Cell lines and animals
CHO-S cells (Invitrogen, Zug, Switzerland) were cultured adherent in RPMI 1640 (Gibco, Zug, Switzerland) supplemented with 10% fetal bovine serum (Gibco), 2 mM ultraglutamine (Lonza, Basel, Switzerland) and antibiotics/antimycotics (Gibco) or in suspension in PowerCHO-2CD (Lonza) with 8 mM ultraglutamine, HT supplement (Gibco) and antibiotics/antimycotics in shaker incubators. Lung murine fibroblasts (CCL-1.3; ATCC, Molsheim Cedex, France) were cultured adherent in Dulbecco’s modified Eagle’s medium (Gibco) with 10% fetal bovine serum and antibiotics/antimycotics. MC/9 cells (murine mast cells, CRL-8306; ATCC) were cultured according to the supplier’s protocol in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 2 mM ultraglutamine, 10% rat-T-STIM (BD Becton Dickinson, Allschwil, Switzerland) and 0.05 mM β-mercaptoethanol (Gibco). Murine F9 teratocarcinoma cells (CRL-1720; ATCC) were cultured on 0.1% gelatin-coated tissue flasks in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum and antibiotics/antimycotics. Male DBA/1J mice were obtained from Janvier (Le Genest-St-Isle, France). Female 129/SvEv mice were obtained from Charles River (Sulzfeld, Germany).
Cloning of murine fusion proteins
For cloning of muTNFR-Fc the murine TNFR gene (extracellular domain of TNFRII, amino acids 23 to 258) was amplified from previously cloned F8(scFv)-TNFRII  by polymerase chain reaction (PCR) using the primer pair 5′-CCTGTTCCTCGTCGCTGTGGCTACAGGTGTGCACTCGGTGCCCGCCCAGGTTGTCTT-3′ and 5′-CAATCCCTGGGCACGCCACCCTTGGTACTTTGTTC-3′ appending part of a signal sequence at the N-terminus and an overlapping fragment to muFc at the C-terminus. The gene for murine Fc (hinge, CH2, CH3; amino acids 98 to 324) was amplified from a commercial cDNA (Source BioScience, Berlin, Germany) using the primer pair 5′-CCAAGGGTGGCGTGCCCAGGGATTGTGGTTGTAAGC-3′ and 5′-TTTTCCTTTTGCGGCCGCTCATTAAGCTATTTACCAGGAGAGTGGGAGAGG-3′ appending an overlapping fragment to muTNFR at the N-terminus and a stop codon and Not I restriction site to the C-terminus. muTNFR and muFc sequences were PCR-assembled using the primer pair 5′-CCCAAGCTTGTCGACCATGGGCTGGAGCCTGATCCTCCTGTTCCTCGTCGCTGTGGC-3′ and 5′-TTTTCCTTTTGCGGCCGCTCATTAAGCTATTTACCAGGAGAGTGGGAGAGG-3′ appending the second part of the signal sequence and a Hind III restriction site to the N-terminus. The assembled fragment was double digested with Hind III/Not I (New England BioLabs, Allschwil, Switzerland) and ligated into the mammalian cell-expression vector pcDNA3.1(+) (Invitrogen) (for full sequence see Additional file 1).
F8-muIL10 was cloned using the sequence for F8 in diabody format  with the primers 5′-CCTGTTCCTCGTCGCTGTGGCTACAGGTGTGCACTCGGAGGTGCAGCTGTTGGAGTCTGGG-3′ and 5′-GATGAGCCGGAAGAGCTACTACCCGATGAGGAAGATTTGATTTCCACCTTGGTCCCTTGGCCGAA-3′ introducing part of a signal sequence at the N-terminus and part of a C-terminal (SSSSG)3 linker. The sequence for murine IL10 (amino acids 19 to 178) was amplified from a commercial cDNA (Sino Biological Inc., Beijing, China) using the primer pair 5′-GGGTAGTAGCTCTTCCGGCTCATCGTCCAGCGGCAGCAGGGGCCAGTACAGCCGGG-3′ and 5′-TTTCCTTTTGCGGCCGCCTAGCTTTTCATTTTGATCATCA TG-3′ appending a complementary part of the (SSSSG)3 linker at the N-terminus and a stop codon and Not I restriction site to the C-terminus. F8 diabody and murine IL-10 sequences were PCR-assembled using the primers 5′-CCCGCTAGCGTCGACCATGGGCTGGAGCCTGATCCTCCTGTTCCTCGTCGCTGTGGC-3′ and 5′-TTTCCTTTTGCGGCCGCCTAGCTTTTCATTTTGATCATCATG-3′ adding the rest of the signal sequence and a Nhe I restriction site to the N-terminus. The assembled fragment was double digested with Nhe I/Not I (New England BioLabs) and ligated into the mammalian cell-expression vector pcDNA3.1(+) (for full sequence see Additional file 2).
