Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The interest in targeting tumor necrosis factor (TNF) in the treatment of Crohn's disease (CD) (an inflammatory bowel disease) and rheumatoid arthritis (RA) originated from a series of studies that revealed higher than normal levels of TNF in patients with these diseases (Hopkins and Meager, 1988), the apparent role of TNF in driving the early cytokine cascade at sites of inflammation (Brennan et al., 1989), TNF-mediated induction of adhesion molecules and angiogenic factors that contribute to synovitis (Furuzawa-Carballeda and Alcocer-Varela, 1999), and the benefit of anti-TNF therapy in animal models such as mouse colitis (Mackay et al., 1998; Kontoyiannis et al., 1999), collagen-induced arthritis in mice (Piguet et al., 1992; Williams et al., 1992), and human TNF-transgenic mice (Keffer et al., 1991; Siegel et al., 1995). Clinical trials involving two distinct TNF antagonists, infliximab (Remicade; Centocor, Inc., Malvern, PA) (Targan et al., 1997; Maini et al., 1999; Lipsky et al., 2000) and etanercept (Enbrel; Immunex, Seattle, WA) (Moreland et al., 1997; Weinblatt et al., 1999; Bathon et al., 2000), later confirmed that inhibiting TNF activity conferred pronounced clinical benefits. Infliximab and etanercept have both received marketing authorization in the United States and Europe for treatment of RA. Infliximab has received United States and European approval for treatment of CD. In addition, infliximab and etanercept have been studied in small, placebo-controlled trials of psoriasis patients with different levels of clinical benefit (Mease et al., 2000; Chaudhari et al., 2001).

Although both infliximab and etanercept are potent neutralizers of TNF bioactivity, there are fundamental differences in their molecular structures, their binding specificities, and the manner in which they neutralize TNF. Infliximab is a chimeric monoclonal antibody (mAb) with murine variable regions and human IgG1 and  constant regions (Fig. 1A) (Knight et al., 1993). The size (149 kDa) and structure of infliximab are therefore similar to those of naturally occurring antibodies. Etanercept is a fusion protein made up of the extracellular domain of the p75 TNF receptor (CD120b) and the hinge and Fc domains of human IgG1 (Mohler et al., 1993), a structure distinct from any known naturally occurring molecule. Importantly, infliximab is not known to bind to any antigen other than TNF, whereas etanercept binds equally well to both TNF and lymphotoxin  (LT), consistent with observations reported for the cellular p75 TNF receptor (Fig. 1A) (Schall et al., 1990; Smith et al., 1990). Each infliximab molecule is capable of binding to two TNF molecules, and up to three infliximab molecules can bind to each TNF homotrimer (G. Heavner, personal communication), thereby blocking all receptor binding sites on TNF (Fig. 1). In contrast, it is believed that the bivalent etanercept molecule forms a 1:1 complex with the TNF trimer in which two of the three receptor binding sites on TNF are occupied by etanercept, and the third receptor binding site is open. In addition, the p75 TNF receptor is known to have fast rates of association and dissociation with TNF (Evans et al., 1994), which suggests that etanercept may only transiently neutralize the activity of an individual TNF molecule. Given these differences between infliximab and etanercept, we sought to compare key TNF-binding characteristics, particularly the relative stability of infliximab/TNF and etanercept/TNF complexes. This was a relevant consideration given recent data that indicate a distinct difference in clinical effectiveness of these two agents in patients with CD (Targan et al., 1997; Sandborn et al., 2001b) or psoriasis (Mease et al., 2000; Chaudhari et al., 2001).

