The CC chemokine, eotaxin1 (CCL11) is an important regulator of eosinophil function. A marked accumulation of eosinophils in tissues has been correlated with the up-regulation of eotaxin1 expression in several diseases. The potential therapeutic value of neutralizing the effects of eotaxin1 in inflammatory conditions (including asthma) is under investigation. A human single-chain fragment variable antibody that neutralizes human eotaxin1 (CAT-212) was produced using antibody phage display and converted to whole antibody IgG4 format (CAT-213). A novel approach to lead optimization in which the length of the variable heavy chain complementarity-determining region 3 was reduced by one amino acid resulted in an increase in potency of >1000-fold compared with the parent anti-eotaxin1 antibody. The optimized antibody binds eotaxin1 with high affinity (80.4 pM) and specificity. CAT-213 and CAT-212 do not bind or neutralize a range of other human proteins including human monocyte chemoattractant protein-1, a structurally similar chemokine. CAT-213 neutralizes the ability of eotaxin1 to cause an increase in intracellular calcium signaling (with an IC50 value of 2.86 nM), migration of CCR3-expressing L1.2 cells (with an IC50 value of 0.48 nM), and inhibition of the eotaxin1-evoked shape change of human eosinophils in vitro (with an IC50 of 0.71 nM). Local administration of CAT-213 to mice (1–100 μg kg–1) attenuates dermal eosinophilia induced by human eotaxin1, achieving >90% inhibition of eosinophil influx. CAT-213 may therefore be of therapeutic value in inhibiting diseases in which eotaxin1 and eosinophils play a major role, for example, severe asthma.
Chemokines are a subfamily of cytokines that mediate leukocyte extravasation into tissues in both homeostatic and pathological states. Eotaxin1 (CCL11) is a CC chemokine that was first purified from the bronchoalveolar lavage fluid of allergic guinea pigs (Griffiths-Johnson et al., 1993; Jose et al., 1994a). The first functional role that was elucidated for eotaxin1 and generated its name was eosinophil chemotaxis mediated via CCR3 agonism (Griffiths-Johnson et al., 1993; Jose et al., 1994b). Eotaxin1 has also been shown to have a variety of effects associated with eosinophil activation, including lipid mediator synthesis, integrin activation, and degranulation (for review, see Bandeira-Melo et al., 2001). Taken together these data implicate eotaxin1 as an important regulator of eosinophil function.
Many diseases are characterized by a marked accumulation of eosinophils (eosinophilia). These diseases include allergic disorders such as asthma, rhinitis, conjunctivitis, and Churg-Strauss syndrome (a rare form of serious systemic vasculitis), various inflammatory diseases (pneumonia and inflammatory bowel disease), and some cancers (most commonly in adult T-cell leukemia/lymphoma and Hodgkin's disease) (for review, see Giembycz and Lindsay, 1999).
Eosinophilia has been correlated with the up-regulation of eotaxin1 expression in several diseases. Eotaxin1 expression is increased in the bronchial mucosa, sputum, and bronchoalveolar lavage fluid of patients with asthma compared with that in normal individuals, and this correlates with local eosinophil numbers (Taha et al., 2001). The sputum eotaxin1 level is correlated to the severity of disease, with the greatest level of eotaxin1 being in sputum from severe asthmatics (Dent et al., 2004). Furthermore, asthma exacerbations result in further increases in plasma eotaxin1 concentrations (Lilly et al., 1999) and associated sputum eosinophilia and are associated with chronic impairment of lung function (Nakamura et al., 1999). Eotaxin1 expression is also up-regulated in the nasal tissue of patients with allergic rhinitis, in the conjunctiva of patients with active allergic conjunctivitis (Eperon et al., 2002), and in a number of other diseases that are known to involve some eosinophilic element, e.g., ulcerative colitis (Hogan et al., 2001) and atopic dermatitis (Morita et al., 2001).
The majority of the eotaxin1 literature focuses on its effects on eosinophils; however, it is important to note that eotaxin1 also mediates effects on mast cells, basophils, T helper 2 lymphocytes and dendritic cells via CCR3 (Heinemann et al., 2000; de Paulis et al., 2001; Beaulieu et al., 2002). Furthermore, although the main receptor for eotaxin1 is CCR3, recent evidence indicates that it also interacts with CCR2, CCR5, and CXCR3 (for review, see Bandeira-Melo et al., 2001), although it is noted that these interactions require high concentrations of eotaxin1, which is an antagonist/weak agonist of CCR2, a weak agonist of CCR5, and an antagonist of CXCR3. Thus, particularly given the reported effects of eotaxin1 on mast cells and T helper 2 cells, future data may reveal more associations between eotaxin1 and allergic diseases.
The pleiotropic effects of eotaxin1 on eosinophil function and its association with various diseases have led several investigators to examine the effect of neutralizing eotaxin1 in vivo. Antibodies given either locally or systemically have been shown to substantially decrease eosinophilia induced by either allergen or eotaxin1 (Gonzalo et al., 1996; Humbles et al., 1997). These findings support the potential therapeutic value of eotaxin1 neutralization in allergic conditions.
