Methods

TNFα inhibitors

ESBA105 was expressed in and purified from E coli as described elsewhere43 and used in 25 mM sodium phosphate pH 6.5. Infliximab (Remicade) and etanercept (Enbrel) were purchased in a local pharmacy.

Monoarthritis model

ESBA105, infliximab or an scFv consisting of the same variable domain framework as ESBA105 but with irrelevant specificity (termed “naïve” scFv; ESBATech, Schlieren, Switzerland) in 40 μl phosphate buffered saline (PBS) followed 5 min later by rhTNFα in 10 μl PBS were injected intra-articularly through the infrapatellar ligament of the knee of female 10-week-old Lewis rats (n = 3 per treatment group; The Jackson Laboratory, Bar Harbor, Maine, USA) anaesthetised with 50 mg/kg ketamine using a 28-gauge needle according to the method of Bolon et al.22 Rats were monitored before and during the study. Knee diameters were measured with calipers (Dyer, Lancaster, Pennsylvania, USA) before the study and 48 h after injection of recombinant human TNFα (rhTNFα). The rats were killed at 48 h and decalcified knee sections were evaluated following staining with H&E or toluidine blue. Sections were scored for inflammation (0–4) and cartilage (0–4) as described by Bolon et al.22

All statistical testing was performed using GraphPad Prism Software 4.03 (GraphPad, San Diego, California, USA). Differences between treatment groups and controls were determined using an unpaired t test. For analysis of scoring results the non-parametric Mann-Whitney U test was used which resulted in non-significant differences between treatment groups and controls owing to the small number of animals per group per single experiment. However, increasing group sizes by linear combination of data from different experiments resulted in significant differences (see fig 1 in online supplement). A p value <0.05 was considered significant.

Cartilage penetration studies

Cartilage dissected from bovine femur (Zentralschlachthof, Hinwil, Switzerland) was mounted in perfusion chambers (PermeGear, Hellertown, Pennsylvania, USA) as previously described.42 Briefly, the apical cartilage surface was exposed to 0.3 ml ESBA105-FITC (1 mg/ml) or infliximab-FITC (1 mg/ml or 2.2 mg/ml) in PBS, pH 7.4. In these penetration chambers the total volume circulating through the receptor compartment, tubing and reservoir was 5 ml. After incubation, specimens of cartilage tissue were washed three times in 20 ml PBS and subsequently embedded in TissueTek OCT (Digitana, Horgen, Switzerland), snap frozen in liquid nitrogen and stored at −20°C. Using a MICROM cryostat HM 560 (Microm International, Walldorf, Germany) at −18°C with disposable microtome knives (Feather N35, Osaka, Japan) at −20°C, 14 μm sections were cut and mounted. The sections were analysed and photographed under a UV microscope (Leica DM RE equipped with a digital camera Leica DC 500; Leica Microsystems, Heerbrugg, Switzerland) at a magnification of 40–100×. Signal intensities on the photographs were analysed using IMAGE QUANT 5.0 software (GE Healthcare, Glattbrugg, Switzerland). A series of penetration studies were performed and a final data set is presented here.

FITC labelling of ESBA105 and infliximab

Fluorescein isothiocyanate (FITC) (F1906; Invitrogen, Basel, Switzerland) was dissolved in anhydrous DMSO (Sigma-Aldrich, Buchs, Switzerland) immediately before use and added to ESBA105 or infliximab to reach a ratio of 40 μg FITC/mg. The reaction mixture was kept in the dark on a rotator wheel over night. Separation of unbound FITC from labelled proteins was performed by dialysis using 5 ml dialysis cassettes (Socochim, Lausanne, Switzerland) against four changes of 5 litres PBS (1×, pH = 6.5) at 4°C during 48 h. FITC labelling of ESBA105 and infliximab resulted in comparable specific activity.

Biodistribution studies

ESBA105 was labelled with 125-iodine (¹²⁵I) to a starting specific activity of 18.6 MBq/mg using the Chloramin T method by MDS Pharma Services Switzerland AG (Fehraltorf, Switzerland).

