Use of uteroglobin for the engineering of polyvalent, polyspecific fusion proteins

Abstract

We report a novel strategy to engineer and express stable and soluble human recombinant polyvalent/polyspecific fusion proteins. The procedure is based on the use of a central skeleton of uteroglobin, a small and very soluble covalently linked homodimeric protein that is very resistant to proteolytic enzymes and to pH variations. Using a human recombinant antibody (scFv) specific for the angiogenesis marker domain B of fibronectin, interleukin 2, and an scFv able to neutralize tumor necrosis factor-alpha, we expressed various biologically active uteroglobin fusion proteins. The results demonstrate the possibility to generate monospecific divalent and tetravalent antibodies, immunocytokines, and dual specificity tetravalent antibodies. Furthermore, compared with similar fusion proteins in which uteroglobin was not used, the use of uteroglobin improved properties of solubility and stability. Indeed, in the reported cases it was possible to vacuum dry and reconstitute the proteins without any aggregation or loss in protein and biological activity.

FIGURE 1.

Central part of the figure depicts the ribbon structure of the oxidized homodimer of UG (adapted with permission from Ref. 4). A–E show the schemes of the various fusion proteins produced using UG as a central core. L19 is an scFv specific for the angiogenesis-associated FN isoform, and D2E7 is an scFv able to neutralize TNF-α.

FIGURE 2.

a, scheme of the cDNA constructs of L19-UG (left) and L19-UG-L19 (right). Sp, signal peptide sequence. b, size exclusion chromatography profiles (Superdex 200) and SDS-PAGE analysis of the purified fusion proteins L19-UG (left) and L19-UG-L19 (right). Nonreducing (NR) and reducing (R) conditions, respectively. ST, molecular mass standard. c, size exclusion chromatography profiles of L19-UG and L19-UG-L19 reconstituted in distilled water after vacuum drying. d, comparative biodistribution experiments in F9 teratocarcinoma-bearing mice of three radioiodinated L19 formats: L19-UG, L19-UG-L19, and L19-SIP. The % ID/g in the tumor at the indicated times after intravenous injection is shown on the left (mean ± S.D.). To the right the tumor to blood ratio of the % ID/g at different times after injection of the radioiodinated proteins are plotted. VL, variable light region of immunoglobulin; CMV, cytomegalovirus; mAu, milli-absorbance units.

FIGURE 3.

A, scheme of the cDNA construct of L19-UG-IL2. Sp, signal peptide sequence. B, SDS-PAGE analysis (nonreducing (NR) and reducing conditions, respectively); C, size exclusion chromatography profile (Superdex 200) of purified L19-UG-IL2. D, proliferation assay on CTLL cells using [³H]thymidine. Equimolar amounts of L19-IL2 and L19-UG-IL2 induced identical thymidine incorporation into cells, indicating that the two proteins share identical IL2 activity. Abs, antibodies. E, tumor growth inhibition of equimolar amounts of L19-IL2 and L19-UG-IL2 or only saline. Each group consisted of four mice. Therapy started 7 days after tumor implantation. Arrows indicate the days of treatment. Tumor size was determined as reported under “Experimental Procedures.” S.E. are indicated. VL, variable light region of immunoglobulin; CMV, cytomegalovirus; mAu, milli-absorbance units.

FIGURE 4.

A, scheme of the cDNA construct of L19-UG-D2E7. Sp, signal peptide sequence. CMV, cytomegalovirus. B, SDS-PAGE analysis under nonreducing (NR) and reducing (R) conditions; ST, molecular mass standard. C, size exclusion chromatography profile (Superdex 200) of purified L19-UG-D2E7 (left panel); size exclusion chromatography profile (Superdex 200) of purified L19-UG-D2E7 after vacuum drying (right panel). The binding ability to the antibody-specific antigens was studied in ELISA using different concentrations of D. L19-UG-D2E7 on ED-B of FN (using L19-UG as control); E, against TNF-α (using divalent D2E7-UG as control). F, inhibitory activity of hTNF-α cytotoxicity of different concentrations of L19-mUG-D2E7 and D2E7-mUG on LM mouse fibroblasts treated with 1 pm hTNF-α. The mean ± S.D. are reported. Abs, antibodies; mAu, milli-absorbance units.

FIGURE 5.

A, ELISA using 0.8 nm D2E7-UG-L19 on the ED-B of FN (white bars) or TNF-α (dark gray bars) as antigen (Ag). The reaction was performed with or without 100 μm ED-B. B, schematic drawing; C, results of ELISA performed using TNF-α as antigen and L19-UG-D2E7 as primary antibody; after removing the excess antibody the FN fragment 7B89 was added and was detected using an antibody specific for the FN repeat 9. D, L19-mUG-D2E7 bound to the ED-B neutralizes TNF-α cytotoxicity (in situ neutralization). The cytotoxic activity of 1 pm human TNF-α was evaluated on LM cells using plates pre-coated with 7B89 and preincubated with different concentrations (0.01–500 pm) of L19-mUG-D2E7 (□) or D2E7-mUG (●). After removing unbound antibodies, hTNF-α was inhibited only by L19-mUG-D2E7, because it bound to the ED-B of FN with which the wells were pre-coated. Abs, antibodies.