Differential Expression of Cytokines in UV-B–Exposed Skin of Patients With Polymorphous Light Eruption:  Correlation With Langerhans Cell Migration and Immunosuppression FREE

Ultraviolet radiation (UV), especially UV-B radiation (280-315 nm), can modify proteins and other organic molecules that may act as neoantigensto provoke an autoimmune reaction.^1- 2 This adverse immune reaction can be suppressed by another effect of UV radiation,ie, the induction of cellular immunosuppression.^3- 4 A defect in this immunosuppression might therefore allow an adverse immune reactionto UV radiation, and polymorphous light eruption (PLE) could thus arise.

Key players in the UV-induced modulation of the skin immune system are cytokines. UV-B radiation can, directly or indirectly, induce the releaseof cytokines in the skin, resulting in the efflux or influx of several cell types. Interleukin (IL) 1β, tumor necrosis factor (TNF)-α, andIL-18 release, for instance by keratinocytes, can induce Langerhans cell migrationout of the skin^5- 7;and IL-10 is also involved in Langerhans cell migration, although it is still not clear whether it enhances⁸ or impairs themigration.⁹

Concurrently with the depletion of Langerhans cells after UV exposure, macrophages expand in the dermis and infiltrate the epidermis.¹⁰ Kanget al^11- 12 showed that these CD11b⁺ cells produce the immunosuppressive cytokine IL-10. Not only IL-10 from macrophages but also IL-10 from keratinocytes¹³ andTNF-α from dermal mast cell¹⁴ contribute to an immunosuppressive environment in the UV-exposed skin.¹⁵ Additionally,IL-4 was shown to be involved in the UV-induced suppression of delayed-type hypersensitivity¹⁶ and contact hypersensitivity.¹⁷

The development of a T_H2 milieu in the UV-B–exposedskin is supported by the influx of IL-4–expressing neutrophils¹⁸ and by the production of IL-10, which counteractsIL-12 activity and inhibits the activation of T_H1 cells but allows activation of T_H2 cells.¹⁹ The productionof IL-12 by monocytes/macrophages is decreased after UV-B irradiation,^12,20 as is the production of interferon (IFN) γ by T cells. Conversely, the production of IL-12p40 homodimers,the natural antagonist of IL-12, by keratinocytes and Langerhans cells is increased after UV exposure.^21- 22 Mostimportantly, a decrease in IL-12 activity and an increase in IL-4 and IL-10activity favors the differentiation of T_H0 cells into T_H2cells.

As a whole, the cytokine balance in the UV-exposed skin is shifted toward an immunosuppressive reaction in healthy individuals.

One of the cellular mechanisms related to this immunosuppression is the disappearance of Langerhans cells. In previous experiments, we showedthat Langerhans cells persisted in the skin of patients with PLE after UV-Boverexposure, in contrast to what was observed in healthy individuals.²³ The main mechanism responsible for Langerhans celldepletion in healthy individuals, ie, migration, was found to be impaired in patients with PLE. Only when patients with PLE were exposed to a largerUV dose did Langerhans cells disappear from the epidermis, thus revealing a disturbance in the balance between Langerhans cell migration and the erythemalresponse.

In the context of a potential defect in UV-induced immunosuppression in patients with PLE, these observations evolve into 2 hypotheses. First,the expression of the cytokines involved in Langerhans cell migration (eg, IL-1 and TNF-α) is reduced in the skin of patients with PLE. Second,we speculate that patients with PLE show an imbalance between proinflammatory and T_H1-skewing cytokines (eg, IL-12 and IFN-γ), and thecytokines involved in T_H2-skewing and T_H1 suppression (eg, IL-4 and IL-10), ultimately shifting the balance toward proinflammatoryand T_H1-skewed immune responses in the skin.

To test these hypotheses, unaffected buttock skin of 6 patients with PLE and 5 healthy volunteers was overexposed to UV-B. Skin biopsy specimenswere obtained from the UV-B–exposed skin 24 hours and 48 hours after irradiation , and from the unexposed skin as well as control specimens. Immunohistochemicalstaining was performed on frozen skin sections for a series of cytokines to locate and identify cytokine-expressing cells.

