Isolation of human anti-idiotypic antibodies by phage display for clinical immune response assays.

The clinical development of therapeutic proteins requires assays that measure the pharmacokinetic (PK) profile of, and the potential immune response (IR) to, the protein agent. Each assay requires reagents that are highly specific for the therapeutic protein. For therapeutic monoclonal antibodies, anti-CDR-specific, or anti-idiotypic (anti-id), antibodies are an ideal class of reagents suitable for these assays because of their high specificity and affinity to the drug antibody. We generated anti-ids to two human antibodies by antibody phage display using the MorphoSys HuCAL GOLD Fab library. To selectively target the CDR regions, serum and a framework-matched mAb were included as competitors during the phage selection process. Panels of CDR-specific Fabs, with low to sub-nM affinities, were isolated against both targets. The CDR specificity of these Fabs was shown by their lack of binding to a framework-matched control mAb and by competition of this binding with the soluble antigens of the respective therapeutic mAb targets. The candidate anti-id Fabs were able to detect both immobilized and soluble target Ab without being affected by serum, a requirement for both PK assay and the IR bridging assay format. Combinations of the Fabs for PK detection assays were identified by pairwise binding studies, although the pair for one target mAb lacks the desired sensitivity for PK assays. To evaluate their potential as anti-drug antibodies (ADAs), the best Fabs for one of the targets were converted and produced as the required bivalent human mAbs. In comparison to rodent mAbs and primate polyclonal serum, the phage display derived human mAbs were equally effective as reference standards. Our results demonstrate that competition-based phage selection can be an effective method for the isolation of anti-idiotypic antibodies for PK and IR assay development, and in this latter case, overcome limitations of current methods using rodent derived anti-ids.

Identification of a selective nonpeptide antagonist of the anaphylatoxin C3a receptor that demonstrates antiinflammatory activity in animal models.

The anaphylatoxin C3a is a potent chemotactic peptide and inflammatory mediator released during complement activation which binds to and activates a G-protein-coupled receptor. Molecular cloning of the C3aR has facilitated studies to identify nonpeptide antagonists of the C3aR. A chemical lead that selectively inhibited the C3aR in a high throughput screen was identified and chemically optimized. The resulting antagonist, N(2)-[(2,2-diphenylethoxy)acetyl]-L-arginine (SB 290157), functioned as a competitive antagonist of (125)I-C3a radioligand binding to rat basophilic leukemia (RBL)-2H3 cells expressing the human C3aR (RBL-C3aR), with an IC(50) of 200 nM. SB 290157 was a functional antagonist, blocking C3a-induced C3aR internalization in a concentration-dependent manner and C3a-induced Ca(2+) mobilization in RBL-C3aR cells and human neutrophils with IC(50)s of 27.7 and 28 nM, respectively. SB 290157 was selective for the C3aR in that it did not antagonize the C5aR or six other chemotactic G protein-coupled receptors. Functional antagonism was not solely limited to the human C3aR; SB 290157 also inhibited C3a-induced Ca(2+) mobilization of RBL-2H3 cells expressing the mouse and guinea pig C3aRS: It potently inhibited C3a-mediated ATP release from guinea pig platelets and inhibited C3a-induced potentiation of the contractile response to field stimulation of perfused rat caudal artery. Furthermore, in animal models, SB 290157, inhibited neutrophil recruitment in a guinea pig LPS-induced airway neutrophilia model and decreased paw edema in a rat adjuvant-induced arthritis model. This selective antagonist may be useful to define the physiological and pathophysiological roles of the C3aR.

Chimeric receptors of the human C3a receptor and C5a receptor (CD88).

