Innate immunity mediated by epidermal keratinocytes promotes acquired immunity involving Langerhans cells and T cells in the skin

Abstract

Skin is an immunological organ consisting of epidermal cells, i.e. keratinocytes and Langerhans cells (LCs, antigen-presenting dendritic cells), and both innate and acquired immune systems operate upon exposure of the skin to various external microbes or their elements. To explore the relationship between innate and acquired immunities in the skin, we investigated whether Toll-like receptor (TLR) ligation of epidermal cells enhances the ability of LCs to present a specific antigen to T cells in mice. LC-containing epidermal cells were incubated with CpG oligonucleotide (TLR9 ligand) modified with trinitrophenyl hapten, and cultured with hapten-primed CD4⁺ T cells. TLR9 ligand was capable of enhancing the hapten-presenting ability of LCs when LC-enriched epidermal cells, but not purified LCs, were used as the LC source, suggesting that bystander keratinocytes play a role in the enhancement of LC function. Cultivation of freshly isolated epidermal cells with CpG promoted the expression of major histocompatibility complex (MHC) class II and CD86 molecules on LCs. CpG enhanced the production of interleukin (IL)-1α, granulocyte–macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor (TNF)-α by primarily cultured keratinocytes. The addition of a cocktail of neutralizing antibodies against these cytokines abrogated the CpG-promoted, antigen-presenting ability of LC-enriched epidermal cells. Moreover, the addition of culture supernatants from CpG-stimulated keratinocytes restored the ability of purified LCs. Our study demonstrated that although the direct effect of CpG on LCs is minimal, LC function can be up-regulated indirectly by cytokines released by CpG-stimulated keratinocytes. This also implies that innate immunity evoked by TLR ligation of keratinocytes enhances acquired immunity comprising LCs and T cells.

Introduction

Conserved microbial structures, termed pathogen-associated molecular patterns are recognized by the immune system consisting of TLRs [1,2]. TLRs recognize microorganisms such as cell wall lipid, lipopolysaccharide (LPS), peptidoglycan and CpG oligonucleotide (CpG) [3]. The presence of TLRs on skin resident immune cells may assist in mounting a rapid and efficient host defence against invading pathogens or tissue injury-causing events. TLRs are expressed on epidermal cells (ECs), including keratinocytes [4,5] and Langerhans cells (LCs) [6] and mast cells [7]. It has been shown recently that TLRs expressed on human and murine ECs are functional and play an important part in skin innate immunity [4,6]. Murine LCs have been shown to express mRNA for TLR9 [6], but the expression and function of TLR9 on murine keratinocytes remains controversial.

Although TLRs expressed at sites of host–pathogen interaction probably serve to protect the host from pathogens, unnecessary immune responses to commensal bacteria may harm the host. This concept has been illustrated by several studies of TLR expression in LCs. In comparison with dendritic dells (DCs), LCs respond differently to microbial TLR ligands. For example, LPS is capable of inducing CD83 expression on DCs but not monocyte-derived LC-like cells [8]. Another group of investigators has also demonstrated a difference in TLR activation between LCs and DCs; stimulation with a ligand for TLR9 matures splenic DCs but not LCs [6]. These findings have suggested that LCs are beneficially low-responsive to skin commensals.

Even if LCs respond poorly to TLR ligands, as assessed by their expression of major histocompatibility complex (MHC) class II and co-stimulatory molecules, this does not exclude the possibility that TLR agonists modify the function of LCs when the skin is exposed to the agonists. Not only LCs but also keratinocytes are involved in the epidermal immunity. The latter cells are a pivotal constituent, participating profoundly in the both innate and acquired immunity [9]. Thus, the possibility arises that the innate immunity evoked via TLRs on keratinocytes enhances the acquired immunity consisting of LCs and antigen-specific T cells. To explore this issue, we investigated whether the ability of epidermal LCs to present a hapten to T cells is altered by CpG oligonucleotide, a TLR9 ligand, in the presence or absence of keratinocytes. Results suggest that the co-existence of LCs and keratinocytes is required for augmentation of LC function upon exposure of ECs to CpG.

