K. G. Haanstra, Department of Immunobiology, Biomedical Primate Research Centre, PO Box 3306, 2280 GH Rijswijk, the Netherlands. E-mail: email@example.com
Blockade of co-stimulation signals between T cells and antigen-presenting cells could be an important approach for treatment of autoimmune diseases and transplant rejection. Recently a series of small compound inhibitors which bind human CD80 (B7-1) and inhibit T cell co-stimulation has been described. To investigate their potency for clinical use, one of these compounds, RhuDex™, was evaluated for reactivity with rhesus monkey CD80. The in vitro biological effect on rhesus monkey lymphocytes, the potency for suppression of an inflammatory recall response and the protein-induced delayed type hypersensitivity (DTH) response in the skin were studied. In a rhesus monkey T cell co-stimulation assay RhuDex™ inhibited proinflammatory cytokine release and cellular proliferation with micromolar potency. Systemic administration of RhuDex™ to rhesus monkeys inhibited the DTH response significantly, indicating that this compound may inhibit autoimmune mediated inflammatory processes where the target, CD80, is up-regulated.
The full stimulation of T cells requires activation via the T cell receptor and via co-stimulatory molecules. Of the various receptor–ligand pairs involved in co-stimulation the CD80/CD86–CD28/CD152 [cytotoxic T lymphocyte antigen 4 (CTLA-4)] pathway is pivotal to the initiation and prolongation of T helper 1 (Th1) responses, which has been demonstrated in various rodent and non-human primate (NHP) studies. Thus the CD80/CD86–CD28/CD152 pathway could be an ideal target for suppressing immune-mediated imflammatory disorders where Th1 cell reactivity often predominates, while preserving the capacity of the immune system to combat pathogens [1–3].
CTLA-4 coupled to the Fc domain of an immunoglobulin (Ig)G1 molecule, CTLA4-Ig (also known as abatacept, or Orencia™), which blocks this pathway, is already used for the treatment of rheumatoid arthritis . Another blocker of this pathway, anti-CD80, Galiximab, has shown efficacy in clinical trials [4,5].
Recently, a series of small compound inhibitors that bind human CD80 and block T cell co-stimulation were reported . Here we describe the efficacy of one such compound: RhuDex™. RhuDex™ inhibited rhesus monkey T cell inflammatory cytokine production and proliferation in an in vitro CD80-dependent T cell stimulation assay. The effect of RhuDex™ on the ovalbumin (OVA)- and tetanus toxoid (TT)-induced delayed type hypersensitivity (DTH) response was investigated. CD80 expression was detected in keratinocytes, the non-professional antigen-presenting cells (APCs) of the skin, after hapten-induced contact hypersensitivity . No or only low-level CD80 expression was detected on Langerhans cells after hapten challenge [8,9]. A role for CD80 in DTH responses was confirmed in transgenic mice expressing CD80 in keratinocytes, which had exaggerated DTH responses to a hapten , but such a response was not detected after Candida albicans infection of the skin . We report here the significant inhibition of the OVA-induced DTH response by RhuDex™.
Materials and methods
Captive-bred 5–13 kg rhesus monkeys (Macaca mulatta) were either born and raised at the Biomedical Primate Research Centre or were purchased from a licensed breeder. Monkeys of Indian and Chinese origin were used. All procedures were performed in compliance with guidelines of the Animal Care and Use Committee in accordance with Dutch law.
Raji cells (1 × 105/well) were preincubated for 30 min at 37°C with RhuDex™ (a kind gift from Medigene, Martinsried, Germany). CD28/Fc chimera (342-CD; R&D Systems, Abingdon, UK) was added at the inhibitory concentration (IC)50 of CD28/Fc chimera binding to Raji cells (4·2 µg/ml). After incubation for 20 h at 37°C cells were washed and stained with an anti-human IgG monoclonal antibody (mAb) (709-116-14S; Jackson ImmunoResearch Laboratories, Cambridge, UK) for 30 min at 4°C. Cells were washed, fixated in formaldehyde and measured by fluorescence activated cell sorter (FACS). CD28/Fc chimera stains CD80 and CD86 on Raji cells. CD86 staining (typically somewhat less than 50%) with an anti-CD86 (clone IT2·2; BD Biosciences, San Diego, CA, USA) was subtracted and maximum staining was normalized to 100%.