Expression, purification and characterization of murine fusion proteins
The fusion proteins were expressed in a stable monoclonal cell line as reported before . Briefly, PEI-mediated transient gene expression  was used to generate a polyclonal batch of protein. An aliquot of the transient gene expression culture was used to produce a stable cell line using geneticin (G418, 0.5 g/l; Santa Cruz, Heidelberg, Germany) for selection. Monoclonal cells were screened for high expression of protein by ELISA, using L19-TNFα (produced in our laboratory) as coating antigen and a goat anti-mouse IgG (Fc-specific)-peroxidase antibody (Sigma-Aldrich, Buchs, Switzerland) for detection in the case of muTNFR-Fc. For F8-muIL10, recombinant EDA was used as antigen and protein A–horseradish peroxidase (GE Healthcare, Glattbrugg, Switzerland) for detection. The best producing clone for each construct was grown in PowerCHO-2CD medium in suspension for large-scale production of protein. The proteins were purified from cell culture supernatant by protein A affinity chromatography and analyzed by SDS-PAGE, size exclusion chromatography (Superdex200 10/300GL; GE Healthcare) and for F8-muIL10 additionally by surface plasmon analysis (BIAcore) on an EDA-coated CM5 sensor chip (GE Healthcare).
The biological activity of muTNFR-Fc was determined by its ability to inhibit TNFα-induced killing of mouse fibroblasts . Lung murine fibroblast cells were seeded in a 96-well plate (30,000 cells/well) in 100 μl culture medium and incubated for 24 hours at 37°C, 5% CO2. Medium containing actinomycin D (final concentration 2 μg/ml; Sigma-Aldrich), TNFα (final concentration of trimer 5 pM; eBioscience, Vienna, Austria) and different concentrations of muTNFR-Fc (serially diluted from 50 nM to 0.1 pM) was added to the cells. After incubation at 37°C for 24 hours cell viability was determined by addition of 20 μl Cell Titer Aqueous One Solution (Promega, Dübendorf, Switzerland), and after 2 hours absorption was measured at 490 nm.
For determination of the biological activity of the IL-10 moiety in the F8-muIL10 fusion protein, an IL-4-dependent proliferation assay of MC/9 cells was used . Cells were seeded in a 96-well plate (40,000 cells/well) with 200 μl culture medium (without rat-T-STIM) containing 5 pg (0.05 units)/ml murine IL-4 (eBioscience) and varying concentrations of F8-muIL10 or recombinant murine IL-10 (eBioscience) starting at a concentration of 100 ng/ml IL-10 equivalents. After incubation at 37°C for 48 hours 20 μl/well Cell Titer Aqueous One Solution was added, and after 2 hours absorption was measured at 490 nm.
The in vivo targeting of F8-muIL10 was tested by quantitative biodistribution analysis using radiolabeled protein as described before . For this analysis 129/SvEv mice were implanted subcutaneously (s.c.) with F9 tumor cells (25 × 106 cells) in the flank. Purified F8-muIL10 (15 μg/mouse) was radioiodinated with 125I and injected intravenously (i.v.) into the lateral tail vein of mice (n = 3) grafted with F9 tumors. Mice were sacrificed 24 hours after injection. Organs were excised, weighed and radioactivity was counted using a Cobra γ counter (Packard Instrument Company, Meriden, CT, USA). Radioactivity content of representative organs was expressed as percentage of injected dose per gram of tissue.
