Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
DOI: 10.1093/brain/awg114 Brain (2003), 126, 1026±1035 Muscle ®bres and cultured muscle cells express the B7.1/2-related inducible co-stimulatory molecule, ICOSL: implications for the pathogenesis of in¯ammatory myopathies Heinz Wiendl,1 Meike Mitsdoerffer,1 Dagmar Schneider,1 Arthur Melms,1 Hanns Lochmuller,2 Reinhard Hohlfeld3 and Michael Weller1 1Department of Neurology, University of Tu È bingen, Medical School, TuÈbingen, 2Genzentrum, Friedrich-Baur Institut, Department of Neurology and Ludwig Maximilians University and 3Institute for Clinical Neuroimmunology, Ludwig Maximilians University, Munich, Germany Summary Inducible co-stimulator ligand (ICOSL), a member of the B7 family of co-stimulatory molecules related to B7.1/2, regulates CD4 as well as CD8 T-cell responses via interaction with its receptor ICOS on activated T cells. Here we examined the expression and the functional relevance of ICOSL in human muscle cells in vivo and in vitro. We investigated 25 muscle biopsy specimens from patients with polymyositis, dermatomyositis, inclusion body myositis, Duchenne muscular dystrophy and non-myopathic controls for ICOSL expression by immunohistochemistry. Normal muscle ®bres constitutively express low levels of ICOSL. However, ICOSL expression is markedly increased in muscle ®bres in in¯ammatory myopathies. Cell surface staining was most prominent in the contact areas between muscle ®bres and in¯ammatory cells, which in turn show expression of ICOS as a marker of T-cell activation. Muscle endothelial cells show constitutive expression of ICOSL under normal and pathological conditions. We also detected mRNA and cell surface protein expression Correspondence to: Dr Heinz Wiendl, Department of Neurology, University of TuÈbingen, Hoppe-Seyler-Strasse 3, D-72076 TuÈbingen, Germany E-mail: [email protected] of ICOSL on myoblasts cultured from control subjects and patients as well as in TE671 muscle rhabdomyosarcoma cells. ICOSL expression was upregulated by tumour necrosis factor-a (TNF-a), whereas interferon-g (IFN-g) had no such effect. Co-culture experiments of major histocompatibility complex (MHC) class II-positive myoblasts with CD4 T cells together with superantigen demonstrated that the expression of musclerelated ICOSL has functional consequences: the production of Th1 (IFN-g) and Th2 cytokines [interleukin (IL)-4 and IL-10] by CD4 T cells was markedly reduced in the presence of a neutralizing anti-ICOSL monoclonal antibody (mAb HIL-131), thus showing the importance of ICOSL co-stimulation for T-cell activation. Taken together, our results demonstrate that human muscle cells express ICOSL, a functional co-stimulatory molecule distinct from B7.1 and B7.2. ICOSL±ICOS interactions may play an important role in in¯ammatory myopathies, providing further evidence for the antigen-presenting capacity of muscle cells. Keywords: co-stimulation; muscle immunobiology; B7 family; myositis; B7-H2 Abbreviations: ICOS = inducible co-stimulator; ICOSL = inducible co-stimulator ligand; IFN = interferon; MHC = major histocompatibility complex; NCAM = neural cell adhesion molecule; PBMC = peripheral blood mononuclear cell; PCR = polymerase chain reaction; SAg = superantigen; TNF = tumour necrosis factor Introduction A considerable body of evidence suggests that muscle actively participates in local immune reactions (Hohlfeld and Engel, 1994). Notably, upregulation of major histocompatibility complex (MHC) class I and II antigens under Brain 126(5) ã Guarantors of Brain 2003; all rights reserved pathological conditions (Karpati et al., 1988; Emslie-Smith et al., 1989; Wiendl et al., 2000) is an essential prerequisite for the interaction of muscle with CD8+ cytotoxic T cells and CD4+ T helper cells, as observed in the idiopathic in¯amma- Inducible co-stimulator ligand and muscle tory myopathies (reviewed in Hohlfeld and Engel, 1994; Dalakas, 1995, 2002). Human muscle cells can present antigens to CD4+ T cells (Goebels et al., 1992) and express as yet unidenti®ed co-stimulatory molecules associated with the B7 family, but distinct from CD80 (B7.1) and CD86 (B7.2) (Behrens et al., 1998; Murata and Dalakas, 1999). Inducible co-stimulator ligand (ICOSL) is a member of the B7 family of co-stimulatory ligands which shares 19±20% sequence identity with CD80 and CD86 (reviewed in Coyle and Gutierrez-Ramos, 2001; Carreno and Collins, 2002; Liang and Sha, 2002; Sharpe and Freeman, 2002). In humans, cell surface expression of ICOSL has been described on B cells, dendritic cells, monocytes/macrophages, T cells and endothelial cells (Coyle and Gutierrez-Ramos, 2001; Carreno and Collins, 2002; Khayyamian et al., 2002; Liang and Sha, 2002). ICOSL mRNA expression has been detected in a variety of lymphoid and non-lymphoid organs (Ling et al., 2000), but the functional signi®cance of ICOSL on nonlymphoid cells has remained unclear. ICOSL binds to ICOS, a T cell-speci®c co-stimulatory molecule homologous to CD28 and CTLA-4 (Hutloff et al., 1999; Yoshinaga et al., 1999; Beier et al., 2000). ICOS±ICOSL