Differential Expression of Cytokine mRNA in Skin Specimens from Patients with Erythema Migrans or Acrodermatitis Chronica Atrophicans

Differential Expression of Cytokine mRNA in Skin Specimens from Patients with Erythema Migrans or Acrodermatitis Chronica Atrophicans Robert R. MuÈllegger,*² Gail McHugh,* Robin Ruthazer,³ Barbara Binder,² Helmut Kerl,² and Allen C. Steere*

Divisions of *Rheumatology/Immunology and ³Clinical Care Research, Tufts University School of Medicine, New England Medical Center, Boston, Massachusetts, U.S.A.; ²Department of Dermatology, Karl-Franzens University School of Medicine, Graz, Austria

Erythema migrans, the characteristic skin manifestation of acute Lyme borreliosis, is a self-limited lesion. In contrast, acrodermatitis chronica atrophicans, the typical cutaneous manifestation of late Lyme borreliosis, is a chronic skin condition. In an effort to understand pathogenic factors that lead to different outcomes in dermatoborrelioses, skin biopsy samples from 42 patients with erythema migrans and 27 patients with acrodermatitis chronica atrophicans were analyzed for mRNA expression of ®ve pro-in¯ammatory cytokines (tumor necrosis factor a, interleukin-1b, interleukin-6, interferon-g, and interleukin-2) and two anti-in¯ammatory cytokines (interleukin-4 and interleukin-10) by in situ hybridization with cytokine-speci®c riboprobes. Among the 27 patients who had erythema migrans alone with no associated signs or symptoms, the major cytokines expressed in perivascular in®ltrates of T cells and macrophages were the pro-in¯ammatory cytokine interferon-g and the anti-in¯ammatory cytokine interleukin-10. In the 15 erythema migrans patients who had associated signs and symp-

L

yme borreliosis (LB) is a complex multisystem disease caused by infection with the tick-borne spirochete Borrelia burgdorferi sensu lato (Steere et al, 1983a). Three pathogenic borrelial genospecies, B. burgdorferi sensu stricto, B. afzelii, and B. garinii, are known to cause human LB (Baranton et al, 1992). To date, all North American strains have belonged to the ®rst group, B. burgdorferi sensu stricto. All three genospecies have been found in Europe, but most isolates there have been B. afzelii or B. garinii strains. These differences may account for certain regional variations in the clinical picture of the infection (Anthonissen et al, 1994; Picken et al, 1998). Manuscript received October 12, 1999; revised July 12, 2000; accepted for publication September 26, 2000. Reprint requests to: Dr. Robert R. MuÈllegger, Department of Dermatology, Karl-Franzens University School of Medicine, Auenbruggerplatz 8, A-8036 Graz, Austria. Email: robert.muellegger@kfunigraz. ac.at Abbreviations: ACA, acrodermatitis chronica atrophicans; EM, erythema migrans; IHC, immunohistochemistry; ISH, in situ hybridization; LB, Lyme borreliosis. 0022-202X/00/$15.00

toms, including headache, elevated temperature, arthralgias, myalgias, or fatigue, a larger number of macrophages and greater expression of macrophagederived pro-in¯ammatory cytokines, tumor necrosis factor a, interleukin-1b, and interleukin-6, were also found. In comparison, in®ltrates of T cells and macrophages in the skin lesions of acrodermatitis chronica atrophicans patients had very little or no interferon-g expression. Instead, they usually expressed only the pro-in¯ammatory cytokine tumor necrosis factor a and the anti-in¯ammatory cytokine interleukin-4. Thus, the activation of pro-in¯ammatory cytokines in erythema migrans lesions, particularly interferon-g, seems to be important in the control of the spirochetal infection. In contrast, the restricted pattern of cytokine expression in acrodermatitis chronica atrophicans, including the lack of interferon-g, may be less effective in spirochetal killing, resulting in the chronicity of this skin lesion. Key words: acrodermatitis chronica atrophicans/cytokines/ erythema migrans/in situ hybridization. J Invest Dermatol 115:1115±1123, 2000

Erythema migrans (EM), the characteristic expanding skin lesion of acute LB, is self-limited in weeks to months (Steere et al, 1983b; AÊsbrink and Olsson, 1985; AÊsbrink and Hovmark, 1988). In contrast, acrodermatitis chronica atrophicans (ACA), the typical skin manifestation of late LB, is a chronic and usually progressive skin condition that does not resolve spontaneously. It starts with an in¯ammatory phase and progresses slowly to an atrophic phase (AÊsbrink and Hovmark, 1988; Weber et al, 1993). In a previous study in Austria,1 as in all of Europe, B. afzelii is most often recovered from EM lesions, and ACA is almost always due to infection with this borrelial species (Wienecke et al, 1994; Picken et al, 1998; Strle et al, 1999). A variety of transient, nonspeci®c signs and symptoms may occur with European EM, including elevated temperature, lymphadenopathy, arthralgias, myalgias, or fatigue, but they usually do not suggest spread of the spirochete to speci®c organs (Weber and Neubert, 1986; AÊsbrink and Hovmark, 1988). 1Zo È chling N, MuÈllegger RR, SchluÈpen EM, et al: Molecular subtyping of Borrelia burgdorferi (Bb) in lesional skin from Stryrian (Austrian) patients with various manifestations of dermatoborreliosis (DB). J Invest Dermatol 104:687, 1995 (abstr.)

