On the asymmetric partitioning of nucleocytoplasmic transport – recent insights and open questions

Eukaryotic cells feature a protective double-layered membrane known as the nuclear envelope (NE) that encapsulates the nucleus within the cytoplasm. Segregating the genome from the protein synthesis machinery enables cells to exert control over transcription and translation in space and time. However, this requires key macromolecular cargoes, such as transcription factors and mRNA, to be selectively shuttled into or out of the cell nucleus. Understandably, neurodegeneration (Kim and Taylor, 2017), aging (Cho and Hetzer, 2020), cancer (Cagatay and Chook, 2018; Dickmanns et al., 2015) and viral pathogenesis (Fulcher and Jans, 2011; Miorin et al., 2020; Yarbrough et al., 2014) are associated with a dysregulation of this intracellular trafficking process, which is termed nucleocytoplasmic transport (NCT) (Stewart, 2007; Strambio-De-Castillia et al., 2010) and proceeds through nanoscale conduits in the NE known as nuclear pore complexes (NPCs) (Beck et al., 2007; Eibauer et al., 2015; Kim et al., 2018; von Appen et al., 2015).

NCT is unprecedentedly selective and efficient within the complex biological milieu. To appreciate its importance, range and complexity, at least 17% of all eukaryotic proteins are deemed to be imported into the nucleus (Cokol et al., 2000) with over 1000 cargoes being exchanged through each NPC every second (Ribbeck et al., 1998). In the past three decades, the key soluble factors that orchestrate NCT have been identified (Christie et al., 2016; Görlich and Kutay, 1999; Macara, 2001; Weis, 2003). Intensive efforts have also been devoted to understanding how these factors actively facilitate the speed, selectivity and direction of NCT through the permeability barrier of the NPC (Hoogenboom et al., 2021). These comprise members of the β-karyopherin (Kapβ) family, which include importins that usher diverse cargoes bearing nuclear localization signals (NLSs) into the nucleus (Boulikas, 1994; Cokol et al., 2000), and exportins, which escort cargoes bearing nuclear export signals (NESs) out of it (Xu et al., 2012), respectively (Baade and Kehlenbach, 2019). By convention, NES-containing cargoes are termed NES-cargo and NLS-containing cargoes are termed NLS-cargo. Another essential factor, the 25 kDa GTPase Ran, cooperates with Kapβs to regulate the delivery and accumulation of cargoes in an asymmetric, compartment-specific manner (Görlich et al., 1996; Moore and Blobel, 1993). This results from a steep gradient that separates its two nucleotide-bound forms, Ran-guanosine triphosphate (RanGTP) in the nucleus and Ran-guanosine diphosphate (RanGDP) in the cytoplasm. In this manner, NCT maintains essential functions within the nucleus and the cytoplasm without compromising the compositional integrity of either compartment (Terry et al., 2007).

However, several aspects of NCT function remain obscure. While both Kapβs and Ran freely traverse NPCs, an intriguing feature of NCT concerns how its opposing directional elements (importin versus exportin, and RanGTP versus RanGDP) remain asymmetrically partitioned across the NE to direct nuclear import and export processes (Fig. 1). For instance, exportins lack putative NLSs but yet accumulate in the nucleus. In this Review, we address the multiple layers of control that are centered around the NPC and regulate NCT. Thereafter, we highlight evidence that suggests how the asymmetry of NCT might be regulated by Kapβs based on their cellular, functional and structural properties. These include roles in reinforcing the NPC permeability barrier, preserving the steep Ran gradient, and the compartmentalization of small cargoes.

NPCs exert the primary means of control over NCT as the exclusive sites of nucleocytoplasmic exchange. Each NPC is assembled from multiple proteins known as nucleoporins (Nups) (Cronshaw et al., 2002) that surround an aqueous central channel measuring ∼40–60 nm in diameter (Eibauer et al., 2015; Kim et al., 2018; von Appen et al., 2015). Each NPC is equipped with ∼200 intrinsically disordered, phenylalanine-glycine (FG)-rich Nups (or FG-Nups) that are tethered within its central channel. Collectively, the FG-Nups function as a filter-like permeability barrier that permits small molecules below ∼40 kDa (or 5 nm in diameter) to passively diffuse through the NPC, while suppressing the passage of larger non-specific cargoes, which are not recognized by Kapβs (Paine et al., 1975; Popken et al., 2015; Timney et al., 2016). Nevertheless, up to 50% of FG Nups can be deleted in vivo without a noticeable impact on NPC permeability (Strawn et al., 2004). However, the exact form of the NPC permeability barrier remains unclear (Huang and Szleifer, 2020; Lemke, 2016). This is due in part to the inherent flexibility and dynamic fluctuations of the FG-Nups (Sakiyama et al., 2016), which precludes structural characterization within NPCs. Consequently, NPC barrier models have mainly derived from studies with purified FG-repeat domains whose behavior can vary depending on length scale and experimental design (Hoogenboom et al., 2021).

