Is disturbed clearance of apoptotic keratinocytes responsible for UVB-induced inflammatory skin lesions in systemic lupus erythematosus?

Patients eligible for the study fulfilled at least four ACR (American College of Rheumatology) criteria for SLE [19] and had inactive disease, defined as SLEDAI (SLE Disease Activity Index) of not greater than 4. Patients with active skin disease, and those patients and controls whose buttock skin had been exposed to sunlight or other sources of UVB in the past 6 months, were excluded from the study. The local ethics committee of the University Medical Center Groningen (The Netherlands) approved the study, and all patients and controls gave written informed consent according to the Declaration of Helsinki. Fifteen patients (age 48.7 ± 12.3 years [mean ± standard deviation (SD)]; four males, 11 females) were included. Skin types were determined based on the Fitzpatrick skin typing chart [20]. Skin type distribution (apart from two type 5 or higher scores in non-Caucasian patients) was similar in patients (three type 2, 10 type 3, one type 5, and one type 6) and controls (four type 2 and nine type 3). Table 1 shows patient characteristics and immunosuppressive medication used at the time of the study. Autoantibodies to double-stranded DNA were measured by Farr assay, and antibodies to SSA/Ro, SSB/La, nRNP, and Sm were assessed by counter immuno-electrophoresis. Anti-cardiolipin antibodies, immunoglobulin (Ig) G or IgM, were tested by enzyme-linked immunosorbent assay. Minimal erythemal dose (MED) was determined as described below. Thirteen healthy volunteers (age 33.8 ± 15.8 years [mean ± SD]; five males, eight females) were included as controls. Additionally, to compare results with established cutaneous lesions of patients with SLE, 20 biopsies of LE skin lesions were retrieved from the files of the Department of Pathology (Table 2).

UVB irradiation was performed using the Waldman 800 'sky' lights with TL-12 lamps (Philips, Eindhoven, The Netherlands) at a distance of 15 cm from the buttock skin. A Diffey grid [21] was used to irradiate the skin with 10 different doses (0.026 to 0.200 J/cm²) during one exposure. After 24 hours, MED was determined by two independent observers, with more than 90% agreement. MED is defined as the lowest UVB dose at which erythema can be detected in the skin. In case of disagreement between observers, the mean of the two values was used. The reproducibility of MED assessment was determined by a second irradiation in 10 subjects, five healthy controls and five patients. Inter-test variability ranged from 0% to 22% (median 3%). After assessment of MED, subjects were irradiated with two MEDs of UVB on four small areas (1 × 2.5 cm/area) on the other buttock. After 1, 3, and 10 days, 4 mm skin biopsies were taken from the area of skin irradiated with two MEDs. As a control, a biopsy was taken after 1 day from non-irradiated skin. The distance between the biopsies was at least 2 cm to avoid the influence of wound healing on the reaction to UVB. Biopsies were split, one half fixed in formaldehyde and the other half snap-frozen in liquid nitrogen.

Diaminobenzidine (DAB) solution contained 25 mg DAB and 50 mg imidazol in 50 ml phosphate-buffered saline (PBS) and was filtrated before use. AEC (3-amino-9-ethylcarbazole). (Sigma-Aldrich, St. Louis, MO, USA) stock solution was diluted 100× in acetate buffer (0.05 M, pH 5.0). To both staining solutions, H₂0₂ was added just before incubation of the sections, resulting in a concentration of 0.1% H₂0₂. For haematoxylin staining, Mayer's haemalum solution (Merck, Darmstadt, Germany) was used.

Rabbit antibodies to cleaved caspase-3 (no. 9661S) were purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA). Primary fluorescein isothiocyanate (FITC)-labeled goat F(ab)₂ antibodies directed against human IgM, IgA, and IgG were purchased from Protos Immunoresearch (Burlingame, CA, USA). Primary mouse antibodies directed against CD68 (clone PGGM 1), CD3-FITC-labeled (clone F7.2.38), C3c-FITC-labeled (clone F201), and C1q-FITC labeled (F0254), and all secondary antibodies (that is, goat anti-rabbit IgG-horseradish peroxidase [HRP], rabbit anti-goat IgG-HRP, and rabbit anti-mouse IgG-HRP) were obtained from DakoCytomation (Glostrup, Denmark).