Six patients with PLE and normal sunburn sensitivity (1 man and 5 women aged between 32 and 64 years) and 5 healthy volunteers (2 men and 3 womenaged between 19 and 21 years) participated in the study. Skin types ranged from type I to type III in both groups. In previous studies we found fewerLangerhans cells in the skin of individuals older than 60 years, with a somewhat lessened depletion of Langerhans cells after UV irradiation.²⁴ Wedid not find any age dependence with respect to epidermal Langerhans cell density in healthy volunteers aged between 19 and 55 years, nor did we finda diminished Langerhans cell migration in this age range.²⁵ In this study the diagnosis of PLE was based on patients' history and clinicalphotoprovocation with daily exposure of UV-A 1, UV-B, and visible light on restricted areas of the upper arms. All recruited patients developed papulesor vesicles in the UV-B–irradiated skin area, with or without a similar reaction in the UV-A 1–irradiated skin. Patients who were taking medication(eg, corticosteroids) or who had received phototherapy were excluded from the study. Patients and healthy volunteers whose buttock skin was exposedto sunlight less than 2 months previously were also excluded. The medical ethics committee of the University Medical Center Utrecht approved the study.

The UV-B dose giving a minimal perceptible redness of the skin (minimum erythema dose [MED]) was determined on the buttock skin of each participantusing a Philips TL-12 lamp (Philips, Eindhoven, the Netherlands). The TL-12 lamp emitted 58% of its UV output in UV-B (280-315 nm) and 5% in UV rays lessthan 290 nm. Erythemally weighted, 98% of the lamp's effectiveness stemmed from the UV-B band (18% of rays below 290 nm). For MED determination a testdevice with 9 windows (3 × 10 mm) was used. These windows opened and closed sequentially, exposing the skin to UV-B for different lengths of time(12.4-200 seconds). The erythema was assessed 24 hours after irradiation.In patients with PLE, MEDs ranged from 483 to 945 J/m² (median,654 J/m²; 95% confidence interval [CI], 396-1080 J/m²) and in healthy controls, they ranged from 490 to 980 J/m² (median,734 J/m²; 95% CI, 421-1279 J/m²). Subsequently, the unaffected buttock skin of each subject was exposed to 6 MEDs of UV-B withthe Philips TL-12 lamp, resulting in a sunburn reaction but not in a pathologic PLE-related reaction (ie, there were no papular or vesicular responses inpatients with PLE). Three-millimeter punch biopsy specimens were obtained from the UV-exposed skin 24 hours and 48 hours after irradiation, togetherwith 1 control biopsy specimen from the normally unexposed buttock skin. The specimens were snap-frozen in liquid nitrogen, embedded in ornithine carbamyltransferase compound (Tissue-Tek OCT compound; Sakura, Zoeterwoude, the Netherlands), and stored at −80°C until further processing.

We used as primary antibodies monoclonal antibodies against IL-1α and IL-1β conjugated with biotine (Immunological Research Institute ofSiena, Siena, Italy); IL-4 and IL-6 (Genzyme, Cambridge, Mass; dilution, 1:100 and 1:50, respectively); IL-10 (Instruchemie, Delfzijl, the Netherlands; dilution,1:400); IL-12p40/p70 (BD Biosciences Pharmingen, Erembodegem, Belgium; dilution, 1:150); IL-18 (MBL, Nagoya, Japan; dilution, 1:200); IFN-γ (U-Cytech,Utrecht, the Netherlands; dilution, 1:100); TNF-α (Biosource, Nivelles, Belgium; dilution, 1:50); horseradish peroxidase (HRP)–conjugated elastaseas a neutrophil marker (Dako, Glostrup, Denmark; dilution, 1:10); alkaline phosphatase–conjugated AA1 against tryptase (Chemicon, Temecula, Calif;dilution, 1:50); and fluorescein isothyocyanate–labeled CD36 (Immunotech, Marseille, France; dilution, 1:80). Biotinylated rabbit antimouse immunoglobulin(Ig) (Dako; dilution, 1:400) or biotinylated horse antimouse Ig (Vector, Burlingame, Calif; dilution, 1:800) were used as secondary antibodies. Alkaline phosphate–or HRP-conjugated avidine-biotine complex (Dako; dilution, 1:50), HRP-conjugated rabbit antimouse Ig (IL-10; Dako; dilution, 1:100), or AP-conjugated F(ab)fragments of sheep anti-FITC (Boehringer Mannheim GmbH, Mannheim, Germany) were used as detecting reagents.