A second layer of control is governed by importins, exportins and transportins (collectively termed ‘Kapβs’) (O'Reilly et al., 2011). Kapβs traverse the NPC permeability barrier in a matter of milliseconds (Dange et al., 2008) by engaging in multivalent interactions with the FG-repeats (Allen et al., 2001; Bayliss et al., 2000b; Kapinos et al., 2014; Port et al., 2015). As mentioned above, importins usher NLS-cargoes into the nucleus, whereas exportins deliver NES-cargoes out of it. Furthermore, transportins can exhibit both import and export functionalities (Twyffels et al., 2014). Altogether, 20 Kapβs are known in vertebrates and 14 in Saccharomyces cerevisiae (Chook and Suel, 2011; Kimura and Imamoto, 2014). This limits the number of cargoes assigned to each Kapβ to reduce potential errors during NCT. Although all Kapβs can bind to their cargoes directly, the 100 kDa canonical importin Kapβ1 (also known as importin β1, KPNB1) also recruits Kapα (importin α), which has seven isoforms (KPNA1–KPNA7) that function as cargo-adaptor proteins (Pumroy and Cingolani, 2015). Kapβ1 also recruits snurportin-1 (SPN1, also known as SNUPN) for the import of small nuclear ribonucleoproteins (Mitrousis et al., 2008). In both cases, SPN1 and Kapα bind to Kapβ1 through their N-terminal importin β-binding (IBB) domains (Lott and Cingolani, 2011).

Third, numerous NLSs and NESs greatly expand the repertoire of cargoes being recognized by each Kapβ. The best-characterized ‘classical’ nuclear import pathway consists of NLS-cargoes that typically form transport complexes with Kapα–Kapβ1, that is NLS-cargo–Kapα–Kapβ1 (Lange et al., 2007). Classical NLSs harbor multiple lysine (K) and arginine (R) residues as exemplified by the NLS of monopartite SV40 T-antigen (Kalderon et al., 1984) or the bipartite NLS of nucleoplasmin (Robbins et al., 1991). Nevertheless, substantial sequence variations exist across NLSs (Boulikas, 1994), both in cargoes that utilize the Kapα–Kapβ1 complex (Kosugi et al., 2009) and those that directly bind to Kapβ1 (Cokol et al., 2000; Lee et al., 2003). Some cargoes, such as myocardin-related transcription factors (MRTFs) (Pawlowski et al., 2010) may even harbor individual NLSs that are recognized by different Kapα isoforms (Goldfarb et al., 2004), although with varying affinities (Friedrich et al., 2006; Pumroy and Cingolani, 2015). Certain cargoes can also contain multiple NLSs that associate with different Kapαs or Kapβs, for instance hypoxia-inducible factors (HIFs) (Chachami et al., 2009; Depping et al., 2008). Other Kapβs such as transportin 1 (also termed Kapβ2) recognize cargoes via a consensus NLS-motif that contains proline (P) and tyrosine (Y) residues (termed PY-NLS cargoes) (Lee et al., 2006). In terms of exportins, chromosomal maintenance 1 (CRM1; also known as exportin 1, Exp1 or XPO1) recognizes a consensus leucine-rich NES (Kosugi et al., 2014). This clearly indicates that NLSs and NESs are diverse and that not all comply with consensus motifs (Cokol et al., 2000).