Skin sections (4 μm) on APES (3-amino-propyltriethoxysilane)-coated glass slides were used for all experiments. Sections were deparaffinised by subsequent incubations in xylene (10 minutes), 100% ethanol (5 minutes), and 96% ethanol (2 minutes), twice, followed by ethanol 70% (2 minutes) and distilled water. H&E staining was performed according to standard protocol using the linear stainer from medite Medizintechnik GmbH (Burgdorf, Germany). Antigen retrieval was performed by boiling in 1 mM EDTA (ethylenediaminetetraacetic acid) pH 8.0 (cleaved-caspase-3 staining), 10 mM citrate pH 6.0 (CD68 staining), or 10 mM Tris, 1 mM EDTA pH 9.0 (CD3 staining). After washing in PBS, endogenous peroxidase was blocked by incubation in 0.37% H₂0₂ in PBS for 30 minutes. Slides were incubated with various antibodies (diluted 1:75 for anti-cleaved caspase-3 and 1:50 for CD68 and CD3, in 1% bovine serum albumin [BSA]/PBS) for 1 to 2 hours at room temperature. Subsequently, slides were washed in PBS (three times) and incubated for 30 minutes at room temperature with either goat anti-rabbit IgG-HRP (for cleaved caspase-3 staining) or rabbit anti-mouse IgG-HRP (for CD68 and CD3 staining) (1:50 in 1% BSA/PBS) and then washed again (three times) in PBS followed by incubation with rabbit anti-goat IgG-HRP or goat anti-rabbit IgG-HRP, respectively, for another 30 minutes at room temperature. After washing in PBS (three times), slides were incubated in DAB solution for 15 to 20 minutes and subsequently washed with distilled water (five times). Slides were then counterstained with haematoxylin for 1 minute, washed in distilled water (five times), dehydrated in 96% ethanol and subsequently in 100% ethanol, and then mounted. Using Leica QWin software (Leica Microsystems, Cambridge, UK), morphometry was performed on entire skin tissue sections stained with antibodies against CD68 or CD3. Epidermal and papillary dermal layers (approximately 150 μm of dermal layer directly localised beneath the epidermis) were assessed separately by manually drawing a line around these layers under ×100 magnification.

Using Olympus Soft Pro software (Tokyo, Japan), the surface area of the epidermis was determined by manually drawing a line around this area and calculating the total surface (mm²). Subsequently, numbers of SBCs and nuclear dust were scored in three sequential H&E-stained sections. Nuclear dust was defined as one whole pyknotic nucleus or a group of pyknotic nuclear fragments (as indicated in Figure 1f by black and white arrows, respectively). Numbers of SBCs or extent of nuclear dust per square millimeter was determined by dividing the counted numbers by the epidermal surface area and calculating the mean value of the three sections. Cleaved caspase-3-positive cells were scored accordingly.

Infiltrating cells in the dermis were scored semi-quantitatively and by morphometry. H&E sections were semi-quantitatively scored for the presence of infiltrating cells using a score from 0 to 5. In short, vessels in the papillary dermis were scored blindly for the presence of perivascular infiltrating cells, in three consecutive sections: no infiltrating cells (0), not more than two infiltrating cells (1), not more than one perivascular layer of infiltrating cells (2), two or three layers of infiltrating cells (3), more than three layers of infiltrating cells (4), and more than three layers of infiltrating cells in combination with clear progression outside the perivascular region (5). The final score was determined by averaging the mean vessel score of three consecutive sections. Morphometry was performed on the dermal layer of CD3- and CD68-stained sections. Infiltrates in patients were considered present when the semi-quantitative scores of dermal influx of infiltrating cells in H&E-stained sections and the morphometric scores of either CD3- or CD68-stained OK sections were increased (> mean + two SDs of controls) in biopsies taken on at least two different days, after UVB irradiation.