Materials and methods

Animals and reagents

Female BALB/c mice, 8 weeks old, obtained from Charles River Japan, Inc., Yokohama, Japan, were maintained in our conventional animal facility. 2,4,6-Trinitrochlorobenzene (TNCB) and 2,4,6-trinitrobenzene sulphonic acid were obtained from Tokyo Kasei, Tokyo, Japan. CpG (ODN 1826) was purchased from Invitrogen (San Diego, CA, USA). As blocking antibodies, purified anti-mouse interleukin (IL)-1α, granulocyte–macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor (TNF)-α monoclonal antibodies (MoAbs) were obtained from BD PharMingen (San Jose, CA, USA).

Culture medium

RPMI-1640 (Sigma-Aldrich, St Louis, MO, USA) was supplemented with 10% heat-inactivated fetal calf serum (FCS), 5 × 10⁻⁵ M 2-mercaptoethanol, 2 mM l-glutamine, 25 mM HEPES, 1 mM non-essential amino acids, 1 mM sodium pyruvate, 100 units/ml penicillin and 100 µg/ml streptomycin.

Flow cytometry

The following MoAbs were employed: phycoerythrin (PE)-labelled anti-mouse I-A^d, fluorescein isothiocyanate (FITC)-labelled anti-mouse CD11c, CD4, CD86 (all from BD PharMingen) and TLR9 (eBiosience, San Diego, CA, USA). All MoAbs were used at 1–5 µg/10⁶ cells, and incubation was performed for 30 min at 4°C, followed by two washes in phosphate-buffered saline (PBS, pH 7·4) supplemented with 5% FCS and 0·02% sodium azide. Non-specific stains were performed with the adequate same-class immunoglobulin for specific MoAb. Fluorescent profiles were generated using a FACScan (Becton Dickinson, San Jose, CA, USA). To analyse the expression of MHC class II and CD86 on LCs, ECs were first incubated with anti-mouse Fcγ II/III receptor MoAb for 5 min to prevent non-specific binding of the subsequent reagents to Fc receptors and double-stained with PE-labelled anti-I-A^d and FITC-labelled anti-CD86 MoAbs. Gate was set so that only MHC class II-bearing, size-gated ECs were included in the analysis. 7-Amino-actinomycin D (7-AAD) was added to exclude dead cells.

For intracytoplasmic staining of TLR9, ECs were stained with PE-labelled anti-I-A^d and FITC-labelled anti-TLR9 MoAbs followed by using the Cytofix/Cytoperm plus Fixation/Permeabilization Kit (BD Biosciences, San Diego, CA, USA).

Preparation of LC–ECs and purified LCs

Ears from sacrificed naïve mice were split along the plane of the cartilage, which was then removed together with subcutaneous tissue. The specimens were incubated for 1 h at 37°C in a 0·2% solution of trypsin in PBS. After incubation, the epidermis was separated from the dermis and ECs were dispersed in PBS supplemented with 10% FCS by rubbing the separated epidermal sheets. The cells were filtered and washed twice in PBS.

To examine the effect of TLR9 ligand, EC suspensions were incubated with various concentrations of CpG for 24 h in the culture medium before preparation of LC–ECs. LC–ECs were obtained with Ficoll-Hypaque (specific gravity: 1·083; Sigma-Aldrich, St Louis, MO, USA), as described previously [10]. The percentage of LCs in LC–EC fraction was 15–20%, as assessed by flow cytometric analysis with anti-I-A^d antibody. For further purification of LCs, CD11c⁺ cells were positively selected from LC–ECs using an anti-CD11c MACS microbead antibody (Miltenyi Biotec Inc., Auburn, CA, USA). The purity of LCs was assessed by flow cytometry after staining with anti-CD11c–FITC and anti-I-A^d–PE MoAbs (BD PharMingen) and their purity was 70–80%.