Antibodies are proving to be an important class of drugs for treatment of chronic diseases such as asthma. For example, omalizumab (anti-IgE humanized IgG1) reduces exacerbations and corticosteroid use in patients with allergic asthma (Schulman, 2001). Furthermore, phage display is now a well-established technique for the isolation of human antibodies in vitro (Hoogenboom, 2005). In this study, we have used phage-display technology to isolate a human anti-eotaxin1 antibody. Novel antibody optimization techniques that increased antibody potency 1000-fold have been described. The resulting antibody has been characterized in a range of in vitro assays and in vivo models of eosinophilia in which it demonstrates therapeutic potential.
Materials and Methods
Human Antibody Phage Display. The following processes were used to identify the selective anti-eotaxin1 parental clone, 3G3, from a human antibody phage library. Optimization of this clone, as described below, results in the generation of CAT-212 scFv and CAT-213 IgG4.
Selection of the Parental Anti-Eotaxin1 Antibody. Selections were carried out as described previously (Vaughan et al., 1996). In brief, human eotaxin1 was coated at 10 μg ml–1 on to immunotubes or coupled to disuccinimidyl suberate-activated BSA coated onto microtiter plates. Three rounds of panning selections were performed using a scFv human antibody phage library with approximately 1.3 × 1010 individual recombinants (Vaughan et al., 1996). The selected output was then evaluated for eotaxin1 binding using phage ELISA as described previously (Vaughan et al., 1996). In brief, phage-displayed scFvs were rescued in a 96-well format and transferred to eotaxin1 (0.5 μg ml–1)-coated plates. Anti-gene VIII-horseradish peroxidase conjugate (1 in 5000 in low-fat milk (Marvel)-PBS) and 3,3′,5,5′-tetramethyl benzidine substrate were used to detect bound antibodies. After addition of 0.5 M sulfuric acid, the absorbance was measured at 450 nm, and clones with the highest absorbance were selected for further study as purified scFv.
Purification of scFvs by immobilized metal affinity chromatography was conducted as described previously (Vaughan et al., 1996). scFvs were assessed for functional neutralization of eotaxin1 using the chemotaxis assay (described below). The scFv 3G3 was shown to neutralize eotaxin1 (see Results) and was chosen for potency optimization.
Optimization of Lead Clone 3G3. The overall strategy for the optimization of 3G3 is shown in Fig. 1. Heavy chain CDR3 randomization and light chain shuffling were conducted in parallel.
Construction of Libraries to Randomize Heavy Chain CDR3. Three individual phage libraries were constructed in which the entire sequence of the parent scFv (3G3) was kept constant apart from the terminal five residues of the VHCDR3. Three mutagenic oligonucleotides were designed to incorporate this variation, as shown below. Oligo 1 maintained the VHCDR3 at the same length, oligo 2 incorporated an amino acid insertion, and oligo 3 incorporated an amino acid deletion (N = random nucleotide). The oligos are indicated as follows: oligo 1, 5′-GCC TCC ACC ACT CGA GAC CGT CAC CAT GGT GCC CTG ACC CCA NN(G/C) NN(G/C) NN(G/C) NN(G/C) NN(G/C) CCC GTA GTC CGT ATC TCT CG-3′; oligo 2, 5′-GCC TCC ACC ACT CGA GAC CGT CAC CAT GGT GCC CTG ACC CCA NN(G/C) NN(G/C) NN(G/C) NN(G/C) NN(G/C) NN(G/C) CCC GTA GTC CGT ATC TCT CG-3′; and oligo 3, 5′-GCC TCC ACC ACT CGA GAC CGT CAC CAT GGT GCC CTG ACC CCA NN(G/C) NN(G/C) NN(G/C) NN(G/C) CCC GTA GTC CGT ATC TCT CG-3′.
The parent scFv VH was PCR-amplified using one of the oligos (1, 2, or 3) together with a vector primer, pUC19Rev (Vaughan et al., 1996). The mutagenized VH DNA fragments were cloned into a heavy chain acceptor vector plasmid (p-Cantab6) containing the parent scFv VL and a linker. The plasmid DNA was transfected into electrocompetent Escherichia coli TG-1 cells, and phage was then rescued for selection as described previously (Vaughan et al., 1996). The first round of selection was panning on 10 μM eotaxin1. All subsequent rounds were on soluble-biotinylated eotaxin1 (biotinylated in house) at decreasing concentrations from 250 nM to 10 pM. This was done to bias the selections toward higher affinity scFvs. Outputs from each round of selection were tested for binding to biotinylated eotaxin1 coated to a BIAcore chip (BIAcore, Uppsala, Sweden), and any scFvs that bound were sequenced as described previously (Vaughan et al., 1996).
Construction of Library to Shuffle Light Chains. The parent scFv VH was PCR-amplified using pUC19Rev (Vaughan et al., 1996) and a primer to incorporate a XhoI site at the end of the VH: 5′ACCGCCTCCACCACTCGAGACGGTGACCATTGTCCC(TC)(TC)(GT)GCCCCA-3′.
The DNA was ligated into a p-Cantab 6 plasmid prepared from a spleen phage-display library containing the entire VL repertoire from that library. The phages were then recovered and selected as described for the heavy chain library construction.
Conversion of scFv (CAT-212) to IgG (CAT-213). Recombinant human heavy (γ4) and light (κ) chain genes were assembled from the CAT-212 VH and VL sequences and inserted into a vector carrying a glutamine synthetase selectable marker. The resulting vector was transfected into mouse myeloma NS0 cells, and recombinants were selected in glutamine-free medium (Bebbington et al., 1992). Cell lines producing IgG (CAT-213) were identified by ELISA analysis of cell supernatants. A high-producing cell line was grown to large scale in serum-free medium. Then, pure preparations of CAT-213 were recovered from the cell supernatant using protein A affinity chromatography, ion exchange chromatography, and gel filtration steps.
