clusters (Figure 3B, filled histograms) stayed predominantly CD86^dim/− and CD83^dim/−.

These experiments demonstrate that cytokines present in 1° cultures (ie, GM-CSF plus TNFα) sufficiently promote the maturation of LCs. Thus, the maturation arrest of immature LC in 1° cultures cannot be explained by a lack of DC maturation stimuli.

LCs generated in primary TGF-β1–supplemented cultures remain immature (Figure 3A, bottom). Because TGF-β1 is mainly described as a suppressive cytokine that inhibits immune responses, we next asked whether TGF-β1 might suppress the maturation of LC. To study this possibility, we added TGF-β1 to the 2° cultures supplemented with either GM-CSF plus TNFα or GM-CSF plus CD40L and phenotypically characterized cultured cells. We observed similar CD86 and CD83 expression patterns by cultured cells in the presence or absence of TGF-β1 (Figure 5), demonstrating that TGF-β1 does not inhibit LC maturation in 2° cultures.

The experiments described above revealed that the observed LC maturation arrest in 1° cultures cannot be explained by a lack of maturation-inducing cytokines or by a direct maturation suppressive effect of TGF-β1. Therefore, we investigated an alternative possibility and analyzed whether LC clustering may provide a suppressive signal that prevents the maturation of LC in this differentiation model.

We observed that the mechanical disruption of LC clusters and replating of generated cells to short-term, serum-free 2° cultures at a high cell density (5 × 10⁴ cells in flat-bottom, 96-well plates; see “Materials and methods”), supplemented with identical cytokines as described for 1° cultures (see above), resulted in secondary cell cluster formation. This permitted us to analyze LC clustering in more detail. As can be seen from Figure6, single-cell suspension of LCs (Figure6B) prepared by the enforced pipetting of clustered cells from 1° cultures gave rise within 24 hours of additional culture in identical growth medium (cytokines GM-CSF plus TNFα, SCF, FL, and TGF-β1) to secondary cell cluster formation (Figure 6A). These cell clusters are smaller than those observed in 1° LC generation cultures (compare Figure 6A with Figure 2, upper panel). We applied a scoring system (1 = lowest, 4 = highest, degree of aggregation; see “Materials and methods”) to quantify secondary cell cluster formation in these 2° cultures. Spontaneous LC re-aggregation, shown in Figure 6, panel A, scored 3 (75%-90% of cells are present in aggregates; see “Materials and methods”). This rapid secondary LC clustering phenomenon was found to be calcium and temperature dependent because it could be completely prevented by the addition of EDTA or by incubating cells at 4°C (score 0; data not shown).

Double staining revealed that most generated CD1a⁺cells in primary cultures express the homophilic adhesion molecule E-cadherin (Figure 1). We analyzed cells generated under the various culture conditions for E-cadherin expression. As can be seen in Table1, expression of E-cadherin slightly decreases on subculturing of day 10 generated cells in 2° cultures supplemented with GM-CSF, GM-CSF plus TNFα, or GM-CSF plus CD40L. Conversely, expression intensity of E-cadherin even increases in parallel 2° cultures containing the initial TGF-β1–supplemented cytokine combination.

We further analyzed cells generated in 1° cultures for the expression of additional cytoadhesion molecules. As can be seen in Table2, substantial proportions of generated CD1a⁺ LC express the β1 integrin molecules VLA-4 (CD49d/CD29) and VLA-5 (CD49e/CD29), the β2 integrin LFA-1 (CD11a/CD18), and sialophorin (CD43). Additionally, a subset of generated cells expresses PECAM-1 (CD31) (Table 2).

The short-term re-aggregation characteristics of generated LC at a high-cell density allowed us to study whether the addition of blocking mAbs to candidate cytoadhesion molecules might influence secondary LC cluster formation. For quantification of secondary LC cluster formation in these 2° cultures, we applied the scoring system (see “Materials and methods”). Additionally, we analyzed cells after stimulation in these 2° cultures for expression of CD1a (immature LC, CD1a^bright; mature LC, marked loss of CD1a expression density; see Figure 3). As can be seen from Figure 6, panel C and Figure 7, panel A, the addition of an mAb to E-cadherin in these 2° cultures led to the inhibition of cell cluster formation (score 1), and this effect was associated with increased CD1a expression by stimulated cells compared to control cultures (Figure 7B). Monoclonal antibodies to CD11a or CD31 also strongly inhibited secondary cell cluster formation (score 0 or 1, respectively; Figure 7A), but this effect was associated with a decrease of mean CD1a expression levels (Figure 7B) of cultured cells in 2° cultures. Conversely, mAb ligation of CD43 led to the enhancement of cell clustering (Figures 6D, 7A; score 4 versus 3), and mean CD1a expression by cultured cells decreased only slightly (Figure7B).

