A highly stable polyethylene glycol-conjugated human single-chain antibody neutralizing granulocyte-macrophage colony stimulating factor at low nanomolar concentration.

GM-CSF (granulocyte-macrophage colony stimulating factor) plays a central role in inflammatory processes. Treatment with antibodies neutralizing murine GM-CSF showed significant therapeutic effects in mouse models of inflammatory diseases. We constructed by phage display technology a human scFv, which could potently neutralize human GM-CSF. At first, a human V(L) repertoire was combined with the V(H) domain of a parental GM-CSF-neutralizing rat antibody. One dominant rat/human scFv clone was selected, neutralizing human GM-CSF with an IC50 of 7.3 nM. The human V(L) of this clone was then combined with a human V(H) repertoire. The latter preserved the CDR 3 of the parental rat V(H) domain to retain binding specificity. Several human scFvs were selected, which neutralized human GM-CSF at low nanomolar concentrations (IC50 > or = 2.6 nM). To increase serum half-life, a branched 40 kDa PEG-polymer was coupled to the most potent GM-CSF-neutralizing scFv (3077) via an additional C-terminal cysteine. PEG conjugation had a negligible effect on the in vitro neutralizing potential of the scFv, although it caused a significant drop in binding affinity owing to a reduced on-rate. It also significantly increased the stability of the scFv at elevated temperatures. In mouse experiments, the PEGylated scFv 3077 showed a significantly prolonged elimination half-life of 59 h as compared with 2 h for the unconjugated scFv version. PEGylated scFv 3077 is a potential candidate for development of a novel antibody therapy to treat pro-inflammatory human diseases.

Therapeutic window of MuS110, a single-chain antibody construct bispecific for murine EpCAM and murine CD3.

EpCAM (CD326) is one of the most frequently and highly expressed tumor-associated antigens known and recently has also been found on cancer stem cells derived from human breast, colon, prostate, and pancreas tumors. However, like many other tumor-associated antigens used for antibody-based immunotherapeutic approaches, EpCAM is expressed on normal tissues including epithelia of pancreas, colon, lung, bile ducts, and breast. To assess the therapeutic window of an EpCAM/CD3-bispecific single-chain antibody construct of the bispecific T-cell engager (BiTE) class, we constructed murine surrogate of MT110 (muS110) from single-chain antibodies specific for murine EpCAM and CD3 antigens. Immunhistochemical analysis showed that, with minor differences, the expression of EpCAM protein on a large variety of tissues from man and mouse was similar with respect to distribution and level. MuS110 exhibited significant antitumor activity at as low as 5 microg/kg in both syngeneic 4T1 orthotopic breast cancer and CT-26 lung cancer mouse models. Dosing of muS110 for several weeks up to 400 microg/kg by intraanimal dose escalation was still tolerated, indicating existence of a significant therapeutic window for an EpCAM-specific BiTE antibody in mice. MuS110 was found to have similar in vitro characteristics and in vivo antitumor activity as MT110, a human EpCAM/human CD3-bispecific BiTE antibody that currently is in formal preclinical development.

Strictly target cell-dependent activation of T cells by bispecific single-chain antibody constructs of the BiTE class.

Bispecific antibodies have been extensively studied in vitro and in vivo for their use in redirected tumor cell lysis. A particular challenge of bispecific antibody constructs that recognize the invariant CD3 signaling complex is a controlled polyclonal activation of T cells that, ideally, is exquisitely dependent on the presence of target cells. Otherwise, overt production of inflammatory cytokines and secondary reactions may occur as side effects, as can be observed with constitutively T-cell activating monoclonal antibodies to CD3 or CD28, and with bispecific antibodies bearing Fc gamma portions. Here we analyzed 2 distinct bispecific single-chain antibody constructs of the BiTE class, called MT110 and MT103 (or MEDI-538), for conditional T-cell activation. In the presence of target-expressing cell lines, low picomolar concentrations of the BiTE molecules were sufficient to stimulate a high percentage of peripheral human T cells to express cytokines and surface activation markers, enter into cell cycle, and induce redirected lysis of target cells. However, in the absence of target cells, the 2 BiTE molecules even at high concentrations did not detectably activate T cells. Our data show that T cell activation by monomeric forms of MT110 and MT103 is highly conditional in that it is strictly dependent on the presence of cells expressing the proper target antigen. BiTE molecules therefore qualify for a highly controlled polyclonal T-cell therapy of cancer.