´ Copyright # 2000 by The Society for Investigative Dermatology, Inc. 1115

1116

È LLEGGER ET AL MU

In contrast, in the U.S.A., where EM is caused by B. burgdorferi sensu stricto infection, faster spreading of skin lesions is frequently seen (Strle et al, 1999), often accompanied by signs and symptoms, such as headache and neck stiffness or multiple skin lesions, that suggest hematogenous dissemination of the spirochete to other organ systems (Steere et al, 1983b; Berger, 1984; Strle et al, 1999). In Europe, the most common extracutaneous complications of ACA are arthropathy and peripheral neuropathy restricted to the Ê sbrink and Hovmark, 1988), whereas in the U.S.A. affected limb (A ACA is almost never seen. Cytokines, which are important regulators and effectors of the immune response, in¯uence the outcome of infectious and autoimmune diseases. The importance of these responses in the outcome of B. burgdorferi infection has been shown to have a genetically determined component in inbred strains of mice. When infected experimentally with B. burgdorferi, C3H mice develop severe arthritis, whereas BALB/c mice develop only mild arthritis (Barthold et al, 1990; Schaible et al, 1991; Yang et al, 1994). These strain-related differences are associated with polarized cytokine patterns and differences in spirochete burden in infected tissues. When infected with the spirochete, C3H mice, which have a high spirochete burden (Yang et al, 1994), produce gradually increasing amounts of interferon-g (IFN-g) with low to absent interleukin-4 (IL-4) levels, a Th1 pattern of cytokine production (Keane-Myers and Nickell, 1995; Matyniak and Reiner, 1995; Kang et al, 1997). BALB/c mice initially have high levels of IFN-g, which probably accounts for their low spirochete burden, but they switch to the production of IL-4, a Th2 cytokine pattern (Keane-Myers and Nickell, 1995; Matyniak and Reiner, 1995; Kang et al, 1997), which is important in the resolution of joint in¯ammation. In patients with Lyme arthritis, cultured peripheral blood or synovial ¯uid mononuclear cells produced high amounts of IFN-g and tumor necrosis factor a (TNF-a), but little or no IL-4 (Yssel et al, 1991; Yin et al, 1997; Gross et al, 1998; Chen et al, 1999). The reactivity of T cells and the production of pro-in¯ammatory cytokines were greater in synovial ¯uid than in peripheral blood (Gross et al, 1998). Moreover, the severity of arthritis correlated directly with the ratio of Th1 to Th2 cells in synovial ¯uid, such that the higher the number of Th1 cells, the more severe the joint swelling (Gross et al, 1998). In the synovial membrane of Lyme arthritis patients, mRNA expression of pro-in¯ammatory cytokines, including IFN-g, TNF-a, and IL-1b, was predominant, but mRNA for anti-in¯ammatory cytokines was found as well (Yin et al, 1997; Harjacek et al, 2000). In neuroborreliosis, cultured mononuclear cells from the peripheral blood and cerebrospinal ¯uid produced primarily IFN-g, whereas IL-4 production was strikingly low (Wang et al, 1995; Oksi et al, 1996; Ekerfelt et al, 1997). Secretion of IFN-g was most pronounced in cerebrospinal ¯uid (Wang et al, 1995; Ekerfelt et al, 1997). Cytokine pro®les have not yet been determined in patients with skin manifestations of LB, however, including in those with EM or ACA. In an effort to understand pathogenic factors that lead to different outcomes in dermatoborrelioses, we used in situ hybridization (ISH) with cytokine-speci®c riboprobes to analyze cytokine responses in patients with EM, a self-limited skin lesion, compared to those with ACA, a chronic, progressive lesion that does not resolve spontaneously. MATERIALS AND METHODS Patients In 1997±8, we carried out a retrospective study of cytokine pro®les of lesional skin in patients with EM or ACA. The study subjects consisted of 42 patients with EM (median age 43.5 y; 29 males and 13 females) and 27 patients with ACA (median age 63 y; seven males and 20 females) seen between June 1993 and December 1996 at the Department of Dermatology in Graz, Austria, for whom biopsy specimens and complete clinical data were available. Patients who had been treated with antibiotics prior to sampling or patients with conditions (e.g., atopy) or medications (e.g., nonsteroidal anti-in¯ammatory drugs or corticosteroids) that could potentially in¯uence the status of the immune system were not included.

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

EM was de®ned clinically as an expanding, sharply demarcated erythema, sometimes with partial central clearing, of at least 5 cm in diameter (Stanek et al, 1996). Of the 42 study patients with EM, 16 (38%) had IgM and/or IgG antibody responses to B. burgdorferi, using puri®ed, native ¯agella of B. afzelii, strain DK-1, as the test antigen (Dako Lyme Borreliosis ELISA Kit; Dako, Glostrup, Denmark) (Hansen et al, 1988). In addition, in 14 of 29 patients tested (48%), B. burgdorferi DNA was detected in lesional skin, using nested polymerase chain reaction (PCR) with two oligonucleotide primer pairs, as previously described (Wienecke et al, 1993). Altogether, 34 of 42 patients (81%) with EM had serologic and/or PCR evidence of borrelial infection. The remaining eight patients with negative laboratory results had the characteristic clinical appearance of EM. Speciation of B. burgdorferi sensu lato subtypes was not done in the PCRpositive EM patients. ACA was de®ned clinically as a doughy bluish-red swelling, with or without signs of atrophy, on the extensor surfaces of the extremities (Stanek et al, 1996). All 27 study patients (100%) with ACA had a positive serologic test result for IgG antibodies to B. burgdorferi. PCR testing was not done in these patients. Preparation of skin specimens Four millimeter punch biopsy specimens were taken from all patients from the active border of EM lesions or from the area of ACA lesions with the most prominent signs of in¯ammation. For comparison, we used 4 mm punch biopsy specimens obtained from seven EM patients from clinically uninvolved skin in the area opposite the site of the EM lesion. All samples were placed immediately in 4% formaldehyde/phosphate-buffered saline (PBS) and ®xed overnight, dehydrated in graded ethanol, and embedded in paraf®n. From each sample, serial sections of 4 mm were cut for histopathology and serial sections of 5 mm for immunohistochemistry (IHC) and ISH. Histopathology and IHC In all patients, serial sections were stained with hematoxylin and eosin for histopathologic examination. In a subset of patients with EM or ACA in whom enough tissue was available, the expression of leukocyte differentiation antigens CD3 (T cells), CD20 (B cells), and CD68 (macrophages) was determined, using the avidin-biotinperoxidase complex method (Vectastain ABC Elite Kit; Vector Laboratories, Burlingame, CA). For IHC, tissue sections mounted on Superfrost Plus slides (Fisher Scienti®c, Pittsburgh, PA) were deparaf®nized with xylene and rehydrated through a decreasing ethanol series. Enhancement of immunoreactivity was achieved by microwave heating of the sections in 10 mM sodium citrate (pH 6.0). For CD3 and CD68, additional pretreatment with proteinase K (5 mg per ml) for 20 min was performed, followed by immersion in 4% paraformaldehyde/PBS. Nonspeci®c binding was blocked using normal horse serum diluted 1:100 in PBS for 20 min. Sections were then incubated with a primary polyclonal rabbit antibody to CD3 (1:10 dilution in 2% bovine serum albumin/PBS) or monoclonal mouse antibodies to CD20 (1:100) or CD68 (1:100) for 30 min (Dako USA, Carpinteria, CA). Binding was detected with the universal biotinylated horse antimouse/rabbit IgG secondary antibody (Vector) diluted 1:50 in PBS containing normal horse serum for 30 min. The avidin-biotin-horseradish peroxidase complex (1:250 dilution) was then applied for 30 min. Sections were developed for 5 min with 3,3¢diaminobenzidine tetrahydrochloride as chromogenic substrate in H2O2, counterstained with hematoxylin, and mounted in an aqueous medium. As negative controls, nonspeci®c rabbit or mouse IgG antibodies (Vector) were used instead of the primary antibodies. As positive controls, the respective primary antibodies were applied to tissue sections of in¯amed tonsils (a gift from Dr. P. Kwan, Department of Pathology, Tufts University, Boston, MA). All controls yielded the expected negative or positive results. Evaluation of IHC was performed by one blind observer using a Zeiss light microscope (Carl Zeiss, Thornwood, NY). The entire tissue section was examined under 4003 magni®cation. The percentage of cells positive for a respective leukocyte marker was estimated relative to the total number of in®ltrating cells. Positivity was graded on a scale from 0 to 3 as follows: 0, all in®ltrating cells negative; 1, up to a third of all cells positive; 2, between one-third and two-thirds of all cells positive; 3, more than two-thirds of all cells positive. RNA probes In each patient, the expression of mRNA for various proin¯ammatory (TNF-a, IL-1b, IL-6, IFN-g, IL-2) and anti-in¯ammatory (IL-4, IL-10) cytokines was determined in skin biopsy samples by ISH with cytokine-speci®c riboprobes. To generate the speci®c antisense and sense probes, coding region cDNA fragments, which were kindly provided by Dr. P. Bucy (University of Alabama, Birmingham, AL), were subcloned into pBluescript vectors (Stratagene, La Jolla, CA). The cDNA fragments