Biodistribution studies were performed at Covance Laboratories Ltd (Harrogate, UK). Male New Zealand white rabbits received a single intravenous (n = 3) or intra-articular (n = 4) dose of [¹²⁵I]-ESBA105 at a target ESBA105 dose level of 1000 μg/animal (see table 1 in online supplement). The administered doses were in the range 884–1034 μg/animal, equivalent to radioactive doses of 0.707–0.827 MBq. After dosing, plasma and tissue samples from animals were taken and subjected to gamma counting (Packard Cobra 2 gamma counters, Perkin Elmer Life Sciences, Waltham, Massachusetts, USA). In addition, frozen sections mounted on Invisible Tape (Supapak, Shipley, UK) were freeze-dried in a GVD03 bench-top freeze drier (Girovac Ltd, Norwich, UK) and placed in contact with Fuji imaging plates (type BAS MS, Raytek Scientific Ltd, Sheffield, UK). ¹²⁵I-labelled blood standards of appropriate activity (also sectioned at a nominal thickness of 30 μm) were placed in contact with all imaging plates. Additional experimental details are provided in the online supplement.

Pharmacokinetic parameters were calculated using WinNonLin Professional software (Version 4.0.1, Pharsight Corporation, Mountain View, California, USA).

Results

Mode of action

ESBA105 blocks the TNFα ligand-receptor interaction by competitive binding to the receptor binding site of TNFα. Data from analytical size exclusion chromatography indicate that three monomeric ESBA105 molecules bind to one TNFα trimer, each interacting with one of the three TNFα monomers. ESBA105 binds to rhTNFα with a K_D of 1.57×10⁻¹⁰ M. The binding dynamics of ESBA105 to rhTNFα is characterised by the rate constants k_on and k_off of 4.71×10⁶/M/s and 7.4×10⁻⁴/s, respectively. Thus, the off-rate from human TNFα is in between that of infliximab and etanercept.44

In vitro potency

The ability of ESBA105 to neutralise the biological activity of TNFα in cell culture was demonstrated with mouse L929 fibroblasts. This cell line expresses TNF receptors I and II and, following sensitisation with actinomycin D, undergoes apoptosis when exposed to TNFα. Similar to infliximab, ESBA105 blocks the apoptotic effect of rhTNFα in a concentration-dependent manner with IC₅₀ values of 14.0 and 12.5 ng/ml, respectively (see online fig 2 and further details in online supplement).

Monoarthritis model

Following intra-articular injection of 10 μg rhTNFα, rat knees showed the expected inflammatory reaction22 of knee swelling, synovitis and loss of proteoglycan in cartilage (see rhTNFα controls in fig 1). A naïve scFv with irrelevant specificity had no effect on the severity of the inflammatory reactions (fig 2). In contrast, ESBA105 inhibited rhTNFα-induced inflammatory reactions in a dose-dependent manner (fig 2). Interestingly, an 11-fold molar (16-fold w/w) excess of ESBA105 over rhTNFα resulted in 90% inhibition of knee swelling (fig 2). ESBA105 and infliximab showed similar potency (fig 2). Also, both compounds reduced the inflammation scores and prevented proteoglycan loss in cartilage equally well (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Comparison of the inhibitory potential of intra-articular injections of ESBA105 and infliximab on acute monoarthritis induced by an intra-articular injection of 10 μg recombinant human tumour necrosis factor α (rhTNFα). Effects on joint swelling (quantified by use of calipers), synovitis (H&E staining) and proteoglycan (PG) loss (toluidine blue staining) were assessed. Mean (SD) values are shown. Significant differences between the groups were determined using an unpaired t test; **p<0.01, ***p<0.001. PBS, phosphate buffered saline.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

In vivo dose-response of ESBA105 and inflixmab on joint swelling. Experiments were performed as described in fig 1. Data on synovitis are available in fig 1 in the online supplement. Mean (SD) values are shown. Significant differences between the groups were determined using an unpaired t test; *p<0.05, **p<0.01. PBS, phosphate buffered saline; TNFα, tumour necrosis factor α.

