Ultraviolet radiation–induced injury, chemokines, and leukocyte recruitment: An amplification cycle triggering cutaneous lupus erythematosus

Authors


Abstract

Objective

To investigate the activation and recruitment pathways of relevant leukocyte subsets during the initiation and amplification of cutaneous lupus erythematosus (LE).

Methods

Quantitative real-time polymerase chain reaction was used to perform a comprehensive analysis of all known chemokines and their receptors in cutaneous LE lesions, and the cellular origin of these chemokines and receptors was determined using immunohistochemistry. Furthermore, cytokine- and ultraviolet (UV) light–mediated activation pathways of relevant chemokines were investigated in vitro and in vivo.

Results

In the present study, we identified the CXCR3 ligands CXCL9 (interferon-γ [IFNγ]–induced monokine), CXCL10 (IFNγ-inducible protein 10), and CXCL11 (IFN-inducible T cell α chemoattractant) as being the most abundantly expressed chemokine family members in cutaneous LE. Expression of these ligands corresponded with the presence of a marked inflammatory infiltrate consisting of mainly CXCR3-expressing cells, including skin-homing lymphocytes and blood dendritic cell antigen 2–positive plasmacytoid dendritic cells (PDCs). Within cutaneous LE lesions, PDCs accumulated within the dermis and were activated to produce type I IFN, as detected by the expression of the IFNα-inducible genes IRF7 and MxA. IFNα, in turn, was a potent and rapid inducer of CXCR3 ligands in cellular constituents of the skin. Furthermore, we demonstrated that the inflammatory CXCR3 ligands cooperate with the homeostatic chemokine CXCL12 (stromal cell–derived factor 1) during the recruitment of pathogenically relevant leukocyte subsets. Moreover, we showed that UVB irradiation induces the release of CCL27 (cutaneous T cell–attracting chemokine) from epidermal compartments into dermal compartments and up-regulates the expression of a distinct set of chemokines in keratinocytes.

Conclusion

Taken together, our data suggest an amplification cycle in which UV light–induced injury induces apoptosis, necrosis, and chemokine production. These mechanisms, in turn, mediate the recruitment and activation of autoimmune T cells and IFNα-producing PDCs, which subsequently release more effector cytokines, thus amplifying chemokine production and leukocyte recruitment, finally leading to the development of a cutaneous LE phenotype.

Cutaneous lupus erythematosus (LE) represents an autoimmune disease characterized by photosensitivity, apoptosis of keratinocytes, and an inflammatory infiltrate in superficial and/or deep compartments of the skin (1, 2). Clinically, cutaneous LE lesions often arise at sun-exposed sites, and phototesting with ultraviolet (UV) light induces typical lesions in susceptible patients (3, 4). Furthermore, UV irradiation is a well known trigger of apoptosis in keratinocytes, and there is emerging consensus that abnormalities in the generation and clearance of apoptotic material may be an important source of antigens in autoimmune diseases (5–8).

Although the relevant antigen in LE remains unknown, skin-infiltrating activated leukocytes are thought to play a crucial role in the induction and maintenance of this autoimmune disease. The inflammatory infiltrate in cutaneous LE is dominated by skin-infiltrating memory T lymphocytes, and the majority of these cells display a CD4 phenotype (9, 10). Moreover, decreased numbers of plasmacytoid dendritic cells (PDCs) in the peripheral blood of patients with systemic LE and their presence and accumulation in cutaneous lesions of LE have been reported (11, 12). Recent findings suggest a central role for PDCs and one of their secreted products, interferon-α (IFNα), in the pathogenesis of systemic LE (SLE) (13–15).

Although the pathogenic role of skin-infiltrating leukocytes is undoubted, their recruitment and activation pathways in cutaneous LE remain elusive. Recently, a superfamily of small cytokine-like chemotactic proteins has been shown to regulate leukocyte trafficking under homeostatic and inflammatory conditions (16, 17). To date, 45 human chemokine ligands binding to 18 different G protein–coupled receptors are known. The chemokine superfamily is thought to be among the first of the complete protein families to be characterized at the molecular level (18, 19), offering the first opportunity to identify all relevant members associated with physiologic or pathologic processes.

Here, we show that UV irradiation induces the release and production of a distinct set of PDC- and T cell–attracting chemokines. Furthermore, we demonstrate that within the chemokine superfamily, the CXCR3 ligands CXCL9, CXCL10, and CXCL11 represent the most highly expressed genes in cutaneous LE. We also demonstrate the presence of activated type I IFN–producing blood dendritic cell antigen 2 (BDCA-2)–positive PDCs in cutaneous lesions in patients with LE. In turn, their major product, IFNα, rapidly and markedly induces CXCR3 ligands in cellular constituents of the skin, suggesting an amplification cycle leading to the development of a cutaneous LE phenotype.

PATIENTS AND METHODS

Patients.