interactions have strong impact on CD4 and CD8 T cell-mediated immune responses, as shown in a number of in vivo studies using ICOS knockout mice or by approaches blocking ICOS± ICOSL interactions in animal models (reviewed in Sperling and Bluestone, 2001; Carreno and Collins, 2002; Sharpe and Freeman, 2002). In addition to their important role in T-cell differentiation and T-cell±B-cell collaboration, ICOSL± ICOS interactions play a prominent role in the co-stimulation of CD4 as well as CD8 effector or memory T-cell responses (Hutloff et al., 1999; Coyle et al., 2000; Sperling and Bluestone, 2001; reviewed in Carreno and Collins, 2002; Sharpe and Freeman, 2002). Our results demonstrate that ICOSL is expressed in muscle, and this co-stimulatory molecule affects functional interactions between muscle cells and T cells. Material and methods Antibodies and reagents The following primary antibodies were used: anti-human ICOSL and anti-human ICOSL±phycoerythrin (PE) (HIL131; Khayyamian et al., 2002), anti-ICOS (F44), all generously provided by R.A. Kroczek (Robert-Koch-Insitut, Berlin, Germany), anti-MHC I (W6/32), anti-MHC II (L243), anti-CD3 (OKT3), anti-CD4 (RPA-T4), anti-CD8 (B9.11), anti-CD11b (BEAR1), anti-CD14 (B-A8), antiCD20 (B9E9), anti-CD56 (anti-NCAM, MY31), anti-CD80 (BB-1), anti-CD86 (B-T7) and anti-pan-a/b-T cell receptor (TCR) (T10B9.1A-31) (all BD PharMingen, Heidelberg, Germany). Secondary antibodies and isotype controls were: goat anti-mouse F(ab¢)2±di-chlorotriazinyl-¯uorescein (DTAF), goat anti-mouse-PE IgG (H+L) F(ab¢)2 fragment (Dianova, Hamburg, Germany), mouse IgG1k (MOPC-21, 1027 Sigma, Saint Louis, MO, USA) and mouse IgG (Linaris, Wertheim, Germany). All antibodies were titrated for ¯ow cytometry. A ®nal concentration of 10±30 mg/ml acid-free antibody was used for blocking experiments. Interferon-g (IFN-g) and tumour necrosis factor-a (TNF-a) were from Peprotech EC Ltd (London, UK). Normal goat serum was from Dianova (Hamburg, Germany), human IgG (Alphaglobinâ) was obtained from Grifols (Langen, Germany), glatiramer acetate was obtained from Teva Pharmaceutical Industries (Petah Tiqva, Israel), phytohaemagglutinin was from Biochrom KG (Berlin, Germany) and CD3/CD28 beads were from Dynal Biotech (Hamburg, Germany). Superantigens (staphylococcal enterotoxin B, staphylococcal enterotoxin A, and toxic shock syndrome toxin-1, all from Toxin Technology, Sarasota, FL) were used as a SAg mix (1.5 pg/ml each). Clinical material Diagnostic muscle biopsy specimens were obtained from patients with in¯ammatory myopathies (polymyositis, n = 7: age range 51±89 years; dermatomyositis, n = 5: age range 61±70 years; inclusion body myositis, n = 6: age range 58±80 years) and non-in¯ammatory or non-myopathic controls (n = 7, age range: 20±62 years). Patients were not under immunosuppressive therapy at time of the biopsy. Immunohistochemical studies Flash-frozen muscle biopsy specimens were cut into 8±10 mm cryostat sections and analysed by immunohistochemistry. Acetone-®xed, air-dried sections were blocked and incubated with primary antibodies or corresponding non-immune IgG isotype controls, diluted in phosphate-buffered saline (PBS) containing 2% bovine serum albumin (BSA) for 45 min at the concentrations optimized by titrating the respective primary antibodies. The reaction product was visualized with the streptavidin±biotin method (reagents from DAKO, Hamburg, Germany) using diaminobenzidine (Serva, Heidelberg, Germany) as an electron donor. Alternatively, antibody binding was visualized by immuno¯uorescence microscopy [Chromophore Cy3- or ¯uorescein isothiocyanate (FITC)labelled secondary antibodies; Dianova, Hamburg, Germany]. Umbilical cord (endothelium of the vessels) was used as a positive control. Cell isolation and culture Myoblasts were isolated from normal subjects or patients with in¯ammatory myopathy, puri®ed by magnetic bead separation and cultured as previously described (Wiendl et al., 2002). Written consent was obtained from all donors, and tissue sampling was approved by the local ethics committee (Munich). Myoblasts stained >95% positive for the neural cell adhesion molecule (NCAM) by ¯ow cytometry. TE671 cells were obtained from ATCC. Peripheral blood mono- 1028 H. Wiendl et al. nuclear cells (PBMCs) were isolated from peripheral blood of healthy volunteers by density gradient centrifugation using Biocoll Separating Solution (Biochrom KG, Berlin, Germany). Monocytes were depleted by adhesion to plastic ¯asks for 1 h. CD4 T cells were puri®ed with a negative isolation kit (Dynal Biotech, Hamburg, Germany); only >95% pure CD4 T cells were used in the experiments. Cytokine induction and ¯ow cytometry For analysis of surface molecules and co-culture experiments, myoblasts or TE671 cells were cultured in the presence of 500 U/ml IFN-g and/or 500 U/ml TNF-a for the times indicated. Adherent cells were detached non-enzymatically using cell dissociation buffer (InvitrogenÔ, Heidelberg, Germany). All cells were washed with FACS (¯uorescenceactivated cell sorting) buffer containing PBS with 0.1% BSA and 