Specificity ELISA. The specificity of CAT-212 scFv was assessed by phage ELISA as described above. Plates were coated with the following antigens: eotaxin1, eotaxin2, eotaxin3, macrophage inhibitory protein-1α (CCL3), monocyte chemoattractant protein (MCP)-1 (CCL2), -2 (CCL8), -3 (CCL7), and -4 (CCL13), tumor necrosis factor-α, RANTES (CCL5) (all coated at 10 μg ml–1), IL-1α, IL-1β, IL-5, IL-12, and IL-18 (all coated at 1 μg ml–1) and TGF-β1 and TGF-β2 (both coated at 0.5 μg ml–1).
The specificity of CAT-213 IgG4 was also assessed by ELISA. Again plates were coated with a subset of the human antigens at 4°C overnight. After washing (PBS + 0.1% Tween 20), CAT-213 (1 μg in 2% Marvel-PBS) was added to the plate for 1 h. After washing again, CAT-213 was detected with anti-human IgG4 horseradish peroxidase conjugate (1 h incubation with 1:3000 dilution in 2% Marvel-PBS; The Binding Site Ltd., Birmingham, UK) and quantified using 3,3′,5,5′-tetramethyl benzidine substrate (Sigma Chemical, Gillingham, UK). The reaction was stopped after 10 min with sulfuric acid (100 μl 2 M), and optical density was measured at 450 nM.
Chemotaxis Assay. L1.2 cells stably transfected with human CCR3 (L1.2-hCCR3) were maintained in suspension in RPMI 1640 containing 10% heat-inactivated FCS, 2% l-glutamine, 100 U ml–1 penicillin, 100 μg ml–1 streptomycin, 250 μg ml–1 kanamycin, and 400 μg ml–1 Geneticin. For the assay, the cells were washed once in PBS and resuspended at 107 cells ml–1 in assay buffer (RPMI 1640, 1% endotoxin-free BSA, 100 U ml–1 penicillin, and 100 μg ml–1 streptomycin). Test antibodies, scFvs or IgGs, were preincubated with 6 nM human eotaxin1 (30 min, 37°C) and placed into the lower chambers of 24-well Transwell plates with 3-μm pores (Corning Life Sciences, Schipol-Rijk, The Netherlands); 100 μl of cells (1 × 106 cells) were placed into the upper chamber of each Transwell plate, and the assay was incubated at 37°C for 4 h (5% CO2). Live cells migrating to the lower chamber were counted using a flow cytometer (FACSCalibur, BD Biosciences, Cowley, UK). Results are expressed as percent control (i.e., eotaxin1-induced) migration using the following equation: where M is migration, Mab+eot is migration due to antibody + eotaxin, Mbuffer is mean migration due to buffer alone, and Meot is mean migration due to eotaxin alone. The number of cells migrating in the presence of eotaxin1 is 100%. It should be noted that basal migration was minimal (∼0.5% of the maximal response with eotaxin1).
Experiments were also conducted to assess the ability of CAT-213 to neutralize eotaxin1 from mouse, rat, and monkey (see Materials for details). In these experiments, 6 nM eotaxin1 from each species was used. The assay was conducted as described above except that 2 × 106 L1.2-hCCR3 cells were placed into the upper chamber to improve the cell migration response (conditions were optimized in pilot experiments; data not shown).
Eotaxin1 Binding Assay. A 96-well phosphor-impregnated microtiter plate (Flash Plate SMP200; PerkinElmer Life and Analytical Sciences, Boston, MA) was coated with 40 nM CAT-212 or 1 nM CAT-213 diluted in 0.05 M carbonate-bicarbonate buffer, pH 9.6, and incubated at 4°C for 4 h. The plate was then blocked with PBS containing 1% Marvel overnight at 4°C. The next day it was washed with PBS and serial 2-fold dilutions of [125I]eotaxin1 (GE Healthcare, Buckinghamshire, UK) plus or minus a 100-fold excess of unlabeled eotaxin1 in assay buffer (RPMI 1640 plus 0.5% BSA) were added. This mixture was incubated at room temperature for 2 h and then counted on a Packard Topcount scintillation counter for 2 min/well.
Calcium Flux Assay. L1.2-hCCR3 cells were incubated with 0.5 μg ml–1 sodium butyrate for 24 h before the experiment to increase expression of CCR3. The cells were resuspended at 4 × 106 cells ml–1 in RPMI 1640 containing 2 μM Fluo-3/acetoxymethyl ester, 0.03% Pluronic acid, and 0.1% FCS and then incubated for 45 min at 37°C. Cells were then washed and resuspended in fluorometric imaging plate reader (FLIPR) buffer (125 mM sodium chloride, 5 mM potassium chloride, 1 mM magnesium chloride, 1.5 mM calcium chloride, 25 mM Hepes, 5 mM glucose, and 0.1% FCS, pH 7.4; Molecular Devices, Sunnyvale, CA) at 1 × 106 cells ml–1. Then 1 ×10–5 cells (100 μl of cell suspension per well) were transferred to the wells of a 96-well black plate with a clear base, and the plate was centrifuged to give a monolayer of cells. In a separate plate, 1 to 300 nM CAT-212 or CAT-213 (or irrelevant scFv or IgG4) was preincubated with a final concentration of 10 nM eotaxin1 (EC80) for 10 min. The cell plate was placed in the FLIPR (Molecular Devices), and 50 μl of each treatment was added to each well after a 10-s interval. Fluorescence readings were taken every 1 s for the 1st min and at 5-s intervals thereafter.