All participants provided informed consent. Skin biopsy specimens were obtained from either healthy individuals (n = 14) or from the lesional skin of patients with cutaneous LE (n = 27), dermatitis solaris (n = 2), atopic dermatitis (n = 12), or psoriasis vulgaris (n = 36). The clinical diagnosis of cutaneous LE was confirmed by histologic evaluation, serologic examinations, and phototesting. Seven patients with discoid LE, 4 with subacute cutaneous LE, 9 with LE tumidus, 4 with UV-induced LE, and 3 patients with SLE were enrolled. Phototesting of patients with cutaneous LE during nonlesional phases of the disease was performed as previously described (20). Briefly, nonlesional skin of patients with LE was exposed to UVA radiation (UVASUN 3000 lamp; Mutzhas, Munich, Germany), UVB radiation (UV-800 lamp; Waldmann GmbH, Villingen-Schwenningen, Germany), or UVA/UVB radiation. Subsequently, skin biopsy specimens were obtained after the onset of typical lesions of cutaneous LE in the respective test field (ranging from day 4 to day 18 after the last exposure to UV radiation). Dermatitis solaris was induced by UVB irradiation (1–1.5 minimal erythema doses). Skin biopsy specimens were obtained from erythematous and slightly edematous lesions on day 2 or 3 after the initial exposure to UV radiation. Patients with atopic dermatitis were identified using the criteria described by Hanifin and Rajka (21). Patients with chronic plaque psoriasis in typical locations were enrolled in the study. Skin samples were immediately frozen in liquid nitrogen and stored at −80°C. This study was approved by the appropriate ethics committee.

Immunohistochemistry.

For immunohistochemical analysis of the expression of chemokines and chemokine receptors, skin sections were fixed with acetone and preprocessed with H2O2 followed by an avidin–biotin blocking step (Avidin/Biotin Blocking Kit; Vector, Burlingame, CA). Sections were stained with monoclonal antibodies against CXCR3 (mouse IgG1; R&D Systems, Minneapolis, MN), CXCR4 (mouse IgG2b; R&D Systems), CXCL9 (goat IgG; R&D Systems), CXCL10 (goat IgG; R&D Systems), CXCL12/stromal cell–derived factor 1α (mouse IgG2a, K15C; Unité d'Immunologie Virale, Institut Pasteur, Paris, France), CCL27 (mouse IgG2a; R&D Systems), myxovirus protein A (MxA; a kind gift from Dr. J. Pavlovic, University of Zurich), and BDCA-2 (mouse IgG1; Miltenyi Biotec, Bergisch Gladbach, Germany). Development of the staining was performed with a Vectastain ABC kit and a Vectastain AEC kit (both from Vector). Sections were counterstained with hematoxylin.

Culture and treatment of cell lines.

Human primary epidermal keratinocytes, dermal fibroblasts, and dermal microvascular endothelial cells were purchased from Clonetics (San Diego, CA) and cultured in keratinocyte growth medium, fibroblast growth medium, or endothelial cell growth medium (KGM-2, FGM-2, and EGM-2, respectively), as previously described (22). Cells were treated with tumor necrosis factor α (TNFα; 10 ng/ml)/interleukin-1β (IL-1β; 5 ng/ml) (Serotec, Dusseldorf, Germany, and R&D Systems, respectively), IFNα (0, 100, 500 or 1,000 units/ml; R&D Systems), IFNγ (50 ng/ml; R&D Systems), IL-4 (50 ng/ml; R&D Systems), or were left untreated. Furthermore, human primary keratinocytes in phosphate buffered saline were irradiated with 100 J/m2 UVB from TL12 lamps (Philips, Munich, Germany). The UVB radiation output, which was monitored using an IL1700 research radiometer and a SEE 240 UVB photodetector (International Light, Newburyport, MA), was ∼2.4 W/m2 at a tube-to-target distance of 22 cm. Immediately after exposure to UV radiation, keratinocytes were washed, cultured in appropriate growth medium, and stimulated with the indicated cytokines. The cells were harvested 6–24 hours after cytokine stimulation and/or UVB irradiation. Control cells were subjected to identical procedures without being exposed to UVB radiation.

Quantitative real-time polymerase chain reaction (PCR) analysis.

Quantitative real-time PCR analyses were performed as previously described (16, 22). Skin biopsy specimens were homogenized in liquid nitrogen using a Micro-Dismembrator U (Braun Biotech, San Diego, CA), and RNA was extracted with RNAzol according to the manufacturer's protocol (Tel-Test, Friedensburg, TX). Four micrograms of RNA was treated with DNase I (Boehringer, Mannheim, Germany) and reverse transcribed with oligo(dT)14-18 (Gibco BRL, Gaithersburg, MD) and random hexamer primers (Promega, Madison, WI), using standard protocols. Twenty-five nanograms of complementary DNA (cDNA) was amplified in the presence of 12.5 μl of TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA), 0.625 μl of gene-specific TaqMan probe, 0.5 μl of gene-specific forward and reverse primers, and 0.5 μl of water. As an internal positive control, 0.125 μl of 18S RNA–specific TaqMan probe and 0.125 μl of 18S RNA–specific forward and reverse primers were added to each reaction. Gene-specific probes used FAM as reporter, whereas probes for the internal positive control (18S RNA) were associated with the VIC reporter. Alternatively, 25 ng of cDNA was amplified in the presence of target-specific primer combinations and SYBR Green PCR Master Mix (Applied Biosystems). Chemokine ligand– and chemokine receptor–specific primers and target-specific probes were obtained from Applied Biosystems. Samples underwent the following stages: stage 1, 50°C for 2 minutes; stage 2, 95°C for 10 minutes; and stage 3, 95°C for 15 seconds, followed by 60°C for 1 minute. Stage 3 was repeated 40 times. Gene-specific PCR products were measured with an ABI PRISM 7700 or 5700 Sequence Detection System (Applied Biosystems), continuously during 40 cycles. Target gene expression was normalized between different samples based on the expression values of the internal positive control (18S RNA) or ubiquitin. Shrunken centroid analysis was performed as previously described (23).

