Increased expression of the novel proinflammatory cytokine high mobility group box chromosomal protein 1 in skin lesions of patients with lupus erythematosus




To investigate the role of the novel cytokine high mobility group box chromosomal protein 1 (HMGB-1) in the pathogenesis of cutaneous lupus erythematosus (CLE).


Punch biopsy specimens of lesional and unaffected skin from 10 patients with CLE and 3 healthy control subjects were investigated. Immunohistochemical staining for HMGB-1, tumor necrosis factor α (TNFα), and interleukin-1β (IL-1β) was performed on consecutive sections. Analysis of single-nucleotide polymorphisms of −308 TNF was performed on DNA extracted from peripheral blood mononuclear cells.


An altered expression of HMGB-1 was observed both in the epidermis and in the dermal infiltrates of lesional skin. Expression of HMGB-1 in the epidermis and dermis was increased (P < 0.01 and P < 0.001, respectively, versus unaffected skin), and translocation to the cytoplasm as well as the extracellular presence of secreted HMGB-1 were found. Increased levels of TNFα and IL-1β were also observed in the dermal infiltrates of lesional skin (P < 0.01 and P < 0.05, respectively, versus unaffected skin). The carrier frequency of the −308A TNF polymorphism was 80% in patients with subacute CLE, but this was not related to higher expression of TNFα in biopsy specimens from the CLE group.


The high amount of extracellular HMGB-1 observed in skin lesions indicates that HMGB-1 is involved in the inflammatory process of CLE. TNFα and IL-1β may form a proinflammatory loop with HMGB-1, since they can induce the release of each other. The extracellular HMGB-1 observed by immunostaining of the epidermis indicates that keratinocytes may be an as yet unrecognized source of secreted HMGB-1, underscoring the role of the target organ in the rheumatoid autoimmune inflammatory process.

The pathogenesis of cutaneous lupus erythematosus (CLE) remains to be fully understood, but ultraviolet (UV) radiation can trigger this autoimmune reaction in the skin. The different subsets of CLE include the acute, subacute, and chronic forms, and an individual patient may have more than 1 subset of CLE according to the currently accepted classifications (1). The lesions of subacute CLE have papulosquamous or annular features and are most commonly distributed on sun-exposed areas of the skin, whereas the dermal manifestations of chronic CLE most often consist of coin-shaped scarring plaques, referred to as discoid lupus erythematosus (DLE). DLE lesions can, however, also occur in the skin of patients with systemic lupus erythematosus (SLE) and patients with subacute CLE (1). Subacute CLE is also frequently associated with the serologic presence of Ro/SSA autoantibodies (2–4).

The diagnosis of different forms of CLE is confirmed histologically by the presence of epidermal hyperkeratosis, follicular plugging, and hydropic degeneration of the basal cell layer and a dermal lichenoid or patchy mononuclear infiltrate with perifollicular accentuation. These pathologic changes appear to a different degree in subacute CLE and DLE, and the conditions are difficult to differentiate with only histologic findings (5). The inflammatory cells in CLE lesions have been reported to be predominantly CD3+ T cells, and more CD4+ T cells than CD8+ T cells are present (6). During autoimmune inflammation, pro- and antiinflammatory cytokines are produced. Tumor necrosis factor α (TNFα) and interleukin-1 (IL-1) are proinflammatory cytokines of central importance in several autoimmune conditions and can also be induced by UV radiation (7, 8). TNFα is predominantly synthesized by macrophage/monocytes, although keratinocytes and mast cells also exhibit the capacity to release TNFα in response to UVB (9). Both TNFα and IL-1β have been shown to stimulate the release of the high mobility group box chromosomal protein 1 (HMGB-1).