The effect of CAT-213 on MCP-1-induced calcium flux on human peripheral blood monocytes was also evaluated using the FLIPR. Peripheral blood mononuclear cells (PBMCs) were purified from buffy coat (Blood Transfusion Service, Addenbrooke's Hospital, Cambridge UK) as follows. Fifteen milliliters of buffy coat was layered on an equal volume of Histopaque 1.077 (Sigma Chemical) and then spun at 400g for 40 min (without brake). PBMCs were then aspirated from the resulting interface, washed in PBS, and centrifuged at 300g for 10 min, and the resulting cell pellet was resuspended in 20 ml of ice-cold distilled H2O for 15 s. This was followed by an immediate addition of ice-cold 1.8% sodium chloride. PBMC were then washed with PBS containing 2 mM EDTA, centrifuged (1200 rpm, 5 min), and resuspended in 600 μl of phosphate-buffered saline-2 mM EDTA. Monocytes were then purified from PBMCs using a monocyte isolation kit (Miltenyi Biotec, Bisley, Surrey, UK) according to the manufacturer's protocol. The purified monocytes were resuspended at 3 × 106 cells ml–1 in RPMI 1640 (Invitrogen, Carlsbad, CA) containing 0.1% FCS, 20 mM HEPES, 2.5 mM probenecid, 0.03% Pluronic acid, and 2 μM Fluo-4/acetoxymethyl ester dye and incubated for 1 h at 37°C, 5% CO2. Cells were then washed, centrifuged, and resuspended in 7 ml of FLIPR buffer (125 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1.5 mM CaCl2, 25 mM HEPES, 2.5 mM probenecid, 5 mM glucose, and 1% FBS, pH 7.6). Dye-loaded cell suspension was transferred to a 96-well black-walled plate with a clear base (70 μl/well, ∼3 × 105 cells/well), and the plate was centrifuged to give a monolayer of cells. In a separate plate, a dose titration of CAT-213 (100 μg/ml–150 pg/ml) and/or control antibodies [CAT-001, an irrelevant IgG4 (100 μg/ml–150 pg/ml) and anti-human MCP-1, MAB279 (50 μg/ml–35 pg/ml)] were preincubated with 5 nM human MCP-1 (approximate EC80) for 10 min. The cell plate was placed in the FLIPR (Molecular Devices), and 50 μl of each treatment was added to each well after a 10-s interval. Fluorescence readings were taken every 1 s for the 1st min, and at 5-s intervals thereafter. All plates were then analyzed in the FLIPR according to the manufacturer's protocols.
Eosinophil Shape Change Assay. Eosinophil shape change was carried out essentially as described previously (Sabroe et al., 2000). Blood (50 ml) from normal healthy human volunteers was taken into EDTA, and erythrocytes were removed by dextran sedimentation. Polymorphonuclear leukocytes were then separated from mononuclear cells using discontinuous plasma-Percoll gradients. After separation, the polymorphonuclear leukocytes were resuspended in buffer (PBS containing 1 mM CaCl2, 1 mM MgCl2, 10 mM Hepes, 10 mM d-glucose, and 0.1% BSA, pH 7.3) to a concentration of 5 × 106 cells ml–1 and allowed to rest for 30 min at 37°C. Aliquots of cells (100 μl) were then incubated with varying experimental treatments of buffer with or without human eotaxin1 (0.1–20 nM) or human eotaxin2 (5.5 nM) and either CAT-213 (0.1–300 nM) or negative control antibody (0.1–300 nM) in a total volume of 400 μl. Tubes were returned to 37°C, incubated for 8.5 min, and then transferred to an ice-water bath and 25 μl of 10 × Cellfix buffer (Becton Dickinson) was added. Samples were analyzed immediately. Eosinophils were identified on the flow cytometer (FACSCalibur) by their FL2 autofluorescence. Shape change was calculated as a percentage of the FSC shape change caused by buffer alone.
Air Pouch Eosinophilia Model. The model was carried out essentially as described previously (Das et al., 1997). Procedures were conducted according to the 1986 Animal Procedures Act (United Kingdom).
Female BALB/c mice (17–21 g; Harlan Sera-Lab, Bicester, Oxon, UK) were housed within a small animal barrier unit. Mice were housed three to a cage with a 12-h day/night cycle (lights on 7:00 AM). Animals were allowed to acclimatize to the facility for at least 1 week before experimentation and were allowed food and water ad libitum. Mice were sensitized to ovalbumin by s.c. injection of 100 μg of ovalbumin in aluminum hydroxide gel (0.4 ml of saline containing 3.3 mg of aluminum hydroxide) on days 1 and 8. Air pouches were formed on the backs of mice on day 9 by anesthetizing them with isoflurane and injecting 2.5 ml of sterile air (0.2 μm filtered) s.c. on the back between the scapulae. On day 12, mice were injected with a further 2.5 ml of sterile air to reinflate the air pouch. On day 15, 100 pmol kg–1 of murine IL-5 (i.v.) was administered followed 30 min later by 1 μg of human eotaxin1 intrapouch (i.po.), and measurements were made 6 h later. In pilot experiments, it was established that IL-5 administration was essential for eosinophil migration to the air pouch. A control group treated with saline and vehicle for antibodies was also included in all experiments to give the baseline cell influx (i.e., a sham challenge group). Antibodies (CAT-213 and CAT-212) or their respective control antibodies (irrelevant IgG4 and scFv, respectively) were administered either systemically (i.v., 30 min prehuman eotaxin1) or locally (i.po., concomitant with human eotaxin1). Vehicle control groups were included in all experiments, and control mice received PBS via the appropriate route.