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Schematic illustrations summarizing similarities and differences between infliximab and etanercept. A, infliximab has mouse variable regions (purple) on both heavy and light chains linked to human IgG1 and  constant regions (light blue). Etanercept contains the extracellular domain of human p75 TNF receptor (dark blue) and human IgG1 hinge and Fc domains (light blue). Infliximab binds TNF (yellow) but not LT (green). B, when present at a molar excess over TNF, three molecules of infliximab can bind to each TNF trimer. In contrast, etanercept binds TNF in a 1:1 stoichiometry, even when present in molar excess.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents. Recombinant human TNF and mouse TNF were expressed in transfected mouse myeloma cells and then purified using a recombinant human p55 TNF receptor affinity column. KYM-1D4 cytotoxicity assays (Meager, 1991) were used to show that the specific bioactivity of this TNF was indistinguishable from that of TNF purchased from R & D Systems (Minneapolis, MN). K2 cells, which are Sp2/0 mouse myeloma cells that stably express a mutant, transmembrane version of human TNF (amino acids 1-12 of mature TNF deleted) (Perez et al., 1990), were prepared at Centocor, Inc. (Malvern, PA). A hybridoma cell line expressing V1q, a rat IgD mAb that blocks mouse TNF bioactivity (Echtenacher et al., 1990), was obtained from P. Krammer (German Cancer Research Center, Heidelberg, Germany). DNA encoding the heavy and light chain variable regions of V1q were cloned using conventional molecular techniques into plasmid vectors that encoded human IgG3 and  constant regions, respectively, and then expressed as chimeric V1q (cV1q huG3) at Centocor, Inc. Infliximab and the isotype-matched negative control antibody cM-T412 were obtained at Centocor, Inc., and etanercept was purchased from a pharmaceutical supplier. H18/7 mAb to human E-selectin (Bevilacqua et al., 1987) was obtained from M. Bevilacqua (Brigham and Women's Hospital, Boston, MA). Sodium iodine-125 (¹²⁵I) was purchased from Amersham Biosciences, Inc. (Piscataway, NJ). Infliximab, etanercept, H18/7 mAb, and TNF were iodinated using IODO-GEN reagent (Pierce Chemical, Rockford, IL) as previously described (Knight et al., 1993). Tissue culture media, Hanks' balanced salt solution (HBSS), and phosphate-buffered saline (PBS) were purchased from JRH Biosciences (Lenexa, KS). Fetal bovine serum (FBS) was obtained from Intergen (Purchase, NY). Mycophenolic acid, hypoxanthine, xanthine, and 3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyltetrazolium-bromide (MTT) dye were purchased from Sigma-Aldrich (St. Louis, MO). Goat anti-human Fc-specific antibody was obtained from Jackson Immunoresearch Laboratories (West Grove, PA). A Superose 12 column was purchased from Amersham Biosciences, Inc. (Piscataway, NJ).

Cell Culture. KYM-1D4 cells that endogenously express TNF receptors (Butler et al., 1994) were maintained in RPMI-1640 medium supplemented with 2 mM L-glutamine and 10% FBS (KYM media). Human umbilical vein endothelial (HUVE) cells from Cell Systems (Seattle, WA) were maintained in HUVE cell medium supplied by Cell Systems. K2 cells were maintained in Iscove's modified Dulbecco's medium (IMDM) supplemented with 5% FBS, 2 mM L-glutamine, 0.5 µg/ml mycophenolic acid, 2.5 µg/ml hypoxanthine, and 50 µg/ml xanthine. All cells were cultured in a humidified incubator maintained at 37°C and 5% CO₂.

Binding to Monomer Subunits of TNF. Dimethyl sulfoxide (DMSO) was added to ¹²⁵I-TNF (40-60 µCi/µg) to a final concentration of 10% DMSO and incubated at 20°C for 30 min to allow dissociation of TNF trimers. The mixture was passed over a 10 × 300-mm Superose 12 column equilibrated with PBS, and ¹²⁵I-TNF trimer and monomer were collected separately. Polystyrene 96-well microtiter plates were coated by incubating 50 µl of 1 µg/ml of either infliximab, etanercept, or an isotype-matched, negative control antibody (cM-T412) in the wells overnight at 4°C. After washing with PBS-0.05% Tween 20 (PBS-T), all wells were blocked for 1 h at 37°C with PBS-1% bovine serum albumin and washed three times with PBS-T. Triplicate wells were then incubated with ¹²⁵I-TNF trimer (0.4 µCi, 10 ng/ml) or ¹²⁵I-TNF monomer (0.1 µCi, 2.5 ng/ml) alone or with 5 µg/ml unlabeled TNF. After 1 h at 37°C, the wells were washed with PBS-T and counted for ¹²⁵I.

Binding Assay to Measure Stability of Complexes with Soluble TNF. Each well of a 96-well enzyme immunoassay plate was incubated overnight at 4°C with 100 µl of 0.1 M carbonate, pH 9.6, containing 10 µg/ml goat anti-human Fc antibody. Plates were washed three times with PBS-T and then incubated for 1 h at 37°C in blocking buffer (10 mM HEPES, pH 7.5, containing 0.1% porcine gelatin, 150 µl/well). Wells were incubated for 1 h at 37°C with 100 µl/well of blocking buffer containing 1 µg/ml infliximab or etanercept. Plates were washed three times with PBS-T, and then all TNF binding sites were saturated by incubating the wells for 1 h at 37°C in 100 µl/well of blocking buffer containing 10 ng/ml ¹²⁵I-TNF (40-60 µCi/µg). Wells were washed three times with PBS-T and then filled with 100 µl/well of blocking buffer alone or containing an excess of soluble, unlabeled competitor such as infliximab, etanercept, or human TNF, and subsequently incubated at 37°C. At the indicated time points, triplicate wells were washed three times with PBS-T to remove free ¹²⁵I-labeled TNF. The last wash was aspirated and replaced with 50 µl of scintillation fluid and the entire plate counted in a Packard TopCount gamma counter.