Cartilage penetration studies

ESBA105 and infliximab were labelled with FITC for cartilage penetration studies. To control for quality and ensure comparable labelling, proteins used for cartilage penetration studies were spotted on glass slides to determine signal intensities under UV. Time course studies revealed that ESBA105-FITC but not infliximab-FITC penetrated into bovine cartilage within a few hours (fig 3). Penetration of ESBA105-FITC was time-dependent and in cartilage the concentration of ESBA105-FITC increased linearly with time (fig 3C). In contrast to ESBA105-FITC, infliximab-FITC did not penetrate into cartilage and there was no detectable difference from the PBS control after 8 h of incubation in penetration chambers (fig 3A–C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

ESBA105 but not a full length IgG antibody (infliximab) penetrates into bovine cartilage. (A) Time course of penetration of ESBA105-FITC into bovine cartilage. Bovine cartilage specimens mounted in penetration chambers were exposed to ESBA105-FITC or infliximab-FITC for up to 8 h. The numbers below the photographs represent incubation times. Signal intensities were normalised to control for bleaching. (B) Image Quant software analysis of signal intensities quantified on photographs in (A). The distance on the x axis corresponds to the distance from the apical surface. Signal intensities were normalised to control for bleaching. (C) Image Quant software analysis of signal intensities quantified at a distance of 0.33 mm from the apical surface. FITC, fluorescein isothiocyanate; PBS, phosphate buffered saline.

Biodistribution studies

ESBA105 is designed for local therapeutic use, particularly the intra-articular application to joints. Systemic pharmacokinetics were studied comparing intravenous and intra-articular application of [¹²⁵I]-ESBA105. As shown in fig 4A, intravenous application showed the expected pharmacokinetic behaviour for an scFv.45 Measured peak concentration occurred 2 min post-dose and, thereafter, radioactivity declined in a biphasic manner, probably representing a distributive phase (for about 1 h post-dose), followed by terminal elimination (table 1). In contrast, after intra-articular injection, C_max in plasma is reached only after 6–12 h, suggesting a prolonged absorption phase. However, the area under the curve in plasma is similar for both intravenous and intra-articular administration (fig 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Biodistribution to knee joint tissues following intravenous and intra-articular injection of [¹²⁵I]-ESBA105. (A) Time course of [¹²⁵I]-ESBA105 levels in plasma following intra-articular and intravenous injection. (B) Time course of [¹²⁵I]-ESBA105 levels in knee joint tissues following intra-articular injection. (C) Time course of [¹²⁵I]-ESBA105 levels in knee joint tissues following intravenous injection. Note that samples were taken from different individual animals than those shown in fig 4A (for details see online supplement).

One of the advantages of local treatment is the possibility of achieving high drug levels locally. One hour and 24 h after intra-articular injection of [¹²⁵I]-ESBA105 into the knee, levels of 574 000 and 14 300 ng equivalents/g were observed in the synovial space (fig 4B). Interestingly, ESBA105 levels in the articular cartilage and synovial space were almost identical after 12 h (fig 4B). ESBA105 was absorbed more slowly in the patella with a C_max of 15 100 ng eqiuvalents/g at 6 h; this level is about 20-fold lower than the levels found in the synovial space. Following intra-articular injection, ESBA105 in cancellous bone reached a C_max of 3440 ng equivalents/g after 6 h (fig 4B). From 12 h onwards the radioactivity diminished in most tissues with a T_½ of about 4 h. Longer T_½ were found in plasma (13.5 h), bone marrow (23.0 h), tibia (14.6 h), epimysium (14.3 h), skin (12.1 h) and femur (9.0 h). Interestingly, the levels in synovial fluid and articular cartilage stayed about 20-fold higher than in the patella, cancellous bone and plasma.

In contrast to intra-articular application to the knee, following intravenous application substantially lower levels of ESBA105 were found in knee joints (fig 4C). Terminal half-lives in tissues (when they were measurable) following intravenous application were found to be comparable to those following intra-articular injection (table 1).