0.1% sodium azide. To minimize binding to Fc receptors, PBMCs, puri®ed T cells and myoblast cells were blocked with human immunoglobulins at 1.25 mg/ml for 10 min at 4°C. After one washing step, the unlabelled or labelled ®rst antibody was added. Isotype control monoclonal antibodies (mAbs) were used at the same concentration as the primary antibody. Incubation was done on ice for 45 min, followed by two washes. Goat anti-mouse IgG [F(ab)2±PE (5 mg/ml, Sigma)] was used as secondary antibody. Histograms of stained cells were analysed by calculating the speci®c ¯uorescence index (speci®c geometric mean:geometric mean of isotype). For labelled PBMCs, quadrant analysis of dot blots was performed. Unspeci®c isotype binding was substracted for each sample. Surface expression was quanti®ed by ¯ow cytometry on a ¯uorescence-activated cell sorter (FACS Calibur cytometer, Becton Dickinson BD, Heidelberg, Germany). Data were analysed using CellQuestâ software. Co-culture experiments of myoblasts and T cells Myoblasts were plated in six-well plates (Costar, Bodenheim, Germany) at a density of 5 3 105 and cultured in the absence or presence of IFN-g (500 U/ml) and TNF-a (500 U/ml) for 24±48 h to induce MHC II expression. After washing the cells, 5 3 105 puri®ed CD4 T cells isolated with a negative isolation kit (Dynal Biotech) were added in the presence of a SAg mix (see above). Cells subsequently were co-cultured in RPMI1640 supplemented with 1% glutamine (Gibco Life Technologies, Paisley, UK), 10% foetal calf serum (Fetal Calf Serum Gold, PAA Laboratories, Linz, Austria) and penicillin (100 IU/ml)/streptomycin (100 mg/ml) (Gibco). The purity of the isolated T-cell populations used in the experiments was >95%. Where indicated, anti-MHC II (L243), anti-ICOSL (HIL-131) or an isotype control antibody (MOPC-21; 15 mg/ml each) were added to the co-cultures. For measurement of cytokine production at the indicated times, T cells alone or T cells plus myoblasts were suspended in Trizolâ and subjected to RNA isolation. RNA extraction, cDNA synthesis and polymerase chain reaction (PCR) Total RNA extraction was performed using peqGOLD Tri FastÔ isolation reagent (peqLab, Erlangen, Germany). For ®rst strand cDNA synthesis, 2.5 mg of total RNA were dissolved in 21.5 ml of diethylpyrocarbonate (DEPC)-treated double-distilled water (H2Odd). A 2 ml aliquot of random hexamers (200 ng/ml) was added to each sample prior to incubation at 70°C for 10 min. Samples were cooled on ice and subjected to a mixture consisting of 53 M-MLV reverse transcriptase (RT) buffer (10 ml/sample, Promega, Madison, WI, USA), dNTPs (10 mM, 10 ml/sample), RNasin (40 U/ml, 0.25 ml/sample, Promega), M-MLV RT (200 U/ml, 1 ml/ sample, Promega) and DEPC-treated H2Odd (5.25 ml/sample). This mixture of 26.5 ml was added to each RNA/random hexamer solution. Samples were mixed and incubated for 10 min at room temperature, then for 50 min at 42°C and ®nally for 15 min at 70°C. For conventional PCR, 1 ml of cDNA was used to amplify the products of interest in a 20 ml standard PCR; 18S RNA was ampli®ed as a control. Speci®city of ampli®cation was con®rmed by direct DNA sequencing of extracted bands after reampli®cation. For quantitative real-time PCR, measurement of gene expression was performed utilizing the ABI prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). The HUSAR Genius software package (http://genius.embnet.dkfz-heidelberg.de, DKFZ, Heidelberg, Germany) was used to design the primers for ampli®cation. Primers (BioChip Technologies GmbH, Freiburg, Germany) were designed when possible to span exon±exon junctions in order to prevent ampli®cation of genomic DNA and to result in amplicons of <150 bp to enhance the ef®ciency of PCR ampli®cation. Relative quanti®cation of speci®c gene expression was performed by two-step real-time PCR using cDNA as a template. Templates were multiplied using PE Applied Biosystems (Perkin Elmer) SYBRâ Green PCR Master Mix [containing hot-start-AmpliTaqGoldâ, SYBRâ Green PCR buffer (23), MgCl2 and dNTPs]. RT±PCR of cDNA specimens was conducted in a total volume of 15 ml with 13 Taq Man Master Mix (Perkin Elmer) with primers at optimized concentrations. Thermal cycler parameters were 2 min at 50°C, 10 min at 95°C and 40 cycles of denaturation at 95°C for 15 s followed by annealing/extension at 60°C for 1 min. The ¯uorescence resulting from binding of SYBRâ Green dye to double-stranded DNA was measured directly in the PCR tube. Data were analysed with the ABI PRISMâ Detection System using the comparative CT (threshold cycle) method (PE Applied Biosystems, User Bulletin). Samples were normalized to 18S rRNA, in order to account for the variability in the initial concentration of the total RNA and conversion ef®ciency of the RT reaction. Product speci®city of the PCR products was con®rmed by agarose gel electrophoresis or dissociation curve analysis. The internal reference dye ROX included in the PCR buffer was Inducible co-stimulator ligand and muscle 1029 Fig. 1 Expression of ICOSL and its receptor ICOS in muscle tissue specimens in vivo. Detection