In vitro proliferation of TNCB-immune T cells to trinitrophenyl (TNP)-modified LC–ECs or purified LCs

Mice were sensitized with TNP hapten by painting 0·05 ml of 5% TNCB in ethanol: acetone (3 : 1) onto the clipped abdomens on day 0. On day 5, lymph node cell (LNC) suspensions were prepared from inguinal and axillary lymph nodes. CD4⁺ T cells were isolated negatively with a cocktail of conjugated MoAbs (anti-CD8a, CD45R, CD49b, CD11b and Ter-119) (Miltenyi Biotec Inc.), and their purity was > 96%. TNP-modification of LC–ECs or purified LCs was performed by incubating with trinitrobenzene sulphonic acid as described previously [11]. Immune CD4⁺ T cells (2 × 10⁵ cells/well) were cultured in triplicate with TNP-modified LC–ECs or purified LCs (5 × 10³ cells/well) and various concentrations of CpG in a final volume of 200 µl in 96-well microtitre plates (Corning Glass Works, Corning, NY, USA) for 72 h at 37°C in 5% CO₂. Indomethacin (Sigma-Aldrich, St Louis, MO, USA) was added to the culture at a final concentration of 1 µg/ml. Methyl [³H]-thymidine (Amersham, Arlington, IL, USA) was added (1 µCi/well) 18 h before harvest. The cells were collected on glass fibre filters using a cell harvester (Futaba Medical Inc., Tokyo, Japan) and their radio-uptake was measured in a scintillation counter.

Keratinocyte culture and quantification of cytokines in the supernatants

Freshly isolated BALB/c ECs were suspended in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated FCS, 100 units/ml penicillin and 100 µg/ml streptomycin. Cells (7 × 10⁶/well) were then cultured (1·1 ml/well) for 72 h in the presence or absence of CpG in 24-well plates (Corning Glass Works) at 37°C in 5% CO₂. The culture supernatants were collected, stored at −80°C and measured for IL-1α, GM-CSF, and TNF-α by enzyme-linked immunosorbent assay (ELISA) (R&D Systems Inc., Minneapolis, MN, USA) according to the manufacturer's instructions.

Statistical analysis

Student's t-test was employed to determine statistical differences between means.

Results

TLR9 ligand is unable to stimulate hapten-immune CD4⁺ T cells in the absence of accessory cells

In advance of investigating the effect of CpG on LCs, we first tested its direct effect on T cells. Crude LNCs or purified CD4⁺ T cells from TNP-sensitized mice were cultured for 72 h in the presence of various concentrations of CpG. As shown in Fig. 1, LNCs proliferated well in response to CpG, but CD4⁺ T cells purified from the immune LNCs proliferated only marginally with the same stimulant. Thus, CpG was incapable of stimulating T cells per se, but it could activate T cells in the presence of TLR-bearing bystander accessory cells residing in the lymph nodes.

Fig. 1.

CpG oligonucleotide (CpG) stimulates crude lymph node cells (LNCs) but not purified CD4⁺ T cells. 2,4,6-Trinitrochlorobenzene (TNCB)-immune LNCs or T cells were cultured for 72 h with varying concentrations of CpG. *P < 0·01, compared with the non-CpG-added group. Data represent the mean ± standard deviation (s.d.).

Both keratinocytes and LCs express intracytoplasmic TLR9

It has been reported that LCs express mRNA for TLR9 as assessed by reverse transcription–polymerase chain reaction (RT–PCR) [6] and TLR9 is located in the endoplasmic reticulum in DCs [12]. Although murine keratinocytes were reported not to express TLR9 by RT–PCR [6], we analysed its expression in murine keratinocytes along with LCs by flow cytometry. We could not detect its expression on the surface of keratinocytes or LCs (data not shown), TLR9 was found to be expressed intracytoplasmically in both keratinocytes and LCs (Fig. 2).

Fig. 2.

Both keratinocytes and Langerhans cells (LCs) express Toll-like receptor 9 (TLR9). Epidermal cell (EC) suspensions were analysed for the expression of TLR9 by flow cytometry. For the intracellular detection of TLR9, cell fixation–permeabilization was performed before immunolabelling with anti-TLR9 and anti-I-A^d antibodies. LCs or keratinocytes were gated by I-A^d positivity.