The Ran gradient constitutes a fourth layer of control that regulates NCT directionality, cargo partitioning and Kapβ recycling (Clarke and Zhang, 2008; Görlich et al., 1996; Izaurralde et al., 1997). RanGTP is ∼200 times more highly concentrated (i.e. partitioned) in the nucleus than in the cytoplasm (Görlich et al., 2003; Kalab et al., 2002; Smith et al., 2002). During import, NLS-cargo–importin complexes (including those that contain the adaptor protein; i.e. NLS-cargo–Kapα–Kapβ1) entering into the nucleus are disassembled upon binding of RanGTP to the importin (Jäkel and Görlich, 1998). This serves to retain the NLS-cargo in the nucleus as the NPC permeability barrier hinders its return to the cytoplasm. At the same time, the binding of RanGTP–importin complexes to the FG-Nups in the NPC facilitates its return to the cytoplasm. RanGTP is then hydrolyzed to RanGDP by SUMOylated RanGTPase-activating protein 1 (RanGAP1) together with Ran-binding protein 1 (RanBP1) and Ran-binding protein 2 (RanBP2, also known as Nup358), which constitute the eight cytoplasmic filaments surrounding the NPC cytoplasmic periphery (Koyama and Matsuura, 2010; Lounsbury and Macara, 1997; Monecke et al., 2013; Vetter et al., 1999). Thereafter, RanGDP frees the importin, which is then able to undertake another cargo import cycle (Stewart, 2007). Similarly, GTP hydrolysis mediated by RanGAP1 disassembles ternary NES-cargo–exportin–RanGTP complexes to complete their nuclear exit. RanGDP is then recycled back to the nucleus by its specific carrier nuclear transport factor 2 (NTF2; also known as NUTF2) (Ribbeck et al., 1998).

The Ran loop is finally closed by the chromatin-bound enzyme regulator of chromosome condensation 1 (RCC1; also known as RanGEF), which recharges RanGDP to RanGTP (Klebe et al., 1995b; Renault et al., 2001; Ribbeck et al., 1998). Hence, GTP is the energy source that powers NCT. Accordingly, the interconversion of RanGTP and RanGDP by RanGAP1 and RanGEF constitutes the fifth and final layer of NCT control.

Each of the aforementioned layers constitutes key mechanistic steps of NCT that lead to the partitioning of NLS-cargoes in the nucleus and NES-cargoes in the cytoplasm. However, they do not sufficiently explain the steady-state partitioning of soluble, yet directionally opposed, transport factors (e.g. importin versus exportin) observed in living cells (Kirli et al., 2015) (Fig. 2). As a comparison, RanGEF contains an NLS (Nemergut and Macara, 2000), whereas RanGAP1 contains a single NLS and nine NESs (Matunis et al., 1998), which regulate their localization in the nucleus and cytoplasm, respectively. Clearly, their enzymatic activity dictates how much RanGTP and RanGDP are generated in each compartment (Görlich et al., 2003; Kalab et al., 2002; Smith et al., 2002). Nevertheless, it is not well understood how the inter-compartmental mixing of RanGTP and RanGDP is prevented to preserve the steep RanGTP–RanGDP gradient (Fig. 3A). There are also no known mechanisms that explain the asymmetric partitioning of importins and exportins. In the following section, we discuss potential factors that could influence the NPC to achieve Kapβ partitioning, maintenance of the steep Ran gradient and the partitioning of other small cargoes between the nucleus and cytoplasm (Fig. 3B).

Currently, the NPC permeability barrier is largely modeled after the behaviors of FG-Nups observed in vitro. This ranges from tethered molecular layers (Eisele et al., 2012, 2010; Kapinos et al., 2014; Schleicher et al., 2014; Schoch et al., 2012; Zahn et al., 2016), liquid droplets (Celetti et al., 2020), and gel-like (Frey et al., 2018; Schmidt and Görlich, 2015) to more solid-like hydrogels (Frey and Görlich, 2007, 2009; Milles et al., 2013). Nevertheless, all of the above studies report permeability barrier properties that facilitate Kapβ passage, but exclude non-specific cargoes irrespective of their different material characteristics. The so-called ‘selective phase’ model postulates that the FG-Nups form a cross-linked gel-like meshwork within the NPC. Here, passive diffusion is determined by the mesh size, whereas selective transport occurs through binding of Kapβ to FG repeats that might effectively break individual cross-links (Frey and Görlich, 2007; Hülsmann et al., 2012). Based on the dynamic behavior of the FG Nups (Sakiyama et al., 2016), the ‘polymer brush’ or ‘virtual gating’ model suggests that the entropic fluctuations of surface-tethered FG-Nups excludes non-specific cargoes from the NPC (Lim et al., 2007, 2006; Rout et al., 2003). Finally, the ‘two-gate’ model envisages the central channel to be occupied by a cohesive meshwork, whereas peripheral FG-Nups are brush-like, thus providing spatially distinct pathways for the cargo molecules to translocate in the NPC central channel (Yamada et al., 2010).

