Regulation of functional nuclear pore size in fibroblasts

Molecular exchanges between the nucleoplasm and cytoplasm occur through nuclear pore complexes, which are tripartite structures consisting of (1) a cytoplasmic ring and attached filaments that extend into the cytoplasm (2) a central spoke element and (3) a nuclear ring that is associated with fibers that project into the nucleus and form a conical basket (Allen et al., 2000; Stoffler et al., 1999). The pore complex in vertebrate cells has an estimated mass of 125×10⁶ Da, and is composed of approximately 50 different proteins, referred to as nucleoporins (Ryan and Wente, 2000). Macromolecular transport is restricted to a central channel located within each pore complex (Bayliss et al., 2000; Rout et al., 2000). Proteins targeted for exchange across the nuclear envelope contain nuclear localization signals (NLSs) and/or nuclear export signals (NESs), which serve as binding domains for transport receptors that mediate translocation through the pores (Jans et al., 2000). A total of 14 members of the importin β/karyopherin β receptor superfamily have been identified in yeast, and at least 20 have been reported in vertebrates; thus, different import and export pathways are available for specific proteins or groups of proteins (Nakielny and Dreyfuss, 1999). An essential step in the overall transport process is the controlled assembly and disassembly of the receptor-substrate complexes. This is dependent on the intracellular distribution of the nucleotide forms of Ran (i.e. RanGTP and RanGDP) (Azuma and Dasso, 2000; Yoneda et al., 1999). The GTP form is more concentrated in the nucleus, where it is required for both the dissociation of receptor-substrate complexes imported from the cytoplasm, and also the assembly of transport complexes for export. RanGDP, which does not destabilize the import complex, is the predominant nucleotide form in the cytoplasm.

The functional size of the nuclear pores (or more specifically the central transport channel) is variable, and dependent on the physiological state of the cells (Feldherr and Akin, 1995a). By using colloidal gold particles coated with proteins containing classical NLSs (signals that use the importin α/β pathway) as substrates for signal-mediated transport, it was determined that the upper limit for nuclear import in normal proliferating BALB/c 3T3 cells is approximately 250 Å. In quiescent populations, this value can decrease by up to 100 Å (Feldherr and Akin, 1991), whereas functional pore size increases by about 40 Å in SV40-transformed BALB/c 3T3 cell lines (Feldherr et al., 1992). Changes of this magnitude could significantly alter the transport rates of large mRNP particles and ribosomal subunits.

It was found that the increase in nuclear import in SV40-transformed cells was caused by large T antigen (Feldherr et al., 1992). This conclusion was based on the finding that the microinjection of purified large T antigen resulted in a significant increase in the dimensions of gold particles able to enter the nucleoplasm, comparable with that observed in transformed cells. Experiments using a series of large T mutants suggested that the transport increase is dependent on p53 binding (Feldherr et al., 1994). In the same study, supporting evidence that sequestering p53 alters transport capacity was obtained by transfecting cells with mutant p53, which forms oligomers with wild-type p53, thereby reducing its effective concentration, and also by microinjecting anti-p53 antibodies. Both of these procedures resulted in a significant increase in functional pore size. Subsequently, it was determined that 100,000 g cytosolic extracts from SV40-transformed cells increased functional pore size when microinjected into non-transformed BALB/c 3T3 cells (Feldherr and Akin, 1995b). This activity could be blocked by treating the extracts with a variety of kinase inhibitors, including specific PKC inhibitors. Taken together, the above studies imply that large T antigen initiates a series of events that result in a PKC-dependent increase in nuclear transport, and that p53 is an intermediate in these events.

To further test the above scenario, which implicates p53 as a factor in modulating nuclear transport, we investigated the effects of cycloheximide, and the p53 inhibitor pifithrin-α on nuclear import. Since p53 is a short-lived protein, inhibiting polypeptide synthesis should cause a rapid depletion of endogenous p53, and an increase in functional pore size. The present study shows that this is the case. A 3 hour treatment with cycloheximide resulted in a significant increase in the nuclear import of large gold particles. In contrast to the results obtained with large substrates, cycloheximide had no significant effect on either the transport rates of small substrates (BSA-NLS conjugates), or the passive diffusion of proteins through the pores. The latter results show that the changes in pore size are coupled to events that occur during signal-mediated transport. Incubation of the cells in pifithrin-α also increased the functional size of the pores.