CpG is unable to sufficiently up-regulate the hapten-presenting ability of purified LCs but able to enhance the LC ability in the presence of keratinocytes

EC suspensions freshly isolated from naive mice were first precultured for 24 h with various concentrations of CpG or without it as control, and were subjected to Ficoll gradient separation for LC–ECs. LCs were further purified from the LC–ECs with anti-CD11c microbeads, and 70% CD11c⁺ I-A⁺ cells were obtained (Fig. 3a) and modified subsequently with TNP. As responders, immune CD4⁺ T cells were prepared from LNCs of TNP-sensitized mice. They were cultured with the TNP-haptenized LCs to examine T cell proliferation. The culture was maintained for 72 h in the presence of the same TLR ligand as that used for preincubation of ECs. As shown in Fig. 3a, the purified and haptenized LCs duly induced the proliferation of immune T cells in the absence of CpG. The addition of CpG to the LC and T cell culture slightly enhanced the T cell responses, but the percentage augmentation by CpG was as low as 67% at most. Thus, the exposure of purified LCs to TLR9 ligand did not augment LC function sufficiently.

Fig. 3.

CpG oligonucleotide (CpG) cannot enhance substantially the hapten-presenting ability of purified Langerhans cells (LCs), but can increase that of LC–epidermal cells (ECs). ECs were incubated with or without CpG, and LC–ECs were obtained from the ECs (b; purification of LCs, 19%). LCs were further purified from the LC–ECs positively with anti-CD11c magnetic beads (a; purification of LCs, 70%). 2,4,6-Trinitrochlorobenzene (TNCB)-immune CD4⁺ T cells (2 × 10⁵/well) were cultured with trinitrophenyl (TNP)-modified, purified LCs (a) or LC–ECs (b) (5 × 10³ cells/well) and various concentrations of CpG. Data represent the mean ± standard deviation (s.d.). *P < 0·05, compared with the T + LC or LC–EC groups.

Even if purified LCs were low-responsive to CpG, the possibility remained that their function was augmented by the ligand in the presence of keratinocytes. By using LC–EC fraction, which contains not only LCs but also a high percentage of keratinocytes, we evaluated the effect of bystander keratinocytes on the antigen-presenting LC function. LC–ECs were prepared from EC suspensions precultured for 24 h with or without CpG. LC–ECs contained MHC class II⁺ CD11c⁺ LCs typically up to 19% (Fig. 3b). They were then derivatized with TNP and cultured for 72 h with TNP-primed CD4⁺ T cells in the presence of CpG or in its absence as control. Significant proliferative responses of immune T cells to haptenized LC–ECs were obtained in the absence of CpG (Fig. 3b). Notably, the addition of CpG dramatically enhanced the T cell response to haptenized LC–ECs. The percentage augmentation was approximately fourfold greater than that of the purified LCs. The results indicated that TLR9 ligands have the potential to augment the hapten-presenting ability of LCs when LC–ECs were used as the LC source.

CpG augments CD86 expression on LCs in the presence of keratinocytes

Freshly isolated ECs were cultured for 24 h in the presence or absence of CpG and the expression of CD86 on LCs was monitored. Figure 4 shows representative flow cytometric data, which were gated for LCs by I-A^d expression. LCs usually have two populations in their CD86 expression after 24-h culture. CpG increased the CD86-highly expressed (mature) population of LCs. Three independent series of experiments confirmed this increment of mature LCs by CpG.

Fig. 4.

CpG oligonucleotide (CpG) augments CD86 expression on Langerhans cells (LCs). Epidermal cell (EC) suspensions from naive mice were cultured with or without CpG (1 µg/ml) for 24 h. The cultured cells were subjected to flow cytometric analysis to see the expression of CD86 on LCs, which were gated by I-A^d expression. A representative flow cytometry shows the mature (CD86 high) and immature (CD86 intermediate) populations of LCs in the CpG added group (+) and the non-added one (–). The dotted line represents the isotype-matched control.

CpG enhances cytokine production by keratinocytes

The above findings raised the possibility that CpG first stimulates keratinocytes to produce cytokines, thereby subsequently promoting LC function. To address this issue, keratinocytes cultured primarily from BALB/c mice were incubated for 72 h with CpG, and the culture supernatants were subjected to ELISA to quantify IL-1α, GM-CSF and TNF-α, which are keratinocyte-derived cytokines concerned with the antigen-presenting function of LCs. As shown in Fig. 5, the production levels of IL-1α, GM-CSF and TNF-α were significantly increased by CpG in a dose-dependent manner.