However, Kapβs such as Kapβ1 and CRM1 exhibit a marked enrichment at the NPCs, which is visible as a distinct nuclear rim staining (Heaton et al., 2019; Kapinos et al., 2017; Lim et al., 2015; Lowe et al., 2015) (Fig. 2). On this basis, the NPC permeability barrier might resemble a mixed ternary phase, comprising Kapβs, FG-Nups and water (Zilman, 2018). Thus, Kapβ-binding to FG-Nups could modulate their biophysical behavior to impact on NPC barrier function in a manner that remains incompletely understood (Kapinos et al., 2014; Vovk et al., 2016; Zahn et al., 2016). Indeed, depleting Kapβ1 ex vivo abrogates NPC barrier function against non-specific cargoes, whereas adding back Kapβ1 rescues it (Kapinos et al., 2017). Hence, enrichment of Kapβ might reinforce the barrier-forming qualities of the FG-Nups (Fig. 3B) (Lim et al., 2015). It remains to be seen whether and how different Kapβs might regulate the permeability barrier as integral components of the pore.

Depending on their cargoes, the dwell times of Kapβs in the NPC are between 5 and 20 ms (Kubitscheck et al., 2005; Tu et al., 2013; Yang et al., 2004), but can reach 180 ms for mRNA (Grünwald and Singer, 2010). Moreover, increasing the concentration of Kapβ1 enhances cargo transport efficiency through the NPC and decreases cargo dwell time at the NPC (Yang and Musser, 2006). The latter might be due to a reduction of available FG repeats and the frequency of their interactions with individual Kapβs (Aramburu and Lemke, 2017), which decreases the avidity of Kapβ–FG-Nup binding (Kapinos et al., 2017, 2014; Lowe et al., 2015; Schleicher et al., 2014; Wagner et al., 2015). Nevertheless, import cargo dwell times also depend on the binding of RanGTP to importins and are not a priori equivalent to Kapβ residence times. Thus, successful import depends on the accessibility of RanGTP to importin–cargo complexes on the nuclear side of the NPC, whereas successful export depends on GTP hydrolysis by RanGAP1 on the cytoplasmic side.

Within the NPC, Kapβ complexes exhibit Brownian diffusion that is facilitated by interactions with the FG-repeats, also termed facilitated diffusion, which seems to expedite their translocation through the central channel (Cardarelli et al., 2011; Yang et al., 2004). However, whether and how a crowding of Kapβs within the NPC affects their kinetic interactions with the FG-Nups and ensuing dynamic movements within the pore remains unclear. To gain a physical understanding of such effects, the behavior of Kapβ1-functionalized colloidal beads was studied on surface-tethered FG-Nup layers. The beads transitioned from being immobile to exhibiting two-dimensional diffusion when the amount of soluble Kapβ1 was raised from low to physiologically relevant concentrations, which resulted in an enrichment of soluble Kapβ1 within the FG-Nup layer (Schleicher et al., 2014). In contrast, non-specific control beads exhibited three-dimensional diffusion that transiently impinged on the FG-Nup layer without binding (Schleicher et al., 2014). It remains to be determined how Kapβ complexes can exhibit rapid movements in the NPC while reinforcing the permeability barrier at the same time.

Although not all Kapβ–FG-Nup interactions have been characterized, their known apparent dissociation constants (K_D) typically fall in the sub-micromolar range (Kapinos et al., 2014; Schoch et al., 2012; Tan et al., 2018; Tetenbaum-Novatt et al., 2012). Hence, the amount of each Kapβ that populates the NPC will depend on its cellular concentration, which varies from the nanomolar to micromolar range (Nguyen et al., 2019; Wühr et al., 2015). Indeed, the four most abundant Kapβs are Kapβ1, importin 5 (Imp5, also known as IPO5 or RANBP5), CRM1 and exportin 2 (Exp2 or Xpo2; also known as CAS or CSE1L) (Kirli et al., 2015; Wang et al., 2015) (Fig. 4). Given that Kapβ1 and CRM1 colocalize at NPCs (Fig. 2), this suggests that their presence might modulate the multivalent interactions between the FG-Nups and other Kapβs. Indeed, this so-called binding promiscuity is relevant to how intrinsically disordered proteins interact with multiple partners simultaneously (Uversky, 2013), as has been shown for the binding of Kapβ1 and NTF2 to the FG Nups (Wagner et al., 2015).

