Characterization of chicken epidermal dendritic cells

Abstract

It has been known for 15 years that the chicken epidermis contains ATPase⁺ and major histocompatibility complex class II-positive (MHCII⁺) dendritic cells. These cells were designated as Langerhans cells but neither their detailed phenotype nor their function was further investigated. In the present paper we demonstrate a complete overlapping of ATPase, CD45 and vimentin staining in all dendritic cells of the chicken epidermis. The CD45⁺ ATPase⁺ vimentin⁺ dendritic cells could be divided into three subpopulations: an MHCII⁺ CD3^– KUL01⁺ and 68.1⁺ (monocyte-macrophage subpopulation markers) subpopulation, an MHCII^– CD3^– KUL01^– and 68.1^– subpopulation and an MHCII^– CD3⁺ KUL01^– and 68.1^– subpopulation. The first population could be designated as chicken Langerhans cells. The last population represents CD4^– CD8^– T-cell receptor-αβ^– and -γδ^– natural killer cells with cytoplasmic CD3 positivity. The epidermal dendritic cells have a low proliferation rate as assessed by bromodeoxyuridine incorporation. Both in vivo and in vitro experiments showed that dendritic cells could be mobilized from the epidermis. Hapten treatment of epidermis resulted in the decrease of the frequency of epidermal dendritic cells and hapten-loaded dendritic cells appeared in the dermis or in in vitro culture of isolated epidermis. Hapten-positive cells were also found in the so-called dermal lymphoid nodules. We suggest that these dermal nodules are responsible for some regional immunological functions similar to the mammalian lymph nodes.

Introduction

Langerhans cells (LCs) are the most important antigen-presenting and accessory cell population of the mammalian skin immune system.¹ Deriving from bone marrow, precursors of LCs migrate via the blood circulation to the suprabasal area of the epidermis, where they represent 2–3% of all epidermal cells.² LCs are members of the dendritic cell (DC) family and are well equipped to ingest foreign antigens that reach the skin. After antigen uptake, LCs leave the skin and migrate to the regional lymph nodes to initiate a systemic immune response by presenting the processed antigens to T cells. During migration, LCs increase the expression of major histocompatibility complex class II (MHCII) and costimulatory molecules, which are essential for effective antigen presentation.^3,4

Few studies have been carried out to demonstrate the existence of similar cells in the skin of non-mammalian vertebrates. Carrilo-Farga et al. have found ATPase-positive DCs in the epidermis of bullfrog and chicken using whole-mount epidermis with electron microscopical histochemistry.^5,6 These cells were named as Langerhans-like cells. They also found that these cells are MHCII-positive^7,8 and contain Bierbeck granules.⁹ ATPase⁺ Langerhans-like cells were also reported at the mucosal surfaces of the chicken^10,11 but no detailed studies on the origin, phenotype and function of chicken LCs have been reported until now. Because birds do not posses lymph nodes, the existence of LCs in the epidermis seemed to be theoretical and these data remained isolated because without linking these cells to a function or a target organ, their existence was merely a curiosity.

The aim of the present study was to provide a detailed morphological characterization of chicken epidermal dendritic cells (EDCs) and to investigate whether they could take part in antigenic uptake and transport from the epidermis and to find their destination. We also present data about their ontogenetic appearance in the epidermis. Using a large panel of monoclonal antibodies (mAbs), we consequently identified subpopulations of EDCs. EDCs could emigrate from cultured epidermal sheets, and showed typical features of DCs in vitro. Hapten-containing DCs appear in the culture of epidermal sheets after cutaneous fluorescein isothiocyanate (FITC) administration. In in vivo experiments, haptens reduced the density of EDCs, and hapten-positive DCs appeared in the dermis, but not, for example, in the spleen. We propose a novel destination for the emigration of the EDCs, the so-called lymphatic nodules of the dermis, which are located in the proximity of the dermal veins and lymphatics.

Materials and methods

Animals

Fertilized specific pathogen-free White Leghorn eggs were obtained from BiOvo, Mohács, Hungary. Conventional New Hampshire eggs and birds were purchased from KÁTKI, Gödöllõ, Hungary. The eggs were incubated at 37·7° in a humidified incubator. Tissue samples were taken from 18-day embryos to 8-week-old chickens. At least three birds were used per group. All bird experiments were performed under the control of the Animal Ethical Committee of Semmelweis University, Budapest (Leg. No. 779-000-2005).

Preparation of epidermal sheets

The skin was isolated from a featherless (usually the axillary) region and cut with a surgical blade into small pieces (about 0·5 cm²). Epidermis was isolated from the underlying tissues (1) using the CaCl₂ method (for immunohistochemistry); (2) mechanically (for immunohistochemistry and emigration studies); or (3) using an enzymatic treatment (for immunohistochemistry and EDC isolation). The three methods gave similar results for enzyme histochemistry or immunofluorescence of EDCs but a slightly higher value using the mechanical method. For CaCl₂ isolation, skin was incubated for 10–20 min in 2 m CaCl₂ solution. The epidermis was separated from the dermis with forceps under a dissecting microscope. After washing in Tris-buffer (0·1 m, pH 7·2) the epidermal sheets were fixed in 4% buffered formalin at 4° for 1 hr. During mechanical isolation, epidermis was simply pulled down mechanically from the dermis using Dumont no. 5 forceps in situ, without any previous treatment. For enzymatic isolation we used dispase treatment (see Enrichment and culture of EDC).