Assay for Bioactivity of Dissociated Soluble TNF. Microtiter plates were coated with goat anti-human Fc antibody and used to capture etanercept as described above. Wells were washed three times with PBS-T and incubated with 100 µl of 10 ng/ml unlabeled human TNF in 100 µl/well of blocking buffer for 1 h at 37°C. Wells were washed three times with KYM media, filled with 100 µl of KYM media, and 500 ng/ml mouse TNF was added to each well as a competitor. After a 1-h incubation at 37°C, the soluble fraction was removed and preincubated for 1 h in fresh wells with either no mAb, 10 µg/ml anti-human TNF mAb (infliximab), 85 µg/ml anti-mouse TNF mAb (cV1q huG3), or a combination of 10 µg/ml anti-human TNF and 85 µg/ml anti-mouse TNF mAb. After the preincubation, the soluble fractions were added to cultures of KYM-1D4 cells (50,000 cells/well in a 96-well plate) and the cells incubated for 16 h at 37°C in the presence of 0.5 µg/ml actinomycin D. To quantitate cell viability, MTT dye was added to a final concentration 0.5 mg/ml and the cells incubated at 37°C for 4 h. The medium was aspirated and 100 µl of 100% DMSO was added to the cells. The difference between the absorbance at 550 and 650 nm was then determined.

HUVE Cell Assay to Measure Stability of Complexes with Soluble TNF. Infliximab or etanercept was mixed with 1 µg/ml human TNF at 10:1 or 30:1 M ratios in HUVE cell medium and incubated for 30 min at 37°C. Serial dilutions of the preformed complexes were then added to confluent HUVE cells cultured in 96-well plates. Cells were incubated with the preformed complexes in 100 µl of HUVE cell medium for 4 h at 37°C and then washed three times with HBSS. The cells were then incubated for 1 h at 37°C in HBSS containing 1 µg/ml ¹²⁵I-labeled anti-E-selectin (20 µCi/µg). Cells were washed three times with HBSS, the last wash was aspirated and replaced with 30 µl of scintillation fluid, and the entire plate was counted in a Packard TopCount gamma counter.

Binding Assay to Measure Stability of Complexes with Transmembrane TNF. K2 cells, which stably express an uncleavable and thus permanently transmembrane (tm) form of TNF, were seeded at a density of 5 × 10⁴ cells/well in a 96-well round-bottom plate in 100 µl of IMDM, 5% FBS. Subsequently, ¹²⁵I-labeled infliximab or ¹²⁵I-labeled etanercept (both at 8.5 µCi/µg) was added to a final concentration of 0.5 µg/ml (enough to saturate all TNF binding sites on the cells). After a 1-h incubation at 25°C, unbound infliximab and etanercept were removed by washing three times with IMDM medium. Fresh IMDM, 5% FBS medium (100 µl) alone, or containing 50 µg/ml of an unlabeled soluble competitor, was added to the cells. Soluble competitors were either infliximab or etanercept for samples treated with radiolabeled infliximab and either infliximab, etanercept, or human LT for samples treated with radiolabeled etanercept. The cells were then incubated at 37°C in 5% CO₂. At different time points, cells in selected wells were washed with PBS, and the number of counts bound to the cells determined using a gamma counter (PerkinElmer Wallac, Wellesley, MA).

Characterization of Infliximab and Etanercept Binding to tmTNF. K2 cells or TNF-negative Sp2/0 control cells were seeded in 96-well round-bottom plates at a density of 5 × 10⁴ cells/well in IMDM, 5% FBS. Varying amounts of ¹²⁵I-labeled infliximab (23.4 µCi/µg) or ¹²⁵I-labeled etanercept (22.4 µCi/µg) were added to the cells. After a 16-h incubation at 4°C, cells were washed four times with culture medium (IMDM, 5% FBS), the last wash was aspirated, and 50 µl of culture medium was added to each well. The cells were then removed with cotton swabs and the number of counts per well was determined using a gamma counter (PerkinElmer Wallac). The resulting binding data were analyzed by nonlinear regression using Prism software (GraphPad Software, San Diego, CA).