Mice were killed before lavage by CO2 asphyxiation. The air pouch of each mouse was lavaged with 1 ml of ice-cold PBS without calcium or magnesium containing 5 U ml–1 heparin. An aliquot (10 μl) of each lavage fluid sample was removed for measurement of the total number of leukocytes (total cells, Fast-Read disposable cell counter; Immune Systems Limited, Paignton, UK). A portion of the remaining lavage fluid (∼100,000 cells) was centrifuged to prepare cell cytospins. Cytospins were air-dried, fixed, and stained with modified Wright-Giemsa stain. The differential leukocyte subpopulations were quantified by differential cell counting using light microscopy under oil immersion (1000×). All cell counting was performed in a masked fashion.
Materials. Recombinant human eotaxin1 was purchased from Cambridge Bioscience (Cambridge, UK) and R&D Systems (Abingdon, UK). Mouse eotaxin1 was purchased from R&D Systems and synthetic human, rat, and monkey eotaxin1 from Albachem Limited (Edinburgh, UK). Macrophage inhibitory protein-1α, MCP-1, -2, -3, and -4, IL-1α, and IL-1β were purchased from Cambridge Bioscience, and IL-5 and IL-18 were from R&D Systems, TGF-β1 and TGF-β2 were from ImmunoKontact (Abingdon, UK), and RANTES was from Peprotech (London, UK). IL-12 was a gift from the Genetics Institute (Cambridge, MA), and tumor necrosis factor-α was a gift from BASF Bio-Research Corporation (Boston, MA). All reagents administered in vivo were tested for endotoxin content. These reagents contained less than 1 endotoxin unit/mg protein (measured by the LAL test; Charles River Laboratories, L'Arbresle, France).
Data Analysis. All data are expressed as means ± S.E.M. Data were analyzed using GraphPad PRISM version 2.0 for Macintosh (GraphPad Software Inc., San Diego, CA) either by a one-site binding hyperbola (eotaxin1 binding and calcium flux assays) or using the four-parameter sigmoidal dose-response (variable slope) function (eosinophil shape change and chemotaxis assays). Where appropriate IC50 values were determined for each individual experiment and are shown as the geometric mean (with 95% confidence limits). Statistical analysis was performed using Instat software (version 2.03 for Macintosh; GraphPad) using either one-way analysis of variance with a post hoc Dunnett's test or a Student's t test. Differences between mean values were taken as significant when P < 0.05.
Isolation and Optimization of Lead scFv to Generate CAT-212 scFv and CAT-213 IgG4
Panning selections were carried out on human eotaxin1 using a phage library displaying human scFvs as described. These selections resulted in the isolation of several scFvs that were identified by their ability to bind specifically to eotaxin1 by phage ELISA (data not shown). All scFvs were purified and tested for their ability to neutralize human eotaxin1 in a chemotaxis assay. One scFv, 3G3, that was specific for eotaxin1 and fully blocked eotaxin1-induced chemotaxis, albeit with an IC50 of 800 nM (Fig. 2), was identified.
Heavy Chain Optimization
For the heavy chain optimization, three libraries were constructed, varying the CDR3 length (see Materials and Methods), and were of sizes 2 × 108, 3 × 109, and 7 × 108 individual recombinants. Selection from libraries 1 (VHCDR3 length retained as in 3G3) and 2 (VHCDR3 length increased by one amino acid) did not result in any higher potency scFvs than the parent. However, scFvs from library 3 (VHCDR3 reduced by one amino acid) were found to contain novel sequences that conferred significantly improved potencies. Examples of data obtained with scFvs derived from each strategy are shown in Table 1.
As shown in Table 1, the potency of the selected scFvs improved with increasing rounds of selection at lower concentrations of soluble eotaxin1. For example, the scFv, 4LD6, selected at round 4 had an IC50 of 160 nM in the chemotaxis assay, whereas the scFv, 7SE2, selected at round 7, had an IC50 of 6 nM. 7SE2, which differs from the parent scFv, 3G3, by only three amino acid changes and a single amino acid deletion in the VHCDR3, possesses a >100-fold increase in potency. There seems to be a strong selection pressure toward scFv antibodies containing particular amino acid residues at particular positions in the VHCDR3. An amino acid “consensus sequence” (Kabat positions 100, 100A, 101, and 102) (Kabat et al., 1991) emerged in the higher potency scFv variants in which positions 100 and 101 were consistently acidic, 100A was an uncharged residue, and 102 (the terminal residue of the VHCDR3) was a proline.
The three most potent scFvs generated, as determined in the chemotaxis assay, were 7SE2, 7SF2, and 9LE4 (Table 1). Comparison of the sequences of these three scFvs revealed that 7SF2 and 9LE4 have an additional PCR-induced point mutation from arginine (R) to glycine (G) in one of the vernier residues (Kabat position 94). This is the amino acid immediately before the VHCDR3 that can affect the angle at which the VHCDR3 loop is presented. To confirm that this change was advantageous, 7SF2, 9LE4, and 7SE2 variants were made with either an R or a G at position 94. In each case, a G conferred up to a 5-fold potency advantage compared with the corresponding R (data not shown).