T cell co-stimulation assay
Two rhesus monkey B cell lines with confirmed high expression of CD20, HLA-DR, CD40 and CD80 and with low CD86 expression were irradiated (50 Gy), dispensed at 2 × 105 cells/well and preincubated for 20 min at 37°C with either RhuDex™ or positive control inhibitors: 400 ng/ml cyclosporin A (CsA) (Sandimmune oral solution; Novartis, Basel, Switzerland), 5 µg/ml anti-CD80 mAb (clone L307·4; BD Pharmingen) or 5 µg/ml recombinant human CTLA-4-Ig (325-CT; R&D Systems). CD4+ T cells from peripheral blood mononuclear cells (PBMC) of rhesus monkeys of both Indian and Chinese origin (n = 6) were selected positively by magnetic affinity cell sorting using magnetic beads coated with anti-CD4 antibody (MACS) (Miltenyi Biotec, Bergisch Gladbach, Germany). The purified cell fraction was verified as > 97% CD4+. Responder CD4+ T cells (1 × 105/well) were added to the B cells in conjunction with an anti-CD3 mAb (1 µg/ml, clone SP34; BD Biosciences). Total culture volumes were 200 µl/well. Aliquots of supernatant (50 µl) were taken at 24 and 48 h and analysed for the cytokines interleukin (IL)-2, tumour necrosis factor (TNF)-α (both 24 h) and interferon (IFN)-γ (48 h) by enzyme-linked immunosorbent assay (ELISA) (U-CyTech, Utrecht, the Netherlands). For analysis of T cell proliferation, 0·5 µCi [3H]-thymidine was added to the cultures after 48–72 h. After a further 18 h, cells were harvested onto glass fibre filters for β-scintillation counting (Topcount NXT liquid scintillation counter; Packard, Ramsey, MN, USA).
Design of in vivo studies
Twelve animals of Indian origin were immunized with two intramuscular injections 2 weeks apart, with the antigens TT and OVA in adjuvant (see below). Two weeks after the last immunization animals were challenged (see below) for the first time. No treatment was given. From these 12 animals, eight animals with the strongest DTH responses to challenge were selected for the cross-over treatment study.
The DTH performed to select the eight animals was taken as the baseline response. Subsequently, these eight animals were randomized into two groups of four and a cross-over treatment study was undertaken. One week after the baseline challenge, four animals were treated with RhuDex™ (see below) and four with vehicle. Two weeks later the groups were crossed for the treatment and challenged again. To determine whether the DTH response had waned during the 30-day study period, all eight animals were challenged for a fourth time to establish the baseline response again, wherein no treatment was given.
Immunization and DTH assays
Immunizations consisted of TT (5 LfU/dose; NVI, Bilthoven, the Netherlands or Novartis Behring, Marburg, Germany) and OVA (1 mg/dose; Worthington, Lakewood, NJ, USA). Self-formulated dimethyldioctadecylammonium bromide (DDA, 1·25 mg/dose; Sigma, St Louis, MO, USA) with (D-(+)-trehalose 6,6’-dibehenate (TDB, 0·25 mg/dose; Sigma) was used as adjuvant, to induce a Th1 skewed response .
DTH challenges consisted of 0·1 ml/spot intracutaneous injections of methylated ovalbumin (mOVA, 100 µg/ml and 10 µg/ml in saline, methylation 64%; University of Wageningen, Wageningen, the Netherlands) and TT vaccine (80 IU/ml and 8 IU/ml; NVI) or saline. Read-out of the skin reaction was performed at 48 h. The diameter was determined by taking the mean of the longest diameter of a spot and the diameter at an angle of 90° to the longest diameter (which was usually the shortest diameter), to correct for the fact that the spots were usually not completely circular. Swelling was assessed with a pressure calliper (no. 2046F; Mitutoyo, Veenendaal, the Netherlands) by measuring the thickness of the spot. Thickness of the skinfold immediately next to each respective spot was subtracted. Six-mm skin punch biopsies were taken from the spots injected with the highest concentration of antigen and from the saline spot. Each biopsy was fixed in 4% formaldehyde. Four-µm skin sections were stained with haematoxylin and eosin (H&E). The biopsies were scored blindly by a trained pathologist according to an arbitrary score ranging from 0 to 4.