HMGB-1 is a structural nuclear protein that binds DNA and is involved in the organization of chromatin (10). As such, it is ubiquitously present in the nuclei of mammalian cells and is highly conserved between species. However, HMGB-1 was recently found to act as a proinflammatory cytokine in both acute and chronic inflammatory conditions, such as septic shock, acute lung injury, rheumatoid arthritis, and myositis (11), and to be actively secreted by macrophage/monocytes via inflammatory stimuli (12). During its secretion, HMGB-1 exits the nucleus, is transported through the cytoplasm, and is actively released to the extracellular space (13). HMGB-1 can also be passively released from the nuclei of necrotic or damaged cells, whereas cells undergoing apoptosis are poor HMGB-1 secreters (14). The receptor for advanced glycation end products has been demonstrated to be the most functional receptor for HMGB-1–mediated production of cytokines (15), including synthesis of TNFα and IL-1β (11).

To investigate the role of HMGB-1 in the cutaneous manifestations of lupus, we analyzed the expression and subcellular localization of HMGB-1, TNFα, and IL-1β in punch biopsy skin specimens from Ro/SSA-positive patients with subacute CLE and DLE lesions in affected skin, as compared with the unaffected skin of the patients and of healthy controls. TNF single-nucleotide polymorphism (SNP) analysis was also carried out in these patients to examine the frequency of carriers of the −308A TNF allele, which is suggested to be associated with subacute CLE and increased production of TNFα (16).


Patients' skin samples.

Ten patients with CLE (7 women and 3 men) were consecutively included in the study on the basis of having Ro/SSA autoantibodies, being photosensitive, and displaying active skin lesions during clinical examination. Classification of photosensitivity was based on a review of the patient's medical history. Three-millimeter punch biopsy samples of sun-exposed, affected skin (defined as skin demonstrating subacute CLE lesions [n = 6] or DLE lesions [n = 4]) and non–sun-exposed, unaffected buttock skin of the same patients were obtained. The definitions of subacute CLE and DLE lesions were based on clinical and histopathologic findings. Patients were diagnosed as having subacute CLE (n = 6), SLE (n = 3), or chronic CLE (n = 1). Four of the patients with subacute CLE also fulfilled the criteria for SLE (17). Biopsy samples of the buttock skin of 3 healthy female volunteers were also included as controls. The patients' demographic and clinical characteristics are shown in Table 1.

Table 1. Demographic and clinical characteristics of the patients with cutaneous lupus erythematosus (CLE) and healthy controls
Subject/ age/sexCLE typeOther diagnoses*Therapy for CLEAntinuclear antibodies−308 TNF genotypes
  • *

    SLE = systemic lupus erythematosus; DLE = discoid lupus erythematosus; SS = Sjögren's syndrome; RA = rheumatoid arthritis.

 1/66/FDiscoidOperable breast cancer GG
 2/65/FSubacuteSLE, DLE, hypertensionHydroxychloroquine, beta-carotene+GA
 3/58/FSubacute  GG
 4/50/MSubacuteSLE, SS, Raynaud's phenomenon, psoriatic arthritisCorticosteroids, hydroxychloroquine+GA
 6/37/FDiscoidSLE, RACorticosteroidsGG
 7/77/FSubacuteSS, hypothyroiditisCorticosteroids, hydroxychloroquineGA
 8/63/FSubacuteSLE, hypertension, hyperlipidemiaBeta-carotene+GA
 9/71/MDiscoidSLE, operable aortic aneurysm +GG
 10/76/MDiscoidSLE, vitamin B deficiency +GG
 C1/53/F None  
 C2/52/F None  
 C3/52/F Hypertension, asthma, allergy  

DNA was extracted from the peripheral blood mononuclear cells of the 10 patients with CLE. The DNA samples were analyzed for TNF SNPs according to a previously described procedure (18). The study was approved by the human ethics committee of the North region, and all subjects provided their informed consent.

Immunohistochemical staining technique.