Discussion

The treatment of OA should ideally address signs and symptoms as well as structure modification. However, such treatment is not available at present.25 A pharmacological target dominantly involved in both pathophysiological processes would therefore be desirable. TNFα offers itself as such a target as (1) (persistent) local exposure to TNFα causes (persistent) hyperalgesia,46 47 (b) it is produced by synovial tissue10 11 12 13 and cartilage14 in OA, and (c) it is a driver of inflammatory processes9 48 and cartilage degradation.24 Furthermore, Hill et al found a correlation between change in pain with change in synovitis during the course of knee OA.49 It has been shown that TNFα inhibitors inhibit pain and hyperalgesia,47 50 51 52 reduce inflammatory processes50 53 and can reverse OA cartilage from a catabolic to an anabolic state ex vivo.24

For treatment of mono- or oligoarthritic “classical” inflammatory arthritides (psoriatic arthritis and others) as well as for treatment of OA, systemic TNFα inhibition often seems not to be appropriate. Consequently, local treatment with an agent characterised by potent TNFα inhibition and good synovial and cartilage penetration but only low systemic TNFα inhibition would be attractive. We have characterised the properties of such a candidate (ESBA105) for local treatment in models addressing local neutralisation of TNFα in vivo, cartilage penetration ex vivo and biodistribution to the knee joint tissues following intra-articular injection in vivo.

ESBA105 has subnanomolar binding affinity to TNFα and inhibits TNFα bioactivity comparable to infliximab in celluar assays (see fig 2 in online supplement). In vivo, in the short-term rat monoarthritis model an 11-fold molar (16-fold w/w) excess of ESBA105 over rhTNFα inhibited the TNFα-induced inflammatory knee swelling, synovitis and proteoglycan loss from cartilage by 90% (figs 1 and 2). Although this model is not directly representative of human disease, it is a suitable animal model to show that ESBA105 provides TNFα-neutralising activity that is comparable to infliximab in vivo. To characterise the cartilage penetration capabilities of ESBA105 further, we studied the penetration of ESBA105 into normal bovine articular cartilage. Within hours, ESBA105 penetrated into the cartilage (fig 3). Cartilage penetration of proteins is dependent on the molecular weight and charge.54 55 56 From our results it is apparent that ESBA105 has the appropriate size (26 kDa) and charge for therapeutic intra-articular use. Cartilage penetration of ESBA105 is linear (fig 3C). To date, existing data support the suggestion that ESBA105 will inhibit TNFα in cartilage in vivo following intra-articular injection. This is highly relevant, particularly in OA where reversal of TNFα-driven catabolic cartilage metabolism is to be expected in at least a significant proportion of patients.24 In contrast to ESBA105, an IgG (∼150 kDa) such as infliximab is too large and cannot penetrate into cartilage (fig 3A, B).

Biodistribution studies with [¹²⁵I]-ESBA105 in the rabbit showed that, following intra-articular injection, it was distributed from the knee joint and reached significantly higher levels in all articular tissues than following intravenous administration (fig 4, table 1). This was most pronounced in the synovial fluid (1700-fold), articular cartilage (46-fold) and patella (127-fold). Distribution of radioactivity into the tissues of the leg after intra-articular administration was a protracted process, with peak levels in most occurring at 6–12 h (fig 4A, B).

In contrast, C_max of ESBA105 in plasma was about 10-fold lower following injection into the knee joint than after intravenous injection (fig 4A). In line with this finding, C_max in plasma following intra-articular administration was observed 6–12 h after dosing, suggesting a relatively slow absorption into the circulation from the knee joint space (fig 4A). This, combined with the short T_½ (7 h in rabbits),43 should result in low and only short-term systemic inhibition of TNFα following intra-articular injection of ESBA105. This is in contrast to approved TNFα inhibitors which show much higher peak and steady state levels in plasma and slower clearance from the circulation within days to weeks.57

The favourable tissue penetration properties of ESBA105 have also been observed following topical administration to the eye in vivo.42 43 This has triggered ongoing clinical trials with ESBA105 in ophthalmology (NCT00820014; NCT00823173).

In summary, ESBA105 is an anti-TNFα antibody fragment with entirely novel pharmacokinetic properties. Upon local delivery such as intra-articular injection, ESBA105 shows high local target neutralisation capacity but low systemic drug exposure. It can therefore be expected that locally applied ESBA105 will show superior efficacy in local disease processes accompanied by an improved overall safety profile compared with currently available systemically administered high molecular weight TNFα inhibitors. Intra-articular ESBA105 is currently being evaluated in a first clinical trial in OA of the knee (NCT00819572).