VOL. 115, NO. 6 DECEMBER 2000

correspond to the following nucleotides of the respective human full-length cytokine sequences: TNF-a, 344±917 (574 bp); IL-1b, 445±1055 (611 bp); IL-6, 19±1128 (1110 bp); IFN-g, 220±1193 (974 bp); IL-2, 145±800 (656 bp); IL-4, 64±518 (455 bp); and IL10, 411±1410 (1000 bp). GenBank accession numbers for these sequences are as follows: TNF-a, M10988; IL1b, M15330; IL-6, M14584; IFN-g, X13274; IL-2, V00564; IL-4, M13982; IL10, M57627. The sequences of the probes were con®rmed (personal communication, Dr. P. Bucy, University of Alabama). For replication, cDNA-containing plasmids were transformed into competent Escherichia coli DH10B cells (Life Technologies, Gaithersburg, MD). Plasmid DNA was then puri®ed from transformants (Wizard Plus Minipreps DNA Puri®cation System; Promega, Madison, WI), digested with restriction endonucleases, and fractionated by agarose gel electrophoresis to screen for the presence of the appropriate insert. Following linearization of the plasmids with the appropriate restriction enzymes, transcription reactions were performed using T3 or T7 polymerases in the presence of digoxigenin-labeled UTP (SP6/T7 DIG RNA Labeling Kit; Boehringer Mannheim, Indianapolis, IN) to obtain antisense and sense riboprobes. The size of the transcribed RNA was tested by agarose gel electrophoresis. Labeling ef®ciency was determined on dot blots by comparing serial dilutions of the digoxigenin-labeled probe with a prelabeled control using chemiluminescent detection with antidigoxigenin Fab fragments conjugated to alkaline phosphatase, according to the manufacturer's protocol (Boehringer). ISH ISH was performed with modi®cations of previously described methods (Lan et al, 1996). Serial tissue sections were mounted on silanecoated Superfrost Plus slides (Fisher) and dried for 2 h at 60°C. Sections were deparaf®nized with xylene and rehydrated through graded ethanol. Following treatment with 0.1 M glycine and permeabilization with 0.3% Triton X-100, the slides were placed in 10 mM sodium citrate (pH 6.0) and brought to the boil three times in a microwave oven (600 W). Sections were then incubated with proteinase K (5 mg per ml) for 20 min and immersed in 4% paraformaldehyde/PBS for 5 min, followed by ethanol dehydration. Thereafter, 25 ml of hybridization solution (40% formamide, 2 3 sodium citrate/chloride buffer, 1 3 Denhardt's solution, 10% dextran sulfate, 0.01 M dithiothreitol, 1 mg yeast tRNA per ml, and 1 mg sheared, denatured salmon sperm DNA per ml) containing 5 ng of the respective heat-denatured antisense or sense riboprobe were applied to each tissue section and covered with a cover slip. Hybridization was carried out overnight at 42°C in a microheating system with a moist environment (TruTemp; Robbins Scienti®c, Sunnyvale, CA). All steps were carried out with precautions to prevent RNAse degradation, including use of separate glassware and autoclaved solutions prepared in diethylpyrocarbonatetreated water. After hybridization, sections were washed under stringent conditions including treatment with 20 mg per ml RNAse A (Boehringer). For detection of hybridized probes, sections were blocked with normal horse serum and incubated with a biotinylated monoclonal mouse antidigoxin antibody (1:1000 dilution) (Sigma, St. Louis, MO) for 60 min at 37°C. The next incubation was in avidin-biotin-peroxidase complex (Vector), followed by the substrate, diaminobenzidine-H2O2. Sections were counterstained with hematoxylin and mounted in an aqueous medium. To document hybridization speci®city, the following control experiments were performed. Duplicate sections on the same slide were always stained with antisense or sense probes for the respective cytokine mRNA. Staining with the sense probes gave only uniform background reactions (Fig 4f). Application of hybridization buffer with no probe yielded the expected negative results (Fig 1a). Investigation of the seven samples from normal skin disclosed a minimal constitutive production of cytokines (Fig 1b). As positive controls, tissue sections of in¯amed tonsils (a gift from Dr. P. Kwan, Department of Pathology, Tufts University, Boston, MA) were subjected to ISH with the seven cytokine-speci®c riboprobes. Localization of the respective signals was found to agree with that previously determined at the protein level (Andersson et al, 1994) (Fig 1c). Quanti®cation of cytokine-producing cells Evaluation of all slides was performed in a blind fashion by the same examiner using a light microscope (Zeiss) with a rectangular morphometric grid. First, the entire slide was scanned at 1003 magni®cation proceeding from the top to the bottom of the section to select dermal areas with the most intense in¯ammatory in®ltrates. These areas were then examined at 4003 magni®cation to ascertain the distribution of cytokine-positive cells within the in¯ammatory in®ltrates. Next, the number of cytokineproducing cells in these areas was counted in 15 randomly chosen highpower ®elds per specimen under 10003 magni®cation, the magni®cation at which positive cells were best discernible. Due to the distribution of the

CYTOKINE mRNA EXPRESSION IN DERMATOBORRELIOSES

1117

in®ltrate, about two-thirds of the analyzed ®elds were located within the upper dermis and one-third of the ®elds within the lower dermis, including central as well as peripheral zones of the section. In preliminary studies, we found that the analysis of as many as 50 high-power ®elds yielded similar results to 15 high-power ®elds. Cytokine-positive cells were de®ned as cells with a distinct local perinuclear (ring-like) granular staining. Within the same high-power ®elds, the number n of in¯ammatory cells seen on a line that bisected the microscopic ®eld was counted, and the total number N of cells in that ®eld was calculated according to the formula N = (n 3 p/4)2. This method has previously been shown to be an accurate estimate of the actual number of cells per high-power ®eld (Simpson et al, 1992). For each cytokine, the percentage of cytokine-positive cells per total number of in¯ammatory cells from all 15 high-power ®elds was calculated for each patient, and this value was used for statistical analysis. Statistics For analysis, the patients were divided into three groups: those who had EM alone without associated signs and symptoms, those who had EM with associated signs and symptoms, and those who had ACA. The distribution of values between the groups in duration of EM, expression of leukocyte differentiation antigens, and percentage of cells expressing mRNA for each cytokine was compared by Kruskal±Wallis test, a nonparametric statistical method. In the tables, the median values (half of the samples) are shown as well as the ®rst and third quartile values (onequarter and three-quarters of the values), which is called the interquartile range. In addition, the percentage of patients with at least one cell expressing mRNA for a given cytokine and the correlation of cytokine data with PCR results were compared between groups using the c2 test. Cytokine mRNA expression in lesional skin was compared with disease duration before sampling in both EM groups and the ACA group using the Spearman rank correlation coef®cient. p-values are two-tailed.