Identification of EDCs with ATPase histochemistry

The ATPase staining was performed as described by Robins and Brandon.¹² Briefly, the formalin fixed epidermal sheets or isolated cells were washed in 0·1 m Tris buffer (pH 7·2) and incubated in ATPase medium (0·1 m Tris-maleate buffer pH 7·2, 10 mm MgSO₄, 6 mm Pb(NO₃)₂, 28 mm glucose and 1·6 mm ATP-disodium salt, (Sigma-Aldrich Kft, Budapest, Hungary) for 1 hr at 37°. The reaction product was developed in 1% (NH₄)₂S for 1 min at room temperature and rinsed in distilled water.

Immunohistochemistry

Immunohistochemical staining was performed on whole epidermal sheets, on 10-μm thick frozen sections, and on isolated (after cytospin centrifugation) or emigrated EDCs. Briefly, tissue samples were washed in phosphate-buffered saline (PBS) and incubated with primary antibodies for 45 min. After addition of biotinylated secondary antibodies (Vector Laboratories, Burlingame, CA) endogenous peroxidase activity was quenched using 3% hydrogen peroxide in PBS for 10 min. After addition of avidin-biotinylated peroxidase complex (Vectastain Elite kit, Vector Laboratories) the binding sites of the primary antibodies were visualized using 4-chloro-1-naphthol (Sigma).

For single immunofluorescent staining we used streptavidin conjugated Alexa-488 or Alexa-594 (Molecular Probes, Leiden, the Netherlands). For double immunofluorescent staining, isotype-specific polyclonal antibodies biotinylated anti-mouse immunoglobulin G1 (IgG1) or IgG2a (from Southern Biotechnology Associates, Inc., Birmingham, AL) anti-mouse IgG2b (from Sigma-Aldrich), biotinylated anti-mouse IgM or anti-rabbit IgG (from Vector Laboratories) were used. Between the two stainings, avidin and biotin blocking was performed (Avidin-Biotin Blocking Solutions, Vector Laboratories). Nuclear staining was performed with 4′6-diamidino-2-phenylindole−2HCl (DAPI; Merck Kft, Budapest, Hungary). Double-stained specimens were studied by fluorescence microscopy (Axiophot, Zeiss) or confocal microscopy (Bio-Rad Radiance 2100 Rainbow LCM). For the CD45, CD3, vimentin, cytokeratin, MHCII, integrin α, Bu-1b, GRL-1, GRL-2, LEP 100 IgG (see Table 1) and caveolin-1 staining, the three epidermis isolation methods (tested on spleen) gave similar staining intensities. More than half of the epitopes of the listed antibodies did not resist (partial or full inactivation) the CaCl₂ method for epidermal isolation (E-cadherin, CD4, CD8, T-cell receptor 1–3 (TCR1–3), integrin β₁, CVI 68.1, 68.2, 74.2, 74.3, K1, KUL01). For these mAbs we used mechanically or enzymatically prepared epidermis.

Table 1.