CD1a⁺ LC generated in 1° cultures make up approximately 60% of all cultured cells. To eliminate a possible influence of contaminating CD1a⁻ cells on cell clustering and maturation induction of LC, we performed cell sorting experiments of generated immature CD1a⁺ LC from 1° cultures (see “Materials and methods”). In these experiments (Figure8), we purified CD1a⁺ LC from day 10 to 12 cultures by flow sorting and replated them in secondary cultures in the presence of mAbs as described above. As can be seen from Figure 8, most purified immature CD1a⁺ LC acquire, after further stimulation in control cultures, high expression levels of CD86 and decreased CD1a expression densities indicative of DC maturation. This maturation pattern of LC is associated with secondary cell cluster formation (cluster score 3; data not shown). Parallel cultures supplemented with mAb HECD-1 specific for E-cadherin contained substantially higher percentages of immature CD86^{− to dim}/CD1a^bright LCs compared with control cultures (64% vs 37% in the representative experiment shown in Figure 8) and showed only minimal cell clustering (score 1). These observations were confirmed in 3 additional experiments using flow sorted CD1a⁺ LCs (mean ± SE, 66% ± 5% vs 38% ± 2%,P = .008; n = 4). In comparison, control cultures supplemented with a mAb to LFA-1 (CD11a) contained reduced percentages of immature LC (29% vs 37%, Figure 8) and, in turn, higher percentages of phenotypically mature CD1a^dim/CD86^bright DCs. As observed for anti–E-cadherin mAb, mAb to LFA-1 inhibited secondary cluster formation of purified LCs (score 1). Parallel analyses for CD1a versus CD83 expression in 2 experiments revealed virtually identical results (CD83^dim/−/CD1a^bright LCs, 60% vs 32% and 53% vs 36%, respectively; CD86^dim/−/CD1a^bright LCs, 61% vs 32% and 53% vs 36%, respectively). We further analyzed the capacity of generated LC to induce allogeneic T-cell proliferation and found that anti–E-cadherin mAb-pretreated cells appear to be less potent inducers of allogeneic T-cell proliferation at low stimulator cell numbers than cells from control cultures supplemented with a nonbinding control mAb or cells from cultures without antibody supplementation (Figure9).

Our data suggest an active involvement of E-cadherin in the regulation of LC maturation. We demonstrate here that immature LCs generated in cultures of CD34⁺ cells undergo profound E-cadherin–dependent homotypic cell clustering similar to what has been described recently for in vitro–generated murine LC.26 We also observed that mechanical disruption of these E-cadherin–dependent LC clusters rapidly induces the acquisition of mature DC features (CD86^bright, CD83⁺) by immature LCs. Furthermore, mAb ligation of E-cadherin on the surface of immature LCs after mechanical cluster disaggregation inhibits the maturation of LC. This effect was found to be specific for E-cadherin engagement because mAbs to other adhesion molecule systems that were found to be similarly involved in LC clustering (CD31, LFA-1) failed to inhibit DC maturation when added to parallel cultures. This suggests that anti–E-cadherin mAb induces functional activation of E-cadherin, resulting in the inhibition of DC maturation in this differentiation model. This effect of anti–E-cadherin mAb mimics E-cadherin–dependent homotypic LC clustering because we observed that LCs stay immature if cell clusters remain undisturbed. Thus anti–E-cadherin mAb binding to E-cadherin seems to functionally mimic E-cadherin binding by its homophilic ligand.