Chemotaxis assays.

Transwell chemotaxis assays were performed as previously described (24). To analyze T cell migration, we purified T cells from human blood samples using the Human T Cell Enrichment Column kit (R&D Systems, Abingdon, Oxon, UK). A total of 1 × 106 cells in 100 μl of RPMI 1640 (Invitrogen, Carlsbad, CA) supplemented with 1% bovine serum albumin (Sigma-Aldrich, Seelze, Germany) and 1% HEPES buffer (Cambrex Bio Science Verviers, Verviers, Belgium) was placed in the upper compartments of the top chamber of a 3-μm pore size polycarbonate Transwell culture insert (Costar, Cambridge, MA). Cells were then incubated with the indicated concentrations of recombinant human CXCL10 (hCXCL10) and hCXCL12 (R&D Systems) in the bottom chamber for 3 hours in a 5% CO2 environment at 37°C. The number of migrated cells was determined by flow cytometry using anti-CD8, anti-CD4, and anti–common leukocyte antigen (CLA) antibodies (PharMingen, San Diego, CA). To determine the absolute number of migrated cells, a known number of 15-μm microsphere beads (Bangs Laboratories, Fishers, IN) was added to each sample before analysis. The relationship between the number of cells and the number of beads of the starting population is taken as 100% migration. Based on that, we calculated the percentage of migration in each sample. For each set of experimental conditions, at least 3 separate experiments were performed.

Statistical analysis.

The nonparametric Mann-Whitney U test was used to analyze differences in the gene expression in skin obtained from patients with cutaneous LE versus skin obtained from healthy individuals. The nonparametric Kruskal-Wallis test for multiple independent samples was used to determine whether genes were specifically overexpressed in skin from patients with cutaneous LE compared with skin from healthy individuals or skin from patients with atopic dermatitis or psoriasis. P values less than 0.05 were considered significant. All statistical analyses were performed with SPSS for Windows (version 11.0, 2000; SPSS, Chicago, IL).

To identify differentially expressed genes, results of quantitative real-time PCR analyses in patients with cutaneous LE were compared with data obtained from healthy individuals, using centroid analysis. Analysis of shrunken centroids was performed as previously described (23). This method efficiently finds and ranks genes that can distinguish one class/disease from another. Therefore, the class centroids are shrunk toward the overall centroids after standardizing by the within-class standard deviation for each gene. This method of standardization has the effect of giving higher weight to genes whose expression is stable within samples of the same class/disease.

RESULTS

Distinct chemokine ligand/receptor profile in cutaneous LE.

Because the chemokine superfamily is among the first of the complete protein superfamilies to be identified at the molecular level, we sought to identify members of this family of chemoattractive proteins that are associated with chronic inflammatory or autoimmune skin diseases such as atopic dermatitis, psoriasis, or cutaneous LE.

Quantitative real-time PCR analyses demonstrated that members of the chemokine superfamily show strong differential expression in skin from patients with cutaneous LE compared with healthy skin (Figure 1A). Notably, the Th1-associated chemokines CXCL9, CXCL10, and CXCL11 were among the top 5 most highly regulated chemokines in cutaneous LE (P < 0.01 by Mann-Whitney U test) (Figures 1A and 2A–C). Recently, 2 forms of their shared receptor, CXCR3, have been identified (25). Both the short form, CXCR3-A, as well as the alternatively spliced long form, CXCR3-B, were markedly induced in lesional skin from patients with cutaneous LE (P < 0.01 by Mann-Whitney U test) (Figures 1B and 2D and E) and were ranked among the top 3 most differentially regulated chemokine receptors in cutaneous LE (Figure 1B).

Figure 1.

Results of quantitative real-time polymerase chain reaction, showing differential expression of chemokine ligands (A) and chemokine receptors (B) in lesional skin specimens obtained from 9 patients with cutaneous lupus erythematosus (LE; 5 with discoid LE and 4 with LE tumidus) and normal skin specimens obtained from 14 healthy individuals.

Figure 2.

Quantitative real-time polymerase chain reaction analysis of the Th1-associated chemokines CXCL9 (A), CXCL10 (B), and CXCL11 (C) as well as their corresponding receptors CXCR3-A (D) and CXCR3-B (E) in skin specimens obtained from patients with cutaneous lupus erythematosus (LE), lesional skin specimens obtained from patients with atopic dermatitis, lesional skin specimens obtained from patients with psoriasis, and normal skin obtained from healthy individuals. Among the 9 patients with cutaneous LE, patients 2, 3, 6, 7, and 9 had discoid LE; patients 1, 4, 5, and 8 had LE tumidus. Absolute values of gene expression are expressed as femtograms of the target gene in 25 ng of total cDNA. ∗ = P < 0.01 by Mann-Whitney U test.