RESULTS Clinical features Of the 42 patients with EM, 39 (93%) had solitary homogenous or ring-like erythematous lesions with a median diameter of 14 cm (range, 6±50 cm), and three (7%) had multiple EM lesions. Twenty-seven of the 42 patients (64%) had EM alone with no associated signs or symptoms. The remaining 15 patients (36%) had EM with associated signs and symptoms, including headache, elevated temperature, arthralgias, myalgias, or fatigue. Four of these 15 patients had clinical evidence of disseminated infection: three had multiple EM lesions, in two instances accompanied by intense headache and neck stiffness, and the fourth patient had polyneuritis. The median duration of EM before biopsy was 7 d (range, 3±60 d) in the group without associated symptoms, and 18.5 d (range, 4±60 d) in the group with associated symptoms. This difference was not statistically signi®cant, however (p = 0.7). Of the 27 patients with ACA, eight (30%) had skin changes for 0±3 mo (acute stage), ten (37%) for 4±12 mo (subacute stage), and nine (33%) for more than 12 mo (chronic stage). The range of disease duration in all ACA patients was 1±156 mo (median, 7 mo). Ten patients (37%) had a distal neuropathy within the area of the affected skin, but none had signs or symptoms extending beyond the skin lesion. Histopathologic ®ndings On histopathologic examination, all specimens were consistent with the diagnosis of EM or ACA (Duray and Chandler, 1997). EM lesions were characterized by mild to marked perivascular in®ltrates in the papillary more than the reticular dermis (Fig 2a). The in®ltrates were composed primarily of lymphocytes and macrophages intermingled with a small number of plasma cells. In ACA lesions, band-like or patchy perivascular in®ltrates, or both, were seen primarily in the upper but also in the lower portions of the dermis (Fig 2b). The in®ltrates were composed of lymphocytes and macrophages with moderate numbers of plasma cells. The in®ltrates were most prominent in acute lesions, but were still present around blood vessels, which were often dilated, in subacute and chronic lesions. In the chronic stage, thinning of the epidermis was observed with loss of rete ridges, decreased breadth of the dermis, and reduction or complete loss of pilosebaceous units.

1118

È LLEGGER ET AL MU

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 1. Control experiments for ISH. Hybridization buffer with no probe was applied to a section obtained from a patient with EM with associated signs and symptoms. No cytokine-speci®c cells are visible (a). A section of a biopsy specimen from normal skin was hybridized to TNF-a antisense riboprobe. Note the minimal constitutive expression of cytokine mRNA (arrow) (b). A section of an in¯amed tonsil was hybridized to IFN-g antisense riboprobe. Signals are visible as granular perikaryal staining in T cell rich, extrafollicular areas (c). Scale bars: 12 mm.

Figure 2. Histopathology of lesional skin in patients with EM or ACA. In a patient with EM (a), perivascular in®ltrates of mononuclear cells are seen throughout the dermis. In a patient with ACA (b), there is a more diffuse in®ltration of mononuclear cells in the upper portion of the dermis. The sections are stained with hematoxylin and eosin. Scale bars: 100 mm (a), 50 mm (b).

Table I. Expression of leukocyte differentiation antigens based on a score of 0±3 in lesional skin of erythema migrans or acrodermatitis chronica atrophicansa

Erythema migrans without associated symptoms (n = 6) Erythema migrans with associated symptoms (n = 6) Acrodermatitis chronica atrophicans (n = 10)

CD3 (T cells)

CD20 (B cells)

CD68 (macrophages)

2.0 (2.0±3.0)bc 2.0 (2.0±3.0)b 1.5 (1.0±3.0)

1.0 (1.0±2.0) 1.0 (1.0±1.0) 2.0 (1.0±3.0)f

1.0 (1.0±2.0) 3.0 (2.0±3.0)de 2.0 (2.0±3.0)g

aMedian value (®rst and third quartile values). The percentage of cells positive for a respective leukocyte marker was estimated relative to the total number of in®ltrating cells: 0, all in®ltrating cells negative; 1, up to a third of all cells positive; 2, between one-third and two-thirds of all cells positive; 3, more than two-thirds of all cells positive. bp < 0.05 vs CD20+ cells. cp < 0.05 vs CD68+ cells. dp < 0.05 vs CD20+ cells. ep < 0.05 vs erythema migrans without associated symptoms. fp = 0.05 vs erythema migrans with associated symptoms. gp = 0.05 vs erythema migrans without associated symptoms.

Expression of leukocyte differentiation antigens (IHC) A semiquantitative determination was made of the frequencies of the various in¯ammatory cells in a subset of patients from each disease group (Table I). Among the six patients who had EM alone without associated symptoms, CD3+ T cells were the predominant

cell type, and they were present in signi®cantly higher numbers than CD68+ macrophages or CD20+ B cells. Among the six EM patients who also had associated symptoms, greater numbers of macrophages were found (Fig 3a); T cells were seen in moderate number (Fig 3b), and B cells were infrequent. When the two EM

VOL. 115, NO. 6 DECEMBER 2000

Figure 3. Expression of leukocyte differentiation antigens in lesional skin in patients with EM or ACA. In a patient with EM who had associated signs and symptoms, large numbers of CD68+ macrophages (a) and moderate numbers of CD3+ T cells (b) are seen in a perivascular in®ltrate. In a patient with ACA, CD68+ macrophages (c) and CD20+ B cells (d) are shown. The number of macrophages is greater in the EM lesion than in the ACA lesion. Immunoperoxidase stained with anti-CD68 (a, c), anti-CD3 (b), or anti-CD20 (d) antibodies. Scale bars: 50 mm (a, c), 20 mm (b, d).

Figure 4. Expression of cytokine mRNA in lesional skin in patients with EM with associated signs and symptoms. Perivascular in®ltrates are shown from several representative patients with in¯ammatory cells that hybridized with antisense riboprobes for TNF-a (a, c), IFN-g (b, d), or IL-1b (e). Signals are visible in individual cells as granular perikaryal staining (arrows). For comparison, no signals were found when staining was done with the control sense riboprobe for IFN-g (f). Scale bars: 25 mm (a, b), 10 mm (c-f).