HUVE Cell Assay to Compare Ability to Inhibit tmTNF Bioactivity. K2 cells or Sp2/0 control cells were seeded in 96-well round-bottom plates at a density of 1 × 10⁵ cells/well in IMDM, 5% FBS. Varying amounts of infliximab or etanercept in IMDM, 5% FBS were added and the mixture incubated for 1 h at 37°C. This mixture was then added to confluent cultures of HUVE cells in 96-well plates. The resulting cell-cell mixture was incubated for an additional 4 h at 37°C in a 5% CO₂ incubator. Cells were then washed three times with HBSS and incubated for 1 h with 1 µg/ml ¹²⁵I-anti-E-selectin (20 µCi/µg). Cells were washed three times with HBSS, the last wash was aspirated and replaced with 30 µl of scintillation fluid, and the plate counted in a Packard TopCount gamma counter. Data were analyzed using a paired Student's t test to determine whether there was a statistically significant difference between the capacities of infliximab and etanercept to block the bioactivity of tmTNF.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Binding to Monomer Subunits of Soluble TNF. To compare the binding of infliximab and etanercept to TNF monomers, trimeric soluble ¹²⁵I-labeled TNF was treated with DMSO to partially dissociate the homotrimer into monomer subunits, as previously described (Corti et al., 1992). The trimeric and monomeric forms of ¹²⁵I-TNF were separated by gel filtration chromatography and then incubated with immobilized infliximab, etanercept, or a negative control antibody. Determination of the number of ¹²⁵I counts bound indicated that infliximab bound to both the monomer and trimer forms of TNF, whereas etanercept showed significant binding only to the trimer form (Fig. 2).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Binding to monomer subunits of TNF. Soluble trimeric 125I-labeled TNF was partially dissociated into monomer subunits by incubation with DMSO, and then the solution was passed over a Superose 12 gel filtration column to separate 125I-labeled TNF trimer from 125I-labeled TNF monomer. The 125I-labeled TNF trimer and monomer fractions were used to test for binding to immobilized infliximab (), etanercept (), or the isotype-matched negative control antibody cM-T412 () in the absence or presence of an excess of unlabeled TNF as shown. Samples were tested in triplicate. The error bars represent standard deviations. The results shown are representative of two independent experiments.

Stability of Infliximab/Soluble TNF and Etanercept/Soluble TNF Complexes. The stability of infliximab or etanercept complexes with sTNF trimer was evaluated by monitoring the dissociation of ¹²⁵I-TNF from infliximab or etanercept that had been captured on goat anti-human Fc-coated enzyme immunoassay plates. When excess unlabeled TNF was included as a competitor to prevent any dissociated ¹²⁵I-TNF from reassociating with infliximab or etanercept, no dissociation of ¹²⁵I-TNF from infliximab could be detected after more than 3 h of incubation (Fig. 3). However, 50% of the ¹²⁵I-TNF molecules bound to etanercept dissociated within 10 min, and 90% had dissociated within 2 to 3 h. The absence of unlabeled TNF competitor did not reveal dissociation of ¹²⁵I-TNF from etanercept, presumably because any dissociated ¹²⁵I-human TNF rapidly reassociated with etanercept. Similar results were obtained regardless of whether the competitor was unlabeled TNF, infliximab, etanercept, LT (relevant for etanercept only), or p55 sTNF receptor (data not shown). These results demonstrated that, in the presence of a soluble competitor that keeps any dissociated TNF in the soluble phase, TNF dissociation from etanercept was readily detectable, whereas dissociation from infliximab was undetectable.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Measure of dissociation of sTNF from infliximab and etanercept. Infliximab and etanercept were captured on goat anti-human Fc-coated immunoassay plates and allowed to bind saturating amounts of 125I-TNF. Unbound 125I-TNF was removed by washing. The fraction of 125I-TNF still bound to immobilized infliximab (A) and immobilized etanercept (B) in the presence of either no competitor (open symbols) or unlabeled TNF competitor (solid symbols) was determined at various time points. Data are expressed as the percentage of 125I-TNF counts bound relative to the counts bound at time 0. The results shown are the mean of triplicate wells. Error bars represent standard deviations. Data are representative of four independent experiments.