Primary antibodies used in the study

Recognized cells	Antibody specificity	Antibodies	Source of antibody
Epithelial cell	pan cytokeratin (cross-reactive with chicken)	clone Lu-5 (mouse IgG1)	BMA, Biomedicals AG Augst, Switzerland
Chicken epithelial cells	chicken E-cadherin	7D6 (mouse IgG1)	DSHB, Iowa, IA
Haematopoietic cell	chicken CD45 antigen	clone HIS-C7 (mouse IgG2a)	gift from Dr H. KurzAlbert-Ludwigs University Freiburg, Germany
pan T cell	CD3 antigen (cross-reactive with chicken)	polyclonal anti-human pan T cell(rabbit IgG)	Dako A/S, Glostrup, Denmark
Th cell	chicken CD4	CT-4 (mouse IgG1)	Southern Biotechnology Associates, Birmingham, AL
Tc cell	chicken CD8	CT-8 (mouse IgG1)	Southern Biotechnology
T cell αβ	chicken (Vβ1) TCR-αβ	TCR-2 (mouse IgG1)	gift from Dr Thomas Göbel,Ludwig Maximilian University, Munich, Germany
T cell αβ	chicken (Vβ2) TCR-αβ	TCR-3 (mouse IgG1)	gift from Dr Thomas Göbel, Ludwig Maximilian University,Munich
T cell γδ	chicken TCR-γδ	TCR-1-FITC (mouse IgG1)	Southern Biotechnology
B-cell and macrophage subpopulation	Bu-1b antigen	clone 5-11G2 (mouse IgG1)	Southern Biotechnology
Different cell types	vimentin intermediate filament	VIM 3B4 (mouse IgG2a)	Progen, Biotechnik GmbH, Heidelberg
Different cell types	vimentin intermediate filament	AMF17b (mouse IgG1)	DSHB, Iowa, IA
	chicken IgG	clone CG-106 (mouse IgG1)	Sigma-Aldrich Kft, Hungary
	chicken IgM	clone M-1 (mouse IgG2b)	Southern Biotechnology
Follicular dendritic cells	Unknown	CVI-ChNL-74.3 (mouse IgG1)	CEDI-Diagnostics, Lelystad, the Netherlands
Chicken reticularellipsoidal cells	45 kDa protein	CVI-ChNL-68.2 (mouse IgG1)	CEDI-Diagnostics
Chicken macrophages	Unknown	CVI-ChNL-74.2 (mouse IgG1)	CEDI-Diagnostics
Chicken pan-mononuclear phagocyte	Unknown	CVI-ChNL-68.1 (mouse IgG1)	CEDI-Diagnostics
Chicken mononuclear phagocyte	Unknown	KUL01(mouse IgG1)	Gift from Dr B.M. Goddeeris, KU, Leuven, Belgium
chicken monocyte/macrophage/ thrombocyte	145 kDa protein	K1 (mouse IgG1)	Gift from H. Lillehoj, FDA Washington, DC
B-cell, activated T-cell and dendritic cells	MHC II	clone 2D5 (mouse IgG2b)	DSHB, Iowa, IA
Chicken granulocytes	Unknown	GRL1 (mouse IgG3)	DSHB, Iowa, IA
Chicken granulocytes	Unknown	GRL2 (mouse IgG1)	DSHB, Iowa, IA
Lysosomal membrane	LEP 100	LEP 100 IgG (mouse IgG1)	DSHB, Iowa, IA
Different cell types	chicken integrin α5 subunit	D71E2 (mouse IgG)	DSHB, Iowa, IA
Different cell types	chicken integrin α6 subunit	P2C62C4 (mouse IgG1)	DSHB, Iowa, IA
Different cell types	chicken integrin α7 subunit	H1B4 (mouse, unknown isotype)	DSHB, Iowa, IA
Different cell types	chicken integrin β1 subunit	CSAT (mouse IgG2b)	DSHB, Iowa, IA
Anti-caveolin-1	Caveolin-1	Polyclonal rabbit IgG	Gift from Dr A. L. Kiss, SU, Budapest, Hungary
Chicken skin melanocytes	unknown	MelEM (mouse IgG1)	DSHB, Iowa, IA

The primary antibodies used in this work are listed in Table 1.

Electron microscopy

The tissue samples or isolated epidermal cells were fixed in 0·1% buffered glutaraldehyde solution overnight at 4°. After washing the samples were stained by CD45 immunohistochemistry as described above, but using longer incubation periods (usually 2–4 hr at room temperature). The binding sites of the primary antibodies were visualized with 3,3′-diaminobenzidine (Sigma). After immunohistochemistry, the samples were post-fixed in 1% osmium tetroxide (Polysciences Inc., Warrington, PA) for 2 hr. After dehydration in graded ethanol the tissue blocks were embedded in Polybed/Araldite 6500 (Polysciences Inc.) mixture. For transmission electron microscopy the ultrathin sections were contrasted with uranyl acetate and lead citrate and studied with a Hitachi electron microscope type H-7600.

Turnover of EDCs

The chickens (three per group) were given a single intravenous injection of 5-bromo-2-deoxyuridine (BrdU; Roche Magyarország, Budaörs, Hungary) at 100 mg/kg body weight (in vivo labelling). At different intervals (30 min, 2 hr and 24 hr) skin biopsies were taken from the same animal for epidermal sheet preparation (CaCl₂ method). In other cases (in situ labelling) the epidermis of untreated animals was wiped off with 70% ethanol, removed mechanically and cultured in medium containing BrdU (10 μm in complete medium; see below) for 30 min or 2 hr.

After BrdU labelling, the epidermal sheets were fixed in buffered formalin, washed in PBS and incubated in 2 m HCl solution for 30 min at 37° for DNA denaturation. After neutralization, double immunofluorescent staining was performed on the sheets for detection of incorporated BrdU, using mAb against BrdU (Roche) and for EDC markers (see above).

Enrichment and culture of EDCs

Isolated skin was cut in approximately 0·2 × 2 cm strips and these were incubated in complete RPMI medium (10% fetal bovine serum, 2 mm l-glutamine, minimal essential medium non-essential amino acids and vitamins, 100 IU/ml penicillin, 100 μg/ml streptomycin, 10 mm HEPES, 1 mm sodium pyruvate, 0·05 mm 2-mercaptoethanol; Sigma) containing 2 mg/ml dispase (grade II, Roche) for 1 hr at 37°. Cell clumping was inhibited using DNase I (0·2 mg/ml, Roche). After incubation, the epidermis was lifted from the dermis, washed in sterile PBS, and incubated in 0·25% porcine trypsin (in PBS-ethylenediaminetetraacetic acid; Sigma) for an additional 30 min at 37°. The separated cells were washed, resuspended in complete medium and layered on Histopaque 1077 (Sigma). Centrifugation took place at 1700 g for 15 min. The cells collected from the interface were washed twice in complete RPMI and used for further studies (immunohistochemistry or in vitro culture).