Our observations that anti–E-cadherin mAb inhibits the maturation of LC despite the continuos presence of TNFα in culture suggests an active suppressive effect of E-cadherin on the maturation of LC. Using similar culture conditions, several previous studies have demonstrated that TNFα efficiently promotes the acquisition of mature DC features (CD86^bright, CD83⁺) by immature DCs generated in cultures of CD34⁺ progenitor cells.19-25 We demonstrate here that E-cadherin expression by cultured cells is dependent on TGF-β1 stimulation. Thus, a failure of induction or maintenance of high expression levels of E-cadherin in the absence of exogenous TGF-β1 in the above-mentioned previous studies may be responsible for the observed rapid acquisition of mature DC features by cultured cells. In support of this assumption, recent studies demonstrated that TGF-β1 induces E-cadherin expression by in vitro–generated murine27 and human DC28 along with immature LC features. Furthermore, TGF-β1 stimulates the generation of immature LC-type DCs, whereas it suppresses the generation of mature DCs in TNFα-containing cultures of murine progenitor cells27,29 or human monocytes.30 

According to our differentiation model, the following mechanism may allow maintenance of immature LCs in vivo. TGF-β1 stimulation may lead to sustained expression of high levels of E-cadherin on the surfaces of epidermal LCs over long periods of time. This will allow LCs to undergo continuous homophilic E-cadherin adhesion with surrounding keratinocytes. E-cadherin adhesion in turn may suppress LC maturation and thereby maintain immature LC features in vivo. Such a scenario seems to be supported by in vivo studies. First, LCs are well described to rapidly acquire mature DC features after purification from their epidermal micro-environment.14 Second, keratinocytes constitutively express DC maturation-inducing cytokines including TNFα, which, in the absence of an epidermal-associated counteracting mechanism, may cause the maturation of LCs in vivo.31,32Third, immature epidermal LCs abundantly synthesize TGF-β1,33 and TGF-β1 autoproduction is essential for the presence of immature LCs in the epidermis in vivo.34,35 In line with this, TGF-β1 seems to promote rather than inhibit physiological functions of freshly isolated immature LCs.36 In further support of a suppressive effect of E-cadherin adhesion on the maturation of LCs, decreased E-cadherin expression in response to the topical application of allergens is correlated with the local maturation of LCs in the epidermis,8 whereas unspecific inflammatory conditions do not necessarily induce the same effect. Thus E-cadherin–mediated active maturation suppression of LCs, as suggested by our study results, may prevent the uncontrolled maturation of LCs by proinflammatory cytokines in vivo and thus may enable LCs to respond to specific signals, such as particular pathogens or antigens.

In support of a direct functional involvement of E-cadherin in regulating the maturation of epidermal LCs, E-cadherin is capable of inducing a variety of cellular responses, including the regulation of epithelial37 and hematopoietic38,39 precursor cell differentiation. LCs express the intracellular E-cadherin–binding signaling molecule armadilloβ-catenin,40 which is capable of translocating to the nucleus where it is involved, together with other transcription factors of the lymphocyte enhancer-binding factor 1/T-cell factor family, in the expression of specific genes.41,42 Cadherins such as E-cadherin seem to represent negative regulators of catenin signaling.43,44 An important suppressive role of E-cadherin in cell signaling is also suggested by the observation that E-cadherin expression is down-regulated in many carcinomas, an effect correlating with tumor cell metastasis. In turn, tumor metastasis is suppressed by the overexpression of E-cadherin.45,46 Interestingly, the down-regulation of E-cadherin cell surface expression in LC histiocytosis, a tumor originating from epidermal Langerhans cells in vivo, similarly correlates with the occurrence of distant metastasis.47 How E-cadherin activation on the surface of immature LCs may inhibit the transactivation of genes involved in DC maturation and migration remains to be analyzed.

We demonstrate that mAb binding to E-cadherin on the surfaces of immature LCs inhibits the acquisition of mature DC characteristics. This is in line with the concept that lateral clustering of cadherin molecules is critical for cadherin-mediated functional responses, such as tight cell adhesion.48,49 E-cadherin binding by mAb may induce cross-linkage of E-cadherin dimers on the surfaces of LCs, which may mimic E-cadherin clustering in response to the homophilic ligation of E-cadherin. Recent observations with cell lines support this concept. First, E-cadherin adhesion leads to a rapid increase in tyrosine phosphorylation at sites of cell contact formation, and this effect can be mimicked by anti–E-cadherin mAb.50 Second, anti–N-cadherin mAb-coated beads functionally mimic beads coated with the extracellular domain of N-cadherin in inducing tyrosine phosphorylation, accumulating junction-associated β-catenin molecules, and inducing G1 arrest of N-cadherin overexpressing Chinese hamster ovary cells.51,52 On the other hand, E-cadherin redistribution on the cell surface is well described as regulated by inside-out signals—ie, by protein binding to its juxtamembrane cytoplasmic tail53 or by binding of armadilloβ-catenin to the cytoplasmic cadherin tail.46 Thus, E-cadherin activation or deactivation in response to E-cadherin redistribution on LC surfaces is regulated by intracellular signaling pathways.