CYTOKINE mRNA EXPRESSION IN DERMATOBORRELIOSES

1119

1120

È LLEGGER ET AL MU

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table II. Percentage of in¯ammatory cells positive for a particular cytokine mRNA in erythema migrans or acrodermatitis chronica atrophicans Erythema migrans without associated symptoms (n = 27) Pro-in¯ammatory cytokines, median value (®rst and third quartile values) TNF-a 2.2 (0.0±6.7) IL-1b 0.0 (0.0±1.8) IL-6 0.0 (0.0±1.0) IFN-g 5.7 (2.9±8.6)d IL-2 2.0 (0.0±8.3)d Anti-in¯ammatory cytokines IL-4 0.0 (0.0±1.5) IL-10 4.1 (0.0±6.0)d

Erythema migrans with associated symptoms (n = 15)

Acrodermatitis chronica atrophicans (n = 27)

6.3 (3.8±11.8)a 3.8 (1.4±7.1)ac 2.1 (0.0±5.2)a 9.1 (3.6±10.6)c 5.0 (2.9±10.9)c

4.8 0.0 0.0 0.0 0.0

1.3 (0.0±3.3) 2.9 (0.9±5.9)c

1.8 (0.0±6.4)b 0.0 (0.0±4.0)

(2.9±10.8)b (0.0±0.0) (0.0±3.8) (0.0±0.0) (0.0±1.4)

ap < 0.05

vs erythema migrans without associated symptoms. vs erythema migrans without associated symptoms. vs acrodermatitis chronica atrophicans. dp < 0.05 vs acrodermatitis chronica atrophicans. bp < 0.05 cp < 0.05

Table III. Number of patients with at least one positive cell for a particular cytokine mRNA in erythema migrans or acrodermatitis chronica atrophicans Erythema migrans without associated symptoms (n = 27) Pro-in¯ammatory cytokines, number of patients (percentage) TNF-a 20 (74) IL-1b 10 (37) IL-6 7 (26) IFN-g 25 (93)c IL-2 19 (70)c Anti-in¯ammatory cytokines IL-4 12 (44) IL-10 19 (70)

Erythema migrans with associated symptoms (n = 15)

Acrodermatitis chronica atrophicans (n = 27)

15 (100)a 15 (100)ab 10 (67)a 14 (93)b 12 (80)b

24 6 10 6 9

9 (60) 13 (87)b

19 (70) 12 (44)

(89) (22) (37) (22) (33)

ap < 0.05

vs erythema migrans without associated symptoms. vs acrodermatitis chronica atrophicans. cp < 0.05 vs acrodermatitis chronica atrophicans. bp < 0.05

groups were compared, macrophages were detected signi®cantly more often among patients with associated symptoms. In the ten patients with ACA, macrophages (Fig 3c), T cells, and B cells (Fig 3d) were found in moderate numbers. As in patients with EM, reactivity in the ACA patients was more prominent in the upper portion of the dermis. Most of the cells were observed within the perivascular in®ltrates except for macrophages, which, in smaller numbers, appeared in interstitial areas as well. Macrophages and T cells were found in a scattered pattern and in clusters, whereas B cells clearly tended to form clusters within the in¯ammatory in®ltrates. Expression of cytokine mRNA (ISH) Among the 42 patients with EM, activation of both pro-in¯ammatory and antiin¯ammatory cytokines was found in lesional skin (Tables II, III). Among the 27 patients who had EM alone without other signs or symptoms, the major cytokines expressed were the proin¯ammatory cytokine IFN-g and the anti-in¯ammatory cytokine IL-10 (Tables II, IIII). Twenty-®ve of these patients (93%) had in¯ammatory cells expressing IFN-g, and a median of 5.7% of the cells expressed it. Nineteen patients (70%) had in¯ammatory cells expressing mRNA for IL-10, and a median of 4.1% of the cells expressed it. Most patients also had a smaller number of in¯ammatory cells expressing the pro-in¯ammatory cytokines TNF-a or IL-2, and less than half of patients had a few cells expressing mRNA for IL-1b, IL-6, or IL-4.

Compared with the 27 patients who had EM alone with no associated symptoms, the 15 patients who had EM and associated symptoms had signi®cantly greater expression of macrophagederived, pro-in¯ammatory cytokines. All 15 of these patients (100%) had in¯ammatory cells expressing mRNA for TNF-a (Fig 4a, c) and IL-1b (Fig 4e), and a median of 6.3% and 3.8%, respectively, of the cells expressed these cytokines. Similar to the group without associated symptoms, 14 of the 15 patients with symptoms (93%) had in¯ammatory cells in EM lesions expressing IFN-g, and a median of 9.1% of the cells expressed it (Fig 4b, d). To a lesser degree, anti-in¯ammatory cytokines were expressed as well. Thirteen patients (87%) had in¯ammatory cells expressing mRNA for IL-10, and a median of 2.9% of the cells expressed it. In addition, nine patients (60%) had in¯ammatory cells expressing mRNA for IL-4, and a median of 1.3% of the cells expressed it. There was no statistically signi®cant difference between the expression of cytokine mRNA and the presence or absence of B. burgdorferi-speci®c DNA in EM lesions. There was also no signi®cant correlation between the expression of cytokine mRNA and the duration of EM lesions. In ACA, a more restricted cytokine pro®le was observed. These 27 patients had very little or no IFN-g expression. Instead, 24 of the 27 patients (89%) had in¯ammatory cells expressing mRNA for TNF-a, and a median of 4.8% of the cells expressed it (Fig 5a, c); mRNA for other pro-in¯ammatory cytokines was rarely found. Nineteen patients (70%) had in¯ammatory cells expressing mRNA

VOL. 115, NO. 6 DECEMBER 2000

CYTOKINE mRNA EXPRESSION IN DERMATOBORRELIOSES

1121

Figure 5. Expression of cytokine mRNA in lesional skin in patients with ACA. Perivascular in®ltrates are shown from representative patients with in¯ammatory cells that hybridized with antisense riboprobes for TNF-a (a, c) and IL-4 (b, d). Signals are visible in individual cells as granular perikaryal staining (arrows). Scale bars: 25 mm (a, b), 10 mm (c, d).

for IL-4 and a median of 1.8% of the cells expressed it (Fig 5b, d). Approximately half of the patients had a few in¯ammatory cells expressing IL-10 mRNA. This same, restricted pattern of cytokine expression was seen in patients with acute, subacute, or chronic lesions. In both EM and ACA lesions, approximately one to ten cytokine-positive cells were seen within a given perivascular in¯ammatory in®ltrate, depending on the frequency of the cytokine (Figs 4a, b, 5a, b). Serial sections stained for leukocyte surface markers and then cytokine mRNA expression revealed that cytokine signals occurred in areas of mononuclear cell in®ltrates. DISCUSSION In this study, cytokine patterns were identi®ed in EM and ACA lesions using ISH with speci®c cRNA probes. This approach had several advantages. First, a large collection of archival, formalin®xed, paraf®n-embedded tissue samples could be used. Second, the method permits an analysis of cytokine mRNA expression in the original microenvironment of the skin (Tijssen, 1993; Hoefakker et al, 1995). Moreover, cytokine induction at the primary site of infection may be strongly compartmentalized (Hoefakker et al, 1995). This phenomenon has been demonstrated in the cerebrospinal ¯uid in neuroborreliosis (Wang et al, 1995; Ekerfelt et al, 1997) and in the joint in Lyme arthritis (Gross et al, 1998). Finally, the anatomic localization of mRNA signals could be determined, which is especially important in skin where cytokines may be produced at multiple sites by many different types of cells (Boehm et al, 1994; Boehm et al, 1995; Sauder and Dytoc, 1997). Thus, with this method, cytokine-positive cells can be distinguished from potential background expression in normal tissue, which is a problem with reverse transcriptase PCR. Although the exact cellular origin of the cytokine mRNA signals could not be ascertained without double staining techniques, which were not done here, serial sections in this study, stained by IHC or ISH, showed that cytokine-positive cells were con®ned to areas of T cell and macrophage in®ltrates. Thus, we assume that these cells were the major source of these cytokines. ISH has several disadvantages. It is less sensitive than reverse transcriptase PCR, a potential problem when there are low copy numbers of mRNA. In our protocol, however, the application of