The technique to detect cytokines by immunohistochemistry in the cytoplasm of cells in microscopic sections has been described in detail by Hoefakkeret al.²⁶ Frozen skin sections (thickness, 6 µm) on 3-aminopropyltriethoxysilane–coated glass slides were fixedfor 10 minutes in 65 mL of acetone containing 100 µL of 30% hydrogen peroxide. Endogenous peroxidase activity was additionally blocked by incubatingthe slides for 15 minutes in 65 mL of 0.05M Tris hydrochloride (pH 7.6) containing 26 mg of 4-Cl-1-naphthol dissolved in absolute ethanol.

Primary antibodies for cytokine detection were all diluted in phosphate-buffered saline solution with 0.1% bovine serum albumin (PBS/0.1% BSA) and incubatedovernight. Slides were then incubated for 30 minutes with biotinylated rabbit antimouse Ig diluted in PBS/1% BSA normal human serum or for 60 minutes withHRP-conjugated rabbit antimouse Ig diluted in PBS/1% BSA for IL-10 staining. After incubation with biotinylated rabbit antimouse Ig, slides were incubatedfor 60 minutes with an HRP-conjugated avidin-biotin complex diluted in PBS/1% BSA. Antibody binding was visualized by incubating all skin sections simultaneouslyin 65 mL of 0.05M acetate buffer (pH 5.0) containing 26 mg of 3-amino-9-ethyl-carbazole (AEC; Sigma-Aldrich Corp, St Louis, Mo) and 32.5 µL of 30% hydrogenperoxide, resulting in red staining. The sections were counterstained with Mayer hematoxylin.

The skin sections were evaluated using 30% hydrogen peroxide (original magnification × 250). The numbers of cells expressing cytokines (TNF-αand IL-4) were assessed per square millimeter of epidermis or dermis.

After a 10-minute fixation in 65 mL of acetone containing 100 µLof 30% hydrogen peroxide and a 20-minute preincubation in 10% normal humanserum and 10% normal horse serum, the skin sections were incubated overnight with antibodies against TNF-α, IL-4, or IL-10 diluted in PBS/0.1% BSA.Subsequently, the slides were incubated for 60 minutes with biotinylated horse antimouse Ig diluted in PBS with 1% normal human serum and 1% normal horseserum, followed by a 30-minute incubation with an AP-conjugated avidin-biotincomplex. After incubation with a 10% normal mouse serum, the slides were incubatedfor 60 minutes with HRP-conjugated antielastase. Alkaline phosphatase activity was visualized by incubating the skin sections in a Tris hydrochloride buffer(pH 8.5) containing 25 mg of Fast Blue BB salt (Sigma-Aldrich), 12.5 mg ofnaphtol AS-MX phosphate, and 35 mg of levamisole (Sigma-Aldrich), resultingin a blue staining. For visualization of the peroxidase activity the slides were then incubated with a 3-amino-9-ethyl-carbazole solution, as describedabove.

Immunohistochemical staining for TNF-α and IL-10 was performed with the double staining technique just described, except that a HRP-ratherthan an AP-conjugated avidin-biotin complex was used as a detecting reagent. The slides were then incubated for 20 minutes with 10% normal mouse serum.Subsequently, the skin sections were incubated for 60 minutes with AP-labeled anti-AA1 (against tryptase) for TNF-α double staining or with FITC-labeledCD36 for IL-10 double staining. Incubation of the slides with CD36-FITC was followed by a 60-minute incubation with AP-conjugated F(ab) fragments of sheepanti-FITC. Alkaline phosphatase and HRP activity was visualized as described above.

The antibodies used in this study have been extensively tested for their suitability to stain microscopic skin sections. The primary antibody was omittedor replaced by an irrelevant antibody of the same isotype for negative controls. Sections of an inflamed human tonsil were used as positive staining controls.All antibody incubations were performed in a humidified chamber at room temperature except for incubations with the anticytokine antibodies, which were carriedout at 4°C. After each antibody incubation, slides were rinsed in PBS containing 0.05% polysorbate (Tween) 20.