Chimeras were generated between the human anaphylatoxin C3a and C5a receptors (C3aR and C5aR, respectively) to define the structural requirements for ligand binding and discrimination. Chimeric receptors were generated by systematically exchanging between the two receptors four receptor modules (the N terminus, transmembrane regions 1 to 4, the second extracellular loop, and transmembrane region 5 to the C terminus). The mutants were transiently expressed in HEK-293 cells (with or without Galpha-16) and analyzed for cell surface expression, binding of C3a and C5a, and functional responsiveness (calcium mobilization) toward C3a, C5a, and a C3a as well as a C5a analogue peptide. The data indicate that in both anaphylatoxin receptors the transmembrane regions and the second extracellular loop act as a functional unit that is disrupted by any reciprocal exchange. N-terminal substitution confirmed the two-binding site model for the human C5aR, in which the receptor N terminus is required for high affinity binding of the native ligand but not a C5a analogue peptide. In contrast, the human C3a receptor did not require the original N terminus for high affinity binding of and activation by C3a, a result that was confirmed by N-terminal deletion mutants. This indicates a completely different binding mode of the anaphylatoxins to their corresponding receptors. The C5a analogue peptide, but not C5a, was an agonist of the C3aR. Replacement of the C3aR N terminus by the C5aR sequence, however, lead to the generation of a true hybrid C3a/C5a receptor, which bound and functionally responded to both ligands, C3a and C5a.

Human anaphylatoxin C4a is a potent agonist of the guinea pig but not the human C3a receptor.

The interaction of human anaphylatoxin C4a with the guinea pig (gp) and human (hu) C3a receptors (C3aR) was analyzed using human rC4a, which exhibited C4a-specific activity on guinea pig platelets. A gpC3aR of 475 residues with a large second extracellular loop and a peptide sequence approximately 60% identical to the huC3aR was isolated from a genomic DNA library and found to be expressed in guinea pig heart, lung, and spleen. HEK-293 cells cotransfected with this clone, and a cDNA encoding G alpha-16 specifically bound (Kd = 1.6+/-0.7 nM) and responded functionally to C3a with an intracellular calcium mobilization (ED50 = 0.18+/-0.02 nM). Human rC4a weakly bound to both the hu- and gpC3aR (IC50 > 1 microM). However, only HEK-293 cells expressing the gpC3aR responded functionally to rC4a (ED50 = 8.7+/-0.52 nM), while cells expressing the huC3aR did not (c < or = 1 microM). Thus, through an interaction with the C3aR, huC4a may elicit anaphylatoxic effects in guinea pigs but not in man.

The human C3a receptor is expressed on neutrophils and monocytes, but not on B or T lymphocytes.

The pathophysiological relevance of the complement split product C3a as a proinflammatory mediator is still ill defined. The expression pattern of the human C3a receptor (C3aR) can provide important clues for the role of this anaphylatoxin in inflammation. There is strong evidence for C3aR expression on basophils, and eosinophils, but additionally, only on tumor cell lines of leukemic or hepatic origin. It is unclear whether neutrophils also express the C3aR, but need a costimulus provided by eosinophils for certain biological responses, or whether neutrophils lack the C3aR and respond to C3a via a secondary stimulus generated by eosinophils, i.e., by an indirect mode. In the present study, polyclonal antiserum raised against the second extracellular loop of the C3aR was used to characterize C3aR expression on peripheral blood leukocytes. For high degree purification of neutrophils, a negative selection method was established that decreased the contamination with CD9(bright+) eosinophils down to <0.2%. Flow cytometric analyses, functional assays, and binding assays on highly purified neutrophils confirmed C3aR expression and coupling. Monocytes were identified as an additional C3aR-positive cell population of the peripheral blood. The expression of the C3aR on eosinophils could be confirmed. In contrast, the receptor could not be detected on unchallenged B or T lymphocytes (or lymphocyte-derived Raji cells).

The mouse anaphylatoxin C3a receptor: molecular cloning, genomic organization, and functional expression.