The secondary and tertiary structures of Kapβs are highly conserved across subfamilies and species despite their low sequence similarity (Conti et al., 2006). Kapβs comprise 19 to 21 consecutive HEAT repeats that are arranged as a pair of amphiphilic α-helices. Thus, Kapβs constitute highly flexible right-handed solenoids that vary in curvature, diameter and pitch (Conti et al., 2006; Fukuhara et al., 2004). By this means, Kapβs exhibit a conformational versatility to bind to different ligands, such as NLS-cargoes, NES-cargoes, Kapα and RanGTP (Cingolani et al., 2000; Fukuhara et al., 2004; Kappel et al., 2010; Monecke et al., 2013; Port et al., 2015; Yoshimura et al., 2014). In addition, adjacent HEAT motifs harbor several hydrophobic pockets that facilitate multivalent binding interactions with FG repeats (Bayliss et al., 2000a; Isgro and Schulten, 2005; Port et al., 2015). Taken together, this suggests that the apparent binding affinity (i.e. binding avidity) of Kapβs to FG-Nups may depend on the resulting conformation that each respective Kapβ adopts during cargo loading. Indeed, FG-Nup binding is stronger for NLS-cargo–Kapα–Kapβ1 complexes than for RanGTP–Kapβ1 and Kapβ1 alone (Kapinos et al., 2017, 2014). The binding of CRM1 to FG-Nups also appears to be enhanced in the presence of RanGTP and NES-cargo (Koyama et al., 2017; Port et al., 2015; Roloff et al., 2013). Accordingly, Kapβ–cargo complexes may populate NPCs more than either Kapβ1 alone or RanGTP–Kapβ1 (Kapinos et al., 2017).

GTP- and GDP-bound states of Ran are continuously interchanged between the nucleus and the cytoplasm (Görlich et al., 1996; Moore and Blobel, 1993). However, neither RanGTP nor RanGDP interacts directly with the FG-Nups (Rexach and Blobel, 1995). Moreover, Ran is smaller in size than the estimated 40 kDa passive exclusion limit of the NPC. Yet, RanGTP is ∼200 times more concentrated in the nucleus than the cytoplasm (Görlich et al., 2003; Kalab et al., 2002; Smith et al., 2002). Thus, it is puzzling how an uncontrolled mixing of RanGTP and RanGDP is minimized at the NPC level. Indeed, interfering with this steep gradient impairs NCT directionality (Nachury and Weis, 1999), can alter the cellular distribution of Kapβs (Kuersten et al., 2002), and is associated with apoptosis (Wong et al., 2009), stress (Chan et al., 2010; Kelley and Paschal, 2007) and neurological disease (Eftekharzadeh et al., 2018). Clearly, RanGTP is generated in the nucleus by RanGEF and is hydrolyzed to RanGDP in the cytoplasm by RanGAP1, which, together, form the basis of the Ran gradient (Kalab and Heald, 2008) (Fig. 3A, left panel). This is further enhanced by the action of NTF2, which facilitates the return of RanGDP into the nucleus (Ribbeck et al., 1998). Thus far, the in vitro reaction rates of RanGEF to produce RanGTP and its hydrolysis to RanGDP by RanGAP1 have been determined to be 2.1 s⁻¹ and 5.0 s⁻¹, respectively (Klebe et al., 1995a). However, it remains to be experimentally verified whether these enzymatic reactions alone are sufficient to maintain the observed steep Ran gradient in vivo (Görlich et al., 2003; Kalab et al., 2002; Smith et al., 2002).

To preserve the steep Ran gradient, we hypothesized that nuclear leakage of RanGTP is mediated through binding to the enriched pool of Kapβ1 at NPCs (Fig. 3A) (Barbato et al., 2020). Indeed, we observed a substantial leakage of Ran from the nucleus when NPCs lacked Kapβ1 enrichment. In comparison, we found that enriched Kapβ1 provides a retention mechanism at the pore that is biochemically specific for RanGTP, as passive molecules of comparable size, such as GFP, could still traverse the NPC (Barbato et al., 2020). Such a retention mechanism might further explain the steady-state accumulation of Ran at NPCs (Abu-Arish et al., 2009; Smith et al., 2002; Yang and Musser, 2006). Besides its retention at the NPC, we could also show that the efflux of RanGTP depends on its hydrolysis to RanGDP by RanGAP1 by comparing it to a non-hydrolyzable RanQ69L-GTP mutant that was unable to depart from the NPC (Barbato et al., 2020). Finally, we rationalized that NTF2 is required to provide a separate pathway to shuttle RanGDP back into the nucleus (Barbato et al., 2020) because the binding of RanGTP to Kapβ1 (K_d≈35 nM) (Kapinos et al., 2017) is significantly stronger than RanGDP to Kapβ1 (K_d=2 µM) (Forwood et al., 2008). Taken together, this is consistent with simulations, which show that the Ran gradient is sensitive to changes in the permeability of the NPC (Becskei and Mattaj, 2003; Görlich et al., 2003). More generally, we hypothesize that Kapβ1 enrichment at the NPCs increases the efficiency of NCT by minimizing RanGTP losses from the nucleus.

