Inter-domain Interactions In The Imp1 Oncofetal Protein

The insulin-like growth factor-II mRNA binding protein 1 (IMP1) is one of 3 human paralogues belonging to the highly conserved, multifunctional, RNA-binding VICKZ family [1]. IMP1 is expressed during early embryogenesis; peaking in expression around embryonic day 12.5 (E12.5), followed by a decline towards birth [3]. IMP1 -/- knockout mice present with dwarfism and perinatal lethality, illustrating the critical role of the protein in embryonic cell proliferation and growth [3]. IMP1 is equally critical in neuronal development, through its control of neurite outgrowth [13], and directed motility of neuronal crest cells [14]. Changes in cell morphology and polarisation, which are integral to these cellular processes, are also modulated by IMP1 [3]. Although absent from normal adult tissues, IMP1 expression has been shown to be severely upregulated in various tumours, the most common including the colon, liver, kidney, pancreas, and breast [4]. This biphasic expression of IMP1, which has also been observed in IMP3, can be characterised as oncofetal. IMP1 is a critical oncogene, and has been correlated with enhanced metastasis and overall poor prognosis in some cancers [15]. In the context of neoplastic expression, the aforementioned cellular functions of IMP1 manifest to promote the adhesion, invasion, and migration of tumour-derived cells [15]. This occurs as a result of post-transcri ptional control of a variety of mRNA transcri pts by IMP1, including mRNAs involved in prominent cancer signalling pathways, mitogen-activated protein kinase (MAPK) and phosphatidylinositol (3,4,5)-trisphosphate (PtdIns(3,4,5)P3) [15]. The fact that IMP1 stands at a junction of several important signalling pathways, makes it a target for potentially very potent cancer treatment. IMP1 functions as part of a network; whereby its multiple cellular functions are underpinned by its subcellular control of mRNA localisation, turn over and translation. Immunocytochemical studies have revealed that IMP1 predominantly resides in the cytoplasm in 200-700 nm ribonucleoprotein (RNP) granules [5]. These RNPs bind to and subsequently cage target mRNAs in cytoplasmic mRNPs, which are transported along cytoskeletal structures [5]. The formation of these stable mRNPs prevents the release of mRNA transcri pts from the IMP1-RNA complex, limiting the premature degradation of mRNA transcri pts, and hence enabling their long distance transport in the cell. The presence of cap binding protein 80 (CBP80), and the absence of eIF4E and eIF4F cytoplasmic translation initiation factors in mRNPs indicates that mRNA transcri pts are translationally repressed; a salient feature of RNA localisation [5]. When mRNPs are docked in their final destination, the associated mRNA transcri pts are released, and are either translated or degraded [4]. What controls the antithetic fate of mRNA transcri pts remains unclear, however it is most likely to do with the cis-determinants of mRNA transcri pts. In this scenario, IMP1 is responsible for appropriate spatiotemporal release of mRNA transcri pts, and their subsequent interaction with either translational or decay machinery [4]. Although mainly in the cytoplasm, IMP1 has two putative nuclear export signals, which opens up the possibility of IMP1 attaching to mRNA targets at the site of transcri ption. IMP1 is comprised of six canonical RNA binding domains: two N terminal RNA recognition motifs (RRMs), and four C terminal K homology (KH) domains [4] (figure1). The absence of RRMs in homologous Drosophila and C. elegans IMPs indicates that the two domains are unlikely to be involved in RNA binding [4]. Although, it has been suggested that RRMs play a role in stabilising IMP-RNA complexes, and this is consistent with full length IMP1 having a higher binding affinity (lower Kd) compared with the four KH domain fragment alone [7]. In vitro studies involving truncated GFP-IMP fusion proteins revealed that the four KH domains are necessary and sufficient for RNA binding, as well as granule formation and subcytoplasmic localisation [7]. Each KH domain is composed of three ?-helices packed onto a three-stranded antiparallel ?-sheet, with a conserved GxxG loop that orientates the target mRNA to interact with the hydrophobic groove of the canonical domain (figure 1) [8]. The hydrophobic groove mediates RNA binding by recognising a specific four nucleotide long sequence. Recognition of RNA by KH domains is often combinatorial, involving multiple low affinity KH domain interactions with target mRNA that amalgamate to a high specificity interaction [8]. This is reflected by sequential dimerisation of IMP1 [9], and more recently the discovery that tandem KH domains are capable of inducing the folding of RNA targets through stable inter-domain interaction [10]. The crystal structure of IMP1 KH34, solved by Singer et al. 2010, showed that the tandem domains adopt an anti-parallel pseudodimer conformation, with the putative RNA binding surfaces located at opposing ends of the molecule, and orientated in opposite directions [10]. The six-stranded ?-sheet interface between KH3 and KH4 is largely stabilised by hydrophobic interactions between residues of the two domains, which is extended onto a short linker (479-487) that connects the KH3 ?3 helix to the ?1 strand of KH4 [10]. It has been proposed that the pseudodimer arrangement causes the RNA backbone to undergo an ~ 180? change in direction in order to bind both KH domains simultaneously (figure 1)[10]. This induced RNA looping allows IMP1 to specifically recognise and bind its RNA substrates. While the role of IMP1 KH34 as an RNA binding pseudodimer is arguably well established, the function of KH12 remains unclear. Although, in vitro studies revealing that KH34 is unable to bind RNA below concentrations of 100nM, unlike the full length IMP1, indicates that KH12 also has an RNA binding role [11]. This is supported by the fact that KH12 modulates IMP1 binding of ACTB and MYC RNA targets. Whether KH12 additionally binds RNA as a pseudodimer, however, remains to be elucidated. It is possible that the high conservation between the two KH didomains is reflected by KH12 having the same RNA recognition and binding mechanism as KH34. However, it is important to consider that only one basic residue (lysine) is conserved between the aforementioned KH34 linker and the corresponding short linker that connects KH1 to KH2 [10]. While the precise function of the linker is yet to be defined, the fact that the KH34 linker sequence is highly conserved between IMP1 and its VICKZ family homologues indicates an important functional role in the stability of the pseudodimer. Moreover, the substitution of the KH34 linker with the residues that connect KH1 and KH2 within ZBP1, the chicken homologue with which IMP1 shares >94% homology, resulted in a significant reduction in the RNA binding affinity of ZBP1 KH3-L12-KH4 mutant, compared with wild type ZBP1 [10]. This could be the result of a change in the orientation of the putative RNA binding domains as a result of destablisation of the KH34 pseudodimer, or equally the loss of critical interactions between the linker residues and the RNA target. The evident importance of the KH34 linker, together with its low sequence identity with the KH12 linker, means that the dimerisation of KH12 cannot be assumed; but must be experimentally determined. In the present study, we aimed to determine whether the IMP1 KH12 di-domain adopts a pseudodimer conformation similar to that of KH1-4, in order to elucidate the structural configuration of the KH di-domains in IMP1, and hence contribute to the understanding of how the KH domains are able to recognise and bind a diverse range of specific mRNA targets. To achieve this, we needed to optimise an established expression and purification protocol to express and purify sufficient concentrations of IMP1 KH12 and KH1-4 constructs for NMR analysis. A 15N relaxation measurement was carried out on the KH12 construct, followed by superimposition of the NMR fingerprint spectra of both KH12 and KH1-4 constructs, with the intention of revealing the conformation of the KH12 di-domain, and whether this is influenced by the presence of the KH34 di-domain.