The biopsy samples were snap-frozen on dry ice and stored at −70°C until sectioned (7 μm) in a cryostat. The sections were placed on chrome gelatin-coated slides and air dried for 30 minutes before fixation in 2% formaldehyde in phosphate buffered saline (PBS). The slides were permeabilized in PBS–Saponin for 10 minutes, and thereafter the endogenous peroxidase activity was blocked with H2O2 (1% H2O2, 2% NaN3, 0.1% Saponin in PBS) for 60 minutes in the dark. The slides were rinsed in PBS with 0.1% Saponin (3 times for 3 minutes each) between each new procedure. After the washing procedure, the slides were blocked for 15 minutes with 1% normal horse serum in PBS–Saponin and then blocked using an avidin–biotin blocking kit (catalog no. SP-2001; Vector, Burlingame, CA). Thereafter, a mouse monoclonal anti–HMGB-1 antibody (2G7, 0.625 μg/ml; Critical Therapeutics, Boston, MA), mouse anti-TNFα (3.33 μg/ml, catalog no. H86410M; Biosite Diagnostics, San Diego, CA), or mouse anti–IL-1β (1.67 μg/ml [catalog no. 211-44-531] and 8.33 μg/ml [catalog no. 211-44-131]; Immunocontact, Frankfurt, Germany) was added and incubated with the slides at room temperature overnight in a humidified chamber. Mouse IgG2b and IgG1 antibodies (0.625 μg/ml [catalog no. X0944] and 8.33 μg/ml [catalog no. X0931]; Dako, Carpinteria, CA) of irrelevant specificity were used as controls.

After rinsing, biotinylated horse anti-mouse IgG antibody (5 μg/ml, catalog no. BA-2001; Vector) in PBS–Saponin with 1% normal horse serum was added for 30 minutes. The slides were then treated with peroxidase-conjugated ExtrAvidin (catalog no. E-2886; Sigma, St. Louis, MO) for 45 minutes in the dark and, finally, developed with a diaminobenzidine kit (catalog no. SK-4100; Vector) for 10 minutes. The slides were counterstained with Mayer's hematoxylin and mounted using PBS–glycerol at a ratio of 1:9.

Evaluation of stained sections.

The stained slides were coded and analyzed independently by 2 individuals (KP and ME) in a blinded manner, using a semiquantitative method. The entire section of the biopsy sample for each staining was analyzed in 2 different ways, using a Polyvar II microscope (Reichert-Jung, Vienna, Austria). First, for the evaluation of the degree of cytokine expression, the section was divided into 3 different parts: the epidermis, dermal infiltrate, and dermis (noninfiltrate). The amount of positively stained cells in each part was scored on masked sections by the 2 independent observers. To determine the cellular distribution of HMGB-1, nuclear, cytoplasmic, and extracellular staining was estimated as a percentage of the total staining in each area of the section (epidermis, dermis, and dermal infiltrate), with the summarized total staining in each area equal to 100%. This scoring was also done manually by the 2 observers in a blinded manner. The individual value of increment was 5%, and the mean ± SEM difference in observers' scores was 10.7 ± 5.4 procentual units. A correlation was found between the 2 observers' scores (r = 0.74, P < 0.005). Results are expressed as the mean values of the evaluations from the 2 observers. When evaluations of a single staining differed by >40%, the 2 observers reassessed the staining together.

Statistical analysis.

The nonparametric Mann-Whitney U test was used for comparisons between groups. P values less than 0.05 were considered significant. Correlations were calculated by Spearman's rank correlation test.


Increased cytoplasmic expression and extracellular deposition of HMGB-1 in cutaneous lupus.

To investigate the role of HMGB-1 in CLE, punch biopsy skin samples of clinically typical, spontaneously occurring CLE lesions were obtained from 10 Ro/SSA-positive patients (Table 1), and unaffected buttock skin was obtained from the same individuals. Skin biopsy samples from 3 healthy volunteers were used as controls.

Nuclear HMGB-1 was expressed in both the affected and the unaffected skin specimens from the patients, as well as in unaffected skin from healthy controls (Figures 1A–C and Table 2). Control staining with irrelevant isotype-matched control antibody was negative (results not shown). A consistently higher degree of cellular expression of HMGB-1 was observed in the dermis and epidermis of lesional skin compared with the unaffected buttock skin of the same patient (P < 0.001 and P < 0.01, respectively) (Figure 1D) and compared with healthy control skin. Infiltrates of mononuclear cells dominated the skin lesions and, within the infiltrates, high HMGB-1 expression was noted (Figure 1D). In the noninfiltrated part of the dermis, the expression of HMGB-1 was low and the levels were similar to those in corresponding areas of healthy buttock skin.