of ICOSL (HIL-131) was performed by immunohistochemistry. Transverse serial cryostat sections from a non-myopathic control stained with anti-ICOSL mAb HIL-131 (B) and isotype control antibody (A) show low levels of constitutive ICOSL expression of muscle ®bres and clear staining of capillaries (arrowhead). (C) A cryostat section from one patient with polymyositis (patient 1) stained with anti-ICOSL mAb HIL-131 shows enhanced expression of ICOSL that was found predominantly in areas of in¯ammatory cell invasion and in®ltration. Fibres surrounded or invaded by in¯ammatory cells (asterisks) stain more intensely for ICOSL. A transverse section from another patient with polymyositis (patient 2) stained with ICOSL mAb HIL-131 (D) shows enhanced ICOSL expression of muscle ®bres and blood vessel endothelium (arrowhead) (compare with B). (E) A section from a patient with dermatomyositis shows strong ICOSL expression on endothelial cells of larger blood vessels within the tissue specimen (asterisks). (F) Endothelial cells of the umbilical cord venule served as a positive control for ICOSL staining with HIL-131. (G) Immuno¯uorescence shows the expression of ICOS on T cells (green) lying in apposition to muscle ®bres (asterisks), suggesting that T cells are activated. Magni®cations 3200 (A, B, D, E and F), 3300 (C) and 3400 (G). 1030 H. Wiendl et al. used to check for ¯uorescence ¯uctuations caused by changes in concentration or volume. All PCR assays were performed in duplicate. The following oligonucleotides were used in this study. 18S: 18s-for, 5¢-CGGCTACCACATCCAAGGAA; 18s-rev, 5¢-GCTGGAATTACCGCGGCT; ICOSL (B7H2): ICOSLfor, 5¢-CGTGTACTGGATCAATAAGACGG; ICOSL-rev, 5¢-TGAGCTCCGGTCAAACGTGGCC (Ling et al., 2000); ICOSL (GL50): GL50-for, 5¢-CTTGTGGTCGTGGCGGTG; GL50-rev, 5¢-TCACGAGAGCAGAAGGAGCAGGTTCC (Ling et al., 2000); IFN-g: IFN-g-for, 5¢-TTCAGCTCTGCATCGTTTTG; IFN-grev, 5¢-CTTTCCAATTCTTCAAAATGCC; interleukin-4: IL-4-for, 5¢-CTTTGAACAGCCTCACAGAGC; IL-4-rev, 5¢-AACTGCCGGAGCACAGTC; IL-10: IL-10-for, 5¢-GTTTTACCTGGAGGAGGTGATG; IL-10-rev, 5¢-GGCCTTGCTCTTGTTTTCAC. Results Analysis of ICOSL expression in muscle biopsy specimens by immunohistochemistry To assess whether the B7 family co-stimulatory molecule ICOSL is expressed in muscle tissue, we analysed 25 muscle biopsy specimens from patients with polymyositis, inclusion body myositis, dermatomyositis, non-in¯ammatory myopathic controls (Duchenne muscular dystrophy) and nonmyopathic controls by immunohistochemistry using mAbs against ICOSL (HIL-131) (Khayyamian et al., 2002), MHC I (W6/32) and MHC II (L243) molecules, and T-cell markers (anti-CD4, RPA-T4; anti-CD8, B9.11, anti-CD3, OKT3). As a positive control for ICOSL expression, we used human umbilical cord, where strong endothelial expression was observed (Khayyamian et al., 2002). HLA class I and class II antigens are not detectable on normal muscle ®bres in situ, but HLA class I, and in some instances HLA class II, antigens are strongly upregulated on muscle ®bres in in¯ammatory myopathies (Karpati et al., 1988; Emslie-Smith et al., 1989; Wiendl et al., 2000; and data not shown). Low levels of constitutive ICOSL expression were observed on the surface of muscle ®bres in our muscle biopsy specimens from nonmyopathic controls (Fig. 1A and B) and in non-in¯ammatory myopathies (data not shown). The staining intensity of ICOSL was markedly enhanced on muscle ®bres in in¯ammatory myopathies, especially polymyositis (seven of seven specimens) (Fig. 1C and D). In polymyositis, muscle-related expression of ICOSL was strongest in areas where in¯ammatory cells surrounded or invaded muscle ®bres. Some regenerating or degenerating muscle ®bres also showed ICOSL staining of the cytoplasm (Fig. 1C). It is of note that muscle capillaries showed constitutive expression of ICOSL, regardless of the origin of the muscle biopsy specimens (nonmyopathic or myopathic) (Fig. 1B±E). Further, some of the mononuclear cells within the in¯ammatory tissue sections also stained positive for ICOSL, most probably re¯ecting ICOSL-expressing B cells and monocytes (Fig. 1C and D). The potential relevance of ICOSL±ICOS interactions for disease pathogenesis of in¯ammatory myopathies was delineated further by detecting the expression of ICOS on T cells in some polymyositis specimens (Fig. 1G). Since ICOS is known to be expressed on CD4 or CD8 T cells upon activation (Hutloff et al., 1999), our ®ndings suggest that muscle-derived ICOSL may interact directly with ICOS on activated T cells. Further, ICOSL expression might be upregulated in the presence of in¯ammatory cytokines released by in¯ammatory cells or destroyed muscle ®bres. ICOSL protein expression in cultured human myoblasts and TE671 rhabdomyosarcoma cells To con®rm that muscle cells express ICOSL and to identify potential regulatory cytokines for ICOSL expression, we analysed several lines of cultured myoblasts for ICOSL expression using mAb HIL-131 (Khayyamian et al., 2002). In all experiments, myoblasts were >95% positive for NCAM (CD56), which serves as a reliable marker for myoblasts (Illa et al., 