The anaphylatoxin C3a receptor (C3aR) is unique among the family of G protein-coupled receptors in possessing an unusually large predicted second extracellular loop. To isolate the mouse C3aR, a probe derived from this extracellular loop was used to screen a mouse brain cDNA library. A 3.3-kb cDNA encoding an open reading frame of 477 amino acids was identified. The predicted amino acid contained four predicted N-linked glycosylation sites and was 65% identical to the 482 amino acids comprising the coding region of the human C3aR. Northern blot analysis revealed that this gene was expressed in a variety of mouse tissue and was especially abundant in heart and lung tissues. The mouse C3aR cDNA was used as a probe to isolate a mouse C3aR genomic clone. The nucleotide sequence of the mouse C3aR genomic clone was identical to the cDNA throughout the coding region, indicating that the receptor is encoded on a single exon. The C3aR cDNA was subcloned into a mammalian expression vector and transiently expressed in HEK-293 cells. Binding of radiolabeled C3a to the transfected cells was competed in a dose-dependent manner by increasing concentrations of unlabeled C3a, with a 50% inhibiting concentration of 10 nM. Similar to the human C3aR, RBL-2H3 rat basophilic cells stably expressing this receptor responded in a dose-dependent manner to C3a, a synthetic C3a peptide agonist, but not C4a or C5a, with a vigorous calcium mobilization.

Evidence that the receptor for C4a is distinct from the C3a receptor.

The cDNAs encoding the human (hC3aR) and mouse C3a receptors (mC3aR) were functionally expressed in RBL-2H3 cells. A calcium mobilization assay was utilized to assess the biologic activity of human anaphylatoxins, and C3a synthetic peptide agonists on hC3aR and mC3aR cells and this activity was compared to the activity of the anaphylatoxins on human neutrophils. Both hC3aR and mC3aR cells responded in a concentration-dependent manner with a robust calcium mobilization response to C3a with 50% effective concentrations (EC50s) of 0.24 nM and 1.3 nM, respectively. The response obtained with hC3aR cells was similar to the response elicited by C3a on human neutrophils (EC50 0.77 nM). The potency of a C3a analogue synthetic peptide (WWGKKYRASKLGLAR), derived from the fifteen carboxy-terminal residues (63-77) of C3a, relative to C3a, in stimulating calcium mobilization differed on cells expressing the human vs. mouse receptors. While the peptide was approximately 10 fold less active than C3a in stimulating calcium mobilization on cells expressing the hC3aR (EC50 2.0 nM), the peptide was essentially equipotent to the native ligand when tested on cells expressing the mC3aR. Data obtained with C4a, purified from activated serum, were difficult to interpret due to possible trace contamination of the C4a with C5a. Subsequently, an alternative C4a isolation scheme was utilized, via cleavage in vitro of purified C4. Concentrations of this latter C4a preparation, of up to 3.3 microM, had no effect on calcium mobilization in human neutrophils or in cells stably expressing the cloned C3a receptors, an indication that C4a does not interact with the C3a receptor.

Isolation of a neutralizing human RSV antibody from a dominant, non-neutralizing immune repertoire by epitope-blocked panning.

We isolated a large panel of human Abs directed against the respiratory syncytial virus (RSV) Ag from combinatorial phage display libraries. Following initial differentiation of the Fabs by BstNI restriction patterns, DNA sequence analysis revealed 10 different classes of VH paired with more than 35 different VL genes. All the Fabs bound with high affinity to the F Ag. However, most Fabs competed with the binding of a representative member of this group, suggesting that the Fabs recognized a common epitope on the F Ag, and none of them neutralized virus in vitro. To suppress repetitive isolation of these non-neutralizing Abs, a representative Fab was included during panning to block this common epitope on the F Ag. By this "epitope-blocked panning" approach, two novel Fabs, encoded by unique VH and VL genes, were isolated from a previously screened library. Competition binding analysis confirmed that the Fabs recognized epitopes distinct from that of the previously isolated Fabs. One of these Fabs, 516, neutralized RSV in cell culture. These activities of Fab-516 were retained upon its genetic conversion to a mAb (IgG1) and expression in mammalian cells. Our results suggest that the RSV F glycoprotein presents a dominant, non-neutralizing epitope to the human immune system, which may serve in evasion of host defenses. However, less prevalent, fusion-inhibiting Abs were revealed by blockade of this epitope during the panning process.

Conversion of murine Fabs isolated from a combinatorial phage display library to full length immunoglobulins.

