BiP was prepared as previously described 6 . The protein purity, as assessed by polyacrylamide gel electrophoresis and silver staining, was greater than 95%. Professional assessment of endotoxin contamination showed < 0.3 endotoxin units/μg protein (Associates of Cape Cod, Liverpool, UK).

The RASM/SCID (CB.17/Icr; Charles River Japan, Tokyo, Japan) murine model was set up as described 4 . All RA patients providing tissue during knee joint replacement surgery gave fully informed written consent and the study was approved by the Research Ethics Committee of Toin University of Yokohama Project approval number I-1.

Therapeutic manipulation of the mice was undertaken only if successful engraftment had been achieved 4 weeks after transplantation. BiP (10 μg/mouse, n = 15) or human serum albumin (HSA) (10 μg/mouse, n = 15), as the control protein, were administrated intravenously either alone or in the presence or absence of anti-IL-10 antibody or isotype control antibody as required. The mice were sacrificed 12 days later and implanted tissue was removed for analysis by immunohistology and weight.

The degree of histological synovial inflammation of the implanted tissue was assessed as described by Koizumi and colleagues 7 or by Rooney and colleagues 8 . The scoring features included measurements of synovial hyperplasia, fibrosis, blood vessels, perivascular lymphocytes, lymphoid follicles, and diffuse infiltrating lymphocytes or synovial cells, palisading, giant cells, lymphocytes, granular tissue and fibrosis.

Paraffin-embedded tissue sections were used for immunostaining for CD86 and HLA-DR. Frozen tissue sections were stained for the detection of cytokines (TNFα, IL-6 and IL-10). The tissue sections were blocked for endogenous peroxidase activity with 0.3% hydrogen peroxide in methanol. Nonspecific antibody binding was blocked with 10% normal goat serum. The sections were incubated with specific antibody or normal IgG for 1 hour at 37°C. The sections incubated with anti-CD86 or HLA-DR antibody were treated with biotinylated secondary antibody, and then visualised using the Vectastatin ABC kit (Vector Laboratories, Funakoshi Co, Tokyo, Japan). A visual scoring scale of 0 to 3 was used to assess the expression of the molecules. The other sections were treated with peroxidase-conjugated anti-mouse IgG (Histofine Simple Stain MAX PO; Nichirei, Tokyo, Japan) developed with 3-amino-n-ethylcarbazole (Nichirei) for visualisation. Sections were counterstained with Mayer haematoxylin.

The serum levels of human IL-4, IL-6 and IL-10 were measured by a quantitative sandwich enzyme immunoassay technique (Quantikine HS; R&D Systems, Funakoshi Co Tokyo, Japan) according to the manufacturer's instructions.

Data were compared using the Student t test and expressed as the mean ± standard deviation.

Twelve days following intravenous injection of BiP into the RASM/SCID chimaeric mice, the histological features of the RASM taken from the control mice were unchanged (Figure 1a) compared with the markedly reduced cellular infiltrate in the RASM grafts from the BiP-treated mice (Figure 1b). Histopathogical measurements of the RASM grafts from mice given intravenous BiP showed that although there was not a complete absence of inflammation it was significantly reduced compared with grafts taken from control mice (overall scores: Rooney, BiP 16 ± 6 vs. HSA 27 ± 8.2, P = 0.006; Koizumi, BiP 6.1 ± 2.6 vs. HSA 12 ± 2.2, P < 0.001; BiP n = 15 and HSA n = 14) (Figure 1c and 1d, respectively). The data from this experimental model confirm that BiP acts in an anti-inflammatory fashion and they predict a reduction in pathology in vivo. Indeed, the gross pathology following BiP treatment showed results similar to those seen in the same model following administration of anti-TNFα or anti-IL-6 receptor mAbs 4 5 and also the disease-modifying drug methotrexate 9 .