RhuDex™ treatment and level determination
RhuDex™ is a hydrophobic compound and was dissolved in PEG400 (BASF, Ludwigshafen, Germany). RhuDex™ (10 mg/kg) or vehicle (PEG400) were administered subcutaneously on days −1, 0 and 1 relative to antigen challenges. RhuDex™ and vehicle were administered three times daily on each of these days, approximately at 08:00, 16:00 and 24:00 h.
Plasma was separated from lithium heparin blood by centrifugation and stored at −80°C. Samples were shipped to Charles River Laboratories (Tranent, Edinburgh, UK). Concentrations were determined according to the standard validated liquid chromatography-mass spectrometry (LC-MS/MS) analytical method developed by Charles River Laboratories (method 0415A applied to rhesus monkey plasma) calibrated with nine non-zero standards over the range of the assay.
Ex vivo experiments
IFN-γ enzyme-linked immunospot (ELISPOT) assays and proliferation assays were performed to determine the ex vivo and in vitro effects of RhuDex™. PBMC were isolated by density gradient separation (LSM; ICN, Aurora, OH, USA). PBMC were stimulated with TT (5 LfU/ml; NVI or Novartis Behring) or OVA (10 µg/ml; Worthington). Cultures were set up in the presence or absence of RhuDex™ or anti-CD80 (5 µg/ml; clone L307·4; BD Pharmingen). All cultures contained dimethylsulphoxide (DMSO) in the amount needed to dissolve RhuDex™ at 10 µM.
For IFN-γ ELISPOT assays (U-CyTech), PBMC were cultured in 24-well plates for 17 h at 37°C with TT, OVA, RhuDex™ and/or anti-CD80. Cells were harvested and washed. Cells were put into IFN-γ coated and blocked Maxisorp plates (Nunc, Roskilde, Denmark) in triplicate. The plates were cultured for a further 5 h at 37°C in the presence of TT, OVA, RhuDex™ and/or anti-CD80. Spots were detected and developed according to the manufacturer's instructions. Responses are considered positive when the number of spots outnumbers the number of spots +2 standard deviations of medium controls. Data are given as the number spots/million cells.
For proliferation assays PBMC were cultured in 96-well round-bottomed plates (Greiner Bio-One, Alphen aan den Rijn, the Netherlands) for 90 h at 37°C with stimuli and/or blocking molecules. Cells were cultured in the presence of [3H]-thymidine (0·5 µCi/well) for the last 18 h of culture. Results are expressed as stimulation index (SI) over the medium control. An SI > 3 is considered positive.
Statistical analyses were performed using Prism 4 for Macintosh, version 4·0b (GraphPad Software, Inc., San Diego, CA, USA). Data are given as mean ± standard error of the mean unless indicated otherwise. To determine whether the RhuDex™ or the controls had an inhibitory effect compared to control wells, a one-way analysis of variance (anova) with Dunnett's multiple comparison test was performed. To determine whether the DTH responses had waned over the 30-day experimental period, 95% confidence intervals (CI) of the differences between the first and the last challenge were calculated. If the 95% CI included zero, it was concluded that the response had not waned. Statistical significances between groups were calculated using non-parametric repeated-measures anova (Friedman), with Dunn's correction for multiple comparisons for multiple groups or using the non-parametric paired t-test (Wilcoxon) comparing two groups. A P-value < 0·05 was considered significant.