Furthermore, other T cell– and dendritic cell precursor–attracting chemokine ligand/receptor pairs, including CCL3, CCL4, and CCL5 and their corresponding receptors CCR1 and CCR5, as well as CCL19 and its receptor CCR7, showed increased RNA levels in cutaneous LE. Moreover, CXCL8 and its receptors CXCR1 and CXCR2 were markedly up-regulated in skin from patients with cutaneous LE compared with normal skin (Figures 1A and B).

Systematic analyses of chemokine ligand and chemokine receptor expression in normal skin from healthy individuals (n = 14) or in the cutaneous lesions from patients with psoriasis (n = 36), atopic dermatitis (n = 12), or LE (n = 9) demonstrated that the induction of CXCL9, CXCL10, and CXCL11, as well as that of their corresponding receptors CXCR3-A and CXCR3-B, in cutaneous LE was unparalleled compared with the induction observed in other inflammatory or autoimmune skin diseases (Figures 2A–E). Comparisons of the amount of transcripts for the indicated genes in normal skin, lesional atopic skin, or lesional psoriatic skin did not show significant differences (by Kruskal-Wallis test); however, lesional skin from patients with cutaneous LE showed significantly higher expression of CXCL9, CXCL10, CXCR3-A, and CXCR3-B messenger RNA (mRNA) compared with atopic, psoriatic, and normal skin (P < 0.001 by Kruskal-Wallis test).

To identify the cellular origin and obtain insights into the anatomic location of CXCL9, CXCL10, and CXCL11 and their receptor CXCR3 within the skin, we performed immunohistochemical analyses (Figures 3A–C). Within the epidermis and dermis, cells with dendritic morphology as well as endothelial cells of the superficial dermal plexus were identified as abundant sources of CXCL9 expression (Figure 3A). In contrast, CXCL10 showed marked expression in the basal layers of the epidermis in cutaneous LE as well as in skin-infiltrating leukocyte subsets clustering in perivascular pockets (Figure 3B, inset). In comparison with inflamed skin from patients with cutaneous LE, normal skin demonstrated negligible expression of CXCL9 and CXCL10 protein (results not shown). Their receptor, CXCR3, demonstrated strong expression on skin-infiltrating leukocytes in perivascular, subepidermal, and intraepidermal locations (Figure 3C). Endothelial cells also showed strong staining for CXCR3 (Figure 3C). Taken together, our immunohistologic findings confirm results obtained by quantitative real-time PCR and demonstrate the expression of chemokine ligands in close anatomic proximity with cells expressing their matching receptors.

Figure 3.

Immunohistochemical analysis of chemokine expression in lesional skin specimens obtained from patients with cutaneous lupus erythematosus (LE). A, CXCL9 was expressed by endothelial cells and skin-infiltrating leukocytes with a dendritic morphology in the epidermal and dermal compartments of lesional skin. B, CXCL10 was abundantly expressed in the basal layer of the epidermis and by skin-infiltrating leukocytes (inset). C, The majority of skin-infiltrating leukocytes in cutaneous LE lesions expressed CXCR3. D and E, CXCL12 was expressed by dendritic cells within the epidermis and dermis in cutaneous LE. Endothelial cells displayed strong CXCL12 reactivity. F, The majority of cells within the inflammatory, perivascular infiltrate of lesional skin were CXCR4 positive. Skin-infiltrating cells with a dendritic morphology showed the most abundant expression of CXCR4. G, Basal keratinocytes in normal skin produced large amounts of the homeostatic and skin-specific chemokine CCL27 (left). Lesional skin from a patient with dermatitis solaris shows increased secretion of CCL27 from the epidermal compartment to the dermal compartment (right). Arrows indicate the secretion of CCL27 from epidermal to dermal compartments. H, Enhanced and profound “leakage” of CCL27 (arrows) from the basal epidermis into the papillary dermis in lesional skin of ultraviolet B–induced cutaneous LE. Results are representative of 11 specimens analyzed. I, Loss of CCL27 expression in skin areas displaying profound vacuolar degeneration of basal keratinocytes. Results shown are representative of results in specimens obtained from patients with discoid LE. AF, Results shown are from patients with LE tumidus and are representative of results from 9 patients analyzed (4 with LE tumidus, 3 with discoid LE, and 2 with subacute cutaneous LE). (Original magnification × 250 in A, B, and D; × 400 in C, E, F, G, and H.)

