Both morphological analysis and an accumulating body of biochemical data suggest that the NPC is composed of modular subunits consisting of specific subsets of nups. Several of these subcomplexes have been identified as a consequence of their resistance to extraction conditions that partially disassemble NPCs, whereas in other cases the functional complexes are inferred based on networks of interactions, both biochemical and genetic, detected between groups of nups. The conservation of the structural features of the NPC between different species suggests that the compositions of the subcomplexes are also conserved. However, sequence similarity between nups of yeast and vertebrates is variable and several nups present in one species have no clear counterparts in the other (see Cronshaw et al., 2002
). Thus, it remains to be determined to what degree individual subcomplexes are conserved.
Genetic and biochemical analyses have established that yeast Nup53p is part of a network of interacting nups that includes Nup170p, Nup59p, Nic96p, Nup157p, and Nup192p (Aitchison et al., 1995b
; Marelli et al., 1998
; Kosova et al., 1999
; Fahrenkrog et al., 2000
; Lusk et al., 2002
; Makhnevych et al., 2003
). A model for how these nups are organized within the NPC is beginning to emerge. Nup53p can bind directly to Nup170p and Nic96p (Lusk et al., 2002
; Makhnevych et al., 2003
; Lusk and Wozniak, unpublished data). However, their interactions are cell cycle regulated. Nup53p binds to Nup170p during interphase, which blocks the ability of Nup53p to bind the karyopherin Kap121p. In contrast, during mitosis Nup53p is no longer detected in association with Nup170p but is instead bound to Nic96p and Kap121p (Makhnevych et al., 2003
). How Nup59p, Nup157p, and Nup192p fit into this developing model is less clear. Proposals must consider the fact that Nup59p and Nup157p are likely products of gene duplications that also gave rise to Nup53p and Nup170p, respectively (Aitchison et al., 1995b
; Marelli et al., 1998
). Thus, although these pairs of related proteins are not functionally identical, it is possible that they are structurally interchangeable within the repetitive subunits of the NPC, creating a mosaic of structurally similar subcomplexes composed of different combinations of these related nups. In contrast to the redundancy observed in yeast, vertebrate Nup53 and Nup155 (the ortholog of Nup170p and Nup157p) represent the only forms of these nups. Together with Nup93 (the ortholog of yeast Nic96p) and Nup205 (the ortholog of yeast Nup192p) they likely form a similar, evolutionarily conserved, structure. We interpret the lack of paralogs in vertebrates to suggest that these Nup53-containing complexes are more homogenous and potentially lack certain functions that have evolved in the yeast counterpart (see Makhnevych et al., 2003
). This may include the ability to regulate import controlled by the mammalian counterpart of Kap121p, because vertebrate Nup53 lacks the Kap121p-binding domain found in yeast Nup53p.
On the basis of data presented in this article and by analogy to the yeast NPC, we propose that Nup53 interacts with Nup155 and a previously identified Nup93-Nup205 complex (Grandi et al., 1997
). This latter complex may also be physically associated with Nup188 (Miller et al., 2000
), the yeast ortholog of which has been functionally linked to Nup53p and Nup170p (Aitchison et al., 1995b
; Marelli et al., 1998
). We have shown that both GST-Nup53 and GST-Nup93 are capable of binding Nup155 and Nup205, as well as Nup93 and Nup53, respectively. These data and the observed decreases in cellular levels of these nups in response to depletion of either Nup53 or Nup93 suggest they exist as a complex within the NPC. This complex, however, has been refractory to direct isolation, likely because it is labile under conditions required to disassemble the NPC. For example, extraction conditions that have allowed the identification of complexes containing Nup107-Nup133 or gp210 on sucrose gradients appear to largely produce monomeric Nup53, Nup93, and Nup155 (see Supplementary Figure 1).
This group of nups likely contributes to the symmetrical core structures of the NPC. Recent immunoelectron microscopy data from Krull et al.