DC emigration was also studied in vitro. The untreated epidermis (3 cm²) was disinfected with 70% ethanol, mechanically isolated and then cultured for 24–72 hr in complete RPMI (37°, 5% CO₂). After in vitro culture, the epithelial sheet was removed and the emigrated cells were studied.

In vivo and in vitro activation of EDCs

Solutions of different haptens [1% FITC (Sigma) or chloro-methylbenzamido-dioctadecyl-tetramethyl indocarbocyanine (cell tracke CM-DiI; Molecular Probes)] dissolved in acetone : dibutyl-phthalate (1 : 1 v/v) were painted on feather-free skin of 8-week-old chickens. After 1, 2, 4, 12 or 24 hr or 1 week (three animals per group), skin biopsies were taken from treated and untreated regions. The epidermis was separated from the dermis using the CaCl₂ method as described above and was processed for immunohistochemistry. The results were analysed statistically with paired-samples t-test, using the SPSS 13·0 program.

In the case of in vitro studies, skin was sensitized with haptens (50 μl of 1% sterile FITC solution, for 30 min), and the treated epidermis (3 cm²) was cleared by 70% ethanol, mechanically isolated and cultured as described above. After 24–72 hr of culture, the epithelial sheet was removed from the culture dish and the emigrated adherent cells were studied.

Results

Phenotype and subpopulations of chicken EDCs

EDCs were identified by ATPase enzyme histochemistry. In young adults (8-week-old chickens) there were about 2000 ATPase⁺ cells/mm² epidermis. We found a similar number of CD45⁺(Fig. 1d) and vimentin⁺ cells of dendritic appearance. When ATPase histochemistry was combined with immunofluorescence, there was a complete overlapping for ATPase and CD45 positivity (Fig. 1a) or ATPase and vimentin positivity among the EDCs (Fig. 1b). Similarly, all CD45⁺ cells contained vimentin in their cytoplasm (Fig. 1c). We then used anti-CD45 or anti-vimentin antibodies as one of the primary antibodies in double immunofluorescent staining to demonstrate the presence or absence of the antigens recognized by the listed antibodies on/in the EDCs. The staining pattern using different primary antibodies is summarized in Table 2.

Figure 1.

(a–k) The phenotype of EDCs in the epidermis of 8-week-old chickens. (a) The ATPase⁺ EDCs share on their membrane CD45 molecules; ATPase histochemistry (brown) and CD45 immunofluorescence (red), bar 10 μm. (b) EDCs with ecto-ATPase activity (brown) also contain vimentin intermediate filament (red), bar 10 μm. (c) The two novel EDC markers completely overlap with each other: CD45 (green) and vimentin (red); nuclear staining with DAPI, bar 10 μm. (d) CD45 immunohistochemical staining showed the unequal distribution of EDCs in the epidermis, bar 30 μm. (e) MHCII staining shows lower cell density, bar 30 μm, inset 1 and 2 – around 40% of EDCs are positive for MHCII. Confocal microscopy for MHCII (green) and vimentin (red). Nuclear staining was performed with DAPI, bars 10 μm and 20 μm, respectively. (f) The monocyte/macrophage-specific marker KUL01 showed fewer positive cells compared to (d), bar 30 μm. (g) The chicken EDCs (CD45: green) are almost completely negative for E-cadherin (red). Signal co-localization (yellow spots) could be observed occasionally (arrowheads), bar 10 μm. Inset – CD45 (green) and E-cadherin (red) were stained on enzymatically isolated epidermal cells. Keratinocytes retained their E-cadherin positivity after dispase-trypsine treatment, bar 10 μm. (h) Confocal microscopy with CD45 (green) and CD3 (red). The CD3 is localized in the cytoplasm, but not on the cell surface, bar 10 μm. (i) The CD45⁺ (green) DCs are located between the cytokeratin⁺ (red) keratinocytes, without any overlapping, bar 10 μm. (j) The isolated keratinocytes but not the EDCs (CD45: green) express caveolin-1 (red), bar 10 μm. (k) The CD45⁺ (green) EDCs contain intracytoplasmic Lep100 (red) protein (yellow spots), bar 20 μm.

Table 2.

Epidermal dendritic cell subpopulations

Antigen expression	Type I	Type II	Type III
ATPase	+	+	+
CD45	+	+	+
Vimentin	+	+	+
MHC II	+	–	–
mAb KUL01	+	–	–
mAb 68.1	+	–	–
E-cadherin	–/+	–/+	–/+
Integrin β1 subunit	+	+	+
Integrin α5, α6, α7 subunits	–	–	–
Anti-Caveolin 1	–	–	–
Lep100	–/+	–/+	–/+
CD3	–	–	+
CD4	–	–	–
CD8	–	–	–
TCR-γδ	–	–	–
TCR-αβ	–	–	–
Bu-1b	–	–	–
IgG and IgM	–	–	–
Follicular dendriticcell markermAb 74.3	–	–	–
Monocyte/macrophage subpopulation markers (mAbs 74.2, K1)	–	–	–
mAb 68.2	–	–	–
Granulocyte markers (mAbs GRL1 and GRL2)	–	–	–
Relative occurrence among all EDCs	40–50%	35–45%	5–10%

