Alternative mechanisms of CRM1 mediated nuclear export inhibition

All previously studied CRM1 inhibitors share a common mechanism that is based on interfering CRM1-cargo interactions by the direct blockage of the NES-binding cleft (Dickmanns et al. 2015;

Sendino et al. 2018; Sun et al. 2016). Very recently the compounds DP2392-E10 was predicted by in silico docking to bind in the region between HEAT repeats 9 and 10 (Chutiwitoonchai et al.

2017). In this study, DP2392-E10 interference with CRM1 interactions was revealed to be independent on Cys528 (section 2.3.5), which further strengthened its predicted binding outside the NES-binding cleft. In addition, when tested at similar concentrations to C3 and C10, DP2392-E10 could dissociate the binding of the exceptional CRM1 cargo SPN1 that binds CRM1 in multiple regions besides its NES peptide (Chapter 2 - figure 4). This further implies DP2392-E10-based inhibition mechanism to be of an allosteric nature and independent on direct blockage of the NES cleft which, in the performed assay, requires the competition with tight SPN1 interactions (section 2.3.5). During this study docking simulations have been performed under different settings and using the atomic coordinates derived from experimentally determined crystal structure of inhibitor-bound human CRM1 (^HsCRM1-^HsRanGTP-LMB complex). Docking calculations of DP2392-E10 with different settings defined a binding region at the upper side of HEAT9 and at the base of the acidic loop (section 3.3.5). Furthermore, predicted binding conformation suggest a unique non-covalent interaction mode that involves several residues of HEAT9 loop which extends to form the acidic loop (section 3.3.5). Docking calculations supported by the experimental findings suggest a novel mechanism of CRM1 inhibition that is based on interfering exportin - cargo interactions by the allosteric modulation of the NES-binding cleft. In the case of DP2392-E10, this can be achieved by altering the conformation of the acidic loop which is a key structural element in CRM1 conformational transition and cargo release (Dölker et al. 2013; Koyama and Matsuura 2010; Monecke et al. 2014). These findings define a new aspect of CRM1 inhibition that is not based on reactive compounds. The development of CRM1 targeting compounds with such properties outcompetes the current nuclear export inhibitors by exhibiting less toxicity during clinical evaluation.

4.3.2 Interference with RanGTP interactions

The nuclear export of a CRM1 cargo is strictly dependent on the formation of a ternary export complex that involves RanGTP and the cargo protein. The assembly of a stable complex is mediated by the cooperative binding of RanGTP and the cargo protein to CRM1. In the absence

Discussion

of RanGTP most cargoes exhibit low affinity towards the export receptor and therefore the export process cannot be initiated (Monecke et al. 2014). While all known CRM1 inhibitors interfere with cargo binding, nuclear transport can also be inhibited by interfering transport receptor - RanGTP interaction (Hill et al. 2014). Computational analysis of potential ligand binding sites using human CRM1 from the complex ^HsCRM1-^HsRanGTP-LMB crystal structure identified more than 40 potential binding sites outside the NES-binding cleft (section 3.3.4). Most sites were distributed at the outer surface of the protein, whereas few sites were identified at the inner surface of HEAT repeats 2 and 3. This region is known as the CRIME domain and it represents the binding site of RanGTP (Petosa et al. 2004). The identified potential ligand binding site at the CRIME domain indicates for the possibly of developing CRM1 inhibitors that prevent or disturb RanGTP binding.

Such an approach has been already applied for the identification of nuclear import inhibitors that affect the importin β - RanGTP interaction (Hintersteiner et al. 2010; Soderholm et al. 2011).

Karyostatin 1A is a compound that inhibits importin β mediated nuclear import of GFP-NFAT in HeLa cells. Karyostatin 1A was initially identified in an affinity-based screening by confocal nanoscanning. Further analysis revealed that Karyostatin 1A acts by the selective binding to importin β that disrupts its interactions with RanGTP (Hintersteiner et al. 2010). In another study, FRET-based high throughput screen that detects the interaction between importin β and RanGTP identified an inhibitor named importazole, a 2,4-diaminoquinazoline. Importazole mechanism is based on interfering importin β - RanGTP interaction; it also exhibits high specificity towards importin β and was shown to block the nuclear import in Xenopus egg extracts as well as in cultured cells (Soderholm et al. 2011). These examples provide an experimental evidence that inhibiting RanGTP interaction with the transport receptor can be a considerable approach for the development of anti-tumor and anti-viral drugs that target the nucleocytoplasmic transport machinery.