A t test was performed to ascertain the observed differences (significance level, P<.05). The numbersof cells expressing TNF-α and IL-4 in the UV-exposed skin and in theunexposed control skin were compared. Furthermore, the numbers of cells expressingcytokines (TNF-α and IL-4) in the exposed and unexposed skin of patientswith PLE were compared with the number of the corresponding cells in the skinof healthy controls. A Spearman correlation test was carried out to reveal a possible correlation between the number of cells expressing TNF-αor IL-4 and an individual's absolute UV dose received (MED in joules per square meter).

Previous experiments showed that, in contrast to what happens in healthy individuals, Langerhans cells persist in the epidermis of patients with PLEafter overexposure of the skin to UV-B, and only a small number migrates out of the skin.²³ Therefore, we studied the expressionof the proinflammatory cytokines IL-1α, IL-1β, TNF-α, and IL-18, which are involved in the migration of Langerhans cells out of theskin.

No appreciable change in the number of cells expressing IL-1β was detected in the epidermis after UV exposure, but an increase was observedin the dermis of all participants. This increase, however, was much less in patients with PLE than in healthy controls (data not shown). Unfortunately,the variable and, at times, strong background staining hampered any good overall quantification of IL-1β expression. A semiquantitative scoring from verypositive (+ + ) to negative (–) was too crude to objectify this difference in any measure of statistical significance.

Constitutive expression of IL-1α and IL-18 was observed in theepidermis as IL-1α appeared to be diffusely spread whereas IL-18 wasrestricted to basal cells. The specimens from patients with PLE tended to show less IL-18 staining. A considerable number of cells in the dermis alsostained positive for IL-1α. No change in the number of IL-1α–positive cells was apparent after UV exposure, as only a slightly weaker staining wasobserved. The expression of IL-18 became more diffuse and spread to suprabasal cells in the UV-exposed skin (data not shown). The present technique is, however,not suitable to quantify these observations on Il-1α and IL-18.

In the unexposed skin of patients with PLE and healthy controls, a limited number of cells expressing TNF-α (weak positive staining) were observedscattered throughout the dermis (mean ± SEM, 29 ± 9 cells/mm² and 42 ± 19 cells/mm², respectively)(Figure 1), but not the epidermis. However, 24 hours after UV-B exposurewe observed a significant increase in the number of these cells in the dermisof healthy controls (mean ± SEM, 157 ± 17 cells/mm² [P = .02]) as well as an influx into the epidermis (mean ± SEM, 34 ± 6 cells/mm² [P =.004]). In patients with PLE we observed a small but significant increase in the number of dermal cells expressing TNF-α 24 hours and 48 hoursafter irradiation (mean ± SEM, 64 ± 13 cells/mm² [P = .02] and 68 ± 16 cells/mm² [P = .04], respectively). However, no change could be observed in the number of cells expressing TNF-α in the epidermis 24 hours and48 hours after UV irradiation (mean ± SEM, 12 ± 5 cells/mm² and 7 ± 3 cells/mm², respectively). Twenty-four hours after UV exposure the number of dermal and epidermal cells expressingTNF-α was significantly smaller (P = .001 and P = .01, respectively) in patients with PLE than in healthycontrols. No correlation was observed between the number cells expressingTNF-α and the absolute UV dose to which the individual's skin was exposed.

Only very few epidermal and dermal cells expressing IL-6 were detected in the unexposed skin of patients with PLE and healthy controls, and we didnot observe any difference in the number of cells expressing IL-6 between the unexposed and UV-B–exposed skin of all participants (data not shown).

All epidermal cells constitutively expressed IL-12, although some cells showed a more pronounced IL-12 expression (data not shown). After UV exposurethe epidermal IL-12 expression became more focal. A few cells with a strong expression of IL-12 could be observed in the upper epidermis. The number ofdermal cells expressing IL-12 did not differ between unexposed and UV-B–exposedskin. The above observations were made in patients with PLE as well as inhealthy controls.

The number of epidermal and dermal cells expressing IFN-γ was very low in the unexposed and UV-B–exposed skin of patients with PLEand of healthy controls, and we did not detect any difference in IFN-γ expression between unexposed and UV-B–exposed skin or between patientswith PLE and healthy controls (data not shown).