1992; Fig. 2A). As reported previously, myoblasts constitutively express MHC I. MHC II is induced by IFN-g but not TNF-a (Hohlfeld and Engel, 1990a; Goebels et al., 1992; Fig. 2A). There were low levels of constitutively expressed ICOSL on six of eight tested myoblast cultures derived from different donors (Fig. 2A). No differences were observed between myoblasts derived from patients with in¯ammatory myopathies and non-myopathic control subjects. Speci®city of staining was demonstrated in some experiments by cold blocking: unlabelled anti-ICOSL antibody was added prior to staining with the PE-labelled HIL131 at 1000-fold excess, and abolished staining (data not shown). We next investigated the effects of in¯ammatory cytokines on the expression of ICOSL. Myoblasts were incubated for 24, 48 or 72 h with IFN-g, TNF-a or both. ICOSL expression was upregulated by TNF-a, whereas IFN-g had no such effect (Fig. 2 and data not shown). Further, we investigated TE671 muscle rhabdomyosarcoma cells for ICOSL expression. TE671 cells constitutively express MHC I but no MHC II. These cells exhibited strong constitutive expression of ICOSL which was not modulated further by treatment with in¯ammatory cytokines (Fig. 2B and data not shown). ICOSL RNA expression in cultured human myoblasts and muscle biopsy specimens Two splice variants of human ICOSL have been described and designated B7-H2/B7RP-1/hLICOS (Brodie et al., 2000; Wang et al., 2000; Yoshinaga et al., 2000) and hGL50 (Ling et al., 2000). To corroborate our results and to identify the splice variants expressed in muscle, we analysed total RNA for ICOSL transcripts by RT±PCR using primers that allow the ampli®cation of B7-H2 or hGL50. Direct visualization of the PCR products showed B7-H2 transcripts in unstimulated Inducible co-stimulator ligand and muscle 1031 Fig. 2 ICOSL is expressed in cultured muscle cells in vitro. Flow cytometric analysis of ICOSL cell surface expression on cultured myoblasts. Puri®ed unstimulated NCAM+ myoblasts express MHC I (mAb W6/32) and low levels of ICOSL (mAb HIL-131). Surface expression of ICOSL was enhanced by TNF-a but not by IFN-g, as shown for myoblasts 48 h after induction. Histograms show staining with the designated antibodies (shaded) underlaid with the isotype controls for ICOSL (mIgG1). (B) Flow cytometric analysis of ICOSL surface expression on TE671 human rhabdomyosarcoma cells. Histograms show staining with the designated antibodies (shaded) underlaid with the isotype controls for ICOSL (mIgG1). (C) RT±PCR of ICOSL transcript B7-H2 shows basal expression in unstimulated cultured human myoblasts (myoblasts I, II and III; myoblast III corresponds to myoblasts shown in (A). After stimulation with TNF-a for 48 h, the B7-H2 transcript was upregulated. A moderate RNA upregulation was observed also in response to IFN-g. 18S RNA was used as an internal control for the quality of cDNA (lower panel). Myoblasts I, II and III are derived from three different donors and are shown as representative examples for these experiments. B cells and monocytes were used as positive controls for B7-H2. NTC = no template control. 1032 H. Wiendl et al. expression under any conditions in cultured human myoblasts or TE671 cells (data not shown). It is of note that we investigated total RNA derived from muscle biopsy specimens for expression of ICOSL and found the B7-H2 isoform in 15 of 17 tested samples derived from donors without pathological abnormalities and donors with in¯ammatory myopathies (data not shown). ICOSL on myoblasts co-stimulates Th1 and Th2 cytokine mRNA synthesis of CD4 T cells To demonstrate that the expression of ICOSL on muscle cells has functional consequences, we performed co-culture experiments with allogeneic CD4 T cells and assessed the modulation of T-cell cytokine mRNA synthesis. Myoblasts were pre-induced to express MHC II by addition of IFN-g and to upregulate ICOSL by TNF-a, and subsequently cocultured with puri®ed CD4 T cells and SAg in the absence or presence of the inhibitory anti-ICOSL mAb HIL-131. The production of Th1 (IFN-g) and Th2 cytokines (IL-4 and IL10) was measured by assessing RNA synthesis via quantitative real-time PCR. HIL-131 inhibited the production of Th1 as well as Th2 cytokines between 52 and 61% (IFN-g, 61 6 12%; IL-4, 57 6 12%; IL-10, 52 6 11%) (Fig. 3). In comparison, L243, an anti-MHC II antibody used to block the interaction of the trimolecular complex (MHC±SAg±TCR), reduced cytokine production by ~90% (81±92%). It is of note that no relevant cytokine mRNA synthesis was observed when myoblasts were cultured together with SAg in the absence of CD4 T cells. Discussion Fig. 3 Inhibition of cytokine production in myoblast and T-cell cocultures by blocking ICOSL±ICOS interaction. Myoblasts (Myo) were cultured in the presence of IFN-g (500 U/ml)/TNF-a (500 U/ ml) to upregulate MHC II and ICOSL expression, and co-cultured with SAg and puri®ed CD4 T cells (T) in the presence of an isotype control (MOPC-21), anti-ICOSL mAb (HIL-131) or antiMHC class II mAb (L243). CD4 T cells plus SAg without myoblasts (T + SAg) and