BALB/c 3T3 A31 cells were purchased from the American Type Culture Collection (Manassas, VA), and cultured in 5% CO₂ at 37°C in Dulbecco’s modified Eagle’s medium containing 4 mM glutamine and 4.5 mg/ml glucose. The culture medium was supplemented with 10% calf serum, 100 U/ml of penicillin G, 100 μg/ml of streptomycin and 0.25 μg/ml of amphotericin B (Fungizone). All reagents were obtained from GIBCO/BRL (Gaithersburg, MD).

Cells that were injected with colloidal gold, and analyzed by electron microscopy, were cultured on gridded ACLAR coverslips (Allied Corp., Morristown, NJ) as described (Feldherr and Akin, 1990). Fluorescent analysis was performed on cells grown on glass coverslips.

Stock solutions of MG-132 (10 nM), and pifithrin-α (10 mg/ml) were prepared in dimethyl sulfoxide. Stock solutions of cycloheximide (10 mg/ml) were made up in ion-free water. Cycloheximide was obtained from Sigma (St Louis, MO); the other inhibitors were purchased from Calbiochem (San Diego, CA).

Colloidal gold particles, ranging in diameter from 80 to 360 Å were prepared by reducing gold chloride with sodium citrate (Frens, 1973). The coating protein, BSA-NLS, was prepared by conjugating bovine serum albumin (Sigma) with a synthetic peptide, CGGGPKKKRKVGG, which contains the SV40 large T NLS (underlined). The peptide was synthesized by the University of Florida protein core facility. The conjugation step was performed using the cross-linker m-maleimidobenzoyl-N-hydroxysuccinimide ester (Pierce, Rockford, IL), as reported (Lanford et al., 1986). It was estimated by SDS-PAGE that each BSA molecule was conjugated with an average of 8 peptides. The gold particles were coated with BSA-NLS as outlined previously (Dworetzky et al., 1988). The protein coat increases overall particle diameter by approximately 30 Å.

To prepare fluorescent BSA-NLS, BSA was first labeled with fluorescein isothiocyanate (Sigma) as previously described (Feldherr and Akin, 1999), and subsequently conjugated with NLS peptide as above. Fluorescein isothiocyanate was also used to label BSA, and ovalbumin (Sigma) for diffusion studies; 260 kDa dextran, labeled with FITC, was purchased from Sigma.

Prior to microinjection, all transport substrates were dialyzed against intracellular medium, 117 mM KCl, 10 mM NaCl, 6 mM K₂HPO₄, and 4 mM KH₂PO₄ (pH 7.0).

Microinjections were performed at 37°C in a CO₂-enriched atmosphere using an inverted Nikon Diaphot microscope, a Narishige hydraulic micromanipulator, and a continuous-flow injection system (Feldherr and Akin, 1990). The micropipets had tip diameters of approximately 0.5-0.7 μm.

Details of the EM procedures have been reported previously (Feldherr and Akin, 1990). Briefly, the cells were fixed in 1% glutaraldehyde, postfixed in 1% OsO₄ and analyzed using a JEOL 100CX electron microscope.

The nuclear import rates of fluorescent-labeled transport substrates were determined using an intensified Hamamatsu CCD camera, and a MetaMorph imaging system (Universal Imaging System, West Chester, PA). At the specified times after cytoplasmic injection, images of the cells were collected using a Nikon PlanApo 60× oil immersion lens, and the N/C ratios were determined by measuring fluorescence in equal areas of cytoplasm and nucleoplasm. To avoid UV damage to the cells, exposure at each time point was less than 1 second and a number 16 neutral density filter was used.