Additionally, H&E-stained sections from biopsies taken before and after irradiation and biopsies of LE skin lesions were assessed for the presence of inflammatory lesions and co-localisation with SBCs. Inflammatory lesions were defined as the presence of category 5 (see above) vessel(s) in the dermis, with inflammatory cell infiltration of the epidermal layer coinciding with marked local hydropic degeneration of the basal layer of the epidermis (Figure 2h). Co-localisation was regarded as present when inflammatory lesions coincided with the presence of two or more SBCs in the epidermal layer lying over the infiltrate. Scoring was performed by two independent observers not informed about the clinical findings.

Sequential frozen sections were used for direct immunofluorescent staining of IgM, IgG, IgA, C3c, and C1q, using standard procedures. In brief, sections were washed with PBS and subsequently incubated with the various FITC-labeled specific antibodies at the dilutions indicated. After incubation, sections were washed in PBS again, and nuclei were stained using bisbenzimide (SERVA Electrophoresis GmbH, Heidelberg, Germany) and mounted. Sections were scored by two independent observers for staining at the dermal-epidermal junction (lupus-band) and in the epidermal layer.

Differences between groups were determined using the Mann-Whitney test. The χ² test was used to analyse categorical variables. Comparison of multiple groups was performed by one-way analysis of variance (Kruskal-Wallis). Correlations between numbers of apoptotic cells detected by H&E staining and numbers detected by cleaved caspase-3 staining and between SBCs and pyknotic nuclear debris were analysed using the nonparametric (Spearman) correlation test. To analyse differences in the level of correlation between patients and controls, slopes of linear regression lines were compared (GraphPad Software 3.02; GraphPad Software, Inc., San Diego, CA, USA).

To investigate apoptotic cell clearance, apoptotic cells were quantified in H&E staining by their altered morphology as SBCs and by the detection of cleaved caspase-3 as a specific apoptotic marker. Significant aspecific staining was observed in basal and spinotic epidermal layers and, to a lesser extent, in the dermis of non-irradiated skin using the TUNEL (terminal deoxynucleotidyl transferase-mediated in situ nick-end labeling) detection method (data not shown). However, results from the two detection methods indicated above did highly correlate, r = 0.91, p < 0.0001. One day after irradiation, apoptotic cells were localised mainly in the stratum spinosum (Figure 1a). After 3 days, approximately 70% of SBCs were localised in the stratum granulosum (Figure 1b), and after 10 days, nearly all apoptotic cells were removed from the skin, partially by shedding (Figure 1c). No SBCs or cleaved caspase-3-positive cells were detected in unexposed skin (Figure 1d,e). After 1 day, SBCs could be detected in patients (88.8 ± 51.8 SBCs per mm² [mean ± SD]) and controls (101.9 ± 68.6 SBCs per mm², p = 0.71). At day 3, the number of apoptotic cells was increased three- to nine-fold in patients (358.2 ± 138.8 SBCs per mm²) and controls (321.8 ± 127.3 SBCs per mm², p = 0.42). Ten days after irradiation, patients (11.0 ± 11.6 SBCs per mm²) and controls (6.0 ± 5.6 SBCs per mm²) had decreased, but similar, numbers of apoptotic cells, which resided in the epidermis (p = 0.41).

Nuclear dust, defined as one whole pyknotic nucleus or a group of pyknotic nuclear fragments (Figure 1f), could be detected after irradiation. The extent of nuclear dust strongly correlated with numbers of SBCs (r = 0.91, p < 0.0001). Clearance rate of nuclear dust was comparable with clearance of apoptotic cells and did not differ between patients with SLE and controls (data not shown).

To study the inflammatory response induced by a single dose of UVB irradiation, skin biopsies were taken after 1, 3, and 10 days and stained with H&E. In general, in skin from control subjects, some influx of inflammatory cells was seen after 1 day, decreasing over time with only low influx remaining after 10 days compared with non-irradiated skin. Influx was localised mainly around the dermal blood vessels (Figure 2a–c,g). In five out of 15 patients, influx of cells was increased, especially after 3 days, and persisted after 10 days (Figure 2d–g). In two of these patients, the infiltrate progressed toward the basal layer of the epidermis, which was damaged as indicated by the presence of marked hydropic degeneration. Based on pre-defined criteria, these were considered inflammatory lesions (Materials and methods). Inflammatory lesions were localised only in the vicinity of apoptotic keratinocytes (Figure 2h).