Recently, a novel skin-specific chemokine, CCL27, was identified and was shown to play an important role during the recruitment of memory T cells into the skin (17, 24). This novel CC chemokine is homeostatically and abundantly expressed by basal keratinocytes in healthy skin (Figure 3G). Although quantitative real-time analyses of the homeostatic chemokine CCL27 showed slightly decreased mRNA levels in inflamed LE lesions compared with normal skin (Figure 1A), interesting results were observed at the protein level. In contrast to normal skin showing a relatively sharp border of CCL27 expression at the epidermal–dermal junction, skin specimens from patients with dermatitis solaris (n = 2) or early skin lesions of patients with LE undergoing UV light provocation testing (n = 4) demonstrated profound “leakage” of CCL27 from the basal epidermis into the papillary dermis, with impregnation of CCL27 on endothelial cells of the superficial plexus (Figures 3G and H). A comparison of skin lesions from patients with dermatitis solaris and lesional skin from patients with UV-induced cutaneous LE showed that the release of CCL27 into the epidermis was markedly enhanced in patients with LE (n = 4) (Figures 3G and H). Moreover, immunohistochemical evaluation of CCL27 protein expression in genuine skin lesions of patients with cutaneous LE (n = 9 [3 discoid LE, 4 LE tumidus, 2 subacute cutaneous LE]) showed strong release of CCL27 from the epidermal to the dermal compartment in 5 of the 9 patients analyzed. However, reduced amounts of CCL27 were observed in specimens from patients with advanced and chronic cutaneous LE displaying marked vacuolic degeneration of epidermal keratinocytes (n = 4), supporting the notion that the release of CCL27 is an early event during the initiation of autoimmune skin inflammation (Figure 3I). Therefore, after UV-induced injury, increased amounts of CCL27 are present within the dermis of patients with cutaneous LE and may support the recruitment of distinct leukocyte subsets into the skin.

Chemokine production in keratinocytes triggered by UVB irradiation.

Clinically, cutaneous LE lesions often arise at sun-exposed sites, and phototesting with UV light induces typical lesions in susceptible patients (3, 4, 20). Therefore, we hypothesized that UV irradiation may induce chemokine production, subsequently recruiting leukocytes to sites of injury and initiating an inflammatory process leading to the development of an LE phenotype. Because epidermal keratinocytes are a major target for UV irradiation, we focused on this cell population. Indeed, UVB irradiation resulted in the induction of a distinct set of chemokines in cultured primary keratinocytes. Although UVB irradiation alone did induce chemokine mRNA expression in keratinocytes, the most pronounced effects were observed in cytokine-stimulated cells (Figure 4). UVB irradiation significantly enhanced the expression of the inflammatory chemokines CCL5, CCL20, CCL22, and CXCL8 in epidermal keratinocytes (Figure 4). Notably, this set of chemokines is mainly regulated by primary proinflammatory cytokines such as IL-1 and TNFα, 2 mediators known to be markedly induced after UV-induced injury (7, 26). Furthermore, CCL5 and CXCL8 are among the top 10 most differentially regulated chemokines in cutaneous LE, suggesting a molecular signature of recent UV-induced injury in patients with cutaneous LE.

Figure 4.

Quantitative real-time polymerase chain reaction analysis of cytokine-stimulated (tumor necrosis factor α [TNFα]/interleukin-1β [IL-1β], interferon-γ [IFNγ], or IL-4) and ultraviolet B (UVB)–irradiated human primary keratinocytes. UVB irradiation induced the expression of a distinct set of chemokines. Absolute values of gene expression are expressed as femtograms of the target gene in 25 ng of total cDNA. Shown are representative results from 3 experiments. Bars show the mean and SD.

Accumulation of activated IFNα-producing cells in cutaneous lesions in LE patients.

Patients with SLE have elevated serum levels of IFNα. Accumulating evidence suggests that natural IFNα-producing cells/PDCs are the major source of IFNα in patients with LE (11, 12, 14). Recent findings place IFNα at the center of immunologic abnormalities observed in SLE and suggest that IFNα and IFNα-producing cells are novel targets for therapeutic intervention. However, the number of PDCs in the peripheral blood of patients with SLE is decreased, pointing to the question of where this population of natural IFNα-producing cells accumulates in patients with LE. Here, we confirm previous observations by other groups of investigators (12, 27) and demonstrate the presence and accumulation of PDCs at sites of autoimmune skin inflammation. Although no BDCA-2–positive PDCs were detected in normal skin (results not shown), cutaneous lesions of patients with LE demonstrated a marked increase in the number of PDCs (Figures 5C and D). BDCA-2–positive cells accumulated in perivascular pockets and were evenly distributed in other compartments of the dermis (Figures 5C and D). However, we detected no PDCs within the epidermis of nonlesional or lesional skin from patients with LE (Figures 5E and F).

Figure 5.

Infiltration of cutaneous lupus erythematosus (LE) lesions by activated interferon-α (IFNα)–producing cells/plasmacytoid dendritic cells (PDCs). A and B, Quantitative real-time polymerase chain reaction (PCR) analysis of IFNα (A) or the IFNα-inducible gene IRF7 (B) in subacute cutaneous LE (SCLE), LE tumidus (LET), systemic LE (SLE), discoid LE (DLE), and normal skin from healthy individuals. ∗ = P < 0.001 by Mann-Whitney U test. C and D, Accumulation of blood dendritic cell antigen 2 (BDCA-2)–positive PDCs within the dermis of a patient with cutaneous LE. Results shown are from a patient with discoid LE and are representative of results from 19 patients analyzed (4 with SLE, 7 with discoid LE, and 8 with LE tumidus). E and F, MxA expression in lesional skin from a patient with LE tumidus. GI, Quantitative real-time PCR analysis of CXCL9, CXCL10, and CXCL11 expression, respectively, in dermal fibroblasts, dermal endothelial cells, and cultured human primary keratinocytes stimulated with IFNα (0, 100, 500, and 1,000 units/ml) for 6 hours or 24 hours. Absolute values of gene expression are expressed as femtograms of the target gene in 25 ng of total cDNA. Shown are representative results from 3 experiments. (Original magnification × 250 in C and E; × 400 in D and F.)