digoxigenin labeling and microwave oven heating to ISH techniques (Komminoth, 1992; Lan et al, 1996) provided suf®cient sensitivity to detect even low levels of mRNA. Second, it is dif®cult to quantify results by ISH although the cells producing cytokine mRNA signals can be counted, as was done in this study. Finally, mRNA expression does not prove that a biologically active protein is produced, but mRNA expression is thought to provide an accurate picture of the cytokine responses at sites of in¯ammation (Brennan et al, 1995). In this study, patients with EM usually had mRNA expression of both pro-in¯ammatory and anti-in¯ammatory cytokines in skin lesions rather than a polarized expression only of Th1 or Th2 cytokines. In EM lesions, IFN-g and IL-10 were often found, whereas in ACA lesions only TNF-a and IL-4 were found, but very little or no IFN-g. Spirochetal factors may play a role in this differential cytokine expression. For example, in experimentally infected inbred mice (Isogai et al, 1996) and in human neuroborreliosis (Ekerfelt et al, 1998), different Borrelia genospecies differentially stimulated the synthesis of cytokines. In another report, each species had strains with different cytokine induction, and therefore different cytokine responses could not be ascribed to particular genospecies (Schulze et al, 1996). In a previous study of dermatoborrelioses from Styria, the Austrian state in which the study patients lived, B. afzelii was found to be the causative agent in 74% and B. garinii in 26% of EM patients, and B. afzelii was the sole pathogen in ACA patients.1 Although the Borrelia species causing the infection was not determined in either our EM or ACA patients, we assume that most had B. afzelii infection. Strains of B. afzelii may differ in the cytokine pro®le that they induce or they may ®nd a protected niche in ACA but not EM lesions. The differential cytokine responses observed here could also relate to the duration of infection. For example, the initial responses in EM lesions may lead to expression of IFN-g and IL-10, which may even be produced by the same T cell (Pohl-Koppe et al, 1998). These cytokines may then induce a broader range of proin¯ammatory and anti-in¯ammatory cytokines accompanied by the development of associated signs and symptoms. In this study, EM lesions tended to have been present longer at the time of biopsy in the patients with associated symptoms than in those without symptoms, but the differences were not statistically signi®cant. Furthermore, no statistical correlation was found between duration

1122

È LLEGGER ET AL MU

of EM and cytokine mRNA expression. Although one may wonder whether IFN-g is initially present in ACA lesions, it was not found even in the lesions of shortest duration. Host factors are also likely to be important in this differential cytokine expression. ACA, for example, has been observed mainly in elderly patients, particularly women, and affects primarily sunexposed acral surfaces (AÊsbrink and Hovmark, 1988). Modulation of the immune response by age, sex, and ultraviolet radiation may be factors that in¯uence the course of infection. In B. burgdorferiinfected C3H mice, ultraviolet B irradiation downregulated the Th1 response while leaving the Th2 response intact (Brown et al, 1995). Consistent with this ®nding, the current study shows that ACA patients had very little expression of the Th1 cytokine IFN-g, but considerable amounts of the Th2 cytokine IL-4. EM patients in this study were strati®ed according to whether or not they had associated signs and symptoms, particularly headache, elevated temperature, myalgias, arthralgias, or fatigue. When these symptoms were present, a larger number of macrophages and greater expression of macrophage-derived, pro-in¯ammatory cytokines was found in EM lesions, including TNF-a, IL-1b, and IL-6. These cytokines share a number of biologic properties (Dinarello, 1991), including the induction of a potent acute phase in¯ammatory response with fever and malaise (Sipe, 1985; Dinarello et al, 1986; Geiger et al, 1988; Tracey et al, 1988). The administration of recombinant IL-1 to humans causes headache, nausea, arthralgias, myalgias, and fatigue (Crown et al, 1991). These symptoms do not necessarily connote systemic infection with B. burgdorferi, as only four of the 15 patients with these symptoms had a clinical picture suggesting spread of the spirochete to other organ systems. Thus, local production in the skin of these macrophagederived cytokines may often account for these associated symptoms. The different cytokine patterns expressed in EM and ACA patients are likely to be important in the host's ability to clear the spirochetal infection. EM, with strong expression of IFN-g and IL10, usually resolves spontaneously in weeks to months (Steere et al, 1983b; AÊsbrink and Olsson, 1985). This response seems to be critical for spirochetal killing; only rarely is it possible to isolate B. burgdorferi from sites of spontaneously healed EM lesions (Strle et al, 1995). In contrast, ACA, which seems to lack adequate IFN-g production, is a chronic progressive infection (AÊsbrink and Hovmark, 1988) as evidenced by the cultivation of spirochetes Ê sbrink and Hovmark, from these lesions years after disease onset (A 1985). Thus, a pro-in¯ammatory response early in the infection, particularly with IFN-g, seems to be essential in eliminating B. burgdorferi from the skin, and anti-in¯ammatory cytokines may be important in the appropriate regulation of this response. Several murine studies show the importance of pro-in¯ammatory cytokines in the control of this spirochetal infection. Although the initial studies in inbred mice reported severe arthritis in C3H mice that produced IFN-g and mild arthritis in BALB/c mice that produced IL-4, these studies did not examine the kinetics of the response (Keane-Myers and Nickell, 1995; Matyniak and Reiner, 1995). It is now known that B. burgdorferi-infected BALB/c mice initially produce large amounts of IFN-g, but then switch to IL-4 production (Kang et al, 1997). In contrast, C3H mice produce less IFN-g initially but continue to have a Th1 response (Kang et al, 1997), which may represent a failure of appropriate immunoregulation. Recently, using C3H mice, Zeidner et al (1996) showed that a pro-in¯ammatory response is critical in the control of early infection. When the mice were infected with B. burgdorferi and given recombinant IFN-g, IL-2, or TNF-a, alone or in combination, the mice were protected from infection. Similarly, Pride et al (1998) showed that B. burgdorferi-speci®c IFN-g and IL2-producing T cells from infected C3H mice signi®cantly reduced the infection rate in naive C3H recipients. Thus, in both the human dermatoborrelioses and these murine models, a Th1 cytokine response seems to be critical in eliminating B. burgdorferi early in the infection. In summary, mRNA expression of pro-in¯ammatory cytokines, particularly IFN-g, and anti-in¯ammatory cytokines was found in

