A novel highly potent therapeutic antibody neutralizes multiple human chemokines and mimics viral immune modulation

Abstract

Chemokines play a key role in leukocyte recruitment during inflammation and are implicated in the pathogenesis of a number of autoimmune diseases. As such, inhibiting chemokine signaling has been of keen interest for the development of therapeutic agents. This endeavor, however, has been hampered due to complexities in the chemokine system. Many chemokines have been shown to signal through multiple receptors and, conversely, most chemokine receptors bind to more than one chemokine. One approach to overcoming this complexity is to develop a single therapeutic agent that binds and inactivates multiple chemokines, similar to an immune evasion strategy utilized by a number of viruses. Here, we describe the development and characterization of a novel therapeutic antibody that targets a subset of human CC chemokines, specifically CCL3, CCL4, and CCL5, involved in chronic inflammatory diseases. Using a sequential immunization approach, followed by humanization and phage display affinity maturation, a therapeutic antibody was developed that displays high binding affinity towards the three targeted chemokines. In vitro, this antibody potently inhibits chemotaxis and chemokine-mediated signaling through CCR1 and CCR5, primary chemokine receptors for the targeted chemokines. Furthermore, we have demonstrated in vivo efficacy of the antibody in a SCID-hu mouse model of skin leukocyte migration, thus confirming its potential as a novel therapeutic chemokine antagonist. We anticipate that this antibody will have broad therapeutic utility in the treatment of a number of autoimmune diseases due to its ability to simultaneously neutralize multiple chemokines implicated in disease pathogenesis.

Figure 1. Binding and in vitro activity of murine 18V4F hybridoma antibody.

(A) ELISA binding of original 18V4F hybridoma antibody to a panel of chemokines (determined in triplicate, shown as mean +/− standard deviation). Directly coated chemokines were used here for direct comparisons, however in other experiments CCL3 showed a significantly enhanced signal when biotinylated and coated on streptavidin plates. (B) Chemotaxis inhibition by 18V4F hybridoma antibody of CCR5-transfected Ba/F3 cells to 5 ng/mL of CCL3, CCL4, and CCL5. Data are representative of at least three similar experiments. All chemotaxis data are represented as a percent of maximum migration in the absence of inhibitors and is fit using a standard four parameter dose-response model (GraphPad).

Figure 2. Diagram of phage display selection strategy.

Individual CDR libraries were sequentially panned against CCL3, CCL4 and CCL5. In step 1, phage libraries were combined with biotinylated CCL3 and bound to streptavidin beads. Bound phage were eluted, amplified, and subjected to panning against biotinylated CCL4 and CCL5 in steps 2 and 3, respectively. This process was repeated 4–5 times with increasing stringency to yield sequences with improved affinities.

Figure 3. Chemotaxis inhibition by affinity matured 18V4F variants.

Chemotaxis inhibition by humanized 18V4F Fab, d5 variant, d7 variant, d5d7, and a negative control Fab of CCR5-transfected Ba/F3 cells to 5 ng/mL of (A) CCL3, (B) CCL4, and (C) CCL5. Data are representative of at least two similar experiments. A loss in potency of humanized 18V4F Fab was observed compared with the 18V4F hybridoma shown in Figure 1b and is likely a result of both the humanization process and loss in avidity caused by switching from full IgG to Fab fragment.

Figure 4. Comparison of vCCI and d5d7 binding epitopes.

Competitive binding ELISA examining molecules that can disrupt the d5d7-CCL3 binding interaction using d5d7 as a homologous competitor and vCCI-Fc, commercial anti-CCL3 antibody, and control IgG as heterologous competitors. Data are representative of at least two similar experiments. Competition experiments were also completed to analyze the d5d7-CCL4 and d5d7-CCL5 binding interactions and similar binding competition was observed between d5d7 and vCCI-Fc (data not shown).

Figure 5. Inhibition of chemotaxis induced with mixtures of chemokines by MAb d5d7.

Inhibition of chemotaxis of (A) CCR5 transfectants to a pool of recombinant CCL3, CCL4, and CCL5 and (B) CCR1 transfectants to a pool of CCL3 and CCL5, by MAb d5d7 antibody, vCCI-Fc, individual commercial anti-chemokine antibodies (anti-CCL3, anti-CCL4, and anti-CCL5), and IgG controls. Chosen chemokine concentrations were those that produced 50% maximal chemotaxis when tested individually (a pool of 3 ng/mL CCL3, 10 ng/mL CCL4, and 3 ng/mL CCL5 was used in CCR5 experiments and a mixture of 20 ng/mL CCL3 and 5 ng/mL CCL5 was used in CCR1 experiments).

Figure 6. Inhibition of chemotaxis induced with native chemokines by MAb d5d7.

Chemotaxis inhibition by MAb d5d7 antibody and vCCI-Fc of (A) CCR5 transfectants and (B) CCR1 transfectants to a supernatant containing inflammatory chemokines from LPS-stimulated PBMC. The dilution of supernatant used in the assay was that which produced 50% maximal chemotaxis (1∶80 for CCR5 cells and 1∶20 for CCR1 cells). Data are representative of at least two similar experiments.

Figure 7. Inhibition of chemokine signaling on native leukocytes by MAb d5d7.

(A) Inhibition by MAb d5d7 antibody, vCCI-Fc, and IgG controls of chemokine-induced phosphorylation of CCR5^Ser349 in CD8+ T cells. Phosphorylation was induced with a pool of 50 ng/mL each CCL3, CCL4, and CCL5 (each of which produced ∼80% maximal response when tested individually). Data are expressed as percent of the maximum number of CCR5^pSer349-positive CD8+ T cells after chemokine induction without inhibitor present and are representative of results using PBMC from six different donors. (B) Inhibition of CD11b up-regulation on monocytes induced with a pool of CCL3 and CCL5. Chosen chemokine concentrations were those that produced ∼80% maximal response when tested individually (16 ng/mL CCL3 and 80 ng/mL CCL5). Data are expressed as percent of the maximal increase in mean fluorescence intensity determined after chemokine induction without inhibitor present and are representative of results using blood samples from eight different donors.

Figure 8. SCID-hu mouse model of leukocyte migration.

(A) NSG (NOD/SCID/IL2r-γnull) mice were injected i.v. with human PBMC and allowed to engraft for 10 d. MAb d5d7 was administered i.v. just before chemokines were injected s.c. in Matrigel. After 7 d the skin sites were harvested and single cell suspensions were generated. Human leukocytes were tagged with specific antibodies and analyzed by flow cytometry. (B) Inhibition by MAb d5d7 of skin leukocyte migration into chemokine-embedded Matrigel plugs in NSG mice engrafted with human PBMC. The negative control group consisted of animals treated with s.c. injection of Matrigel + PBS and i.v. administration of control IgG. All other groups had s.c. injections of Matrigel containing CCL3, CCL4, and CCL5 (400 ng each) with i.v. administration of PBS, control IgG, or MAb d5d7 antibody. Data were analyzed using a student t test.