The next important question that arose was whether skin-infiltrating PDCs were activated. Because PDCs have the unique ability to produce large amounts of IFNα upon activation, we sought to analyze the expression of type I IFNs in cutaneous LE lesions (Figure 5A). Although a trend toward increased IFNα production in the setting of autoimmune skin inflammation was observed, initial examinations did not demonstrate a significant up-regulation of IFNα transcripts in cutaneous LE compared with skin specimen from healthy individuals (Figure 5A). However, cutaneous LE lesions consistently showed the presence of an IFNα signature, with significant up-regulation of transcripts for the IFNα-inducible gene IRF7 (P < 0.001 by Mann-Whitney U test) (Figure 5B) and marked induction of the IFNα-inducible protein MxA (Figures 5E and F) when compared with normal skin (results not shown).

Rapid induction by IFNα of CXCR3 ligands in structural cells of the skin.

After demonstrating the presence of activated PDCs at sites of autoimmune skin inflammation, we next sought to evaluate the effects of IFNα on the recruitment pathways of pathogenically relevant leukocyte subpopulations. Human primary keratinocytes, dermal endothelial cells, and dermal fibroblasts were cultured in the presence or absence of different amounts of IFNα. Subsequently, chemokine expression was analyzed 6 hours and 24 hours later. Compared with resting cells, IFNα rapidly and markedly induced expression of the Th1-associated chemokines CXCL9, CXCL10, and CXCL11 in a dose-dependent manner. Only 6 hours after stimulation, maximum levels of chemokine expression were observed (Figures 5G–I). In contrast, IFNγ showed a delayed induction kinetic, with peak expression of CXCL9, CXCL10, and CXCL11 in structural cells of the skin 18–24 hours after stimulation (results not shown). Taken together, IFNα represents a potent and rapid inducer of relevant chemokines in structural cells of the skin, suggesting an important role of IFNα in the recruitment of leukocytes and the amplification of autoimmune skin inflammation.

Cooperation of homeostatic and inflammatory chemokines in attracting lymphocytes.

Previously, Vanbervliet et al showed that the inflammatory CXCR3 ligands regulate the responsiveness of PDCs to the homeostatic chemokine CXCL12 (28). Here, we showed that, similar to PDCs, CXCR3 and CXCR4 ligands cooperate in the recruitment of CLA-positive skin-homing memory T cells (Figures 6A and B).

Figure 6.

The homeostatic chemokine CXCL12 cooperates with the inflammatory chemokine CXCL10 in recruiting skin-homing common leukocyte antigen–positive (CLA+) memory T cells. A and B, Chemokine-mediated directional migration of CD4+,CLA+ T cells (A) or CD8+,CLA+ T cells (B) in response to CXCL10 (10, 100, and 1,000 ng/ml) alone or in combination with a suboptimal dose of CXCL12 (10 ng/ml) using Transwell chemotaxis assays. Results are expressed as the percentage of migrated T cells of the starting population. C and D, Preincubation of T lymphocytes with pertussis toxin (PTX) completely abolished CXCL10- and CXCL12-induced chemotactic activity. AD, Representative results from more than 3 independent experiments. Bars show the mean and SD.

In vivo, CXCR3 and CXCR4 ligands are expressed at similar anatomic locations within epidermal and dermal compartments of lesional skin in patients with LE (Figures 3A–F). The homeostatic chemokine CXCL12 is predominantly expressed by cells with dendritic morphology within the epidermis and dermis of patients with LE. Moreover, this potent chemoattractive chemokine is produced by endothelial cells of the superficial plexus of the skin (Figures 3D and E). The CXCR3 ligand CXCL9 demonstrates a similar cellular distribution, with the predominant expression by epidermal and dermal dendritic cells and strong CXCL9-specific immunoreactivity of dermal endothelial cells. Furthermore, CXCL10 is predominantly produced by epidermal keratinocytes and skin-infiltrating perivascular leukocytes within skin lesions in patients with LE (Figure 3). Hence, CXCR3 and CXCR4 ligands are perfectly positioned to collaborate during the recruitment of receptor-expressing leukocyte subsets. Their corresponding receptors are expressed on the majority of skin-infiltrating leukocytes in cutaneous LE and are known to be abundantly expressed on the cell surface of PDCs as well as memory T cells displaying a Th1 phenotype (18, 28–31).

Transwell chemotaxis assays showed that the CXCR3 ligand CXCL10 induces moderate migration of CD4+ or CD8+ skin-homing CLA-positive memory T cells, leading to the recruitment of 5–10% of the cells of the starting population (Figures 6A and B). In combination with suboptimal doses of CXCL12, however, CXCL10 mediated chemotactic responses of 35–50% of CD4+ or CD8+ skin-homing CLA-positive memory T cells present in the starting population (Figures 6A and B). Similar results were obtained with the CXCR3 ligand CXCL9 (results not shown). Furthermore, experiments using pertussis toxin demonstrated that this synergism between the homeostatic chemokine CXCL12 and the inflammatory chemokine CXCL10 was dependent on Gαi protein–coupled receptor signaling (Figures 6C and D).