No double-positive cell were identified using anti-CD45 and pan-cytokeratin antibodies (Fig. 1i). Fewer than half of the chicken EDCs (on average 40%) were positive for MHCII (Fig. 1e, insets). About 50% of all EDCs were positive for the monocyte/macrophage subpopulation-specific marker mAbs KUL01 and 68.1 (Fig. 1f). E-cadherin expression of the EDCs (among the strongly positive keratinocytes) was studied by confocal microscopy. CD45⁺ EDCs were mostly negative for E-cadherin, but occasional spots (signal co-localizations) could be observed on dendritic processes (Fig. 1g). Such positive spots were not present on enzymatically isolated CD45⁺ EDCs but were found on keratinocytes (Fig. 1g, inset). EDCs express integrin β₁ at a medium level but not the other integrins (data not shown). Caveolin-1, an important component of caveosomes that plays a role in endocytosis, is not present on EDCs, in contrast to the strongly expressing keratinocytes (Fig. 1j). EDCs express lysosomal protein Lep100, but less than the surrounding keratinocytes (Fig. 1k).

We found a minor (5–10%) population of ATPase⁺ CD45⁺ DCs that were CD3⁺. The CD3 positivity appeared in the cytoplasm, but not on the cell surface (Fig. 1h). These cells were negative for MHCII, CD4, CD8, TCR-α/β and TCR-γ/δ.

We failed to demonstrate the presence of immunoglobulin isotypes (IgM or IgG), pan B cell (Bu-1b), follicular dendritic cell-specific (mAb 74.3), other monocyte-macrophage-subpopulation-specific (mAbs 74.2, K1), splenic reticulum cell-specific (mAb 68.2) or granulocyte-specific (GRL1 and GRL2) markers on the EDCs.

Finally we could establish three subpopulations of the ATPase⁺ CD45⁺ vimentin⁺ DCs in the epidermis (Table 2): (1) MHCII⁺ KUL01⁺ 68.1⁺ CD3^–, (2) MHCII^– KUL01^– 68.1^– CD3^– and (3) MHCII^– KUL01^– 68.1^– CD3⁺ (Table 2).

The immunoelectron microscopy of isolated epidermis showed CD45-positive cells with dendritic morphology among the keratinocytes (Fig. 2a). These cells have organelle-rich perinuclear cytoplasm and organelle-poor dendritic processes. It was also possible to find cytoplasmic structures similar to the Bierbeck granules of mammalian LCs (Fig. 2b, inset).

Figure 2.

(a,b) CD45 immune electron microscopy of EDCs. (a) An EDC in the germinal layer of epidermis isolated by the CaCl₂ method (8-week-old chicken). The cell is outlined by a broken line. The arrows show the CD45⁺ positivity on dendritic processes, bar 0·5 μm. (b) EDCs were isolated from epidermis by dispase-trypsin treatment. Rectangle shows a cytoplasmic structure resembling a Bierbeck granule, bar 0·5 μm. Inset shows higher magnification of Bierbeck granule; bar 0·1 μm. (c,d) Isolation of epidermal EDCs by dispase-trypsin treatment. (c) Cytospin preparation of crude epidermal cell suspension was stained for CD45. About 5–8% of total cells are positive for CD45, bar 30 μm. (d) After density gradient centrifugation (on Ficoll–Hypaque) of crude epidermal cell suspension, the DC population can be enriched up to 50%, bar 30 μm. (e,f) DCs emigrated from untreated epidermal sheets during 24 hr of in vitro culture. Cells stained for (e) CD45 and (f) MHCII, bar 8 μm.

Isolation and culture of EDCs

The crude epidermal cell suspensions after dispase/trypsin treatment usually contained 5–8% CD45⁺ ATPase⁺ vimentin⁺ cells (Fig. 2c). The enrichment of DCs could be achieved by centrifugation on Ficoll–Hypaque, where the interface suspension comprised 50% EDCs (Fig. 2d). Pure EDC populations could be obtained from chicken epidermis by culturing the mechanically isolated epidermal sheets for 24–72 hr in complete medium. After a 72-hr incubation, about 500–600 cells with typical dendritic appearance emigrated from 1 mm² epidermis (Fig. 2e,f) and adhered to the plastic surface. The epidermal sheet removed from the culture at this time contained 800–1000 EDCs/mm². All of these emigrated dendritic cells were CD45⁺ ATPase⁺ vimentin⁺ E-cadherin^– integrin β₁⁺. About 50% of them were MHCII⁺ (Fig. 2f).

