FRET-based analysis and molecular modeling of the human GPN-loop GTPases 1 and 3 heterodimer unveils a dominant-negative protein complex.

Abstract

GPN-loop GTPases 1 and 3 are required for RNA polymerase II (RNAPII) nuclear import. Gpn1 and Gpn3 display some sequence similarity, physically associate, and their protein expression levels are mutually dependent in human cells. We performed here Fluorescence Resonance Energy Transfer (FRET), molecular modeling, and cell biology experiments to understand, and eventually disrupt, human Gpn1-Gpn3 interaction in live HEK293-AD cells. Transiently expressed EYFP-Gpn1 and Gpn3-CFP generated a strong FRET signal, indicative of a very close proximity, in the cytoplasm of HEK293-AD cells. Molecular modeling of the human Gpn1-Gpn3 heterodimer based on the crystallographic structure of Npa3, the Saccharomyces cerevisiae Gpn1 ortholog, revealed that human Gpn1 and Gpn3 associate through a large interaction surface formed by internal α-helix 7, insertion 2, and the GPN-loop from each protein. In site-directed mutagenesis experiments of interface residues, we identified the W132D and M227D EYFP-Gpn1 mutants as defective to produce a FRET signal when coexpressed with Gpn3-CFP. Simultaneous but not individual expression of Gpn1 and Gpn3, with either or both proteins fused to EYFP, retained RNAPII in the cytoplasm and markedly inhibited global transcription in HEK293-AD cells. Interestingly, the W132D and M227D Gpn1 mutants that showed an impaired ability to interact with Gpn3 by FRET were also unable to delocalize RNAPII in this assay, indicating that an intact Gpn1-Gpn3 interaction is required to display the dominant-negative effect on endogenous Gpn1/Gpn3 function we described here. Altogether, our results suggest that a Gpn1-Gpn3 strong interaction is critical for their cellular function.