hnRNPK interacts with Abi-1 via its KH2 domain.
(A) Yeast two-hybrid screen. The full length Abi-1 cDNA was cloned as bait to screen a human fetal brain cDNA-library for putative interaction partners. 9 independent partial C-terminal hnRNPK clones were identified and retested for interaction by a yeast two-hybrid assay. Results are shown for the longest (aa248–464) and shortest (aa267–464) prey clone. hnRNPK is a 464 aa long protein that codes for several specific domains: N-terminal NLS, nuclear localization signal, KH1-KH3, K homology domains 1–3 (light grey); KI, K interaction domain (black); KNS, K nuclear shuttling signal. Abi-1 (476 aa) codes for the following domains: WAB, WAVE binding domain; SNARE, HHR, homeobox homology region; PP, proline rich domain; SH3 src homology 3 domain. (B) Schematic illustration of the Abi-1 and hnRNPK clones (and abbreviations) that have been used for further experiments. (C) The hnRNPK KH2 domain colocalizes with Abi-1.Several partial GFP- or Myc-tagged hnRNPK and Abi-1 clones were coexpressed in Cos7 cells to identify the interacting subdomains of the two proteins. In single transfection experiments, hnRNPK full length protein predominantly localizes to the nucleus, whereas Abi-1 shows a typical cytoplasmic staining pattern. When coexpressed, both proteins are localized in identical dotted structures (I). In contrast, the K1-recombinant protein alone is restricted to the nucleus and does not colocalize with Abi-1 after cotransfection (II). K2 fusion protein readily colocalizes with full-size Abi-1 in the cytoplasm (III). As shown for K1, K3 also shows no colocalization with Abi-1 (IV). The cotransfection of hnRNPK K2 with Abi-1 missing the SH3 domain (AbiΔSH3) results in no colocalization (V), the expression of Abi-1 SH3 domain alone (AbiSH3), however, gives rise to a perfect overlay in the perinuclear region (VI). (D) Coimmunoprecipitation experiments with overexpressed and endogenous Abi-1 and hnRNPK proteins. Plasmids encoding full-size hnRNPK-GFP and Abi-1-Myc were cotransfected in Cos7 cells and Abi-1-Myc was immobilized using anti-Myc microbeads loaded on a column. Protein-complexes then were eluted, separated by SDS-Page and hnRNPK-GFP (size 95 kDA) was detected by immunoblot using a specific anti-GFP antibody (I). As controls, beads loaded with lysate only (ctrl) and the input lysate were used. (II) Cos7 cells were transfected with partial hnRNPK-coding constructs K1-GFP (KH1 domain), K2-GFP (KH2 and KI domain), K3-GFP (KH3 domain) and K-full-GFP (full length) as GFP-fusion proteins. The correct expression of the hnRNPK constructs was controlled by using an anti-GFP antibody and a commercial anti-hnRNPK antibody that could detect the GFP fusion protein as well as the endogenous hnRNPK in the lysate (95 kDA and 65 kDA). Moreover, the commercial antibody detects the K2 construct. The correct expression and antibody specificity of the Abi-1-Myc construct was tested by cotransfection with truncated hnRNPK constructs and subsequent immunoblotting with an anti-Myc antibody. Afterwards, precipitation was performed with GFP-tagged microbeads after cotransfection of Abi-1-Myc and hnRNPK constructs. The precipitates were subjected to immunostaining with an anti-Myc antibody. The Abi-1-Myc protein could only be detected within hnRNPK-K2-GFP precipitate but not within K1-GFP and K3-GFP precipitate or within the GFP-only and/or negative controls. (III) Vice versa experiments were done by coimmunoprecipitations using lysates of Cos7 cells cotransfected with a combination of full length hnRNPK-Myc (K-Myc) and AbiΔSH3-GFP or AbiSH3-GFP, respectively. The immunoprecipitation was performed using antibodies directed against GFP and immunoblot-detection was performed using anti-hnRNPK antibodies showing that expression of the Abi-1 SH3 domain is a prerequisite for protein binding. (IV) The hnRNPK antibody was used to precipitate the protein complex from brain lysate as well as from the synaptosomal fraction. In the Western blot, an antibody against Abi-1 could readily detect its antigen in the precipitate. As positive control, brain lysate or synaptosomal material was used (Input lane: 4% of the total lysate used for immunoprecipitation), a negative control was performed with unspecific IgG (ctrl IgG). Scale bars are as indicated.