The number of dermal cells expressing IL-4 in the unexposed skin ofpatients with PLE and healthy controls was very low (mean ± SEM, 4± 2 cells/mm² and 1 ± 0.5 cells/mm²) (Figure 2) and no IL-4–xpressing cells could be detected in the epidermis. The number of dermal cells expressingIL-4 increased significantly 24 hours after irradiation in the skin of patients with PLE and healthy controls (mean ± SE, 208 ±49 cells/mm² [P = .008] and 246 ± 20 cells/mm² [P<.001]). At 48 hoursafter UV exposure the number of dermal cells expressing IL-4 decreased, but not significantly, in the skin of patients with PLE and healthy controls (mean± SEM, 103 ± 28 cells/mm² and 196 ± 46 cells/mm²). At 24 hours and 48 hours after UV exposure cells expressing IL-4infiltrated the epidermis of healthy controls (mean ± SEM, 47 ±13 cells/mm² and 55 ± 26 cells/mm²) and, to alesser extent, of patients with PLE (mean ± SEM, 11 ± 3 cells/mm² and 9 ± 4 cells/mm²) [P =.03 and P = .05]). Twenty-four hours after irradiation,the number of epidermal cells expressing IL-4 differed significantly fromthe number found in the unexposed skin of patients with PLE (P = .02) and of healthy controls (P = .02). To summarize, the number of cells expressing IL-4 increased after UV exposureand was significantly less in the epidermis of patients with PLE 24 hoursafter irradiation than in the epidermis of healthy controls. No correlationwas observed between the number of cells expressing IL-4 and the absoluteUV dose.

Cells expressing IL-10 were present in the basal layer of the epidermis and in the dermis of the unexposed skin (Figure 3) of patients with PLE and healthy controls. At 24 hourspostirradiation the expression of IL-10 in the basal layer had increased, and decreased again at 48 hours in most of the participants. A few cells thatstrongly expressed IL-10 had appeared in the epidermis at 24 hours in mostof the healthy controls or at 48 hours in most of the patients with PLE. Therewas no difference in the number of dermal cells expressing IL-10 in the unexposedand UV-exposed skin of all participants. The diffuse staining of IL-10 inthe epidermis prevented a good quantification of this expression; moreover, the difference seemed to be more in a delayed occurrence of densely stainedcells in the skin of patients with PLE than in the absolute number of thesecells, as is shown in Figure 3.

Because Meunier et al¹⁰ reported that CD36⁺, CD11b⁺, and CD1a⁻cells infiltratethe skin after UV exposure and that the CD11b⁺ CD1a⁻cells produce IL-10¹¹ and are involved in UV-induced immunosuppression,²⁷ we determinedwhether the cells strongly expressing IL-10 in the epidermis of the UV-exposedskin were CD36⁺ macrophages. Double staining was performed forIL-10, CD36, and elastase (a neutrophil marker). At 24 hours after irradiationalmost all epidermal cells expressing IL-10 in healthy controls were neutrophils.However, the epidermal cells with a strong IL-10 expression in the UV-exposedskin of patients with PLE and healthy individuals did not coexpress elastase48 hours after irradiation (Figure 4). Only very few cells expressing IL-10 were CD36⁺ macrophages. Therefore,the cells expressing IL-10 in the basal layer of the epidermis, and probably also a fraction of the cells that expressed IL-10 very strongly, were mostlikely keratinocytes.

Mast cells were shown to release large amounts of TNF-α upon activation by UV-B radiation.¹⁴ However, the pattern ofTNF-α expression that we observed was not indicative of mast cells, as it resembled the pattern of the neutrophil influx. Therefore, double stainingwas performed for TNF-α and tryptase (a mast cell marker) or elastase (a neutrophil marker). All dermal and epidermal cells expressing TNF-αin the skin of healthy controls and patients with PLE proved to be neutrophils (Figure 4). Only a small minorityof the neutrophils did not express TNF-α.

Teunissen et al¹⁸ showed that neutrophils are the main source of IL-4 in the UV-B–exposed skin. Hence, we performeda double staining for IL-4 and elastase to characterize the IL-4–expressingcells. All cells expressing IL-4 in the skin of healthy controls and patientswith PLE were neutrophils (Figure 4), and a minority of the neutrophils did not express IL-4.