In vitro inhibitory capacity of RhuDex™
Binding of CD28/Fc chimera was inhibited by RhuDex™ in a dose-dependent manner (Fig. 1a). B cells were irradiated and pre-incubated with RhuDex™ at different concentrations as well as with anti-CD80, CTLA4-Ig and CsA. CD4+ responder T cells were cultured with the two B cell lines in the presence of anti-CD3. The anti-CD3 antibody binds to B cells through the Fc fragment, while the Fab fragment interacts with the CD3–T cell receptor (TCR) complex on the CD4+ cells. This resulted in high levels of IL-2 (958 ± 308 pg/ml) and TNF-α (444 ± 119 pg/ml) in the supernatant after 24 h of culture and IFN-γ levels were high after 48 h of culture (1514 ± 151 pg/ml). The CD4+ T cells also proliferated vigorously, as determined 48–72 h after the initiation of these cultures (SI: 196 ± 24). Anti-CD80 mAb inhibited secretion of IL-2 (22 ± 6·4%, P < 0·001) and TNF-α (12 ± 2·3%, P < 0·001) of control wells (Fig. 1d,e). CTLA4-Ig had only a marginal additional effect, confirming the functional predominance of CD80 co-stimulation over CD86 co-stimulation in this assay and its suitability for screening CD80 inhibitor compounds. IL-2 production was reduced by RhuDex™ at the highest concentration tested of 100 µM to similar levels (21 ± 4·9%) as the 5 µg/ml anti-CD80 mAb control (Fig. 1d). TNF-α release was reduced to a lesser extent (Fig. 1e; 33 ± 4·0%), whereas proliferation was inhibited much more effectively by RhuDex™ compared with anti-CD80 mAb (9·5 ± 1·9% versus 53 ± 7·4% of control wells, Fig. 1b). Inhibition of IFN-γ in this assay was only marginally effective by anti-CD80 as well as by RhuDex™ (Fig. 1c).
Addition of CsA resulted in variable and incomplete inhibition of cytokine release and T cell proliferation, consistent with previously published results .
RhuDex™ plasma levels
RhuDex™ was administered subcutaneously (10 mg/kg/dose) every 8 h, beginning at 08:00 h of day –1 and ending 24:00 h of day 1 relative to TT and OVA challenge. Eight-hour RhuDex™ trough levels were determined in plasma before each 08:00 h administration. RhuDex™ levels peaked on day 0 (2012 ± 1315 ng/ml), after which RhuDex™ levels declined.
Effects of RhuDex™ on antigen-induced DTH reaction
The responses against the lower doses of TT and OVA were very low. Therefore, only the responses against the highest of the two doses (80 IU/ml TT vaccine and 100 µg/ml mOVA) were evaluated. We determined the 95% CI of the differences of the responses between the first and the last (fourth) challenge (both baseline, no treatment). This was conducted both for swelling and diameter of TT and OVA challenge. If the 95% CI included zero, we concluded that there was no waning of the response over the 30-day experimental period. Indeed, all the 95% CI included zero (TT swelling –1·7 to 0·1 mm; TT diameter –4·7 to 0·8 mm; OVA swelling –0·1 to 0·6 mm; OVA diameter –1·0 to 4·2 mm). However, these data indicate that the response against TT increased slightly over time, whereas the response against OVA waned a little over time.
Because there was no statistically significant waning of the response, the cross-over design allowed for evaluation of the effects of the RhuDex™ treatment using the pooled data of the two groups and comparison with the pooled data of the vehicle treatment. Both swelling and diameter of the indurations at the OVA challenge sites were decreased during RhuDex™ treatment, 0·5 ± 0·2 mm (P = 0·0156, non-parametric paired t-test) and 4·2 ± 1·1 mm (P = 0·0313), respectively (Fig. 2). The redness of the indurations was also less (not shown). The diameters of the indurations of the TT challenge sites were decreased significantly with 2·4 ± 0·6 mm (P = 0·0220, Fig. 2). The swelling of the indurations against TT was decreased with 0·5 ± 0·3 mm, but this was not significant; the redness was similar during RhuDex™ and vehicle treatment.