Light Chain Optimization
Light chain shuffle selections were conducted in parallel with the heavy chain optimization. As described under Materials and Methods, a light chain library (6 × 109) was constructed comprising the heavy chain of 3G3 and a diverse panel of human light chains. Many scFvs with higher potency than 3G3 were selected. These scFvs were ranked in order of potency in the chemotaxis assay, and the most potent two scFvs contained different Vκ1DPκ5-derived light chains designated VL11 and VL23. These two light chains were then each paired with one of the top three heavy chains identified earlier: 7SE2, 7SF2, or 9LE4. Six VH-VL constructs were therefore assessed in total. The most potent scFv antibody as determined in the chemotaxis assay comprised the 7SE2 heavy chain, a G at VH position 94, and the VL23 light chain. This scFv was named CAT-212. The sequences of CAT-212, the parent scFv (3G3), and the closest VH and VL germline sequences are shown in Fig. 3. The scFv CAT-212 was converted to a whole antibody format, IgG4, and named CAT-213. CAT-213 retains specificity and potency equivalent to CAT-212 (see below).
Affinity of the Interaction of CAT-212 with Eotaxin1
Results of the binding assay to determine the affinity of the interaction of CAT-212 with eotaxin1 are shown in Fig. 4. The geometric mean (95% confidence interval) equilibrium dissociation constant (KD) of CAT-212 was 80.4 (27–239) pM.
Neutralization of Eotaxin1 by CAT-212 and CAT-213 in Vitro
Chemotaxis.Figure 2 shows inhibition of eotaxin1 (6 nM)-stimulated chemotaxis of L1.2-hCCR3 cells caused by 3G3, CAT-212, and CAT-213. The IC50 for 3G3 was 784 (481–1280) nM and for CAT-212 was 0.43 (0.26–0.71) nM, demonstrating that lead optimization produced a potency improvement of >3 orders of magnitude. The IC50 for CAT-213 was 0.48 (0.18–1.25) nM, which shows that the antibody retains equivalent potency on conversion from scFv to IgG format.
The cross-reactivity of CAT212 and CAT-213 with eotaxin1 from other species was also evaluated. The ability of CAT-212 and CAT-213 to inhibit chemotaxis induced by monkey, rat, and mouse eotaxin1 is compared in Table 2. CAT-212 and CAT-213 are selective for human eotaxin1, having a reduced but measurable effect on monkey eotaxin1 (13- and 45-fold less activity than for human eotaxin1, respectively) but little cross-reactivity with mouse or rat eotaxin1 (>1000 fold less activity than for human eotaxin1).
Calcium Flux. CAT-212 and CAT-213 both fully inhibited calcium flux induced by 10 nM human eotaxin1 in L1.2-hCCR3 cells. Figure 5 shows the concentration-response curves for CAT-212 and CAT-213. IC50 values were 2.49 (1.23–5.01) nM for CAT-212 and 2.86 (2.21–3.71) nM for CAT-213 (n = 3 for both).
Shape Change.Figure 6A shows the effect of eotaxin1 on eosinophil shape change. Eotaxin1 produced a bell-shaped dose response in three of the five donors; in the other two donors, the response was proportional to dose.
CAT-213 inhibited 3 nM eotaxin1-induced eosinophil shape change with an IC50 of 0.71 (0.21–2.45) nM (n = 5; Fig. 6B). In contrast, an irrelevant IgG4 control antibody did not inhibit eotaxin1-induced eosinophil shape change at any of the concentrations tested (n = 5). CAT-213 or the irrelevant IgG4 control antibody at concentrations up to 300 nM did not inhibit shape change of eosinophils induced by human eotaxin2 (data not shown).
Specificity of CAT-212 and CAT-213 in Vitro
CAT-212 and CAT-213 were specific for eotaxin1 as described by ELISAs conducted on a panel of human cytokines and chemokines (Fig. 7, A and B). This includes the functionally related eotaxin2 and eotaxin3 as well as the most structurally related family member, MCP-1. The high degree of sensitivity of the ELISA assay system, high antigen coating concentrations up to 10 μg ml–1, and the lack of any detectable cross-reactivity with the closest family members indicate that CAT-212 and 213 are highly specific for human eotaxin1. Moreover, CAT-213 at a high concentration (300 nM) was unable to inhibit the eosinophil shape change mediated by eotaxin2 or block monocyte calcium flux caused by MCP-1 (Fig. 7C). Combined, these findings indicate a high degree of specificity of CAT-213 for eotaxin1.
Neutralization of Eotaxin1 by CAT-212 and CAT-213 in Vivo
Mouse Air Pouch Eosinophilia. Administration of human eotaxin1 (1 μg) to the air pouch of ovalbumin-sensitized mice treated with IL-5 caused a significant dose-related increase in the recruitment of leukocytes to the pouch after 6 h. The majority of the cells trafficking to the pouch were eosinophils, although neutrophils and mononuclear cells were also recruited (Table 3).