) suggest that both Nup93 and Nup205 are centrally positioned within the NPC near the pore membrane. Also consistent with this idea, Nup155 is symmetrically distributed on both faces of the NPC (Radu et al., 1993
). In addition, we observed that Nup53 is accessible to cytoplasmically presented antibodies in the presence or absence of the NE membrane (see Supplementary Figure 2). The localization of these nups to the NPC core is also consistent with data showing that Nup93 and Nup205 play a role in the permeability of the NPC (Galy et al., 2003
), a property that is shared with the yeast counterpart of Nup155 (Shulga et al., 2000
Data are also consistent with the idea that this complex of nups lies near the circumference of the NPC at the pore membrane. This conclusion is based both on the localization of Nup93 and Nup205 as discussed above (Krull et al., 2004
) and on the tight association between Nup53 and the NE membrane. We have shown, using various urea extraction conditions, that the release of Nup53 from the NE membrane appears to largely parallel that of the lamins, requiring significantly higher concentrations of urea for extraction from the membrane than other nups. These results are similar to those previously documented with yeast Nup53p (Marelli et al., 2001
). In yeast, the association of Nup53p with the membrane was shown to be dependent on a putative amphipathic helix at its C-terminus, a structure that also appears to be present within the last 13 amino acid residues of vertebrate Nup53. Nup53 also appears to interact with the nuclear lamina. We have observed that Nup53 remains tightly associated with the lamina after removal of the NE membrane with detergent (). Moreover, we have detected lamin B bound to recombinant Nup53 using in vitro binding assays () and we have observed that Nup53 and lamin B cosediment on sucrose gradients (Supplementary Figure 1). In the aggregate, our data support a model in which Nup53 is positioned at the interface between the pore membrane and the lamina where it could anchor other nups to these structures.
Nup53 is one of two nups that have so far been implicated in linking the NPC to the nuclear lamina. Smythe et al.
) have suggested that Xenopus
Nup153 binds to lamin B3
. Nup153 is located on the nuclear side of the NPC (Cordes et al., 1993
; Sukegawa and Blobel, 1993
) and it has been suggested that it links the core structures to the nuclear basket and the nuclear filamentous protein Tpr (Krull et al., 2004
). These data place Nup153 in a separate subcomplex from the core associated Nup53. Considering these data and ours, its would seem possible that different subcomplexes of the NPC may possess distinct lamin binding members, with Nup53 functioning to link the Nup93-containing complex, and perhaps more generally the NPC core, to the nuclear lamina.
At the pore membrane Nup53 is strategically positioned to function in the assembly of nups into a forming NPC. Using RNA interference to deplete endogenous Nup53, we showed that reducing the cellular levels of Nup53 produced a corresponding decrease in Nup93, Nup205, and Nup155, while not affecting other nup complexes (). A likely explanation for this phenomenon is that reduced levels of Nup53 inhibit the assembly of these nups into the NPC, leading to unstable, partially assembled complexes that are more rapidly degraded. Similar results have also been documented for other nup complexes, including the depletion of the Nup107-Nup160 complex, where the stability of the complex is decreased by depletion of individual members (Boehmer et al., 2003
; Harel et al., 2003
; Walther et al., 2003
A striking consequence of depleting Nup53 is the loss of normal nuclear morphology. Nup53-depleted cells lose their spherical shape, generally adopting an elongated lobular or kidney-shape. This observation was intriguing in light of the interactions between Nup53 and the lamina and the similarity of this morphology to that previously detected in cells lacking lamin A (Sullivan et al., 1999
). However, we have not observed any interactions between Nup53 and lamin A/C nor have we detected changes in the levels or the distribution of the lamins or the inner membrane protein emerin (). Both of these classes of proteins are evenly distributed along the irregular nuclear periphery of the Nup53-depleted cells, each following the contour of the underlying chromatin mass. Thus, the relationship between the nuclear morphologies of cells lacking Nup53 and the lamin A–depleted cells is unclear. Not surprisingly, we also observed a similar phenotype in cells depleted of Nup93 where levels of Nup53 are also reduced. This raises the question of whether the altered nuclear morphology is related directly to the loss of Nup53 function or whether it is a result of the codepletion of Nup93 or other interacting nups.
Depletion analysis of the C. elegans
counterparts of the Nup93, Nup205, Nup155, and Nup53, has also been performed in embryos. This study conducted by Galy et al.
