Engineering a Growth Sensor to Detect Antigen-Antibody Interactions in Mammalian Cells

Intracellular antibody (intrabody), an antibody that can function intracellularly, is a promising tool for analyzing protein functions and therapeutic use. However, since the reducing cytoplasmic environment generally leads to low antibody stability, it has been difficult to generate effective intrabodies. 

In this study, we designed an scFv-c-kit as a growth sensor for detecting antigen–antibody interactions directly in the cytoplasm of mammalian cells. This growth sensor is constructed by fusing an scFv fragment and the cytoplasmic domain of c-kit as a receptor tyrosine kinase. The roles of the scFv fragment and c-kit are to detect an antigen and transduce a growth signal, respectively. In principle, as scFv-c-kit binds to an oligomer antigen, it will form dimers and transduce a growth signal. Therefore, detection of the growth signals allows selecting functional intrabodies. Since the selection is performed in interleukin-3-dependent Ba/F3 cells, the cells with antigen–antibody interactions can be selected as growing cells in the culture medium without interleukin-3.

Firstly, we tried to exploit the growth sensor to select an scFv clone against rabies virus nucleoprotein (RV-N) since it is known that RV-N forms oligomer complexes in the cells. We constructed two RV-N expression vectors; one contained RV-N fused to V5 and His tags, and the other had a dimer of RV-N connected by a flexible linker with the same tags. Then, we introduced them into Ba/F3 cells to obtain stable cells expressing RV-N (termed as Ba/N and Ba/N2). Next, we constructed an scFv-c-kit library expression vector, in which the scFv library is derived from the phage-displayed Tomlinson I scFv library after a single-round panning against RV-N. After the transduction of the scFv-c-kit library to Ba/N and Ba/N2, library selection was performed in 24-well plates to select cells having scFv-c-kit growth signals. The scFv sequences from growing cells were analyzed to find twelve different scFv clones. The antigen specificity evaluation revealed that three clones showed RV-N-dependent growth signals, one of which showed selective binding to RV-N.

However, during the library selection, noise of the growth sensor caused selection of false-positive clones. Therefore, it was necessary to improve the growth sensor performance. Using a structure-folding software, we recognized that scFv and c-kit were too close to each other. This may cause inherent interaction between them and subsequent cell growth noise. Hence, we decided to engineer the growth sensor structure by inserting one of three flexible linkers G₄S, (G₄S)₂, and (G₄S)₃ between scFv and c-kit. The results revealed that the insertion of the flexible linkers significantly reduced the growth noise of one of the non-specific clones, especially with the G₄S linker. Moreover, the insertion of the G₄S linker remarkably increased the signal-to-noise ratio of a specific clone for RV-N.

Taken together, we demonstrated that the scFv-c-kit growth sensor could be feasible to select functional intrabodies, and structural engineering could ameliorate its performance. This may develop into a promising intrabody selection method in the future.