CAT-213 (0.01–10 mg kg–1) administered i.v. 30 min before i.po. injection of human eotaxin1 (1 μg) caused a significant dose-dependent inhibition (maximal inhibition 91 ± 7%, n = 8) of eosinophil recruitment in IL-5-treated, ovalbumin-sensitized mice (Fig. 8). CAT-213 also significantly inhibited neutrophil and mononuclear cell influx into the air pouch, which resulted in a dose-related inhibition of total cell influx (>94% inhibition; data not shown). CAT-001, control IgG4 (10 mg kg–1 i.v.), did not significantly affect eosinophil (19 ± 9% inhibition, n = 8), neutrophil, or mononuclear cell influx.
CAT-213 (0.001–1 mg kg–1, n = 7–8) or CAT-212 (0.5–50 μg kg–1, n = 7–8) administered i.po. concurrently with human eotaxin1 (1 μg i.po.) caused a potent dose-related inhibition of eosinophilia (Fig. 8). All doses of CAT-213 and the two higher doses of CAT-212 were effective. Maximal inhibitions of eosinophil recruitment were 94 ± 2 and 94 ± 4% for CAT-213 and CAT-212, respectively. Neither the control IgG4 (1 mg kg–1 i.po.) nor the control scFv (50 μg kg–1 i.po.) had any effect on leukocyte influx (eosinophils; 14 ± 14 and –4 ± 15% inhibition, respectively).
In this article, we describe the isolation, optimization, and in vitro and in vivo characterization of the first human anti-eotaxin1 antibody, CAT-213. CAT-213 is a high-affinity antibody that exhibits no detectable cross-reactivity with other chemokines or cytokines. In functional assays, CAT-213 neutralizes the ability of eotaxin1 to induce responses in human CCR3-expressing cells; i.e., CAT-213 inhibited eotaxin1-induced increases in intracellular calcium levels and chemotaxis (IC50 5 and 0.48 nM, respectively). Furthermore, CAT-213 was shown to neutralize the ability of eotaxin1 to evoke shape change of eosinophils isolated from human blood (IC50 0.71 nM). Finally, CAT-213 was shown to cause a potent dose-related inhibition of eosinophilia in vivo.
It has previously been reported that mutagenesis of the CDR regions can lead to improvements in antibody potency (Schier et al., 1996). This process typically involves point mutations and/or randomization of residues within the CDRs with no alteration in length. The heavy chain CDR3 region has been established as the region of interaction between antibody and antigen with the greatest influence on binding affinity. Therefore, in this study, the VHCDR3 was chosen for randomization, but the effects of altering VHCDR3 length were also examined.
A novel finding in this study was that reducing the length of the VHCDR3 by one amino acid conferred a marked increase in antibody potency. This strategy alone effected an increase in potency of 2 orders of magnitude in a chemotaxis assay. When combined with a new a light chain and incorporating an R to G change at VH position 94 (Kabat), the optimized antibody increased in potency by a further 10-fold to give a total 1000-fold increase over the parent. This method of lead optimization may provide a new route for increasing antibody potency. Interestingly, these observations are supported by a study conducted by de Wildt et al. (1999), who reported that alterations in antibody sequence length can confer an advantage during natural antibody selection. These authors sequenced the heavy and light chain genes from >350 human B cells and found that 6.5% contained somatically introduced insertions or deletions that were clustered at the antigen binding site, suggesting that variations in the CDR length can confer enhanced antibody affinity.
The main potency gains in CAT-212 were obtained not only by decreasing the length but also by selecting four new terminal residues of the VHCDR3. The VHCDR3 changed from DTDYGDAFDI in 3G3 to DTDYGDIDP in CAT-212. It is interesting to note that there are four aspartic acids of a total of nine residues in the VHCDR3, conferring the potential antigen binding region with an overall negative charge. This may be important for the establishment of an electrostatic interaction with eotaxin1, which is lysine-rich and has a basic pI (Ponath et al., 1996).
In Vitro Effects of Eotaxin1 and Neutralization with CAT-212 and CAT-213. Human eotaxin1 evoked dose-dependent calcium flux and chemotaxis in L1.2 cells stably transfected with human CCR3. Inhibition of this response to eotaxin1 was achieved at 10 nM CAT-212 and 6 nM CAT-213 (EC80). The effects of eotaxin1 in this cell line are similar to those in previous studies reporting the effects of eotaxin1 on cells that possess endogenous CCR3. For example, human eotaxin1 has been shown to induce calcium influx in eosinophils obtained from peripheral blood (Tenscher et al., 1996) and to induce chemotaxis of a broad range of CCR3-expressing cells including eosinophils and basophils also derived from peripheral blood (Uguccioni et al., 1997) and mast cells derived from lung parenchyma (de Paulis et al., 2001).
CAT-213 neutralized the ability of recombinant eotaxin1 to both induce chemotaxis and to increase intracellular calcium levels of CCR3-expressing cells, as well as to evoke shape change of human eosinophils in vitro. Furthermore, CAT-213 has been shown to inhibit eosinophil chemotaxis stimulated by processed asthmatic sputum (Dent et al., 2004) as well as L1.2-hCCR3 cell chemotaxis induced by conditioned medium from IL-13-stimulated human lung fibroblast cultures (data not shown) when both stimuli contained endogenous human eotaxin1. The responses to eotaxin1 assessed in this study are mediated via CCR as nontransfected L1.2 cells do not respond to human eotaxin1 (present study, data not shown). The ability of eotaxin1 to also bind CCR2, CCR5, and CXCR3 has been reported (Martinelli et al., 2001; Ogilvie et al., 2001). However, it is difficult to predict whether CAT-213 will neutralize effects driven through these receptors since the epitope on eotaxin required for their activation is currently unknown.