In vivo imaging
To test the targeting properties of the murine and human F8-IL10 fusion proteins, a near-infrared fluorescence imaging study was performed. For this purpose, the proteins (11 nmol F8-muIL10 and F8-huIL10) were incubated for 1 hour with a 20× molar excess of IRDye 750 N-hydroxysuccinimidyl ester (220 nmol; LI-COR, Bad Homburg, Germany) in 10% dimethylsulfoxide/phosphate-buffered saline (PBS), pH 7.4, at room temperature. Protein was purified from free dye using a PD10 desalting column (GE Healthcare), eluted in 5% dimethylsulfoxide/PBS and concentrated to 1.3 mg/ml using Amicon Ultra (10K) centrifugal filter units (Millipore, Zug, Switzerland). Then 200 μg (or 100 μg) of each protein were injected i.v. into the lateral tail vein of mice (n = 1) that had developed arthritis after the second collagen immunization (see section Mouse model of collagen-induced arthritis for more details). Mice were imaged at 1, 4, 24 and 48 hours after the injection under isoflurane anesthesia on their ventral side using an IVIS Spectrum machine (Xenogen, Caliper Life Sciences, Oftringen, Switzerland) with the following imaging parameters: λex = 745 nm, λem = 800 nm, exposure time = 1 second, F/stop = 4, small binning. After 48 hours, mice were sacrificed and paws (arthritic and not affected ones) were photographed and then submitted to fluorescence imaging, using the same parameters.
Mouse model of collagen-induced arthritis
Male DBA/1J mice (8 weeks old) were immunized by subcutaneous injection at the base of the tail with 0.05 ml emulsion of bovine type II collagen emulsified in Complete Freund’s Adjuvant (Hooke Laboratories, Lawrence, MA, USA). Three weeks later, a booster injection of 0.05 ml bovine collagen/Complete Freund’s Adjuvant in the case of the full collagen induction protocol and 0.04 ml for the reduced collagen induction protocol was given to the mice. After the booster injection, mice were inspected daily and disease was monitored using two different scoring systems. To each limb a clinical score was assigned (0 = normal, 1 = swelling of one or more toes of the same limb and 2 = swelling of the whole paw). A maximum score of eight can be reached in this first scoring system [8, 18]. A more diverted clinical score, the modified score, was also used (0 = normal; 1 = one toe inflamed and swollen; 2 = more than one toe, but not entire paw, inflamed and swollen or mild swelling of entire paw; 3 = entire paw inflamed and swollen; 4 = very inflamed and swollen paw; adapted from Hooke Laboratories). A maximum score of 16 can be reached. In addition swelling of affected paws was measured daily with a caliper under isoflurane anesthesia. Paw thickness is expressed as the mean of all four paws of each animal. Animals were included into a therapy group when showing signs of joint inflammation with a score of 1 to 3. When the joint inflammation was too strong at day 1 (more than one paw, score >3) mice were not included into the experiment, because according to our project license (208/2010) we are not allowed to keep a mouse alive with a conventional arthritic score ≥4 for more than 4 days. All animal experiments were performed in agreement with Swiss ethical regulations. Ethical approval for all experiments was given by the state veterinary office (reference number 208/2010; Veterinäramt des Kantons Zürich, Zürich, Switzerland).
Combination therapy of muTNFR-Fc and F8-huIL10
Mice were immunized according to the full collagen induction protocol. Mice with a new clinical score of 1 to 3 were randomly assigned to a treatment or control group and therapy was started (day 1). Mice received intravenous injections of muTNFR-Fc (10 μg) into the lateral tail vein or subcutaneous injections of F8-huIL10 (200 μg) or saline or a combination of muTNFR-Fc (10 μg, i.v.) and F8-huIL10 (200 μg, s.c.), three times on days 1, 4 and 7. Seven mice were analyzed per group in a daily, nonblinded fashion and the arthritic clinical score, the thickness of inflamed paws and weight was monitored. Mice were sacrificed at day 5 (PBS), day 8 (F8-huIL10, muTNFR-Fc) or day 13 (combination) due to arthritic score (≥4 for more than 4 days with conventional arthritic score) and weight loss (>15%), in accordance with local regulations.