myoblasts plus SAg (Myo + SAg) were used as controls. Cytokine mRNA synthesis by T cells was assessed by quantitative real-time PCR after 24 h of co-culture. Cytokine mRNA synthesis of the isotype control was set to 100% and data are expressed as percentages of mRNA relative to the isotype control. Error bars represent the means 6 SEM of three experiments. nm, not measurable. cultured myoblasts (Fig. 2C). Corresponding to the protein expression data, RNA expression was upregulated by addition of TNF-a, and to a lesser extent by IFN-g (Fig. 2C). In contrast to B cells and monocytes, which were used as a positive control, direct visualization of RT±PCR products did not result in the detection of hGL50 mRNA Muscle ®bres and myoblasts can actively participate in immune cell interactions (e.g. Hohlfeld and Engel, 1990b; Goebels et al., 1992; Wiendl et al., 2003) and thus probably play an important role in initiating or perpetuating muscledirected immune responses (reviewed in Hohlfeld and Engel, 1994; Nagaraju, 2001; Dalakas, 2002). (i) Cultured muscle cells induce primary immune responses in allogeneic situations (Curnow et al., 1998) and are killed ef®ciently by cytotoxic T cells and natural killer (NK) cells (Hohlfeld and Engel, 1990b; Wiendl et al., 2003). (ii) In polymyositis and sporadic inclusion body myositis, CD8+ T cells and macrophages invade initially non-necrotic muscle ®bres (Arahata and Engel, 1988). The expression of MHC I in all of the invaded muscle ®bres and some non-invaded ®bres has been taken as evidence for an HLA class I-restricted cytotoxic T cell-mediated response against surface antigens expressed on muscle ®bres (reviewed in Hohlfeld and Engel, 1994; Nagaraju, 2001; Dalakas, 2002). (iii) Myoblasts inducibly express MHC II molecules and have the capability to present antigens to CD4 T cells (Goebels et al., 1992; Curnow et al., 2001). Inducible co-stimulator ligand and muscle The muscle cells armamentarium with co-stimulatory molecules has been the subject of ongoing debate. Presence of 37.5 (CD80/86) is believed to be a prerequisite for antigenpresenting cells to initiate immune responses according to the two-signal model of T-cell activation (Lafferty and Woolnough, 1977). However, muscle ®bres and cultured myoblasts do not express detectable levels of B7.1 (CD80) or B7.2 (CD86) protein under normal or in¯ammatory conditions (Behrens et al., 1998; Bernasconi et al., 1998; Murata and Dalakas, 1999; Nagaraju et al., 1999), but nonetheless are capable of augmenting antigen-speci®c T-cell responses. Muscle cells therefore have been postulated to express other co-stimulatory molecules including an as yet unidenti®ed B7-related co-stimulatory protein (BB-1) that interacts with CD28/CTLA-4 and believed to exhibit strong co-stimulatory function (Behrens et al., 1998; Murata and Dalakas, 1999). The important advance of our present work is the demonstration of a recently identi®ed and molecularly de®ned member of the B7 family, ICOSL, functionally expressed in muscle under physiological and pathological conditions in vivo as well as in cultured human myoblasts in vitro. Despite the existing data showing that muscle cells can act as potent antigen-presenting cells, the lack of CD80/ 86 has questioned the potential of muscle to prime naive T cells per se and maintained the debate on how and when muscle cells would interact with immune cells during the initiating or perpetuating events of in¯ammatory myopathies. In this view, the characterization of novel members of the B7 family of co-stimulatory molecules that exhibit a broader tissue distribution than B7.1/2 importantly advances our understanding of immune reactions directed against or derived from non-lymphoid tissues (reviewed in Coyle and Gutierrez-Ramos, 2001; Carreno and Collins, 2002; Liang and Sha, 2002; Sharpe and Freeman, 2002). ICOSL is a member of the B7 family of co-stimulatory ligands that has strong impact on T cell-mediated immune responses by interacting with ICOS on T cells (reviewed in Coyle and Gutierrez-Ramos, 2001; Carreno and Collins, 2002; Liang and Sha, 2002; Sharpe and Freeman, 2002). Two splice variants of human ICOSL have been described and designated hGL50 (Ling et al., 2000) and B7-H2/B7RP-1/hLICOS (Brodie et al., 2000; Wang et al., 2000; Yoshinaga et al., 2000). hGL50 showed a more lymphoid-restricted expression pattern (spleen, lymph node), whereas B7-H2/B7RP-1/ hLICOS mRNA was expressed in a broader variety of organs tested (Ling et al., 2000). We report here B7-H2 mRNA and protein expression in muscle tissue, cultured myoblasts and TE671 rhabdomyosarcoma cells. ICOSL expression was enhanced in muscle cells under in¯ammatory conditions in vivo and in vitro, and the in¯ammatory cytokine TNF-a augmented cell surface expression of ICOSL on cultured myoblasts. We further showed that ICOSL expression on cultured muscle cells is functional in that it modulates the cytokine production by activated T cells. Thus, since ICOSL± ICOS interactions have strong impact on CD4+ and CD8+ 1033 T-cell responses, this pathway may be of critical