We investigated whether the diminished migration of Langerhans cellsfrom the UV-exposed epidermis of patients with PLE²³ couldbe related to a lack of essential cytokines such as IL-1, TNF-α, andIL-18. The number of cells expressing IL-1β in the UV-exposed skin appearedto be lower in patients with PLE than in healthy controls. Some investigators reported that the UV-induced increase in IL-1β was mainly seen in keratinocytesand Langerhans cells,²² but we observed an increase in cells expressing IL-1β scattered throughout the dermis. Theincrease in cells expressing TNF-α after UV exposure was significantly higher in healthy controls than in patients with PLE. In contrast to whathappened in healthy controls, hardly any cells expressing TNF-α infiltrated the epidermis of patients with PLE. It has been reported that TNF-αis produced by keratinocytes²⁸ and mast cells²⁹ upon UV exposure. However, the TNF-α–expressing cells that we observed were neutrophils. Overall, the reduced expression ofIL-1β and TNF-α derived from neutrophils supports our hypothesis that the cytokines inducing Langerhans cell migration are lacking in the UV-exposedskin of patients with PLE. This could explain the impaired Langerhans cell migration that we observed in patients with PLE using the same irradiationprotocol as in the present study.

Interestingly, exposure of the skin of patients with PLE to 8.4 MEDs resulted in a near complete depletion of Langerhans cells 48 hours after irradiation.In addition, the expression of IL-1β, TNF-α, and IL-4 was considerably higher than with 6 MEDs and more cells expressing TNF-α and IL-4 infiltratedthe epidermis (data not shown). These observations indicate that Langerhans cells are not defective in their ability to migrate in patients with PLE butthat the balance between erythemal and immunosuppressive responses is disturbed.

Another hypothesis that we wanted to investigate was whether the T_H1-skewing cytokines were expressed in larger amounts in the UV-exposedskin of patients with PLE in comparison with healthy controls, thereby contributingto the pathogenesis of PLE. There are limited data on the cytokine profile of patients with PLE, and only on lesional skin. Norris et al³⁰ showedan increased activity of IL-6, IL-8, and, possibly, IL-1 in experimentally provoked PLE lesions. However, these changes need not be identical to thosepreceding the development of a lesion. We did not observe any difference inthe number of cells expressing IL-6 between the unexposed and the UV-exposedskin, nor between patients with PLE and healthy controls. Interleukin 12 was constitutively expressed by keratinocytes, which has been described by Yawalkeret al.³¹ The dose of UV radiation did not appear to change the level of IL-12 expression; it changed the expression pattern,however, as expression became more focal and stronger in some epidermis cells. Only very few IFN-γ–expressing cells were observed in the UV-exposedor unexposed skin of all participants. Collectively, these findings show that the expression of T_H1-skewing cytokines is similar in the UV-B–exposedskin of patients with PLE and of healthy individuals.

To further investigate the balance between T_H1-skewing and T_H2-skewing cytokines we performed immunohistochemical stainingsfor IL-4 and IL-10. In line with the results from Teunissen et al¹⁸ we found a UV-induced increase in dermal IL-4–expressingneutrophils and an influx of these cells into the epidermis of healthy controls.The number of IL-4–expressing neutrophils in the UV-exposed epidermisof patients with PLE was significantly lower than in healthy controls, indicatingthat the balance is somewhat shifted toward a T_H1 response in patients with PLE. UV radiation can induce the production of IL-10 by keratinocytes^15,32 and macrophages.¹¹ Treatmentwith IL-12 to inhibit IL-10 release³³ or treatment with anti-CD11b to deplete macrophages²⁷ reversedthe UV-mediated immunosuppression. Our results showed that IL-10 is predominantly expressed by cells in the basal layer of the epidermis. This expression isincreased after UV exposure and cells that strongly express IL-10 can be observed in the epidermis of patients with PLE (mostly at 48 hours) and healthy controls(mostly at 24 hours). It was less the number of cells expressing IL-10 that differed between patients with PLE and healthy controls than the timing oftheir appearance in the epidermis. Only very few IL-10–expressing cellswere CD36⁺ macrophages. The epidermal cells that strongly expressedIL-10 in healthy controls 24 hours after UV exposure were neutrophils. However, 48 hours after irradiation and in the UV-exposed skin of patients with PLE,cells expressing IL-10 coexpressed neither CD36 nor elastase and hence were most probably keratinocytes. CD11b is also expressed by neutrophils, and anti-CD11btreatment²⁷ will therefore also deplete neutrophils expressing IL-10, resulting in reversion of the immunosuppression. To summarize,our findings show that there was less expression of the T_H2-skewingcytokine IL-4 in the epidermis of patients with PLE than in the epidermisof healthy controls, thereby shifting the balance toward a T_H1 response. Furthermore, the observed difference in IL-10 expression betweenpatients with PLE and healthy controls was restricted to the infiltration of neutrophils in the epidermis of healthy controls.

