Postmitotic NPC assembly is thought to be regulated by the phosphorylation of several nucleoporins by mitotic kinases (Onischenko et al., 2005; Glavy et al., 2007) as well as by the small GTPase Ran and its counterplayer importin-β. Importin-β binds to several nucleoporins during mitosis, and their release by RanGTP in the vicinity of chromatin is crucial for postmitotic NPC assembly (Walther et al., 2003b). Assembly is initiated by the Nup107–160 complex in early anaphase, when it binds to chromatin via its DNA-binding subunit embryonic large molecule derived from yolk sac (ELYS)/Mel28 (Franz et al., 2007). This complex is essential for pore formation, and pore-free nuclei form in its absence in Xenopus laevis egg extract (Harel et al., 2003; Walther et al., 2003a; Franz et al., 2007; Gillespie et al., 2007). The association of membrane nucleoporins (Pom121) with the newly forming pores is the second phase of NPC assembly, followed by addition of the Nup93 and Nup62 complexes and finally more peripheral nucleoporins (Bodoor et al., 1999; Haraguchi et al., 2000; Dultz et al., 2008). Although many of the mechanistic interactions remain to be understood, the order and timeline of events that lead to the formation of NPCs after mitosis are thus rather well defined.

By comparison, the process of NPC assembly in interphase in metazoan cells is poorly characterized and has only recently become the focus of research (D’Angelo et al., 2006; Doucet et al., 2010; Iino et al., 2010). Although assembly during both phases of the cell cycle is thought to produce indistinguishable pores, and some of the regulatory requirements such as activity of the small GTPase Ran are conserved between interphase and mitosis (Walther et al., 2003b; D’Angelo et al., 2006; Iino et al., 2010), the assembly takes place under fundamentally different conditions. At the end of mitosis, a large number of NPCs assemble simultaneously. NPC assembly is initiated on naked chromatin and occurs concomitant with the reformation of the nuclear envelope, and the cytoplasm is in the biochemical state of mitotic exit with significant remaining mitotic kinase activity. In contrast, interphase assembly events are singular and rare. NPC insertion into the closed nuclear envelope requires fusion between inner and outer nuclear membranes, and there is no mitotic kinase activity in the cytoplasm. A very recent study has also reported molecular differences in the requirements of some nucleoporins for the two assembly processes (Doucet et al., 2010). ELYS, which mediates the interaction of the Nup107–Nup133 complex with chromatin, appears dispensable for interphase assembly. On the other hand, the membrane nucleoporin Pom121 as well as a membrane association domain of Nup133 were found to be required.

To estimate at what rate new NPCs form during the cell cycle, we next measured the growth of the nuclear surface during interphase. Because NPCs are anchored in the nuclear lamina and do not move relative to each other in the plane of the nuclear envelope (Daigle et al., 2001), individual NPCs can be tracked over long intervals. Measuring the area between tracked NPCs (Fig. 1, B and D), we determined the growth rate of the nuclear surface to be 3.9 ± 0.7% per hour (24 cells from four independent experiments). We also independently determined the growth rate of the entire nucleus by imaging NRK cells expressing fluorescent nuclear markers and reconstructing the nuclear volume (Fig. S1). The growth rate of the nuclear surface area determined in this manner was 4.1 ± 1.4% per hour (24 cells from four independent experiments), which is indistinguishable from the growth rate of the bottom surface measured by pore tracking (P = 0.7 for unpaired t test). This indicates that nuclear growth occurs homogenously over the entire surface of the nucleus. Interestingly and in agreement with previous observations (Maul et al., 1973), nuclear growth was independent of DNA replication because treatment with aphidicolin to inhibit DNA polymerase did not affect nuclear growth rates while effectively blocking cells at the G1/S boundary (Fig. S1).