Ontogenetic appearance of EDCs

Emergence of the EDCs in the skin was followed in two chicken strains: New Hampshire and White Leghorn. We analysed the epidermis from 18-day-old embryos (ED18) to 8-week-old chickens (Fig. 3). The epidermis of New Hampshire chickens contained 2–10 EDCs/mm² before hatching (at ED18, Fig. 3a). The epidermis of the 10- to 14-day-old New Hampshire birds already showed the EDC density typical for the adults (around 2000 cells/mm², Fig. 3b). On the other hand, the first EDCs appeared in the epidermis of White Leghorn chickens around 10–14 days after hatching (Fig. 3e). The White Leghorn chickens showed great fluctuations according to the number of EDCs. We have also found 5-week-old White Leghorn chickens with an EDC density of between 20 and 30 cells/mm². The epithelial EDC number of the New Hampshire strain was more stable, and so this strain was used exclusively in all experiments. In the embryos and young animals, only ATPase histochemistry and CD45 and vimentin immunohistochemistry were performed (Fig. 3a–f). All EDCs were ATPase⁺ CD45⁺ vimentin⁺.

Figure 3.

(a–f) Time–course of EDCs appearance in two chicken strains followed by CD45 immunohistochemistry, bar 30 μm (a) In New Hampshire chickens a few cells/mm² can be detected before hatching (18th embryonic day, 18 ED). (b) The epidermis of the 14-day-old New Hampshire birds already showed the EDC density typical of adults. (c,f) In the 8-week-old birds of both strains the number of EDCs stabilized around 2000 cells/mm². (d) At 18 ED no EDCs were present in White Leghorn embryonic epidermis. (e) The first EDCs appear in the epidermis of White Leghorn chickens 14 days after hatching. (g) Eight-week-old chicken was treated with BrdU intravenously. Epidermis was isolated with the CaCl₂ method at 2 h after administration and stained with CD45 (green) and BrdU (red). Arrowheads point to proliferating EDCS, bar 30 μm. Inset shows a now divided EDC, bar 10 μm. (h) A cryostat section of skin from 8-week-old chicken. The dermal lymphoid nodules, located beneath the epidermis (arrowhead), were double-stained for B cells (mAb Bu-1; grey-blue) and T cells (CD3, red), bar 50 μm. (i) After skin sensitization with FITC, the FITC positivity appears in the cytoplasm of KUL01⁺ cells (red; arrowhead) and (j) in the central part of the nodule (mAb 74.3; red), bar 20 μm. (k) In vitro emigration of EDCs from sensitized skin of 8-week-old chicken. FITC was painted onto axillary skin and after 30 min epidermis was removed and cultured in vitro. After 24 hr of culture the adherent cells were collected (cytospin centrifugation) and stained for CD45 (red), bar 10 μm.

In vivo studies using BrdU revealed that 4·9% of all epidermal CD45⁺ cells were labelled after 30 min of intravenous administration. This proportion did not change significantly 2 or 24 h after BrdU administration (Table 3, Fig. 3g). To clarify whether extra-epidermal precursors or the epidermal EDCs thermselves were mitotically active, in situ epidermal labelling (isolated sterile epidermal sheets were labelled in culture dish) was performed. This gave similar labelling rates (compared to in vivo results) after 30 min or 2 hr (Table 3).

Table 3.

BrdU labelling of epidermal dendritic cells

Time after BrdU administration	In vivo BrdU labelling	In situ BrdU labelling
30 min	4·86%	4·42%
2 hr	5·38%	5·12%
24 hr	3·99%	Not done

Functional study on epidermal dendritic cells

EDC dynamics after antigen loading was studied both in vivo and in vitro. In in vivo experiments, skin of the apteric region was painted with different haptens (FITC or DiI) dissolved in acetone/dibutylphthalate and the number of EDCs was determined at different time intervals by CD45 immunohistochemistry. After 1 hr, many EDCs became rounded and their number started to diminish. The cellular density of EDCs reached a minimum value at 16–24 hr after antigen loading: it was reduced by 60–70% (to 600–800 EDCs/mm²) compared to untreated skin regions of the same bird. The solvent itself did not result in any significant change of EDC density (Fig. 4a). The phenotype of the remaining EDCs however, did not change: they are uniformly CD45⁺ ATPase⁺ vimentin⁺ and we could not observe any shift in the frequency of MHCII, CD3, LEP-100, caveolin-1, KUL01 or 68.1 positivity. The cellular density of EDC was normalized about 1 week after cutaneous antigen administration.

Figure 4.

(a) Changes in EDC number after skin sensitization. The chart represents data from three different experiments: 24 hr after hapten stimuli. (b) Schematic presentation of EDC migration. In mammals LCs migrate to the nearest peripheral lymph node after antigen uptake. In birds the emigration route of EDCs is not obvious because they lack peripheral lymph nodes. On the basis of our experimental results the terminals of EDCs migration should be the lymphoid nodules in the dermis. We could not find labelled EDCs in spleen, but this does not rule out the possibility that some EDCs could reach the spleen. Other possible terminal(s) could exist (question mark). BM, basement membrane; GL, germinal layer; SP, spinous cell layer; HL, horny layer; LC, Langerhans cell.