The use of combinatorial Ig libraries displayed on the surface of bacteriophage has advantages over traditional hybridoma techniques for the generation of mAbs but in many instances full length Igs may be more desirable than Fab fragments. Two murine Fabs reactive with the human complement component C5a, recovered from a combinatorial library, were converted to full length IgG2a mAbs. The VH and VL domains of these antibodies were removed from the bacterial expression vector used for the combinatorial library construction, and subcloned into individual mammalian expression vectors containing the corresponding Ig heavy and light chain constant regions. The subcloning relied on 5' restriction endonuclease sites encoded by the oligonucleotide primers originally used to amplify the Ig cDNAs and 3' sites conserved in CH1 and C kappa. These vectors were co-transfected into COS cells yielding full length IgG2a versions of the anti-C5a antibodies. The mAbs, purified from the culture supernatant, retained the full activity of the Fabs, binding specifically to and neutralizing human recombinant C5a. Refined versions of the mammalian expression vectors have been constructed for single step conversion of murine recombinant Fabs, recovered from combinatorial libraries, to IgG2a mAbs.

Neutralizing murine monoclonal antibodies to human IL-5 isolated from hybridomas and a filamentous phage Fab display library.

Conventional hybridomas and combinatorial Ab libraries were used to develop neutralizing murine mAbs to human IL-5. Mice were immunized with rIL-5. Spleens from two mice were used to generate hybridomas. Spleens from an additional three mice were used to construct a combinatorial library. In both instances, Abs were identified and selected by ELISA using 96-well plates coated with rIL-5. These Abs were tested for the ability to block binding of iodinated rIL-5 to the alpha-chain of the human IL-5 receptor (IL-5R alpha) and to inhibit proliferation of IL-5-dependent cells. By hybridoma technology, 16 mAbs were obtained, 11 of which blocked binding to IL-5R alpha, including three that inhibited proliferation. Quantitative binding assays and sequence analysis revealed that these latter three mAbs were closely related. Combinatorial cloning and selection by phage display was used to isolate 24 bacterial colonies secreting Fabs that bound to 125I-rIL-5 and to rIL-5-coated plates. Sequencing of 10 of the Fabs indicated that four unique Abs were obtained, comprising one predominant VH paired with one of two different VL. The sequence of the Fabs was distinct from the sequences of the neutralizing mAbs. In contrast to the mAbs, none of the Fabs blocked binding of 125I-IL-5 to IL-5R alpha or neutralized the biologic activity of IL-5. The inability to identify neutralizing Fabs was shown not to result from their monovalency, because a Fab derived from one of the neutralizing mAbs, by cloning and expression of its Fd and kappa light chains, retained neutralizing activity. By chain shuffling, pairing of the Fd fragment of the heavy chain of one of the neutralizing mAbs (2B6), with the light chain library derived from the IL-5-immunized mice, neutralizing Fabs were obtained. These Fabs contained light chain sequences closely related to the original light chain of 2B6. Hence, chain shuffling allowed detection of a light chain sequence that was not evident upon two-chain combinatorial selection. The results reveal differences in the Abs obtained from a combinatorial library vs hybridomas and demonstrate how these approaches can be used in concert to select mAbs with neutralizing activity.

Isolation of neutralizing anti-C5a monoclonal antibodies from a filamentous phage monovalent Fab display library.

A panel of mAbs against the activated complement component C5a was obtained from a filamentous phage M13-Fab display library generated from mice immunized with human rC5a. Fabs isolated from the library after iterative selection vs rC5a bound to both rC5a and purified C5. To isolate Fabs specific for neoepitopes expressed on C5a but not on the native complement component C5 the library was rescreened in a competitive manner. The phage Fab library was first incubated with immobilized C5 to deplete C5 reactive Fabs. The C5 nonadherent phage were then incubated with immobilized rC5a in the presence of soluble C5. Bound phage were eluted and subjected to two additional cycles of subtraction with immobilized C5 and selection with immobilized rC5a in the presence of soluble C5. After three cycles of this competitive biopanning four Fabs reactive with rC5a were isolated. Two bound preferentially with and neutralized C5a. Competitive biopanning of phage display libraries may increase the probability of identification of Abs of the desired specificity.