Comparison of F8-huIL10 and F8-muIL10
Mice were immunized according to the reduced collagen induction protocol. Therapy was performed as described before. Mice received either intravenous injections of muTNFR-Fc (30 μg) or subcutaneous injections of F8-huIL10 (200 μg), F8-muIL10 (200 μg) or saline or a combination of muTNFR-Fc (30 μg, i.v.) and F8-huIL10 (200 μg, s.c.). Ten mice were analyzed per group in a daily, nonblinded fashion and the arthritic clinical score, the thickness of inflamed paws and weight was monitored. Mice were sacrificed at day 8 (PBS), day 9 (F8-huIL10, muTNFR-Fc, F8-muIL10) or day 13 (combination) due to arthritic score and weight loss, in accordance with local regulations.
Detection of anti-F8-huIL10/anti-F8-muIL10 antibodies in plasma of treated mice
Blood was obtained at the start of the therapy (day 1, n = 4) from the vena saphena or at the end of therapy from sacrificed mice through cardiac puncture, processed to plasma and stored at -20°C. The immunogenicity of F8-huIL10 and F8-muIL10 was assessed by surface plasmon resonance (BIAcore 3000) screening of mouse plasma samples. F8-huIL10 or F8-muIL10 at a concentration of 50 μg/ml were immobilized on a CM5 sensor chip (GE Healthcare) using an amine coupling kit (GE Healthcare). Surface density of 2,600 RU and 2,900 RU was achieved for F8-huIL10 and F8-muIL10, respectively. On a control flow cell, activation by 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS) was performed and immediately blocked by injecting ethanolamine. For binding analysis, positive control samples of anti-human IL-10 and anti-murine IL-10 antibodies (1 and 4 μg/ml; eBioscience) and serum samples diluted 500-fold in HBS-EP buffer (GE Healthcare) were passed over the different flow cells with a flow rate of 30 μl/minute for 3 minutes. The response was recorded 30 seconds after the end of the injection. The positive control was run again at the end of the analysis to confirm binding capacity of the immobilized protein. To regenerate the surface, 10 mM glycine, pH 2.0, was run over the flow cells for 40 seconds at 30 μl/minute.
Analysis of cytokine levels in plasma of mice
Blood was obtained at the end of therapy from each mouse (see above), processed to plasma and stored at -20°C. To quantify cytokine levels in plasma of treated and control mice, a multiplex bead-based flow cytometry analysis was performed using the Mouse Th1/Th2/Th17/Th22 13plex FlowCytomix Multiplex (eBioscience) following the supplier’s protocol. Fluorescence-activated cell sorting analysis was performed on a BD FACS Canto (BD Bioscience, Allschwil, Switzerland) and data evaluated with FlowCytomix Pro 3.0 software (eBioscience). The experiment was repeated on a different day in order to have an independent replicate of the assay (Additional files 3 and 4). Using standard curves generated by the FlowCytomix Pro 3.0 software with positive control samples, a level of quantification was assigned to every cytokine (Additional files 5 and 6).
Incubation experiment of radiolabeled immunocytokines with whole blood
The ability of F8-huIL10 and F8-muIL10 to interact with blood cells was determined by a centrifugation-based assay with radiolabeled preparations. Purified F8-huIL10 and F8-muIL10 were radioiodinated with 125I as described before  and different concentrations of labeled protein were incubated with fresh human and mouse blood. Human blood was collected in S-Monovette tubes (Kalium-EDTA (ethylenediamine tetraacetic acid), Sarstedt, Sevelen, Switzerland) and mouse blood taken from DBA/1J mice via cardiac puncture after sacrifice using Microtainer LH tubes (lithium heparin, BD Bioscience) to prevent coagulation. After 10 minutes of incubation, tubes were centrifuged for 3 minutes at 2,000 × g. Plasma was separated from the cell pellet and radioactivity of both was counted using a Cobra γ counter.
Data are expressed as the mean ± standard deviation or standard error of the mean. Differences in arthritic outcome between therapeutic groups were compared using GraphPad Prism (GraphPad Software Inc., La Jolla, CA, USA) grouped two-way ANOVA multiple-comparison (Bonferroni-corrected) analysis, with P <0.05 considered significant. Differences in cytokine levels were compared using a Mann–Whitney test, with P <0.05 considered significant.