In Vivo Effects of Eotaxin1 and Neutralization with CAT-212 and CAT-213. Rodent, human, and murine eotaxin1 have been shown to be equieffective in terms of inducing eosinophil chemotaxis in vivo (Kudlacz et al., 1999). In this study, human eotaxin1 caused a marked increase in leukocyte recruitment after 6 h when administered to the air pouch of ovalbumin-sensitized mice pretreated with IL-5. In pilot experiments, it was shown that IL-5 and ovalbumin sensitization were prerequisites for significant recruitment of leukocytes (data not shown). The dependence of this response of eotaxin1 on prior administration of IL-5 has been reported (Das et al., 1997). Moreover, administration of IL-5 has been shown to potentiate responses to eotaxin1 (Mould et al., 1997).
The present study has shown the importance of eotaxin1 in mediating eosinophil influx to a subcutaneous mouse air pouch. This is in agreement with previous work in rodents (Teixeira et al., 1997; Kudlacz et al., 1999), monkeys (Ponath et al., 1996), and man (Menzies-Gow et al., 2002). Interestingly, in mice, recruitment of neutrophils and mononuclear cells was concomitant with eosinophil influx. It may be tempting to attribute these other responses to indirect/secondary effects of eotaxin1 recruitment of eosinophils. However, direct effects of eotaxin1 on these cell types cannot be ruled out. There is evidence for CCR3 on human neutrophils (Bonecchi et al., 1999; Menzies-Gow et al., 2002); however, this is controversial (Hochstetter et al., 2000), and no functional effect of eotaxin1 agonism has been established (Menzies-Gow et al., 2002).
In mice, CAT-213 administered either systemically or locally or CAT-212 administered locally caused a dose-related inhibition of eosinophil influx induced by human eotaxin1. Local CAT-213 and CAT-212 were equipotent in their ability to prevent eosinophilia in vivo (ED70 values for CAT-213 and CAT-212 are ∼80 and 100 pmol kg–1, respectively), which equates to their equipotent effects in vitro in the chemotaxis assay. The null control IgG4 and scFv had no effect on the response to eotaxin1, eliminating the possibility of a nonspecific antibody response. Taken together, the in vitro and in vivo information supports the use of CAT-213 at much lower doses in humans (in the range of 1 to 10 mg kg–1).
The human pharmacology of CAT-213 is currently being investigated. In a phase I single-dose clinical study, CAT-213 has been shown to be safe and well tolerated. In this study, CAT-213 was also shown to be present in serum in an active form for >60 days after intravenous injection. In a more recent clinical study, submucosal eosinophil infiltration induced by allergen challenge in subjects with rhinitis (Pereira et al., 2003; Salib et al., 2003) was reduced by CAT-213.
It is likely that antibodies will be an important class of drugs in allergic diseases and in particular in the treatment of severe asthma. Omalizumab (anti-IgE), the first antibody to be approved for the treatment of asthma, reduces exacerbations and corticosteroid use in patients with allergic asthma (Schulman, 2001). Furthermore, mepolizumab (anti-IL-5) has recently been studied in a bronchial challenge model (Leckie et al., 2000). Somewhat surprisingly, no changes in response to allergen challenge or bronchial hyper-activity were observed despite marked reductions in sputum and blood eosinophilia. However, it should be noted that tissue eosinophilia has been shown to be less markedly reduced (Flood-Page et al., 2003). This suggests that inhibition of IL-5 alone is insufficient to eradicate tissue eosinophil levels. However, an anti-eotaxin1 antibody may have advantages over anti-IL-5 therapy in that eotaxin1 may have significance in asthma beyond its action on eosinophils given that CCR3 is also expressed on other cells known to play an important role in allergy, and eotaxin1 can bind other receptors (see Introduction).
In conclusion, we have shown that the scFv, CAT-212, and the IgG4 antibody, CAT-213, are potent human anti-eotaxin1 antibodies that neutralize the effects of human eotaxin1 both in vitro and in vivo. CAT-213 may therefore be of therapeutic value in diseases in which eotaxin1 plays a major role.
We thank Chris Traher, Sue Pearce, Ann Traher, Tamara Baker, Elaine Derbyshire, Antonia Banyard, Thor Holtet, Tauqir Ahmad, Matt Sleeman, and Elise Martin (Charles River Laboratories) for expert assistance and advice.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: CCL, CC chemokine ligand; CCR, CC chemokine receptor; CXCR, CXC chemokine receptor; scFv, single-chain variable fragment(s); BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; PBS, phosphate-buffered saline; VH, variable heavy chain; VHCDR3, variable heavy chain complementarity determining region 3; PCR, polymerase chain reaction; VL, variable light chain; MCP, monocyte chemoattractant protein; IL, interleukin; RANTES, regulated on activation, normal T-cell expressed and secreted; TGF, transforming growth factor; L1.2-hCCR, L1.2 cells stably transfected with human CCR3; FCS, fetal calf serum; FLIPR, fluorometric imaging plate reader; PBMC, peripheral blood mononuclear cell; i.po., intrapouch.
↵1 These authors contributed equally to this work.
- Received July 26, 2006.
- Accepted September 12, 2006.
- The American Society for Pharmacology and Experimental Therapeutics