Immunohistochemistry appears to be a suitable method for localizing cytokines in the cytoplasm and identifying the cells expressing certain cytokinesin vivo.^26,34 However, we are aware of the limitations of this technique. Most of all, it will not detectsecreted cytokines, only cytokines bound in the cytoplasm or on the cellular membrane. A clear intracellular staining and the difficulty of detecting verylow levels of cytokines lead to a visualization of cytokine-producing cells rather than receptor-bound cytokines on target cells, although this possibilitycannot be excluded completely.

A first exploration of the cytokine patterns in UV-B–exposed skinof patients with PLE vs healthy controls demonstrated that, in conjunctionwith diminished Langerhans cell migration, the skin of patients with PLE contains fewer migration-inducing cytokines. And although the expression of T_H1-skewing cytokines IL-12, IFN-γ, and IL-6 was not found to bemore prevalent in the UV-exposed skin of patients with PLE, we found a smaller number of cells expressing IL-4 in their epidermis, indicating a shift inthe cytokine profile toward a T_H1 response. The scarce expression of TNF-α and IL-4 in their UV-exposed skin appears to be largely attributableto a strongly reduced influx of neutrophils.

Following this exploratory study, future experiments on cytokine profiles in patients with PLE should focus on the level of expression of cytokinesin blister fluids using an enzyme-linked immunosorbent assay to detect cytokines, as has been described for healthy individuals by Barr et al.³⁵ However,to measure multiple cytokines, this technique would require many blisters. Modern techniques with bead-coupled antibodies appear to be excellently suitedto pursue the measurement of multiple cytokines in small volumes of fluid. Alternatively, one could determine cytokine mRNA levels by quantitative polymerasechain reaction. A disadvantage of this technique is that gene expression of cytokines does not necessarily lead to protein synthesis.³⁴

Based on the present cytokine data and on previous experiments we speculate that the following events occur in the skin of patients with PLE upon UV exposure:Langerhans cells residing in the epidermis capture UV-induced neo-antigens. Subsequently, these cells migrate out of the skin but their migration is impairedby low levels of IL-1β and TNF-α. The Langerhans cells that migrate increase their expression of costimulatory molecules and HLA-DR. Because ofthe impaired migration, activated Langerhans cells move slowly and accumulate in the dermis where they can stimulate memory T cells. A reduced expressionof IL-4, due to a failure of neutrophils to infiltrate in adequate numbers the skin of patients with PLE, favors the development of a T_H1response, in contrast to a pronounced T_H2 response in healthy controls.Together, these events may contribute to the pathogenesis of PLE. Clearly,this model of PLE pathogenesis needs further substantiation but provides a framework for well-targeted experiments.

Corresponding author: Wendy Kölgen, MSc, Free University Medical Center, Department of Pathology, Room 2E47, De Boelelaan 1117, 1081 HV Amsterdam,the Netherlands (e-mail: w.kolgen@vumc.nl).

Accepted for publication September 2, 2003.

This work was supported by the Stiching Instituut Voor Dermatologie en Venereologie Utrecht and by European Community grant QLK-CT-2001-000115.

We thank P. Ghiara, MD, from the Immunobiological Research Institute of Siena for providing the monoclonal antibodies against IL-1α and IL-1βconjugated with biotine.

We thank Ina Sybesma from the Department of Dermatology of the UniversityMedical Center Utrecht for the UV irradiation of all participants.