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

EM lesions, which seems to be important in the control and resolution of this spirochetal infection. In contrast, the restricted pattern of cytokine expression in ACA lesions, consisting primarily of TNF-a and IL-4 but little or no IFN-g, may be less effective in spirochetal killing, which may be a factor in the chronicity of this skin lesion over a period of years. The authors wish to thank Dr. P. Bucy for providing the cytokine-speci®c cDNA constructs used in this study and Dr. P. Kwan for providing tonsil sections. We thank Dr. N. ZoÈchling for the performance of B. burgdorferi-speci®c PCR and A. Shepard-Barry for capable technical assistance. We thank Drs J. Coburn, S. DiazCano, L. Glickstein, M. Harjacek, and M. Stadecker for valuable advice. Dr. R. MuÈllegger was on leave from the Department of Dermatology, Karl-Franzens University School of Medicine, Graz, Austria, and was supported by a Max Kade Postdoctoral Research Grant. Additional grant support for this work was from the National Institutes of Health (AR 20358), the Centers for Disease Control and Prevention (CCU 110291), the Mathers Foundation, and the Eshe Fund.

REFERENCES Ê gren K, Andersson U: Andersson J, Abrams J, BjoÈrk L, Funa K, Litton M, A Concomitant in vivo production of 19 different cytokines in human tonsils. Immunology 83:16±24, 1994 Anthonissen FM, De-Kesel M, Hoet PP, Bigaignon GH: Evidence for the involvement of different genospecies of Borrelia in the clinical outcome of Lyme disease in Belgium. Res Microbiol 145:327±331, 1994 Ê Asbrink E, Hovmark A: Successful cultivation of spirochetes from skin lesions of patients with erythema chronicum migrans Afzelius and acrodermatitis chronica atrophicans. Acta Pathol Microbiol Immunol Scand B 93:161±163, 1985 Ê Asbrink E, Hovmark A: Early and late cutaneous manifestations in Ixodes-borne borreliosis (erythema migrans borreliosis, Lyme borreliosis). Ann NY Acad Sci 539:4±15, 1988 Ê Asbrink E, Olsson I: Clinical manifestations of erythema chronicum migrans Afzelius in 161 patients. Acta Derm Venereol (Stockh) 65:43±52, 1985 Baranton G, Postic D, Saint Girons I, Boerlin P, Piffaretti JC, Assous M, Grimont PAD: Delineation of Borrelia burgdorferi sensu stricto, Borrelia garinii sp. nov. & group VS 461 associated with Lyme borreliosis. Int J Syst Bacteriol 42:378±383, 1992 Barthold SW, Beck DS, Hansen GM, Terwilliger GA, Moody KD: Lyme borreliosis in selected strains and ages of laboratory mice. J Infect Dis 162:133±138, 1990 Berger BW: Erythema chronicum migrans of Lyme disease. Arch Dermatol 120:1017± 1021, 1984 Boehm KD, Yun JK, Garner C, Strohl KP, Elmets CA: In situ detection of cytokine messenger RNAs in the eccrine sweat gland of normal human skin. Lymphokine-Cytokine Res 13:9±13, 1994 Boehm KD, Yun JK, Strohl KP, Elmets CA: Messenger RNAs for the multifunctional cytokines interleukin-1a, interleukin-1b, and tumor necrosis factor-a are present in adnexal tissues and in dermis of normal human skin. Exp Dermatol 4:335±341, 1995 Brennan FM, Maini RN, Feldman M: Cytokine expression in chronic in¯ammatory disease. Br Med Bull 51:368±384, 1995 Brown EL, Rivas JM, Ullrich SE, Young CR, Norris SJ, Kripke ML: Modulation of immunity to Borrelia burgdorferi by ultraviolet irradiation: differential effect on Th1 and Th2 immune responses. Eur J Immunol 25:3017±3022, 1995 Chen J, Field JA, Glickstein L, Molloy P, Huber BT, Steere AC: Association of antibiotic treatment-resistant Lyme arthritis with T cell responses to dominant epitopes of outer-surface protein A of Borrelia burgdorferi. Arthritis Rheum 42:1813±1822, 1999 Crown J, Jakubowski A, Kemeny N, et al: A phase I trial of recombinant human interleukin-1b alone and in combination with myelosuppressive doses of 5¯uoruracil in patients with gastrointestinal cancer. Blood 78:1420±1427, 1991 Dinarello CA: Interleukin-1 and interleukin-1 antagonism. Blood 77:1627±1652, 1991 Dinarello CA, Cannon JC, Mier JW, et al: Multiple biological activities of human recombinant interleukin-1. J Clin Invest 77:1734±1739, 1986 Duray PH, Chandler FW: Lyme disease. In: Connor DH, Chandler FW, Schwartz DA, Manz HJ, Lack EE: eds. Pathology of Infectious Diseases Appleton and Lange, Stamford, 1997:pp 635±646 Ekerfelt C, Ernerudh J, Bunikis J, et al: Compartmentalization of antigen speci®c cytokine responses to the central nervous sytem in CNS borreliosis: secretion of IFN-g predominates over IL-4 secretion in response to outer surface proteins of Lyme disease Borrelia spirochetes. J Neuroimmunol 79:155±162, 1997 Ekerfelt C, Ernerudh J, Forsberg P, BergstroÈm S: Augmented intrathecal secretion of interferon-gamma in response to Borrelia garinii in neuroborreliosis. J Neuroimmunol 89:177±181, 1998 Geiger T, Andus T, Klapproth J, Hirano T, Kishimoto T, Heinrich PC: Induction of rat acute-phase proteins by interleukin-6 in vivo. Eur J Immunol 18:717±721, 1988 Gross DM, Steere AC, Huber BT: T helper 1 response is dominant and localized to