Fig. 5.

CpG oligonucleotide (CpG) enhances keratinocyte production of cytokines. Freshly isolated keratinocytes were cultured for 72 h with various concentrations of CpG. Tumour necrosis factor (TNF)-α, interleukin (IL)-1α and granulocyte–macrophage colony-stimulating factor (GM-CSF) in culture supernatants were quantified by enzyme-linked immunosorbent assay (ELISA). Data represent the mean ± standard deviation (s.d.).

CpG up-regulates the antigen-presenting ability of LCs by promoting keratinocyte production of cytokines

We dissected the role of IL-1α, GM-CSF and TNF-α in the T cell response to haptenized LC–ECs under stimulation with CpG by blocking these cytokines with a cocktail of specific MoAbs. CD4⁺ T cells were cultured with haptenized LC–ECs and CpG in the presence or absence of anti-IL-1α, anti-GM-CSF and anti-TNF-α MoAbs. Figure 6a shows that the T cell proliferation was significantly lower in the MoAb-added group than in the non-added group. Therefore, these keratinocyte-derived cytokines mediated the promoted antigen-presenting ability of LCs.

Fig. 6.

Increasingly produced keratinocyte cytokines by CpG oligonucleotide (CpG) up-regulate the hapten-presenting ability of Langerhans cells (LCs). Neutralizing antibodies to keratinocyte cytokines, Interleukin (IL)-1α, granulocyte–macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor (TNF)-α, reduced the hapten-presenting ability of CpG (10 µg/ml)-stimulated LC–epidermal cells (ECs) (a). EC suspensions were incubated with CpG, and LC–ECs were prepared and haptenized with trinitrophenyl (TNP). TNP-immune CD4⁺ T cells were cultured with TNP-modified LC–ECs in the presence of CpG. A cocktail of neutralizing monoclonal antibodies (MoAbs) to IL-1α (2·5 µg/ml), GM-CSF (2·5 µg/ml) and TNF-α (2·5-µg/ml) were added to both the starting EC suspensions and the final culture with CD4⁺ T cells. *P < 0·05, compared with the MoAbs non-added culture. Data represent the mean ± standard deviation (s.d.). Culture supernatants from TLR ligand-stimulated keratinocytes enhanced the hapten-presenting ability of purified LCs (b). Culture supernatants from keratinocytes stimulated for 72 h with 10 µg/ml CpG were used in this study. LCs were purified positively with anti-CD11c magnetic beads. TNP-immune CD4⁺ T cells (2 × 10⁵/well) were cultured with TNP-modified LCs (5 × 10³ cells/well) in the presence of varying dilution of keratinocytes supernatants. The numbers in parentheses indicate the final dilution. *P < 0·01, compared with T + LC group. Data represent the mean ± s.d.

To examine the contribution of each cytokine to the antigen-presenting ability of LCs, TNP-immune T cells were cultured with TNP-modified LC–ECs and 10 µg/ml CpG in the presence or absence of two of anti-IL-1α (2·5 µg/ml), anti-GM-CSF (2·5 µg/ml) and anti-TNF-α (2·5 µg/ml) MoAbs. Percentage suppression with each combination of the two MoAbs and all the three MoAbs were as follows: anti-IL-1α + anti-GM-CSF, 23·3%; anti-IL-1α + TNF-α, 10·8%; anti-GM-CSF + anti-TNF-α, 23·2%; and anti-IL-1α + anti-GM-CSF + anti-TNF-α, 40·8%. These data suggested that all three cytokines are required for the activation of LCs and GM-CSF may be the most critical.

To further confirm the augmentative role of keratinocyte cytokines, immune CD4⁺ T cells were cultured with anti-CD11c-purified, TNP-modified LCs in the presence of varying dilution of supernatant from CpG-stimulated keratinocytes that were used in Fig. 5. As shown in Fig. 6b, the supernatant strongly increased LC function, and thus replaced the effect of keratinocytes residing in the LC–EC fraction. These findings suggested that LC function was up-regulated indirectly by CpG via cytokines released by CpG-stimulated keratinocytes.