Immunohistological assessment of the RASM tissue sections showed significantly reduced expression of the co-stimulatory molecule CD86 (mean visual score ± standard error: HSA 1.86 ± 0.46, n = 7 vs. BiP 0.89 ± 0.35, n = 9; P = 0.05) and a trend to the downregulation of HLA-DR (HSA 1.67 ± 0.37, n = 9 vs. BiP 1.37 ± 0.37, n = 8, P < 0.06 (NS)) (Figure 2a), thus reducing the capacity for antigen presentation by monocytes/macrophages/dendritic cells. This observation complements our in vitro data showing that BiP has a significant inhibitory effect on HLA-DR expression and downregulates CD86 expression by human peripheral blood monocytes 1 and monocyte-derived dendritic cells 10 . One mechanism of action by which BiP might reduce T-cell activation, and thus inflammation, is therefore by restricting the antigen-presenting abilities of monocytes 1 .

Additionally, a profound suppression of inflammatory cytokines TNFα and IL-6 was noted in RASM retrieved from the BiP-treated animals (Figure 2a). The suppression of TNFα in the RASM by BiP is particularly pertinent in diseased tissue where TNFα is known to be the major inflammatory mediator, and gives credence to such disease-related changes noted in in vitro human peripheral blood mononuclear cell cultures.

Only human IL-6 was produced at levels detectable in the sera of control mice injected with HSA, but not in the sera of mice given intravenous BiP. There was a significant reduction in background levels of circulating human IL-6 in these mice (HSA 1.13 ± 1.1 pg/ml vs. BiP 0.32 ± 0.13 pg/ml, P = 0.01) (Figure 2b). This is an important correlation. Since IL-6 is a pleiotropic cytokine this inhibition may have important consequences if translated to RA patients receiving BiP, including: reduced stimulus for the acute-phase response, leading to a reduction in C-reactive protein and the erythrocyte sedimentation rate; inhibition of development of Th17 cells, which are important proinflammatory cells in RA 11 for which IL-6 is a pre-requisite 12 ; and downregulation of other pathogenic mechanisms, such as osteoclast activity 13 .

Although BiP stimulates the production of IL-10 from peripheral blood mononuclear cells in vitro 1 , we lacked histological evidence for the upregulation of IL-10 protein in the BiP-treated RASM/SCID mouse. This may have been due to the single time-point chosen for the experiment. To test the hypothesis that the therapeutic effect of BiP was at least partially mediated via IL-10, BiP or HSA were therefore simultaneously administered with either a neutralising anti-IL-10 antibody or an isotype antibody control.

The weight of the transplant was used as a surrogate measure of inflammation. Weight loss, indicative of the loss of inflammatory cells and reduced tissue oedema, signified a reduction of tissue inflammation. The explants from the mice given HSA, whether co-injected with the anti-IL-10 antibody or the isotype control, did not differ in weight (HSA + anti-IL-10, 0.33 ± 0.06 g; HSA + isotype, 0.27 ± 0.07 g; P = NS). In contrast, the explants from the mice given BiP + isotype control were significantly lighter in weight from those explants given anti-IL-10 (BiP + isotype control, 0.07 ± 0.04 g vs. BiP + anti-IL-10, 0.26 ± 0.04 g; n = 4, P = 0.007) (Figure 3). Importantly, there was no difference in weight between the two HSA groups and the group given BiP + anti-IL-10 (Figure 3). These data provide indirect evidence of IL-10 involvement. Undoubtedly the production of IL-10 facilitates the attenuation of TNFα production 1 , but data also indicate that the positive deactivation of monocytes 14 is important for the immunomodulatory effect of BiP, with downregulation of CD86 and HLA-DR and upregulation of soluble TNF receptor II and IL-1 receptor antagonist production 1 . These latter functions may be independent of IL-10 production 1 15 . It is certainly evident that, although BiP and IL-10 have a similar effect on monocytes, the kinetics are different 1 - therefore, at present, the extent to which IL-10 defines BiP activity is unclear.

Consideration should also be given to the extended activity of BiP treatment in the absence of protein. In vivo adoptive transfer experiments clearly indicate that the induction of regulatory T cells is important 2 , and the identification of the regulatory cell subpopulation involved is currently under investigation. Likewise, the BiP-induced attenuation of IL-6 is a novel finding from this model and we are presently investigating this in relation to BiP.