Bioactivity of Soluble TNF that Dissociated from Etanercept. The observation that p55 receptor prevented dissociated TNF from rebinding to etanercept suggests that the dissociated TNF may also bind to cell surface p55 receptor, which would result in cell activation. To determine whether the TNF that had dissociated from etanercept in the above-mentioned experiment was bioactive, we performed a similar experiment to test whether the dissociated TNF could cause killing of human KYM-1D4 cells. To distinguish killing activity due to dissociated TNF from killing activity due to competitor TNF, mouse TNF (which binds etanercept with an affinity comparable to that of human TNF) was used as the competitor in this experiment instead of human TNF. Specific mAbs for either human TNF or mouse TNF were then used to selectively block each TNF species in the mixture before the addition of the soluble fraction to the KYM-1D4 cells. As shown in Fig. 4, KYM-1D4 cells that were treated with TNF-containing supernatants that had not been preincubated with either anti-TNF mAb showed almost complete cell death. However, KYM-1D4 cells that were treated with supernatant that had been preincubated with the anti-human TNF mAb showed approximately 50% cell viability, indicating that there was indeed bioactive human TNF that was contributing to the complete cell killing that was observed in the absence of an anti-TNF mAb. As controls, supernatant that had been incubated with anti-mouse TNF mAb also showed partial inhibition of cell killing, and supernatant preincubated with both anti-human TNF and anti-mouse TNF mAbs showed maximum cell viability (Fig. 4). These results indicated that the human TNF that had dissociated from etanercept was bioactive.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Cell cytotoxicity assay for bioactivity of dissociated sTNF. Human TNF was captured on immobilized etanercept as described for the previous experiment and incubated with an excess of soluble mouse TNF competitor. After 1 h, the soluble fraction was collected and assayed for the presence of bioactive human TNF by adding a portion of it to cultures of KYM-1D4 cells in the presence of either no antibody or an anti-human TNF mAb. As controls, other portions of the soluble fraction were added to KYM-1D4 cells in the presence of either anti-mouse TNF mAb or a combination of anti-human TNF mAb and anti-mouse TNF mAb. After an overnight incubation, KYM-1D4 cell viability was determined by adding MTT dye and measuring optical density values. Higher optical density values indicate higher cell viability. Samples were tested in duplicate. Error bars mark the two values obtained for each test sample. The results shown are representative of two independent experiments.

Measure of Endothelial Cell Activation by Preformed Soluble Complexes. A more physiologically relevant approach for comparing the stability of infliximab/TNF and etanercept/TNF complexes is to measure the activation of human endothelial cells incubated with preformed infliximab/TNF or etanercept/TNF complexes. In this experiment, TNF receptors on HUVE cells effectively act as a competitor for any TNF that dissociates from the complexes. TNF-induced activation was measured as an increase in expression of the adhesion molecule E-selectin on the surface of the HUVE cells. As shown in Fig. 5, incubation of various dilutions of preformed infliximab/TNF complexes with HUVE cells did not lead to cell activation, regardless of whether the infliximab/TNF molar ratio used to prepare the complex was 10:1 or 30:1. In contrast, etanercept/TNF complexes prepared at a molar ratio of 10:1 and, to a lesser extent, at a molar ratio of 30:1 resulted in TNF-induced activation of HUVE cells, further indicating that bioactive TNF had dissociated from etanercept. These data again show that infliximab, but not etanercept, can stably maintain inhibition of the proinflammatory activity of TNF.

View larger version (28K): [in this window] [in a new window]   Fig. 5.   Comparison of the relative stability of infliximab/TNF and etanercept/TNF complexes as measured by endothelial cell activation. Infliximab and TNF, or etanercept and TNF, were mixed together at molar ratios of 10:1 (A) or 30:1 (B) to allow formation of infliximab/TNF () or etanercept/TNF () soluble complexes. These preformed complexes were then serially diluted and added to cultures of HUVE cells to determine whether TNF-mediated induction of E-selectin on the cells could be detected. Cell-surface E-selectin was measured 4 h after adding the complexes to the cultures using an 125I-labeled anti-E-selectin mAb. Each sample was tested in triplicate. Error bars represent standard deviations. These data are representative of three independent experiments.