A similar retention mechanism may also apply to other small NLS-cargoes that accumulate in the nucleus. For example, most histones and ribosomal proteins (Table 1) have molecular masses that lie below the NPC size-exclusion limit. In this manner, small NLS-cargoes may be prevented from returning to the cytoplasm by binding to importins that are enriched within the NPC.

Another striking and, perhaps least-understood hallmark, of NCT concerns the asymmetric partitioning of Kapβs themselves. Kapβs lack NLS or NES signals, yet most importins tend to localize in the cytoplasm [with the exception of importin-11 (Imp11, also known as IPO11)], whereas exportins reside in the nucleus, and transportins can be evenly distributed in the nucleus or cytoplasm, depending on their function (Fig. 2). Quantitative analysis by compartment-based mass spectrometry of Xenopus laevis oocytes revealed that the nuclear-to-cytoplasmic ratio (N:C) of Kapβ1 is ∼1:10, while the N:C ratio for both CAS and CRM1 is almost 2:1 (Fig. 2B) (Kirli et al., 2015).

As a case in point, the mechanism(s) that regulates the partitioning of exportins in the nucleus remains elusive despite noted associations between exportins and cancer (Cagatay and Chook, 2018). For example, CRM1 is involved in the export of NES-cargos (Johnson et al., 2002) including mRNA complexes and ribosomal subunits (Chao et al., 2012; Jäkel and Görlich, 1998; Spits et al., 2019; Sutherland et al., 2015), as well as tumor suppressor and regulatory proteins such as BRCA1 (Brodie and Henderson, 2012) and p53 (Kanai et al., 2007). In cancer, CRM1 overexpression enhances the nuclear export of such tumor suppressor proteins, resulting in their mislocalization and functional inactivation in the cytoplasm (Azmi et al., 2021). This has led to the development of selective inhibitors of nuclear export (SINE) that prevent the binding of such NES-cargoes to CRM1 (Azizian and Li, 2020; Parikh et al., 2014; Sun et al., 2016). CAS, whose role is to export Kapα back to the cytoplasm to sustain nuclear import, is another exportin that is overexpressed during cancer progression and metastasis (Jiang, 2016). It is therefore pertinent to account for how exportins are asymmetrically partitioned in the nucleus and to address how interfering with this behavior leads to downstream defects in NCT with relevance to disease. Thus far, only one study has linked the nuclear localization of exportin-T (Xpo-t) to the RanGTP gradient (Kuersten et al., 2002) whereby Xpo-t was mislocalized when its interactions with RanGTP was impaired. Evidently, the lack of any further explanation underscores the little we know about the mechanism(s) that regulates the accumulation of importins and exportins in the cytoplasm and nucleus, respectively. For now, we hypothesize that the dissociation of Kapβ-cargo complexes at the peripheries of the NPC allows for the Kapβs to be rapidly re-captured and circulated back through the NPC (Fig. 3B).

The asymmetric partitioning of NCT and its directionally opposed transport factors (NLS-cargo versus NES-cargo, importin versus exportin and RanGTP versus RanGDP) is achieved by a complex interplay between: (1) the nature of the permeability barrier, (2) cellular abundance of each Kapβ, (3) binding to FG-Nups, and (4) the Ran gradient. We hypothesize that this is further mediated by Kapβ enrichment at the NPC permeability barrier (Fig. 3), which serves to (1) facilitate signal-specific cargo transport, (2) prevent the unsolicited entry of non-specific cargoes, and (3) prevent the leakage of Ran and other small specific cargoes between compartments. Moreover, we speculate that the release of both NLS- and NES-cargo occurs in close proximity to the NPCs so that Kapβs are rapidly re-captured by the FG-Nups and can be circulated back through the NPC. Taken together, these attributes constitute a puzzling causal dilemma – what forms the underlying basis for asymmetry during NCT, transport or partitioning? Certainly, this along with several open questions motivate further basic research in NCT (Box 1). Moving forward, future studies would benefit from adopting a more systems-based approach (Becskei and Mattaj, 2003; Görlich et al., 2003; Kopito and Elbaum, 2009; Smith et al., 2002) to resolve the fascinating complexities of NCT asymmetry and partitioning behavior.