Figure 1.

Immunohistochemical staining for high mobility group box chromosomal protein 1 (HMGB-1) in cutaneous lupus erythematosus (CLE) lesions. Staining for HMGB-1 in the lesional skin of a patient with subacute CLE (A) was compared with the nonlesional buttock skin of the same patient (B) and with the buttock skin of a healthy control subject (C) (original magnification × 10; scale bars = 100 μm). The results of semiquantitative analysis of cells staining positive for HMGB-1 are represented as box plots (D) for epidermal lesions versus the unaffected epidermis of the same patients (n = 10) (P < 0.01), and for dermal infiltrates of affected skin versus the unaffected dermis of the same patients (n = 10) (P < 0.001).

Table 2. Mean percentage of cells staining positive for HMGB-1 and cytokine expression, and results of DNA analysis for the −308 TNF GG and GA genotypes in skin specimens*
CytokineLE lesional skinUnaffected skinHealthy control skin
EpidermisDermal infiltrateDermisEpidermisDermisEpidermisDermis
  • *

    HMGB-1 = high mobility group box chromosomal protein 1; LE = lupus erythematosus; IL-1β = interleukin-1β; TNFα = tumor necrosis factor α.

−308 TNF       

In the unaffected buttock skin of the patients and healthy control skin, HMGB-1 was expressed mainly in the epidermis (Figure 1). The intracellular localization of HMGB-1 was predominantly cytoplasmic in all patients, both in the dermis and in the epidermis (Figures 2A–C). However, translocation of HMGB-1 to the extracellular space was detected almost exclusively in the lesional dermis biopsy samples from lupus skin (Figure 2D), and this difference between lesional and nonlesional skin was highly significant (P < 0.001). Staining without obvious restriction by cell membranes, suggestive of the extracellular presence of HMGB-1, was also detected almost exclusively in the lesional epidermis biopsy samples from lupus skin (Figure 2D). In healthy control subjects, no extracellular staining for HMGB-1 was observed.

Figure 2.

Increased extracellular expression of HMGB-1 in CLE lesions. Lesional skin from a patient with subacute CLE showed extracellular staining of HMGB-1 (arrow) (A), whereas nonlesional buttock skin from the same patient (B) and buttock skin from a healthy control (C) showed mainly cytoplasmic (black arrow) and nuclear (white arrow) staining (original magnification × 25; scale bars = 50 μm). The results of semiquantitative analysis of cells staining positive for extracellular HMGB-1 are represented as box plots (D) for the epidermis from lesional versus unaffected skin (n = 10), and for dermal infiltrates of affected skin versus unaffected dermis (n = 10) (P < 0.001 for both). See Figure 1 for definitions.

TNFα and IL-1β coexpression in areas of extracellular HMGB-1.

TNFα was detected in the skin biopsy samples from all subjects, but the degree of expression in the infiltrates of lesional dermis was higher than that in nonlesional dermis from the same patient (P < 0.01) (Table 2). The localization of TNFα was mainly intracellular in the nonlesional dermis and epidermis of all patients, but in the lesional infiltrates of the dermis, secreted TNFα was expressed to almost the same degree as that of intracellular TNFα (Figures 3A and B).

Figure 3.

Increased expression of tumor necrosis factor α (TNFα) (A and B) and interleukin-1β (IL-1β) (C and D) in lupus erythematosus lesions compared with unaffected skin. Expression of TNFα in lesional (A) and nonlesional (B) buttock skin of the same patient. Expression of IL-1β in lesional (C) and nonlesional (D) buttock skin of the same patient. (Original magnification × 25; scale bars = 50 μm.)