Stability of Infliximab/tmTNF and Etanercept/tmTNF Complexes. As depicted in Fig. 6, both infliximab and etanercept are believed to be capable of binding to tmTNF. Infliximab is capable of bridging two molecules of tmTNF (Scallon et al., 1995), whereas etanercept presumably can bind only a single molecule of tmTNF. Given the dramatic differences in stability between complexes of infliximab or etanercept with sTNF (Fig. 3), we sought to compare the stability of complexes with tmTNF. This was done by allowing radiolabeled infliximab or etanercept to bind to tmTNF-expressing K2 cells and then monitoring their dissociation over time by measuring radiolabel released into the soluble fraction. As shown in Fig. 7A, no dissociation of radiolabeled infliximab was detected during the 2-h incubation, even when excess unlabeled infliximab was used as a competitor to prevent dissociated radiolabeled infliximab from reassociating with the K2 cells. Similar results were seen for infliximab when etanercept or TNF were used as competitors. In contrast, 90% of the radiolabeled etanercept had dissociated from the cells after only 10 min when unlabeled etanercept was used as competitor (Fig. 7B), and similar results were seen when infliximab or sTNF was used as competitor. These data show that the stability of infliximab/TNF complexes and the relative instability of etanercept/TNF complexes are similarly observed for both the transmembrane and soluble forms of TNF.

View larger version (41K): [in this window] [in a new window]   Fig. 6.   Schematic illustration depicting infliximab and etanercept binding to cell surface tmTNF. Etanercept is believed to bind in a 1:1 stoichiometry with tmTNF, leaving one receptor binding site free, even when etanercept is present in excess. Each molecule of infliximab is believed to be able to bind either one or two molecules of trimeric and/or monomeric tmTNF, and up to three molecules of infliximab are believed to bind each tmTNF trimer, leaving few or no receptor binding sites free when infliximab is present in excess. The variable regions of infliximab are indicated in purple and the p75 TNF receptor portion of etanercept is indicated in dark blue. Both the heavy and light chains of human IgG1  constant regions (infliximab) and human IgG1 hinge and Fc domains (etanercept) are shown in light blue. tmTNF is shown in yellow.

View larger version (26K): [in this window] [in a new window]   Fig. 7.   Measure of dissociation of infliximab or etanercept from tmTNF-expressing cells. 125I-Labeled infliximab (A) or 125I-labeled etanercept (B) was allowed to bind to tmTNF-expressing K2 cells, unbound material was removed by washing, and a competitor was added to selected samples to prevent any dissociated molecules from reassociating with the cells. At different time points, the amount of 125I-labeled infliximab or 125I-labeled etanercept still bound to the cells in the absence of competitor () or in the presence of the competitors infliximab (), etanercept (), or human TNF () were determined. The data are expressed as the percentage of counts bound relative to the counts bound at time 0. Samples were tested in duplicate. Error bars mark the two values obtained from the duplicate samples. The results shown are representative of two independent experiments.

Characterization of Binding to tmTNF. To further compare the binding of infliximab and etanercept to tmTNF, K2 cells expressing uncleavable tmTNF on their surface were incubated with varying amounts of radiolabeled infliximab or etanercept. The amount of drug bound under equilibrium binding conditions was measured by monitoring the level of radioactivity associated with the cells. As shown in Fig. 8, when added at saturating concentrations, nearly 4 times more molecules of infliximab than etanercept were bound to the tmTNF-expressing cells. In addition, infliximab had a slightly lower dissociation constant of 0.45 nM, compared with 1.15 nM for etanercept. These data indicated that a notably greater number of infliximab molecules bind tmTNF on K2 cells than etanercept, and that infliximab has a slightly greater avidity for tmTNF relative to etanercept.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Comparison of avidity and number of molecules bound per cell for infliximab and etanercept on tmTNF-expressing K2 cells. K2 cells were incubated with varying amounts of 125I-labeled infliximab () or 125I-labeled etanercept () and the number of counts bound to the cells was determined after removing unbound material. Data were analyzed by nonlinear regression to determine the dissociation constant (KD) values for infliximab and etanercept. The number of molecules bound per cell is shown. Each sample was tested in duplicate. Error bars mark the two values obtained for the duplicate samples. These data are representative of three independent experiments.

Inhibition of tmTNF-Mediated Cell Activation. The experiments described above revealed that, compared with etanercept, infliximab shows higher overall binding, higher avidity, and much greater stability of binding to tmTNF. To determine whether these characteristics of infliximab translate into more effective neutralization of tmTNF signaling, K2 cells were preincubated with varying amounts of infliximab or etanercept, and the mixture added to HUVE cells. As shown in Fig. 9, both infliximab and etanercept inhibited tmTNF-dependent E-selectin expression in a concentration-dependent manner. However, a statistical analysis showed that infliximab was significantly (p < 0.05) more potent than etanercept at blocking tmTNF-mediated E-selectin expression, requiring approximately 1/10 the concentration required of etanercept to achieve comparable levels of inhibition. These biological results are consistent with the different binding characteristics described above for these two TNF antagonists.