4.3.3 Interfering CRM1 translocation through the nuclear pore complex

Interfering the association of the export complex with the NPC represents an alternative mechanism that has not been yet addressed or considered for inhibiting CRM1 mediated nuclear export. This is most likely due to the insufficient knowledge on CRM1 interaction with the NPC from a structural perspective. The crystal structure of human CRM1 in complex with a Nup214 fragment that contains 8 FG-repeat motifs was recently published, and it is the first crystal structure that provided an insight into CRM1 interactions with the NPC (Port et al. 2015). Soon after, another structure of ^ScCRM1 in complex with SxFG/PxFG repeat peptide of the yeast Nup42 was released (Koyama et al. 2017). Both structures revealed that CRM1 interactions with

FG-Discussion

repeats is mediated by the binding of FG motifs as the main anchor points of FG-repeat peptides (Koyama et al. 2017; Port et al. 2015). FG motifs were shown to bind CRM1 in specific pockets that are distributed at the outer surface of the transport receptor at the N- and C- termini (section 1.2.4). This characteristic binding of FG-repeats is important to maintain the export complex stability during transportation and to allow the complex translocation through the NPC (Koyama et al. 2017; Port et al. 2015). During this study, human CRM1 from the complex ^Hs

CRM1-HsRanGTP-LMB crystal structure was subjected to computational analysis for the identification of potential ligand binding sites. Most of the identified binding sites are distributed at the outer surface of CRM1 (Chapter 3 - figure 6) which strongly indicated the possibility to inhibit the export receptor by disrupting its translocation through the NPC.

Figure 11: Structural alignment of human CRM1 showing potential ligand binding site with CRM1-bound Nup214 FG repeat fragment in (A) and with CRM1-CRM1-bound Nup42 FG repeat fragments in (B).

binding sites (filled with orange spheres) were detected using the Alpha Site Finder function of the program MOE. FG repeat fragments are shown in cartoon representation (Nup214 in red, Nup42 in green) with FG motifs depicted as spheres. FG motif binding sites are indicated as P1-P8 for Nup214 fragment and as S1-S5 for Nup42 fragments. Binding pockets that align with potential binding sites are circled with a dashed line.

Discussion

Interestingly, the alignment of potential ligand binding sites, identified by Site Finder application available in MOE software package, with CRM1-NUPs FG-repeats complex structures unveiled overlap in case of several FG binding pockets (Figure 11) (Anon n.d.; Tøndel, Anderssen, and Drabløs 2006). Six out of eight binding pockets (P1-P8) of Nup214 (Figure 11 A) fragment FG motifs as well as two out of five binding pockets (S1-S5) of Nup42 FG motifs were identified as potential ligand binding sites at the outer surface of CRM1 (Figure 11 B). Most of the identified pockets are highly conserved among distantly related species what indicates their functional importance (Koyama et al. 2017; Port et al. 2015). Furthermore, the identified pockets are suggested to be involved in the binding of other NUPs which was experimentally proven for Nup62 and RanBP3, an FG-repeat containing nucleoporin-like protein (Port et al. 2015). These findings together with the biochemical and functional analysis performed in both studies strongly indicate that the blocking of FG motifs binding pockets interferes with transport receptor translocation through the NPC. Interfering CRM1 interaction with the NPC may reduce CRM1-mediated export which might be a more reliable and less toxic approach to counteract its overexpression in several cancer diseases. This concept has been already considered for the inhibition of nuclear import mediated by importin β (Ambrus et al. 2010; Gasiorowski and Dean 2003), for which several structures with FG repeats were reported over the last two decades (Bayliss et al. 2002, 2000; Isgro and Schulten 2005; Liu and Stewart 2005). A peptidomimetic inhibitor that mimics the FXFG motif was identified as an inhibitor of the importin-α/β mediated import. (Ambrus et al. 2010) The identified peptidomimetic compound exhibited a selective inhibition of importin α/β mediated transport (Ambrus et al. 2010), indicating that this approach has potentials for developing nuclear transport inhibitors with high specificity.

4.4 Structure-based methods in the discovery and development of CRM1