The NPC density and nuclear surface area measurements allowed us to calculate the number of NPCs on the nucleus during the cell cycle. Our data predict a continuous increase in the number of NPCs throughout interphase with 60–80 new NPCs forming per nucleus in 1 h. Nuclei of NRK cells had 1,960 ± 390 NPCs in early G1 and 2,830 ± 740 NPCs in late G2. We did not detect a phase of higher NPC formation rate during early S phase as reported for HeLa S3 cells (Maul et al., 1972). Our data indicate that the total number of NPCs at the end of G2 does not account for all NPCs formed in both daughter nuclei by early G1. Postmitotic NPC assembly therefore either has to rely on preexisting nucleoporin pools that were not assembled into NPCs in the preceding interphase or nucleoporins newly synthesized immediately after mitosis. Previous work in fixed cells showed an approximate doubling of NPC numbers from ~2,000 in G1 to ~4,000 in G2 in HeLa cells (Maul et al., 1972). The differences in our findings are likely a result of variations between different cell lines in NPC assembly during interphase, which may be related to cell cycle length and metabolic rates.

In our long-term high resolution observations of NPCs on the nuclei of individual cells (Fig. 1 A), new NPCs appearing between preexisting ones could be identified. We next wanted to determine whether interphase assembly occurred with the same kinetics as postmitotic assembly. Our recent characterization of the kinetics of postmitotic NPC assembly in the same cell type (Dultz et al., 2008) provided a quantitative basis for a direct comparison. To determine the kinetics with which individual new NPCs are assembled from their subcomplexes in interphase, NPCs of cells expressing GFP-tagged nucleoporins were imaged on the bottom surface of the nucleus. In agreement with previous work (D’Angelo et al., 2006), we observed new NPC assembly events at previously pore-free sites independent of the position of preexisting NPCs (Fig. 2) and could not find any evidence for “pore-splitting” events in our analysis of a total of ~150 single NPC assembly events.

We first analyzed the assembly kinetics of two major soluble NPC subcomplexes, the Nup107–160 complex (represented by GFP fusions of Nup107 and Nup133) and the Nup93 complex (represented by GFP fusion of Nup93). The Nup107–160 complex is essential for NPC formation both after mitosis and in interphase (Walther et al., 2003a; D’Angelo et al., 2006). We observed that the accumulation of the Nup107–160 complex at sites of new NPC assembly from background nuclear envelope levels to levels comparable to neighboring preexisting pores was completed within ~15 min (Fig. 2 A). Similar results were found for the Nup93 complex (Fig. 2 B), which is also essential for correct NPC assembly and provides interaction between soluble and transmembrane components of the NPC (Mansfeld et al., 2006; Hawryluk-Gara et al., 2008). Fitting of the assembly kinetics with a sigmoid function allowed us to determine characteristic assembly times. These were not significantly different for the Nup107–160 complex and Nup93 with 12 ± 8 min (n = 21) and 9 ± 10 min (n = 14), respectively (P = 0.2 for unpaired t test; Fig. 3, A and B), indicating that these two essential soluble NPC subcomplexes assemble into NPCs with similar kinetics.

We then measured the assembly kinetics of the integral membrane nucleoporin Pom121, which is also essential for NPC assembly (Antonin et al., 2005). In contrast to the relatively rapidly assembling soluble components, the majority of NPC assembly events observed for Pom121 were characterized by a gradual accumulation process that could last several hours before the newly forming NPCs reached similar intensities as neighboring preexisting NPCs (Fig. 2 C). The characteristic assembly time determined for Pom121 was 29 ± 15 min (n = 28). Interestingly, a high variability in the assembly kinetics of Pom121 between individual newly assembling NPCs was apparent (range of characteristic assembly time from first to third quartile was 15–44 min; Fig. 3, A and B). We occasionally observed fast events also for Pom121 consistent with an individual rapid NPC formation event reported previously for this nucleoporin (D’Angelo et al., 2006).

Overall, the NPC assembly process during interphase is thus significantly slower than after mitosis, in which all NPCs of the daughter nucleus form within ~10 min (Dultz et al., 2008). We were able to directly compare the kinetics between interphase and postmitotic assembly by fitting our previously published quantitative data for the assembly kinetics of Nup133 (representing the Nup107–160 complex), Nup93, and Pom121 at the end of mitosis (Dultz et al., 2008) with the same sigmoid function. This yielded mean characteristic postmitotic assembly times of 1.9 ± 0.5 min for Nup133, 0.75 ± 0.17 min for Nup93, and 1.5 ± 0.2 min for Pom121 (n = 5 for all nucleoporins; Fig. 3 C and Fig. S2). Although the precision of the comparison is somewhat hampered by the lower time resolution necessary to capture the slow and rare single NPC assembly events in interphase, the difference in the overall duration of the assembly process between interphase and mitosis by about one order of magnitude is striking (Fig. 3, B and C). This longer time frame of NPC assembly in interphase might be, for example, a result of the requirement for fusion between inner and outer nuclear membranes.