View larger version (19K): [in this window] [in a new window]   Fig. 9.   Comparison of infliximab and etanercept inhibition of tmTNF-mediated induction of E-selectin on endothelial cells. tmTNF-expressing K2 cells were preincubated with varying amounts of infliximab or etanercept and then coincubated with HUVE cells (without removing unbound infliximab or etanercept). The amount of E-selectin on the HUVE cells induced by the K2 cells in the presence of infliximab () or etanercept () was determined using 125I-labeled anti-E-selectin mAb. Each sample was tested in triplicate. Error bars represent standard deviations. Asterisks mark those data points for infliximab that showed a statistically significant difference (p < 0.05; Student's t test) compared with etanercept. The data shown are representative of three independent experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The ligands TNF and LT bind to the p55 and p75 TNF receptors (Idriss and Naismith, 2000). Activation of these receptors initiates intracellular signal transduction cascades that lead to cell proliferation, up-regulation of proinflammatory mediators, or apoptosis. Although TNF participates in normal immune system function, excessive TNF receptor activation can lead to severe inflammatory conditions and tissue damage (Bazzoni and Beutler, 1996). Therefore, inhibiting activation of these receptors could provide benefit by down-regulating a chronic inflammatory response.

We sought to better understand differences between two therapeutic modalities that block TNF-mediated inflammatory responses: the chimeric mAb infliximab and the TNF receptor fusion protein etanercept. These agents not only have different structures but also have different binding specificities. Infliximab is specific for TNF, whereas etanercept binds and neutralizes both TNF and LT. We report herein that another specificity difference is that infliximab binds to monomer subunits of TNF, whereas etanercept does not. Etanercept would not be expected to bind to monomeric TNF because receptor binding site(s) on TNF is in the cleft formed between two TNF subunits, as demonstrated conclusively for LT (Banner et al., 1993). It is possible that, by binding to TNF monomers, infliximab may slow or even prevent association of monomeric subunits of TNF to form bioactive trimeric TNF.

The distinction between infliximab and etanercept extends beyond differences in their structures and binding specificities. We have shown that infliximab maintains a stable complex with sTNF, whereas etanercept does not. The relative instability of etanercept/TNF complexes reported herein is in excellent agreement with results of a similar experiment using a p75 IgG fusion protein (Evans et al., 1994). Those authors suggested that the capacity of soluble p75 receptor to efficiently bind and then release TNF may serve to prolong the circulating half-life of TNF. The most physiologically relevant "competitors" for TNF molecules that have dissociated from etanercept are cell-surface p55 and p75 TNF receptors. We have shown that sTNF that dissociates from etanercept is bioactive, i.e., dissociated TNF can bind and signal through these cell-surface receptors. As a result, the relative instability of etanercept/TNF complexes may not allow complete down-regulation of inflammatory processes at tissue sites where there is insufficient etanercept to prevent dissociated TNF from binding and activating cell-surface receptors. In contrast, the greater stability of infliximab/TNF complexes suggests that infliximab may prevent binding of TNF to these cellular receptors more efficiently.

In addition to binding sTNF, infliximab and etanercept also bind to tmTNF (Fig. 8). We have shown that infliximab has a slightly higher avidity for tmTNF than etanercept as measured under equilibrium binding conditions. However, the rates at which infliximab and etanercept dissociate from tmTNF are markedly different. Complexes of infliximab and tmTNF were significantly more stable than complexes of etanercept and tmTNF. In addition, a greater number of infliximab molecules than etanercept molecules bound to tmTNF-expressing cells. This may be partially explained by our data showing that, unlike etanercept, infliximab binds to both monomeric and trimeric forms of sTNF, and so probably monomeric and trimeric forms of tmTNF as well. In addition, three molecules of infliximab can simultaneously bind a single TNF trimer (G. Heavner, personal communication), whereas it is believed that only one molecule of etanercept can bind any particular TNF trimer. Therefore, the greater stability of infliximab/tmTNF complexes and the greater degree of binding to tmTNF by infliximab are the most likely reasons why infliximab was observed to be more effective than etanercept at neutralizing tmTNF bioactivity. These are important observations because tmTNF is also a potent signaling molecule (Perez et al., 1990; Georgopoulos et al., 1996).