Our data suggest that the loss of E-cadherin adhesion rapidly induces the maturation of LC. Despite rapid secondary E-cadherin–mediated cluster formation after the disruption of primary LC clusters, most LCs have already undergone maturation (within 24 hours). This indicates that DC maturation in response to loss of E-cadherin adhesion may represent a rapid and irreversible process. These observations may mimic the in vivo situation. In vivo, the down-regulation of E-cadherin expression by LCs seems to precede the emigration of LCs from the epidermis.8 

We performed a side-by-side comparison of the effects of soluble CD40L and TNFα on the acquisition of mature DC features by dispersed immature LCs. We observed that both stimuli rapidly induce the acquisition of CD83 expression and the up-regulation of CD86 expression by LCs. However, LCs stimulated in the presence of CD40L showed consistently higher CD86 and CD83 expression densities than those stimulated in the presence of TNFα. In line with this they showed more mature morphologic characteristics. Therefore, our defined serum-free culture model seems to recapitulate 3 consecutive maturation stages of LCs in vivo: immature LCs undergoing E-cadherin adhesion (primary culture, Figure 3A bottom) and resembling immature/nonactivated LCs in epidermis; intermediate-stage LCs that have lost E-cadherin adhesion, are released from a possible active suppressive effect of E-cadherin (Figure 3A, top), and resemble migrating LCs; and LCs stimulated by CD40L (Figure 3A, second panel) that correspond to DC and are activated by antigen-specific T-helper cells by CD40 engagement in secondary lymphoid organs in vivo.54-56 These experiments also demonstrate that the maturation of LCs does not require serum/plasma supplementation of the culture medium, thus confirming recent observations in the murine system.57 

As mentioned, our data suggest that E-cadherin–mediated suppression of LC maturation may prevent the uncontrolled maturation of LCs induced by pro-inflammatory mediators. This seems to be in conflict with murine studies, which demonstrated that the same pro-inflammatory mediators that induce LC maturation simultaneously down-regulate E-cadherin expression and E-cadherin–mediated adhesion and, thus, seem to reduce the proposed maturation inhibition by E-cadherin.58Although several parameters vary between this study and ours, it remains to be analyzed whether exogenous TGF-β1 counteracts E-cadherin down-regulation on murine LC induced by pro-inflammatory cytokines.

Although most cells generated in primary TGF-β1– supplemented cultures show LC features (CD1a⁺, E-cadherin⁺, Lag⁺), these cultures also contain approximately 40% of cells that lack expression of CD1a. Interestingly, most of these CD1a⁻ cells share phenotypic features with myelomonocytic precursor cells (CD33⁺, cytoplasmic CD68⁺, CD34⁻), with fractions of them also expressing CD14, lysozyme, or myeloperoxidase.16 We observed that purified fractions of these CD1a⁻ cells can be induced in response to stimulation with TGF-β1 plus GM-CSF, SCF, FL, and TNFα (ie, the initial cytokine combination) to differentiate into CD1a⁺cells (data not shown). Future studies should analyze further the molecular and functional characteristics of these cells.

In conclusion, active maturation suppression of LCs by TGF-β1–induced E-cadherin adhesion, as suggested by our study, may represent an important epithelium-specific mechanism by which LCs retain their immature features over long time periods and resist uncontrolled activation by pro-inflammatory cytokines.

We thank A. Renner for his invaluable contribution in cell separation and flow sorting, M. Merad and R. Smith for critically reading the manuscript, and all the collaborating nurses and doctors of the gynecology departments at Sozialmedizinisches Zentrum Ost and Kaiser Franz Josef Spital for providing cord blood samples.