IL-1β was expressed in both affected and unaffected skin specimens (Figures 3C and D), with the most intense staining in the epidermis. When comparing IL-1β expression in lesions with IL-β expression in unaffected buttock skin, an increased expression in dermal infiltrates was observed (P < 0.05). The localization of IL-1β was mainly intracellular in all subjects, both in the dermis and in the epidermis. Secreted IL-1β, however, was observed in only the dermal infiltrates of lesions (Figures 3C and D). Control staining with irrelevant isotype-matched control antibody was negative (results not shown).

Analysis of TNF SNPs.

DNA was extracted from the peripheral blood mononuclear cells of the investigated patients and analyzed for the previously defined −308 TNF SNPs. Five patients (50%) had a GG genotype, while the carrier frequency of the A allele was 50% (Table 1). Patients carrying the A allele did not show a higher expression of TNFα in either the affected or the unaffected skin when compared with that in the patients with the GG genotype (Table 2). All 5 patients carrying the GA allele had subacute CLE with either SLE and/or Sjögren's syndrome, and accordingly, 5 of the 6 patients with subacute CLE carried an A allele.


Cutaneous lupus erythematosus is the most common form of lupus, and the common involvement of the skin in lupus is reflected by mucocutaneous symptoms that constitute 4 of the 11 American College of Rheumatology criteria for SLE (17). Although lupus is a heterogeneous disease that may affect any organ of the body, study of the pathogenesis in skin biopsy specimens is an attractive model because it offers direct access to affected tissue. The appearance of lesions is commonly triggered by UV radiation, which induces the production of TNFα. TNF-based approaches to treatment, however, do not seem to work efficiently in all patients with CLE, and have also been demonstrated to cause flares in a few cases (19–21). To investigate other potential factors of importance in activation of the target T cells in lupus, we studied the expression and release of the proinflammatory cytokine HMGB-1 in skin biopsy samples from patients with subacute CLE and DLE lesions.

In the present study we found that both keratinocytes and dermal mononuclear inflammatory cells in the skin samples from patients with CLE exhibited an increased amount of cytoplasmic and suggestive extracellular secreted HMGB-1 compared with healthy buttock skin of the same patients. The extracellular HMGB-1 staining thus indicates release of cytoplasmic HMGB-1 either from activated macrophages or from necrotic cells. Since cell death from apoptosis and not necrosis is typically seen in CLE, one may conclude that the extracellular HMGB-1 observed in our lupus patients was secreted from activated inflammatory cells. However, it is also possible that the keratinocytes released HMGB-1 and may thus constitute a novel source contributing to the extracellular pool of this proinflammatory cytokine.

A biallelic polymorphism at position −308 within the human TNF promoter region has been described in subacute CLE (16) and related to increased TNFα production after UVB and IL-1 exposure of 3T3 fibroblasts transiently transfected with the −308A and −308G TNF promoter region in a reporter assay system (16). In our study, none of the patients had the rare AA genotype, although 50% carried the −308A allele and as many as 80% (5 of 6) of the included patients with subacute CLE carried an A allele. No increased TNFα expression, however, was noted in the A allele–carrying patients compared with the other patients, although increased TNFα was observed in all lesions compared with unaffected skin. This may reflect a difference between in vivo and in vitro conditions or may be a result of different lapsed times between the triggering event and analysis, since the lesions analyzed in our study were manifest, spontaneously occurring lesions. UV radiation causes the release of TNFα and IL-1 from keratinocytes (7, 8). Both TNFα and IL-1β can induce secretion of HMGB-1, which in turn can stimulate the synthesis of TNFα and IL-1β. Accordingly, while UV radiation may initiate formation of the lesions, HMGB-1 may appear at a later stage and be of importance in sustaining inflammation, leading to a more chronic disease.

In conclusion, this study offers novel insight into the pathogenesis of CLE. Our study is, to the best of our knowledge, the first to describe the expression of the cytokine HMGB-1 in lesions of patients with CLE. Our results suggest that HMGB-1 plays an important role in the pathogenesis of CLE and constitutes a potential future therapeutic target.