When actin and cofilin were fused to Venus fragments that were split at residues 144, 154, and 171, near the sites that are generally used (1-5)(18, 26-28), no distinct difference in fluorescence intensity was observed between cells expressing cofilin(WT) and cofilin(S3E) (Figure 2A and Supplementary Figure S2). When CFP-tagged LIMK1 was cotransfected simultaneously with actin-VC210 and VN210-cofilin(WT) in cells, the Venus fluorescence intensity was significantly lower in LIMK1-expressing cells than in the surrounding LIMK1-nonexpressing cells (Figure 2C). Cotransfection of LIMK1(WT), but not kinase-dead LIMK1(D460A), did reduce the fluorescence intensity (Supplementary Figure S3). These findings indicate that LIMK1 suppresses the cofilin-actin interaction in cells by phosphorylating cofilin and inactivating the actin binding activity of cofilin. In contrast, the fluorescence intensity was not decreased when the cells expressing actin-VC210 and VN210-cofilin(WT) were microinjected with the plasmid coding for LIMK1-CFP at 24 h after the transfection of BiFC probes (data not shown). These results indicate that the complex of actin-VC210 and VN210-cofilin(WT) is difficult to dissociate once the Venus fragments associate to regenerate a mature fluorescent protein by complementation.
Figure 2D depicts the Venus-based BiFC assay for visualizing the cofilin-actin interaction, in which the recovery of Venus fluorescence depends on the non-phosphorylation state of cofilin at Ser-3. To measure the interaction between VN210-cofilin(WT or S3E) and actin-VC210 in the cell-free system, His6-tagged BiFC probe proteins were expressed in Sf9 cells by the baculovirus expression system and purified through a nickel agarose column (Figure 2E). When purified actin-VC210 was incubated with VN210-cofilin(WT), the fluorescence intensity was gradually increased and saturated after 180 min (Figure 2F). The average rate of fluorescence recovery (t1/2 = 56 min) was comparable to the reported rate of fluorescence recovery of YFP fragments that were split at 154 (t1/2 = 50 min) (2) and was extremely slower than the reported rate of formation of the actin-cofilin complex (t1/2 < 0.01 s) (29). In contrast to VN210-cofilin(WT), incubation of actin-VC210 with VN210-cofilin(S3E) did not exhibit fluorescence even after 600 min incubation, indicating that the recovery of fluorescence intensity reflects the specific interaction between actin and an active (nonphosphorylated) form of cofilin. Moreover, our cell-free BiFC assay is useful for detecting the cofilin-actin interaction with low background fluorescence.
We next examined whether the pair of VN210 and VC210 fragments is applicable to the detection of interactions of other protein pairs. In mitotic signaling, the small GTP binding protein Ras binds to and activates Raf1, a protein kinase that activates mitogen-activated protein kinase (MAPK) through MEK activation, in a manner dependent on the GTP-bound state of Ras. To visualize the interaction between Ras and Raf in cells, we constructed plasmids encoding VN210 and VC210 fused to the N or C terminus of H-Ras and Raf1. Since active RasV12, but not dominant-negative RasN17, binds to Raf1, we compared the fluorescence intensity in cells coexpressing RasV12 and Raf1 with that in cells expressing the corresponding pair of RasN17 and Raf1. Of the eight pairs of fragments examined, the pair of RasV12-VC210 and VN210-Raf exhibited a distinct difference in fluorescence intensity from the corresponding pair of RasN17-VC210 and VN210-Raf (Figure 3A and Supplementary Table S2). Raf1 binds to H-Ras through the Ras binding domain (RBD), corresponding to amino acids 51–131 of Raf1 (19). Therefore, we examined the fluorescence recovery after the coexpression of VN210-Raf-RBD with RasV12- or RasN17-VC210. As expected, RasV12-VC210, but not RasN17-VC210, exhibited Venus fluorescence by coexpression with VN210-Raf-RBD (Figure 3B). Immunoblot analysis revealed the comparable level of expression of these probe proteins (Supplementary Figure S4). These results indicate that the fluorescence recovery reflects the specific interaction between active Ras and Raf or Raf-RBD.
Ras proteins are known to undergo posttranslational modifications near the C terminus; S-farnesylation of Cys of the C-terminal CAAX motif, endoproteolytic cleavage of the last three amino acids, and carboxymethylation of the isoprenylated Cys. These modifications are involved in membrane association of Ras proteins. Ras-VC210 appeared to diffusely distribute in the cytoplasm (Figure 3, A and B), and immunoblot analysis showed that Ras-VC210-His6 (calculated molecular weight, 26 kDa) was not cleaved in cells (Supplementary Figure S4). These results suggest that the fusion of VC210 at the C terminus may interfere with the C-terminal modifications of H-Ras.