In addition to NPC assembly events, we occasionally observed events of apparent NPC disassembly in which a fluorescent signal of indistinguishable appearance from neighboring NPCs disappeared (Fig. 2 D and Fig. S3). Disappearance of the signal in all observed cases occurred very rapidly, but the events were too rare to characterize their kinetics in detail. Such “disassembling” pores were also observed in cells coexpressing Nup107 and Pom121 tagged with different fluorescent proteins, indicating that they indeed represent NPCs (Fig. S3 B). In these double-labeled events, both signals decreased simultaneously (at a 15-min time resolution), indicating that disassembly is more rapid than assembly also in interphase, similar to our previous observations in mitosis (Dultz et al., 2008). Previous analysis in live cells indicated that NPCs are extremely stable in interphase (Daigle et al., 2001; Rabut et al., 2004) and can remain assembled and functional for the entire lifetime of differentiated cells in Caenorhabditis elegans (D’Angelo et al., 2009). Our observations suggest that NPCs may be able to disassemble spontaneously during interphase at low frequency. It will be interesting to study this process further in the future when advances in imaging technology will allow capturing these rare and transient events reliably.

To gain insight into not only the kinetics but also the order of interphase NPC assembly, we next asked whether the slow accumulation of the integral membrane protein Pom121 at the assembly site is an early or late step of assembly. At the end of mitosis, NPCs assemble in a stereotypic and highly ordered process in which the different nucleoporin subcomplexes are sequentially recruited to the reforming nuclear envelope (Bodoor et al., 1999; Dultz et al., 2008). Assembly is initiated by the binding of the Nup107–160 complex to chromatin upon anaphase onset, where it provides the assembly seeds for NPCs that then recruit transmembrane nucleoporins like Pom121 (Dultz et al., 2008; Rasala et al., 2008). By analyzing cells expressing Pom121 and Nup107 tagged with GFP and mCherry, respectively, resulting in double-labeled pores, we found that this order is inverted in interphase. Of 34 double-labeled NPC assembly events, 28 (82%) started with the signal of Pom121 appearing before the signal of Nup107, 5 showed simultaneous appearance of both nucleoporins (at a time resolution of 15 min), and only once was the signal of Nup107 above the significance threshold (see Materials and methods section Measurements of pore formation kinetics) before Pom121 (Fig. 4 C). There was no difference in this assembly order after swapping the colors between Pom121 and Nup107, excluding an influence of the different fluorescent proteins on the detection of the tagged proteins. Fitting of the assembly kinetics and temporal alignment using the half-maximal intensity of Nup107 as a reference clearly shows the assembly of the transmembrane protein Pom121 before Nup107 with a mean delay of 45 ± 36 min between the half-maximal intensities of Pom121 and Nup107 (Fig. 4 B). Our data thus provide direct evidence that during interphase, Pom121 assembles into newly forming NPCs before Nup107, as proposed by Doucet et al. (2010) based on indirect immunofluorescence data. Surprisingly, the order of recruitment of these two essential subcomplexes for NPC assembly is thus the opposite of postmitotic assembly, demonstrating fundamental mechanistic differences between NPC assembly in the two different cell cycle states. After mitosis, NPC assembly is initiated by the large, soluble Nup107–160 subcomplex by binding to chromatin via ELYS, thus defining the assembly position and acting as a platform for NPC assembly. Our data show that, in interphase, this role is taken over by other NPC components, likely by the transmembrane nucleoporin Pom121. Interestingly, both Nup107–160 and Pom121 have been implicated in providing a checkpoint for nuclear envelope closure of in vitro assembled nuclei (Antonin et al., 2005). They might play a similar role, but in reverse order, during interphase assembly and are then followed by the assembly of peripheral and central channel nucleoporins.