Histology of induration biopsies
Biopsies were taken from the OVA-, TT- and saline-challenged sites during RhuDex™ treatment and during vehicle treatment. Pathological changes were scored blindly according to an arbitrary scoring system ranging from 0 to 4. The overall pathology scores confirmed the macroscopic scores (Fig. 3). In general, the DTH response following TT injection spots showed more severe pathology, including necrotizing dermatitis with oedema and haemorrhage, leading to higher arbitrary scores, while the OVA DTH spots showed usually mild perivascular infiltrates and dermatitis. Most saline spots showed no significant changes or few perivascular infiltrates. No significant differences were found between biopsies taken during RhuDex™ treatment and during vehicle treatment.
Ex vivo OVA and TT responses
At the time of DTH evaluation (48 h after challenge) blood was obtained to investigate the ex vivo response to OVA and TT using PBMC. Proliferation and IFN-γ production (ELISPOT) were evaluated. In spite of a good DTH response to OVA, PBMCs did not show significant proliferation or IFN-γ production when restimulated in vitro with OVA (results not shown). In contrast, significant TT-specific proliferation and/or IFN-γ responses were found in all animals following immunization. However, no difference was seen between PBMC from animals treated with RhuDex™ or vehicle (Fig. 4a,b). The effect of addition of RhuDex™ and anti-CD80 mAb in vitro on TT-specific responses was investigated in all animals with positive proliferative (SI > 3) and IFN-γ ELISPOT responses (no. of spots/million PBMC > 25). Data from RhuDex™- and vehicle-treated animals were pooled. RhuDex™ blocked the proliferation of PBMC at concentrations of both 10 µM and 50 µM. Anti-CD80 mAb (5 µg/ml) did not inhibit TT-specific responses (Fig. 4c). IFN-γ ELISPOT responses were inhibited by neither RhuDex™ nor anti-CD80 mAb (Fig. 4d).
Blocking the co-stimulation pathway, CD80/CD86–CD28/CD152 is a valid therapeutic target for autoimmune disease and prevention of graft rejection . Antibodies or constructs such as CTLA4-Ig blocking these pathways have the disadvantage that these require intravenous application. Therefore, small molecules that can be taken orally will have treatment advantages . RhuDex™ is a small molecule specifically blocking CD80-dependent immune mechanisms. Because of a low oral bioavailability in NHP, we administered RhuDex™ via the subcutaneous route in this study. However, RhuDex™ has higher bioavailability and longer half-life in man than in rhesus monkeys (unpublished results) and is therefore developed for oral administration in humans.
To show that RhuDex™ binds and blocks rhesus monkey CD80 and shows comparable in vitro inhibition profiles to human CD4+ T cells, rhesus monkey B cell lines expressing high levels of CD80 were used to stimulate rhesus CD4+ T cells in the presence of anti-CD3. Significant inhibition of release of the major inflammatory cytokines, IFN-γ, TNF-α and IL-2, was found with RhuDex™ (Fig. 1).
CD80, as well as CD86, bind to CD28, providing the co-stimulatory signal for T cell activation together with the T cell antigen receptor. After T cell activation CD152 (CTLA-4) is up-regulated, serving as a negative regulator for T cell activation. CD152 interaction with CD80 or CD86 gives a down-regulatory signal to the T cell. The expression kinetics, binding avidity and dissociation rates indicate that CD86 may bind and activate CD28 preferentially, while CD80 may bind and activate CD152 preferentially [15,16]. The effects of blocking CD80 depend upon the model and type of blockade used. Blockade of CD80 with an antibody in the mouse allograft model does not have an effect  or even enhances graft rejection [18,19], which could be explained by the fact that CD86 is expressed constitutively on APCs, in contrast to CD80, which is up-regulated only after activation. The T cells may therefore be activated via CD86. Blockade of CD80 with CD80-binding competitive antagonist peptides, blocking only the interaction of CD80 with CD152 in the collagen-induced arthritis model in mice, reduced disease severity . Plasmids encoding for a CD80 mutant blocking the interaction with CD28, but not CD152, had a down-regulatory effect on the immune response in the non-obese diabetic mouse model . Whether RhuDex™ selectively blocks the interaction with CD28 or with CD28 and CD152 has not yet been established fully, but given the fact that it inhibits T cell activation in vitro, the inhibitory effects in vivo were investigated further.