DISCUSSION

Chemokines represent a superfamily of small cytokine-like proteins that mediate directional migration in vitro and control leukocyte trafficking in vivo (19). Recent studies have demonstrated that members of this family of chemoattractive proteins play an important role in the organization of innate and adaptive immune responses (32, 33).

Here, we provide the first comprehensive analysis of chemokines and their receptors in cutaneous LE, demonstrate the induction of a distinct set of genes, and provide novel insights into the recruitment and activation pathways of relevant leukocyte subsets to peripheral sites of autoimmune inflammation in patients with LE. Notably, the Th1-associated chemokines CXCL9, CXCL10, and CXCL11 as well as their receptor CXCR3 were among the top 5 most highly regulated chemokine ligands or receptors, respectively, in cutaneous LE. Among the memory T cell subsets, CXCR3 is predominantly expressed on the surface of IFNγ-producing Th1 cells (31). The findings in the present study demonstrate that the majority of cells infiltrating the skin display CXCR3 on their cell surface. Furthermore, IFNγ markedly induces these top-ranked chemokine ligands both in vitro and in vivo (34, 35), suggesting that IFNγ may play a role in cutaneous LE and defining cutaneous LE as a putatively Th1-associated disease. In fact, recent studies suggest that a Th1/Th2 imbalance with a predominance of Th1 cytokines, including IFNγ, is of pathogenic importance in autoimmune diseases such as SLE (36–38).

Furthermore, other T cell– and DC precursor–attracting chemokine ligand/receptor pairs such as CCL3, CCL4, and CCL5 and their corresponding receptors CCR1 and CCR5 showed increased RNA levels in cutaneous LE (29, 39–41). Moreover, CCL19 and its receptor CCR7 were markedly induced in cutaneous LE, suggesting the recruitment of CCR7-positive central memory T cells to sites of autoimmune skin inflammation. Comparing the expression of CXCR3 ligands in cutaneous LE with that in other chronic inflammatory or autoimmune skin diseases, including atopic dermatitis and psoriasis, or normal skin demonstrated that the induction of CXCL9, CXCL10, and CXCL11 as well as their corresponding receptors CXCR3-A and CXCR3-B in cutaneous LE was unparalleled. Although these transcripts were significantly overexpressed in cutaneous LE, some degree of variation between patients was observed. These differences in chemokine and chemokine receptor expression may reflect different stages and severity of the disease and/or be the result of interindividual variation.

Our quantitative real-time PCR results are consistent with previous qualitative results by Flier et al (42), in which immunohistochemical analysis and in situ hybridization demonstrated the expression of CXCL9 and CXCL10 in skin lesions of patients with LE. The abundant production of CXCR3 ligands in cutaneous LE lesions points to the question of their induction pathways. The type I effector cytokine IFNγ is a potent inducer of CXCR3 ligands both in vitro and in vivo (34, 35). In turn, antigen-specific stimulation of skin-infiltrating CXCR3-positive memory T cells may lead to the production and release of large amounts of IFNγ at sites of LE antigen exposure, suggesting that skin-infiltrating memory T cells may initiate chemokine production.

Nonetheless, the question remains regarding which activation pathways may lead to the recruitment of the first wave of T lymphocytes into the skin of patients with LE. Clinically, cutaneous LE lesions often arise at sun-exposed sites, and diagnostic UV irradiation regimens induce typical lesions in susceptible patients (3, 4, 20). Findings in the present study suggest that UV irradiation may provide the clue to explaining the early recruitment of leukocytes into the skin of patients with LE. Two different UV-induced phenomena might be of importance. First, UV irradiation may directly induce chemokine production by epithelial cells, and second, UV-induced apoptosis or necrosis of keratinocytes may lead to the release of homeostatic chemokines (e.g., CCL27 from epidermal stores into the dermis). Recently, CCL27 has been shown to bind with high affinity to extracellular matrix components and impregnate on the surface of dermal fibroblasts and endothelial cells (17). Furthermore, the novel skin-specific chemokine CCL27 has been shown to play an important role during the recruitment of CCR10-positive memory T cells into the skin (24). Hence, the early release of CCL27 in patients with cutaneous LE may provide matrix-bound sustained gradients mediating the recruitment of CCR10-positive skin-homing leukocyte subsets to sites of UV-induced injury.