relevance for immune reactions under different conditions in muscle. The expression patterns, inducibility characteristics in vitro and receptor interactions of ICOSL make it unlikely that this molecule represents the postulated B7-related co-stimulatory molecule BB-1 in muscle cells, for the following reasons (Behrens et al., 1998; Murata and Dalakas, 1999): the putative B7 family co-stimulatory molecule BB-1 shows no constitutive expression of ICOSL on cultured myoblasts nor on capillaries (Behrens et al., 1998); further, BB-1 has been suggested to interact with CTLA-4 (Behrens et al., 1998; Murata and Dalakas, 1999), a condition that does not hold true for ICOSL interacting with ICOS on activated T cells. Current models propose that ICOSL±ICOS interactions play a more prominent role in the co-stimulation of effector or memory T-cell responses (Hutloff et al., 1999; Coyle et al., 2000), whereas CD28 co-stimulates primary T-cell functions. It is tempting to speculate that muscle-associated ICOSL augments the production of Th1 and Th2 cytokines by interaction with CD4 effector/memory T cells also in vivo, thus augmenting the muscle-directed antigen-speci®c immune responses in in¯ammatory myopathies. Further, ICOSL±ICOS interactions in¯uence the clonal expansion and the cognate destruction of tumour cells by CD8 cytotoxic lymphocytes, thus extending the immunological role of ICOSL±ICOS interactions to CD8 T-cell functions (Liu et al., 2001; Wallin et al., 2001). This may be of particular importance for MHC I-expressing muscle cells presenting the target antigen(s) on the muscle ®bre surface to cytotoxic T cells in polymyositis or inclusion body myositis. We also observed ICOSL expression on endothelial cells in muscle. It is therefore conceivable that ICOSL is of relevance for the immune pathogenesis of dermatomyositis where endothelial cells represent the primary target structure for the autoimmune process. ICOSL could play an important role in the (re)activation of effector/memory T cells on the endothelium either controlling the entry of immune cells into in¯amed tissue or augmenting antibody production by B cells. It should be noted that the signi®cance of ICOSL expression in muscle reaches beyond its implications for the pathogenesis and treatment of in¯ammatory myopathies. Muscle is the site for many other immune reactions, including muscle infections, graft-versus-host disease, conventional intramuscular vaccination and, intramuscular gene transfer by injection of genetically engineered myoblasts or naked DNA (Hohlfeld and Engel, 1994; Blau and Springer, 1995; Miller and Boyce, 1995; Pardoll and Beckerleg, 1995). In all these reactions, the expression of ICOSL on muscle ®bres could play an important role. Acknowledgements This work was supported by grants from the Deutsche Forschungsgemeinschaft (Wi 1722/2-1 to H.W., SFB571 A-1 to R.H., and Lo 549/2-3 to H.L.) and the IZKF TuÈbingen (M.W.). The Institute for Clinical Neuroimmunology is 1034 H. Wiendl et al. supported by the Hermann and Lilly Schilling Foundation. Human myoblast cultures were obtained from the Muscle Tissue Culture Collection at the Friedrich-Baur-Institute (Department of Neurology, Ludwig-MaximiliansUniversity, Munich, Germany). The Muscle Tissue Culture Collection is supported by the `Association Francaise contre les Myopathies' (Paris, France) and the German Ministry for Education and Research (BMBF; MD-NET, project S1). Emslie-Smith AM, Arahata K, Engel AG. Major histocompatibility complex class I antigen expression, immunolocalization of interferon subtypes, and T cell-mediated cytotoxicity in myopathies. Hum Pathol 1989; 20: 224±31. References Arahata K, Engel AG. Monoclonal antibody analysis of mononuclear cells in myopathies. IV: cell-mediated cytotoxicity and muscle ®ber necrosis. Ann Neurol 1988; 23: 168±73. Hohlfeld R, Engel AG. Lysis of myotubes by alloreactive cytotoxic T cells and natural killer cells. Relevance to myoblast transplantation. J Clin Invest 1990b; 86: 370±4. Behrens L, Kerschensteiner M, Misgeld T, Goebels N, Wekerle H, Hohlfeld R. Human muscle cells express a functional costimulatory molecule distinct from B7.1 (CD80) and B7.2 (CD86) in vitro and in in¯ammatory lesions. J Immunol 1998; 161: 5943±51. Beier KC, Hutloff A, Dittrich AM, Heuck C, Rauch A, Buchner K, et al. Induction, binding speci®city and function of human ICOS. Eur J Immunol 2000; 30: 3707±17. Bernasconi P, Confalonieri P, Andreetta F, Baggi F, Cornelio F, Mantegazza R. The expression of co-stimulatory and accessory molecules on cultured human muscle cells is not dependent on stimulus by pro-in¯ammatory cytokines: relevance for the pathogenesis of in¯ammatory myopathy. J Neuroimmunol 1998; 85: 52±8. Blau HM, Springer ML. Muscle-mediated gene therapy. N Engl J Med 1995; 333: 1554±6. Brodie D, Collins AV, Iaboni A, Fennelly JA, Sparks LM, Xu XN, et al. LICOS, a primordial costimulatory ligand? Curr Biol 2000; 10: 333±6. Goebels N, Michaelis D, Wekerle H, Hohlfeld R. Human myoblasts as antigen-presenting cells. J Immunol 1992; 149: 661±7. Hohlfeld R, Engel