The DTH response is used commonly to test the effect of immunomodulatory compounds [22–25]. It can be used to study secondary immune responses . Significant proliferation and differentiation occur at the challenge site . Up-regulation of CD80 in keratinocytes during a DTH response in humans was demonstrated by Wakem et al.. As proof of principle to test whether RhuDex™ could inhibit CD80-dependent immune reactions in vivo, the antigen-induced DTH reaction in the skin was used. Both the OVA- and TT-induced DTH reaction could be inhibited by RhuDex™, although the TT-specific response was inhibited to a lesser extent (Fig. 2). In the study reported here, the OVA-induced DTH reaction in rhesus monkeys was characterized by mild dermatitis showing perivascular lymphocytic infiltrates. The TT-induced DTH reaction showed a more pronounced dermatitis, sometimes accompanied by eosinophil and neutrophil infiltrates. The strong swelling and more severe pathology after TT challenge may be caused by the fact that we used TT vaccine (TT in aluminium phosphate). This is in contrast to the report by Cordoba Castro et al., who also challenged with TT vaccine, although they used TT in aluminium hydroxide. They found an infiltrate dominated by mononuclear cells (80–90%). Our method of intracutaneous challenge with TT vaccine may even have caused a booster effect, which was evident from the increased swelling and diameter upon TT challenge in the fourth DTH compared to the first DTH. This may also explain the lesser effect of RhuDex™ on TT challenge compared with OVA challenge. Possibly, the TT-specific DTH reaction was mediated by other cells that could not be blocked by a T cell co-stimulation blocker such as RhuDex™. Pure TT cannot be used as challenge antigen, because it diffuses away from the injection site and the response is not localized, but spreads to a large part of the arm (unpublished observations). The OVA challenge was performed with methylated OVA. Methylated proteins have fewer negative charges, which prevent the proteins from diffusing away from the injection site. Biopsies of the OVA challenge site showed mainly lymphocytic infiltrates, which were more sensitive to RhuDex™.
No effect of RhuDex™ treatment was seen on the ex vivo TT-specific response (Fig. 4a,b). A possible explanation for this could be that the responses measured in PBMC are not representative for the responses in vivo, as also evidenced by the absence of OVA-specific response in vitro. A second explanation could be that inhibition by RhuDex™ requires the exposure of cells to RhuDex™, which was not sufficient during the ex vivo examination, and that exposure in vivo did not induce a state of anergy. A non-mutually exclusive third explanation could be that RhuDex™ does not inhibit the T cell-mediated recall response in vitro, but in vivo other effects, such as the activation of the dendritic cells of the skin, the Langerhans cells, are prevented. This would lead to a reduced recruitment of other inflammatory cells. Histology of biopsies of the OVA-challenged sites showed a trend towards a reduction in the arbitrary pathology score (P = 0·1756). The absence of an effect of RhuDex™ex vivo cannot be explained by the fact that RhuDex™ was unable to block protein antigen-induced responses, as the in vitro addition of RhuDex™ could inhibit TT-specific proliferation (Fig. 4c), although the inhibition was not complete.
In conclusion, RhuDex™ can block an in vivo T cell recall response to a soluble antigen. These results are promising, and provide evidence that RhuDex™ can be expected to inhibit other inflammatory responses involving CD80 in the skin and elsewhere in the body. RhuDex™ could be a useful addition to the therapeutic armamentarium used currently to treat autoimmune diseases such as psoriasis and rheumatoid arthritis. Trials to test the effectiveness of RhuDex™ for the treatment of arthritis in man are currently under way.
The authors thank Dr B. ‘t Hart for critical review of the manuscript.
Dr J. Endell is employed by MediGene. The other authors declare no conflict of interest.