To determine whether the inflammatory lesions seen in the vicinity of SBCs after irradiation might also be present in established LE lesions, biopsies of 20 LE skin lesions were assessed for co-localisation of infiltrate and SBCs (Table 2). In 16 out of 20 LE biopsies, SBCs could be detected, and in 10 of these biopsies, co-localisation of inflammatory lesions and local accumulation of SBCs was seen (Figure 3). The latter group of patients could not be distinguished from the other patients by any of the characteristics depicted in Tables 1 and 2.

To characterise and quantify the infiltrating cells, sections were stained using a T-cell (CD3) and monocyte/macrophage (CD68) marker. Subsequently, staining in the papillary dermis and epidermis was quantified by morphometry. In the dermis of non-irradiated skin, low numbers of T cells (0.25% ± 0.20% in patients versus 0.27% ± 0.17% in controls) and macrophages (0.69% ± 0.45% in patients versus 0.70% ± 0.33% in controls) were present. Influx of T cells and macrophages increased in all subjects 1 day after irradiation, declined slowly, and was only slightly increased after 10 days compared with non-irradiated skin (Figure 4a,b). Macrophages were increased in the skin of patients with SLE after 1 day as compared with controls (1.19% ± 0.55% and 0.66% ± 0.35%, respectively, p = 0.02). T-cell influx was not significantly different at any time point between patients and controls. Neutrophils were accidentally (one to five cells per 4 mm section) detected in H&E staining in both controls and patients (data not shown). Five patients developed infiltrates of inflammatory cells detected by semi-quantitative analysis in H&E-stained sections and by morphometric analysis in CD3- or CD68-stained sections on at least two different days after irradiation. This group of patients could not be distinguished from the other patients by any of the patient characteristics listed in Tables 1 and 2. In the patients who developed infiltrates, levels of T-cell and macrophage influx were in the same range as seen in 20 skin biopsies from patients with cutaneous LE lesions (Figure 4). This group of patients with cutaneous LE lesions could not be distinguished from the patients who got UVB irradiation to the skin by any of the patient characteristics listed in Tables 1 and 2.

Influx of T cells and macrophages was not limited to the dermal layer of the skin. T cells and, especially, macrophages were detected in the epidermis as well. In the epidermis of non-irradiated skin, almost no T cells (0.010% ± 0.011% in patients, 0.028% ± 0.023% in controls) (Figure 4c) or macrophages (0.026% ± 0.047% in patients, 0.0005% ± 0.0008% in controls) (Figure 4e) could be detected. However, 3 days after irradiation, a substantial number of macrophages were observed in the epidermal compartment. Influx into the epidermis was higher in patients (0.038% ± 0.074% for CD3, p = 0.02, and 0.26 ± 0.31 for CD68, p = 0.04) compared with controls (0.0023% ± 0.0057% for CD3 and 0.061% ± 0.046% for CD68). Furthermore, epidermal influx was higher in patients with infiltrates compared with patients without these infiltrates and controls (p = 0.0009 for CD3-positive T cells and p = 0.009 for CD68-positive macrophages) (Figure 4d,f).

Deposition of Igs and complement factors was studied to assess their potential involvement in the inflammatory response. Most intense staining of Igs and complement in the epidermal layer was seen near local accumulations of epidermal apoptotic cells. However, depositions were not restricted to patients and could also be detected in all healthy controls (data not shown).

Numbers of apoptotic cells were compared between patients with and without infiltrates to evaluate whether differences in clearance rate of apoptotic cells might have been responsible for the development of infiltrates. Patients with infiltrates in the skin did not have increased numbers of apoptotic cells at any time point compared with patients without infiltrates and controls (Figure 5a). Also, extent of nuclear dust did not differ at any time point between patients with infiltrates and the remaining patients without these infiltrates and controls (data not shown).

Macrophages in the epidermis were often localised in the vicinity of apoptotic cells. Only in two patients with infiltrates, a large proportion of the epidermal macrophages contained multiple large vacuoles (Figure 5b). The morphology of these vacuoles indicated ingestion of apoptotic cells. This was confirmed by counterstaining with H&E which showed that the apoptotic bodies ingested by macrophages had an eosinophilic stained cytoplasm (Figure 5c).