Discussion

It has been shown that LCs, differing from DCs, are incapable of responding to TLR ligands, as the expression of MHC class II and co-stimulatory molecules on LCs are not augmented by the TLR-mediated signal [6]. Although this notion was confirmed by the present study, we also found that CpG is capable of stimulating LCs in a certain condition, resulting in enhancement of their hapten-presenting function. This stimulatory effect was exerted only when LCs co-existed with keratinocytes, as purified LCs mounted the CpG-induced enhancement at a lower level than did LC–ECs. The mediation of keratinocytes was clearly proved by the following findings: (1) CpG stimulated keratinocytes to secrete the three cytokines promoting the antigen-presenting function of LCs; (2) blocking of these cytokines with specific MoAbs abrogated the CpG-induced enhancement of LC–EC function; and (3) the addition of culture supernatant from CpG-exposed keratinocytes had the same augmentative effect as bystander keratinocytes on purified LCs. Therefore, TLR9 ligand is able to enhance LC function indirectly but substantially by promoting keratinocyte production of cytokines. It should be noted that TLR9 was expressed by normal BALB/c keratinocytes in primary culture, because this is different from the previous RT–PCR observation in murine keratinocytes [6]. However, information on the culture condition and passage of keratinocytes are missing in their report. The expression of TLR9 in normal human keratinocytes [5,13] may support our finding.

TLR ligand-enhanced production of cytokines/chemokines has been reported in several epithelial cells [14]. Intestinal and uterine epithelial cells express functional TLRs and their expression may be increased upon activation of cells [15,16]. Stimulation of these epithelial cells with TLR agonists triggers the increased expression of a number of cytokines and chemokines as seen in our study. Therefore, TLR ligands initiate innate immune responses by inducing the production of proinflammatory cytokines and chemokines as well as anti-microbial peptides, which are essential for host defence and survival. Furthermore, the present study suggests that the stimulation of keratinocytes with TLR9 ligand results in the activation of bystander LCs or DCs as a sequential event.

CpG stimulated keratinocytes to produce IL-1α, GM-CSF and TNF-α in a dose-dependent manner. The positive feedback in the interaction between TLR expression and cytokine production has been suggested in keratinocytes [5]. Among EC-derived cytokines, GM-CSF maintains the viability and potentiates the antigen-presenting function of LCs [17]. Koch et al. reported that IL-1 enhances LC function twofold when combined with GM-CSF [18]. This seems to be accordance with our findings that GM-CSF neutralizing antibody is the most suppressive for the antigen-presenting activity of LCs. IL-1 further augments the T cell stimulatory activity of LCs induced by GM-CSF [19], and exposure to TNF-α enhances the viability of LCs [18]. GM-CSF may not be the predominant regulator of MHC class II and B7 expression on LCs [20], while IL-1 and TNF-α are the candidates that up-regulate MHC class II, CD54 and CD86 on LCs [21,22]. Thus, it is difficult to define actual regulators of the expression of MHC class II and co-stimulatory molecules on LCs in the epidermal milieu where many biologically active substances are released [23]. Nevertheless, it is most likely that the enhanced expression of co-stimulatory molecules on LCs observed in this study results from interactions among these cytokines.

Our finding is clinically significant. The skin is considered to be frequently exposed to TLR ligands derived from bacteria [24], or possibly viruses. Keratinocytes are the initiator to respond to TLR9 ligands and enhance the antigen-presenting function of LCs with their augmentatively produced cytokines. In parallel with this study, we also found that other TLR ligands derived from Gram-positive or negative bacteria, such as peptidoglycan and LPS, enhance the hapten-presenting ability of LCs by augmenting of keratinocyte cytokine production. Therefore, in addition to releasing anti-microbial peptides [25], keratinocytes serve as an up-regulator of acquired immunity upon exposure to microbes. Even if LCs are low-responsive to skin commensals, they can be stimulated with microbial elements indirectly via keratinocytes. On one hand, this may be a mechanism underlying exacerbation of atopic dermatitis by skin-colonized Staphylococcus aureus as well as the superantigen mechanism [10]. On the other hand, the subsequently activated acquired immunity may further eliminate microbes.