In assessing the physiological relevance of these observations, the ultimate question is whether there are differences in the capacities of infliximab and etanercept to block TNF-mediated disease processes. Both agents have been effective in providing symptomatic improvement and inhibition of structural damage progression in RA patients (Bathon et al., 2000; Lipsky et al., 2000). However, this is in marked contrast to results observed from controlled clinical trials in treatment of CD in that infliximab, but not etanercept, was demonstrated to be clinically effective (Targan et al., 1997; Sandborn et al., 2001b). Table 1 summarizes results of recently completed, controlled clinical trials in CD with infliximab and etanercept. In addition, results of a controlled clinical trial with CDP571, a humanized, anti-TNF mAb of the IgG4 class (Sandborn et al., 2001a) are also shown. The three trials each studied patients with active, moderate-to-severe CD and used the same definition for clinical response, i.e., a 70-point decrease in the CD activity index. The clinical response rates for the two mAbs infliximab and CDP571 were superior to those observed after placebo treatment, whereas an improved clinical response relative to placebo was not demonstrated for etanercept. Although the two mAbs both demonstrated clinical efficacy, a calculation of number of patients that would need to be treated to observe a clinical benefit from therapy suggests that twice as many patients would need to be treated with CDP571 (3.7 patients) compared with infliximab (1.6 patients) to observe this benefit. This difference between infliximab and CDP571 might be attributed to differences in effector functions such as complement-mediated cell lysis that may occur with IgG1 antibodies but not with IgG4 antibodies. The odds ratio calculations also demonstrated notable differences between these therapies in the treatment of CD. As shown in Table 1, the 95% confidence intervals did not overlap between infliximab and etanercept, indicating a statistically significant difference between these two agents.

Preliminary data from two controlled clinical trials indicate that there may also be differences in clinical benefits observed with infliximab and etanercept in treatment of psoriasis. In one trial (Mease et al., 2000), 26% (5 of 19) of psoriatic arthritis patients with psoriasis who received etanercept (25 mg given twice weekly as a subcutaneous injection) achieved at least 75% improvement in the psoriasis area and severity index compared with none of the placebo-treated patients (p = 0.015). In a separate trial (Chaudhari et al., 2001), 82% (9 of 11; p = 0.0089 versus placebo) of plaque-type psoriasis patients that received 5 mg/kg infliximab and 73% (8 of 11; p = 0.03 versus placebo) of patients that received 10 mg/kg infliximab achieved at least 75% improvement in the psoriasis area and severity index, compared with 18% (2 of 11) of placebo-treated patients. Although infliximab and etanercept both showed significant improvement compared with placebo in the two patient populations studied, a greater proportion of patients responded to infliximab than to etanercept.

Clearly, differences in infliximab and etanercept binding to TNF as described in this report must be considered as potential factors that contribute to the different clinical results in CD and psoriasis clinical trials. Maintaining stable neutralizing complexes with TNF may be key to realizing clinical benefits in CD patients. But there are other possible reasons for the different clinical outcomes. Different dosing regimens and routes of administration were used for infliximab and etanercept in the CD clinical trials, although the total doses administered were similar. Infliximab was administered as a 5-mg/kg intravenous infusion (400 mg of total dose or an average of 200 mg/month if patients were retreated every 8 weeks) and etanercept was administered as a subcutaneous injection of 25 mg twice weekly (also 200 mg/month). However, the less frequent but larger dose of infliximab results in higher peak serum concentrations compared with more frequent, smaller doses of etanercept. This could in turn result in higher infliximab concentrations in diseased tissues.

The explanation for why infliximab and etanercept show similar clinical benefit in the treatment of RA despite having different profiles in CD and psoriasis is unknown. Factors that might contribute to the similar clinical outcomes in RA versus different outcomes in CD and psoriasis may be accessibility of involved tissues to these products (at doses administered), potential role of LT in different disease processes, amount of TNF in involved tissues, or relative sensitivity of diseased tissues to effects of TNF.

There are also important distinctions in the safety profiles of these two products, most notably an increased risk of specific types of infection for infliximab (Keane et al., 2001) or etanercept (Fisher et al., 1996; Song et al., 2001), or possibly an increased risk of neurological disorders for etanercept (Mohan et al., 2001). It is possible that differences in stability of complexes with TNF, antigen specificity, and cell lysis properties may contribute to differences in their overall safety profiles in addition to differences in clinical outcomes.

In conclusion, we have demonstrated that, compared with etanercept, infliximab forms more stable complexes with sTNF and tmTNF. The dissociation of complexes between etanercept and TNF could allow the dissociated TNF to bind and activate cell surface receptors, as we have demonstrated with cytotoxicity assays using human tumor cells and cell activation assays using human endothelial cells. In contrast, the binding characteristics of infliximab are consistent with conferring more complete and sustained neutralization of TNF, which may be reflected in the greater responses among Crohn's disease and psoriasis patients treated with infliximab.