VOL. 115, NO. 6 DECEMBER 2000

the synovial ¯uid in patients with Lyme arthritis. J Immunol 160:1022±1028, 1998 Hansen K, Hindersson P, Pedersen NS: Measurement of antibodies to the Borrelia burgdorferi ¯agellum improves serodiagnosis in Lyme disease. J Clin Microbiol 26:338±346, 1988 Harjacek M, Diaz-Cano S, Alman BA, Coburn J, Ruthazer R, Wolfe H, Steere AC: Prominent expression of mRNA for proin¯ammatory cytokines in synovium in patients with juvenile rheumatoid arthritis or chronic Lyme arthritis. J Rheumatol 27:497±503, 2000 Hoefakker S, Boersma W, Claassen E: Detection of human cytokines in situ using antibody and probe based methods. J Immunol Methods 185:149±175, 1995 Isogai E, Isogai H, Kimura K, et al: Cytokines in the serum and brain in mice infected with distinct species of Lyme disease Borrelia. Microb Pathog 21:413±419, 1996 Kang I, Barthold SW, Persing DH, Bockenstedt LK: T-helper-cell cytokines in the early evolution of murine Lyme arthritis. Infect Immun 65:3107±3111, 1997 Keane-Myers A, Nickell SP: Role of IL-4 and IFN-g in modulation of immunity to Borrelia burgdorferi in mice. J Immunol 155:2020±2028, 1995 Komminoth P: Digoxigenin as an alternative probe labeling for in situ hybridization. Diagn Mol Pathol 1:142±150, 1992 Lan HY, Mu W, Ng YY, Nikolic-Paterson DJ, Atkins RC: A simple, reliable, and sensitive method for nonradioactive in situ hybridization: use of microwave heating to improve hybridization ef®ciency and preserve tissue morphology. J Histochem Cytochem 44:281±287, 1996 Matyniak JE, Reiner SL: T-helper phenotype and genetic susceptibility in experimental Lyme disease. J Exp Med 181:1251±1254, 1995 Oksi J, Savolainen J, PeÁne J, Bousquet J, Laippala P, Viljanen M: Decreased interleukin-4 and increased gamma interferon production by peripheral blood mononuclear cells of patients with Lyme borreliosis. Infect Immun 64:3620± 3623, 1996 Picken RN, Strle F, Picken MM, Ruzic-Sabljic E, Maraspin V, Lotric-Furlan S, Cimperman J: Identi®cation of three species of Borrelia burgdorferi sensu lato (B. burgdorferi sensu stricto, B. garinii, and B. afzelii) among isolates from acrodermatitis chronica atrophicans lesions. J Invest Dermatol 110:211±214, 1998 Pohl-Koppe A, Balashov KE, Steere AC, Logigian EL, Ha¯er DA: Identi®cation of a T-cell subset capable of both IFN-g and IL-10 secretion in patients with chronic Borrelia burgdorferi infection. J Immunol 160:1804±1810, 1998 Pride MW, Brown EL, Stephens LC, Killion JJ, Norris SJ, Kripke ML: Speci®c Th1 cell lines that confer protective immunity against experimental Borrelia burgdorferi infection in mice. J Leukoc Biol 63:542±549, 1998 Sauder DN, Dytoc MT: Cytokines. Curr Probl Dermatol 9:1±32, 1997 Schaible UE, Kramer MD, Wallich R, Tran T, Simon MM: Experimental Borrelia burgdorferi infection in inbred mouse strains: antibody response and association of H-2 genes with resistance and susceptibility to development of arthritis. Eur J Immunol 21:2397±2405, 1991 Schulze J, Breitner-Ruddock S, von Briesen H, Brade V: High- and low-level cytokine induction in human peripheral blood mononuclear cells by different Borrelia burgdorferi strains. Med Microbiol Immunol 185:31±37, 1996 Simpson JF, Dutt PL, Page DL: Expression of mitoses per thousand cells and cell density in breast carcinomas: a proposal. Hum Pathol 23:608±611, 1992

CYTOKINE mRNA EXPRESSION IN DERMATOBORRELIOSES

1123

Sipe JD: Interleukin-1 as the key factor in the acute-phase response. In: Gordon AH, Koj A: eds. The Acute-Phase Response to Injury and Infection Elsevier, Amsterdam, 1985:pp 23±36 Stanek G, O'Connell S, Cimmino M, et al: European Union concerted action on risk assessment in Lyme borreliosis: clinical case de®nitions for Lyme borreliosis. Wien Klin Wochenschr 108:741±747, 1996 Steere AC, Grodzicki RL, Kornblatt AN, et al: The spirochetal etiology of Lyme disease. N Engl J Med 308:733±740, 1983a Steere AC, Bartenhagen NH, Craft JE, et al: The early clinical manifestations of Lyme disease. Ann Intern Med 99:76±82, 1983b Strle F, Cheng Y, Cimperman J, et al: Persistence of Borrelia burgdorferi sensu lato in resolved erythema migrans lesions. Clin Infect Dis 21:380±389, 1995 Strle F, Nadelman RB, Cimperman J, et al: Comparison of culture-con®rmed erythema migrans caused by Borrelia burgdorferi sensu stricto in New York State and by Borrelia afzelii in Slovenia. Ann Intern Med 130:32±36, 1999 Tijssen P: In situ hybridization. In: Van der Vliet PC: ed. Laboratory Techniques in Biochemistry and Molecular Biology, Vol 24, Hybridization with Nucleic Acid Probes Part II, Probe Labeling and Hybridization Techniques Elsevier, Amsterdam, 1993:pp 495±524 Tracey KJ, Wei H, Manogue KR, et al: Cachectin/tumor necrosis factor induces cachexia, anemia, and in¯ammation. J Exp Med 167:1211±1227, 1988 Wang WZ, Fredrikson S, Sun JB, Link H: Lyme neuroborreliosis: evidence for persistent up-regulation of Borrelia burgdorferi-reactive cells secreting interferong. Scand J Immunol 42:694±700, 1995 Weber K, Neubert U: Clinical features of early erythema migrans disease and related disorders. Zbl Bakt Hyg A 263:209±228, 1986 Weber K, P®ster HW, Reimers CD: Clinical overview. In: Weber K, Burgdorfer W: eds. Aspects of Lyme Borreliosis. Springer, Berlin-Heidelberg-New York, 1993:pp 93±104 Wienecke R, Neubert U, Volkenandt R: Molecular detection of Borrelia burgdorferi in formalin-®xed, paraf®n-embedded lesions of Lyme disease. J Cutan Pathol 20:385±388, 1993 Wienecke R, ZoÈchling N, Neubert U, SchluÈpen EM, Meurer M, Volkenandt M: Molecular subtyping of Borrelia burgdorferi in erythema migrans and acrodermatitis chronic atrophicans. J Invest Dermatol 103:19±22, 1994 Yang L, Weis JH, Eichwald E, Kolbert CP, Persing DH, Weis JJ: Heritable susceptibility to severe Borrelia burgdorferi-induced arthritis is dominant and is associated with persistence of large numbers of spirochetes in tissues. Infect Immun 62:492±500, 1994 Yin Z, Braun J, Neure L, et al: T-cell cytokine pattern in the joints of patients with Lyme arthritis and its regulation by cytokines and anticytokines. Arthritis Rheum 40:69±79, 1997 Yssel H, Shanafelt MC, Soderberg C, Schneider PV, Anzola J, Peltz G: Borrelia burgdorferi activates a T helper type 1-like T-cell subset in Lyme arthritis. J Exp Med 174:593±601, 1991 Zeidner N, Dreitz M, Belasco D, Fish D: Suppression of acute Ixodes scapularisinduced Borrelia burgdorferi infection using tumor necrosis factor-a, interleukin2, and interferon-g. J Infect Dis 173:187±195, 1996