The emigration of EDCs from the epidermis was also demonstrated by in vitro experiments. After 30 min of cutaneous administration of FITC in acetone/dibutylphthalate, the treated epidermal sheet was prepared mechanically and cultured in vitro. After 24 hr of culture, the epidermis was removed from the culture dish and the adherent cells were stained. All the cells were CD45⁺ ATPase⁺ vimentin⁺ MHCII⁺ and had a typical dendritic appearance. Ninety per cent of them contained FITC in the cytoplasm (Fig. 3k). We never found CD3⁺ FITC⁺ emigrants in these cultures. The in vitro emigration rate of treated skins was significantly accelerated (1000–1200 EDCs emigrated from 1 mm² of treated epidermis during 24 hr of culture) compared to that of untreated skin (500–600 emigrated cells from 1 mm² of epidermis during 72 hr of culture; see above).

To see which organ could be the target of the emigration of EDCs, we analysed different lymphoid organs such as spleen, thymus, caecal tonsil and bone marrow after in vivo cutaneous FITC administration. During the time interval (30 min to 24 hr) after cutaneous antigen administration we found no FITC⁺ cells – either in lymphoid organs or in the blood cells. This finding raised the possibility that the emigrated DCs do not leave the skin. Investigation of the treated skin regions with serial sections revealed the presence of FITC⁺ cells scattered throughout the dermis as well as accumulated inside the so-called dermal lymphoid nodules. These structures are mini cell clumps of T and B lymphocytes (containing several hundred cells) often located in close proximity to dermal veins (Fig. 3h). At the periphery of the nodules, KUL01⁺ cells accumulate (Fig. 3i). The centre of the nodule contains a well-developed reticulum of 74.3⁺ cells (Fig. 3j). The FITC signal appeared intracytoplasmic in the KUL01⁺ cells, as well as in the 74.3⁺ cells (Figs 3i,j).

Discussion

Using combined histochemistry (enzyme histochemistry with immunohistochemistry or double immunofluorescence) we showed that all ATPase⁺ EDCs also share CD45 and vimentin, similar to mammalian LCs,^13,14 while the keratinocytes are negative for these markers. These cells represent the majority of DCs in the epidermis. Other cells with dendritic morphology, such as melanocytes, occur in the pigmented New Hampshire strain at a very low density (2–3 melanocytes/mm² measured by mAb MelEM; data not shown). The melanocytes are CD45^– and are therefore excluded from our further investigations. CD45 expression of chicken EDCs also supports their haematopoietic (bone marrow) origin. Double immunofluorescence also revealed at least three different DC subpopulations in the chicken epidermis. These could be distinguished simply on the basis of their MHCII and CD3 expression: type I EDCs are MHCII⁺ CD3^–, type II EDCs are MHCII^– CD3^– and type III EDCs are MHCII^– CD3⁺.

Our results on MHCII expression contradict the data from Perez-Torres et al.⁸ who found that all ATPase⁺ DCs express MHCII. This could be a difference caused by methodology. Perez-Torres et al. did not apply simultaneous double labelling, but performed the second labelling after photographing the first. We used double-labelling, and could clearly show that only 35–50% (on average 40%) of all epidermal CD45⁺ DCs possess MHCII. These type I epidermal cells could represent the avian analogue of mammalian LCs – capable for antigen uptake and presentation. This suggestion is supported by the fact, that during in vitro culture of FITC-treated epidermis, nearly all FITC⁺ emigrated cells were MHCII⁺. The MHCII⁺ EDC population was also positive for two monocyte markers (recognized by mAbs KUL01 and 68.1) suggesting that chicken LCs (chicken type I EDCs) may belong to the myeloid lineage. The distribution of MHCII⁺ EDCs is not uniform; a significant fraction of EDCs often forming clusters stained more brightly for MHCII (Fig. 1e).

The type II MHCII^– DCs represent about one-third of all CD45⁺ dendritic cells. Melanocytes should be precluded, because they are relatively infrequent in the avian epidermis and they never express CD45. There are currently no data on their origin or function. We suggest that they could be the resting form of type I DCs or their precursors. Interestingly, many MHCII^– cells emigrate from the unstimulated epidermis. In contrast, FITC appears mainly in MHCII⁺ epidermal emigrant EDCs. In humans, MHCII expression is up-regulated by LC activation.^13,15

The phenotype of type III (CD3⁺) EDCs completely meets the published phenotype of the avian natural killer (NK) cells: both are TCR-α/β^–, TCR-γ/δ^–, CD4^–, but are occasionally CD8⁺(α), MHCII^– cells with cytoplasmic CD3 positivity.¹⁶ In contrast to mammals, for which very few data were published about sporadic cells having NK cell markers in normal epidermis (e.g. for humans¹⁷), we report the relatively frequent occurrence of NK cells in the chicken epidermis. The level of adaptive immunity of skin (i.e. regional lymph nodes) developed by mammals is not seen in birds. Therefore, the 5–10% NK cells (among all EDCs) should represent an important fraction of defence (against for example epitheliotropic viruses or mutated epithelial cells). A clue to their NK function is that we never found CD3⁺ FITC⁺ emigrants from cultured epidermal sheets, i.e. they are not antigen-sampling cells. The epidermal NK cells in the avian epidermis have a typical dendritic appearance. This dendritic nature of epidermal NK cells raises important questions. Is this feature simply the hallmark of cellular motility or of cytoskeletal polarization during immunological contacts (like the immunological synapses between the NK cells and their targets)¹⁸. In contrast to mucosal (intestinal) epithelium, where the NK cells and T cells usually coexist as distinct intraepithelial lymphocyte populations,¹⁹ in the avian skin we could not find any T cells.