) suggests that distinct morphological phenotypes are associated with the depletion of these nups. Although depletion of each causes an embryonic lethal phenotype, depletion of the counterparts of Nup53 or Nup155 appears to block nuclear formation after mitosis. The cause of this block is yet to be determined. In contrast, embryos lacking Nup93 and Nup205 exhibit abnormal peripheral chromatin condensation and what appears to be an altered distribution of NPCs in the NE. We have not detected similar phenotypes in HeLa cells depleted of either Nup53 or Nup93 (Figures and ; unpublished data). Moreover, unlike the depleted HeLa cells, the C. elegans
embryos lacking counterparts of Nup93 or Nup53 do not exhibit any alterations in nuclear shape (Galy et al., 2003
A disruption of nuclear morphology similar to that detected in cells depleted of Nup53 was also observed in cell lines depleted of Mad1 (Luo et al., 2002
). The most well-defined function of Mad1 is its role in the spindle assembly checkpoint, a quality control feature of chromosome segregation that monitors spindle attachment to kinetochores during mitosis (reviewed in Lew and Burke, 2003
). Intriguingly, Mad1 is localized to the NPC during interphase in both mammalian and yeast cells (Campbell et al., 2001
; Iouk et al., 2002
). However, it remains to be determined what function the association of Mad1 with the NPC plays in the spindle assembly checkpoint or to what degree Mad1 contributes to the function of the NPC. In yeast, the docking of Mad1p to the NPC has been shown to occur, in part, through its interaction with the Nup53p-containing complex (Iouk et al., 2002
). Consistent with this, yeast cells exhibit diminished levels of Mad1p at the NPC in nup53
Δ cells (Iouk et al., 2002
). Our data are consistent with a similar interaction occurring in mammalian cells, where the levels of Mad1 are reduced, but still detectable, at the NE in the Nup53-depleted cells (). We conclude from these observations that Mad1 may interact with more than one nup at the NPC. Of note, siRNA induced depletion of Nup93, which reduces NE levels of Nup53, does not appear to alter cellular levels of Mad1. A potential explanation for these results is that Mad1 interacts with a separate pool of Nup53 distinct from that bound to the Nup93-containing complex.
Previous studies of cell lines specifically depleted of Mad1 using RNA interference have shown that these cells exhibit defects in spindle checkpoint function (Martin-Lluesma et al., 2002
). We have performed similar experiments on Nup53-depleted cells to determine whether the reduced NPC association and cellular levels of Mad1 are sufficient to abrogate spindle checkpoint function. However, we failed to detect an obvious chromosome segregation defect (unpublished data) that would suggest that this pathway is inhibited. Presumably, sufficient levels of Mad1 remain to conduct its essential role in spindle checkpoint function. What role the NPC association plays in the function of this remaining pool of Mad1 will await the identification of its other NPC binding sites.
During prometaphase Mad1 is recruited to kinetochores (Chen et al., 1998
; Campbell et al., 2001
). In contrast, we have not detected Nup53 (using either indirect immunofluorescence or GFP fusions) or Nup93 at kinetochores during mitosis, suggesting that any interaction of Mad1 with Nup53 is not maintained at the kinetochores. Of note, members of the Nup107-containing complex and Nup358 (RanBP2), although largely distributed throughout the cytoplasm during mitosis, have also been detected at kinetochores (Belgareh et al., 2001
; Joseph et al., 2002
; Loïodice et al., 2004
). The functional role of the Nup107-Nup160 complex at kinetochores is not established and no specific defects in the function of this structure have been attributed to the depletion of these nups. In contrast, depletion of Nup358 leads to severe defects in kinetochore structure and a reduction in the recruitment of Mad1 and Mad2 to kinetochores (Salina et al., 2003
; Joseph et al., 2004
). However, the spindle checkpoint remains active in these cells. Thus it appears unlikely that these nups play a critical role in basal levels of spindle checkpoint activity. A more plausible scenario is that the association of nups with kinetochores, as well as interactions of the Mads with the NPC, may function to regulate the fidelity of the spindle checkpoint response.