AG. Induction of HLA-DR expression on human myoblasts with interferon-gamma. Am J Pathol 1990a; 136: 503±8. Hohlfeld R, Engel AG. The immunobiology of muscle. Immunol Today 1994; 15: 269±74. Hutloff A, Dittrich AM, Beier KC, Eljaschewitsch B, Kraft R, Anagnostopoulos I, et al. ICOS is an inducible T-cell co-stimulator structurally and functionally related to CD28. Nature 1999; 397: 263±6. Illa I, Leon-Monzon M, Dalakas MC. Regenerating and denervated human muscle ®bers and satellite cells express neural cell adhesion molecule recognized by monoclonal antibodies to natural killer cells. Ann Neurol 1992; 31: 46±52. Karpati G, Pouliot Y, Carpenter S. Expression of immunoreactive major histocompatibility complex products in human skeletal muscles. Ann Neurol 1988; 23: 64±72. Khayyamian S, Hutloff A, Buchner K, Grafe M, Henn V, Kroczek RA, et al. ICOS-ligand, expressed on human endothelial cells, costimulates Th1 and Th2 cytokine secretion by memory CD4+ T cells. Proc Natl Acad Sci USA 2002; 99: 6198±203. Lafferty KJ, Woolnough J. The origin and mechanism of the allograft reaction. Immunol Rev 1977; 35: 231±62. Carreno BM, Collins M. The B7 family of ligands and its receptors: new pathways for costimulation and inhibition of immune responses. Annu Rev Immunol 2002; 20: 29±53. Liang L, Sha WC. The right place at the right time: novel B7 family members regulate effector T cell responses. Curr Opin Immunol 2002; 14: 384±90. Coyle AJ, Gutierrez-Ramos JC. The expanding B7 superfamily: increasing complexity in costimulatory signals regulating T cell function. Nat Immunol 2001; 2: 203±9. Ling V, Wu PW, Finnerty HF, Bean KM, Spaulding V, Fouser LA, et al. Cutting edge: identi®cation of GL50, a novel B7-like protein that functionally binds to ICOS receptor. J Immunol 2000; 164: 1653±7. Coyle AJ, Lehar S, Lloyd C, Tian J, Delaney T, Manning S, et al. The CD28-related molecule ICOS is required for effective T cell-dependent immune responses. Immunity 2000; 13: 95±105. Curnow SJ, Willcox N, Vincent A. Induction of primary immune responses by allogeneic human myoblasts: dissection of the cell types required for proliferation, IFNgamma secretion and cytotoxicity. J Neuroimmunol 1998; 86: 53±62. Curnow J, Corlett L, Willcox N, Vincent A. Presentation by myoblasts of an epitope from endogenous acetylcholine receptor indicates a potential role in the spreading of the immune response. J Neuroimmunol 2001; 115: 127±34. Dalakas MC. Immunopathogenesis of in¯ammatory myopathies. Ann Neurol 1995; 37 Suppl 1: S74±86. Dalakas MC. In¯ammatory myopathies. Recent advances in pathogenesis and therapy. Adv Neurol 2002; 88: 253±71. Liu X, Bai XF, Wen J, Gao JX, Liu J, Lu P, et al. B7H costimulates clonal expansion of, and cognate destruction of tumor cells by, CD8(+) T lymphocytes in vivo. J Exp Med 2001; 194: 1339±48. Miller JB, Boyce FM. Gene therapy by and for muscle cells. Trends Genet 1995; 11: 163±5. Murata K, Dalakas MC. Expression of the costimulatory molecule BB-1, the ligands CTLA-4 and CD28, and their mRNA in in¯ammatory myopathies. Am J Pathol 1999; 155: 453±60. Nagaraju K. Immunological capabilities of skeletal muscle cells. Acta Physiol Scand 2001; 171: 215±23. Nagaraju K, Raben N, Villalba ML, Danning C, Loef¯er LA, Lee E, et al. Costimulatory markers in muscle of patients with idiopathic in¯ammatory myopathies and in cultured muscle cells. Clin Immunol 1999; 92: 161±9. Inducible co-stimulator ligand and muscle 1035 Pardoll DM, Beckerleg AM. Exposing the immunology of naked DNA vaccines. Immunity 1995; 3: 165±9. Hohlfeld R, et al. An autoreactive gd TCR derived from a polymyositis lesion. J Immunol 2002; 169: 515±21. Sharpe AH, Freeman GJ. The B7-CD28 superfamily. Nature Rev Immunol 2002; 2: 116±26. Wiendl H, Mitsdoerffer M, Hofmeister V, Wischhusen J, Weiss EH, Dichgans J, et al. The non-classical MHC molecule HLA-G protects muscle cells from immune-mediated lysis: implications for myoblast transplantation and gene therapy. Brain 2003; 126: 176±85. Sperling AI, Bluestone JA. ICOS costimulation: it's not just for TH2 cells anymore. Nat Immunol 2001; 2: 573±4. Wallin JJ, Liang L, Bakardjiev A, Sha WC. Enhancement of CD8+ T cell responses by ICOS/B7h costimulation. J Immunol 2001; 167: 132±9. Wang S, Zhu G, Chapoval AI, Dong H, Tamada K, Ni J, et al. Costimulation of T cells by B7-H2, a B7-like molecule that binds ICOS. Blood 2000; 96: 2808±13. Wiendl H, Behrens L, Maier S, Johnson MA, Weiss EH, Hohlfeld R. Muscle ®bers in in¯ammatory myopathies and cultured myoblasts express the nonclassical major histocompatibility antigen HLA-G. Ann Neurol 2000; 48: 679±84. Wiendl H, Malotka J, Holzwarth B, Weltzein HU, Wekerle H, Yoshinaga SK, Whoriskey JS, Khare SD, Sarmiento U, Guo J, Horan T, et al. T-cell co-stimulation through B7RP-1 and ICOS. Nature 1999; 402: 827±32. Yoshinaga SK, Zhang M, Pistillo J, Horan T, Khare SD, Miner K, et al. Characterization of a new human B7-related protein: B7RP-1 is the ligand to the co-stimulatory protein ICOS. Int Immunol 2000; 12: 1439±47. Received October 29, 2002. Revised November 27, 2002. Accepted November 11, 2002