Human or mouse LCs are E-cadherin expressing cells. Confocal microscopy revealed that all CD45⁺ EDCs are negative for chicken E-cadherin except for occasional small spots on some dendritic processes. The in vitro emigrated cells, as well as the isolated epidermal CD45⁺ cells, are completely E-cadherin^–. Enzymatic isolation did not degrade the surface E-cadherin, because the isolated keratinocytes retained their positivity (Fig. 1g). Cadherin-mediated adhesion is held as a prerequisite for the LCs to stay among the keratinocytes. On the other hand, all EDCs in the chicken express integrin β₁, like mammalian LCs,²⁰ but are negative for the other available chicken integrin-specific mAbs. The chicken epidermis is much thinner compared to the mammalian epidermis (only one or two spinous layers) and most DCs are located among the cells of the germinal layer, near the basal lamina. We suggest, that integrin β₁ on EDCs could mediate their adhesion to the basal lamina constituent laminin.

Up to now, there have been no studies on the ontogeny of EDCs in chickens. In contrast to mammals, where the LC number usually reaches its maximum level during late embryonic stages,²¹ in New Hampshire chickens only a few cells/mm² (about 1% of the adult) can be detected before hatching. The number of EDCs slowly rises and in the 8-week-old chickens it stabilizes around 2000 cells/mm². The EDCs appeared somewhat later in White Leghorn chickens, and the number of EDC in the young adults fluctuates more significantly than in New Hampshire chickens. This difference could reflect the genetic diversity of the two strains. Another possible explanation could be the origin of the eggs (specific pathogen-free versus conventional).

The kinetics of LCs has been studied extensively in humans²² and mice.²³ These studies proved that LCs are able to undergo mitosis in the epidermis. Using BrdU labelling we proved for the first time, that under normal steady-state conditions (i.e. without skin immunization) the chicken EDCs could also proliferate in the skin. A single intravenous BrdU injection resulted in a relatively constant labelling index throughout the day (30 min to 24 hr). We obtained similar indices for short-term in situ epidermal labelling. These data provide evidence similar to that obtained in mammals, EDC proliferation occurs in the epithelium to maintain the equilibrium against emigration. Although the effect of mechanical isolation and the presence of, for example, bovine proteins (e.g. fetal calf serum) should be a stimulus for the emigration of EDCs, the emigration rate of EDCs from isolated, unstimulated epidermis should stay for some orientation. (The EDC emigration from undisturbed epidermis could not be measured in this study.) Using this rate and the BrdU-labelling indices, we calculate that the whole EDC population should be renewed not faster than every 10–12 days. All these data point to the constitutive emigration of avian LCs from the epidermis, which – in the absence of inflammatory stimuli – should be responsible for the induction of peripheral tolerance.²⁴

Transcutaneous immunization experiments clearly showed that the chicken EDCs could respond to different antigens (haptens) by taking them up and migrating from epidermis. It is difficult to identify the destination of migrating EDCs because lymph nodes are absent in chickens. With the immunofluorescent method we were unable to demonstrate antigen-bearing descendants of EDCs either in primary or in secondary lymphoid organs. We therefore prepared serial sections from treated skin, to determine the EDCs migration pattern. FITC⁺ cells could be found scattered not only in the dermis but also in special lymphoid accumulations of the chicken dermis. These accumulations, the so-called dermal lymphoid nodules, were described histologically by Biggs in the 1950s²⁵ but until now there were no data on their cellular composition. These nodules are located in the dermis, often in the neighbourhood of small veins or lymphatics. We found T and B cells in these nodules. The lymphocytes are surrounded by a well-developed reticulum of DCs. The central part of this reticulum contains mAb 74.3⁺ cells, but typical B-cell follicles do not occur in the nodules. At the periphery of the nodule, the reticulum contains KUL01⁺ cells. FITC label could be detected in the cytoplasm of both KUL01⁺ and 74.3⁺ cells. This structure also suggests a possible differentiation pathway for EDCs: they arrive as KUL01⁺ cells at the periphery of the nodule and differentiate into 74.3⁺ cells. Their proximity to small veins/lymphatics and their capacity to collect FITC⁺ cells after cutaneous hapten administration suggests that they could be analogues of mammalian lymph nodes. The dermal lymphoid nodules can function like regional lymph nodes in mammals. We propose that these nodules could be possible terminals for chicken EDC migration. Further studies are needed to clarify the roles of these nodules in the skin-associated defence system. On the other hand, the presence of other, unidentified target organs could not be ruled out (Fig. 4).

Abbreviations

BrdU

5-bromo-2-deoxyuridine

DC

dendritic cell

EDC

epidermal dendritic cell

LC

Langerhans cell