In addition to CCL27, UV radiation–induced chemokines such as CCL5 and CCL20 are also potent attractants for lymphocytes and are associated with the pathogenesis of several other inflammatory or autoimmune skin diseases, including atopic dermatitis and psoriasis (22, 40). Furthermore, CCL5 and CXCL8 are among the top 10 most differentially regulated chemokines in cutaneous LE, suggesting a molecular signature of recent UV–induced injury in patients with cutaneous LE. In addition to skin-infiltrating CD4+ and CD8+ T cells, we also observed the presence of significant numbers of activated BDCA-2–positive PDCs in cutaneous LE lesions. Recent findings suggest that PDCs acquire CCR6 and CCR10 cell surface expression and chemoattractive responsiveness after stimulation with IL-3 or viral activation (Bendriss-Vermare N, et al: personal communication). Thus, we speculate that UV irradiation may induce chemokine production and release, subsequently recruiting a first wave of skin-homing memory T cells and PDCs via CCR5-, CCR6-, and CCR10-driven pathways to sites of UV-induced injury.

UV irradiation is a well-known trigger of keratinocyte apoptosis and necrosis (8, 43, 44). Furthermore, patients with LE have been shown to display an increased susceptibility toward apoptosis induction (43, 45), and a defect in the clearance of apoptotic cells has been suggested (6, 46, 47). Clinically, these phenomena may explain the marked vacuolar degeneration of keratinocytes within the epidermis of patients with cutaneous LE. Recent findings by Lovgren et al (48) demonstrate that necrotic and late apoptotic cells release material that, combined with LE IgG, induces the production of IFNα in PDCs. Those authors suggest that IFNα inducers probably consist of immune complexes containing RNA and possibly DNA as essential interferogenic components.

Consistent with previous reports by Farkas et al (12) and Blomberg et al (27), we demonstrate the accumulation of PDCs at sites of autoimmune skin inflammation in patients with LE. Although we could observe only a trend and no statistically significant induction of IFNα transcripts in chronic cutaneous LE lesions, we provide the first quantitative evidence that the IFNα-inducible gene IRF7 is significantly up-regulated in cutaneous LE lesions. Furthermore, cutaneous LE lesions demonstrated the marked expression of the IFNα-inducible protein MxA. Hence, these observations suggest that PDCs in cutaneous LE are activated, and IFNα production may represent an early phenomenon during the initiation of autoimmune skin inflammation.

Findings of recent studies suggest a central role for IFNα in the break of peripheral tolerance and the induction of autoimmunity (15). The level of this cytokine is elevated in patients with SLE and correlates with disease activity and severity, pointing to a pivotal role for type I IFN in the immunopathogenesis of LE (13, 49, 50). After establishing the activation of PDCs and the presence of IFNα-inducible proteins, the question arises of how this potent inflammatory cytokine influences leukocyte recruitment pathways in cutaneous LE. This study is the first to show that IFNα is a potent and rapid (6 hours) inducer of CXCR3 ligands in structural cells of the skin such as primary keratinocytes, dermal fibroblasts, and dermal endothelial cells. In contrast, the T cell effector cytokine IFNγ mediates a delayed induction pattern of CXCL9, CXCL10, and CXCL11 in structural cells of the skin, with peak expression between 18 hours and 24 hours after cytokine stimulation (data not shown). These observations suggest that PDC-derived IFNα, in concert with T cell–derived IFNγ, may promote chemokine production in cutaneous LE and amplify the recruitment of leukocytes to sites of autoimmune skin inflammation.

Until recently, the recruitment pathways of PDCs to peripheral sites remained elusive. Vanbervliet et al (28) demonstrated that inflammatory CXCR3 ligands regulate the responsiveness of PDCs to the homeostatic chemokine CXCL12. PDCs abundantly express the chemokine receptors CXCR3 and CXCR4 on their cell surface but do not show significant migratory responses to CXCR3 ligands alone. However, when presented in combination with suboptimal doses of CXCL12, CXCR3 ligands dramatically enhanced migratory responses of PDCs (28). This is the first study to demonstrate that the activity of homeostatic chemokines can be controlled by inflammatory chemokines. Here, we show that CXCR3 ligands and the CXCR4 ligand CXCL12 are produced at similar anatomic locations in cutaneous LE, and broaden the general concept of chemokine cooperation by showing the synergism of chemokines during the recruitment of CLA-positive skin-homing memory T cells.

Taken together, the findings in the present study suggest an amplification cycle, with UV radiation injury inducing apoptosis and necrosis of keratinocytes, together with the production and release of a distinct set of chemokines. Subsequently, chemokines mediate the recruitment of the first wave of leukocytes, including skin-homing memory T cells and PDCs, into the skin of patients with LE. Within the skin, T cells may encounter their specific antigen and secrete effector cytokines (e.g., IFNγ), while PDCs undergo activation processes induced by apoptotic or necrotic cells plus LE serum IgG, leading to the production of IFNα. In turn, IFNα and IFNγ may induce the abundant production of the CXCR3 ligands CXCL9, CXCL10, and CXCL11, which in concert with the homeostatic CXCR4 ligand CXCL12 recruit more CLA-positive memory T cells and PDCs to sites of autoimmune skin inflammation and perpetuate an amplification process that finally leads to the development of a cutaneous LE phenotype.

Currently, intense efforts focus on the development of small molecule antagonists or neutralizing biologics against chemokines and their receptors. Indeed, several drugs, including a small molecule antagonist against CXCR3, are currently in the process of clinical development. Although further evidence will be needed to validate the proposed model, the findings in the present study provide insights into the immunopathogenesis of cutaneous LE and may lead to the initiation of evidence-based clinical trials.