BSA-NLS-coated gold particles, 80 to 360 Å in diameter (not including the thickness of the protein coat) were injected into either untreated BALB/c cells (controls), or cells that had been pretreated with 50 μg/ml cycloheximide for 1 to 18 hours. This concentration of cycloheximide was found to reduce the incorporation of [³H]leucine by approximately 94% in cells that were incubated in the drug for either 6 or 24 hours. Thirty minutes after injection, the cells were fixed and subsequently analyzed using the electron microscope. Both the size distribution of the particles that entered the nucleus, and the N/C gold ratios (based on counts made in equal and adjacent areas of nucleoplasm and cytoplasm) were determined. The results are given in Table 1. Since 200 Å (plus the 30 Å coat) is close to the exclusion limit for normal proliferating fibroblasts, only the percentage of nuclear particles above and below this size are shown. After 1 hour in cycloheximide there was a statistically significant increase in the percentage of large particles in the nucleus. By 3 hours, both the uptake of large particles and the N/C ratio increased significantly. Interestingly, the transport capacity returned to control levels in cells treated with cycloheximide for 6 hours or longer. To test the effect of cycloheximide on cell viability, cultures were incubated in the drug for 6 or 18 hours, at which time they were returned to normal growth medium. Subsequent cell counts demonstrated that the effects of cycloheximide treatment, even for extended periods, was reversible.

As the size of the particles that enter the nucleoplasm approaches the diameter of the transport channel, their passage is increasingly impeded by the frictional resistance or drag at the channel wall. This resistance is a function of the ratio of particle diameter to channel diameter, and increases rapidly with this ratio. A drag factor has been defined as the ratio of the drag on a particle in a pore to the drag in an unrestricted environment. Drag factors for spherical particles in cylindrical channels have been reported (Paine and Scherr, 1975). From the size distribution of gold particles (including an estimated 30 Å coating) injected into the cytoplasm, possible nuclear distributions were calculated using drag coefficients based on trial diameters of the channels. For control cells, the best fit to the data is for pores of diameter 310 Å, and for the cycloheximide-treated cells, the best fit is 370 Å. The fit is fairly robust with channels 10-20 Å larger or smaller clearly not fitting as well. Since drag is always due to interactions at the molecular level, it applies to both mediated transport and passive diffusion; thus, it could be the result of either multiple sequential specific binding or nonspecific interactions.

To study the effect of protein turnover on cycloheximide-dependent increases in transport capacity, cells were incubated for 3 hours in 50 μg/ml cycloheximide alone, or in cycloheximide plus 10 nM MG-132, a proteasome inhibitor. The control cells were not pretreated. Nuclear transport was analyzed using colloidal gold, as described above. As can be seen in Table 2A, the increase in transport capacity that resulted from cycloheximide treatment was blocked by inhibiting proteasome activity. These results indicate that the effect of cycloheximide on pore size is due to the depletion of a short-lived protein; thus, if turnover is prevented by inactivating proteasomes, there is no significant change in transport capacity.

To determine if more than one factor is required to increase the nuclear import of large particles, experiments were performed in which cycloheximide and MG-132, at the above concentrations, were added to the cells sequentially rather than simultaneously. Gold import was assayed in untreated cells (control), cells that were incubated in cycloheximide alone for 6 hours, and cells in which MG-132 was added 1 hour after the start of a 6 hour cycloheximide treatment. The data are shown in Table 2B. Consistent with the results reported above (Table 1), nuclear import returned to the control level after 6 hours in cycloheximide. With the addition of MG-132, the percentage of large particles present in the nucleus remained significantly higher than in the controls (i.e. equivalent to the uptake at 3 hours in cycloheximide alone) and, although not statistically significant, there was an increase in the N/C ratio. These data support the view that the increase in functional pore size is initiated by a short-lived protein during the first hour of cycloheximide treatment (Table 1), but requires at least one additional protein which, in the absence of MG-132, is degraded by 6 hours.

The intracellular distributions of FITC-BSA-NLS, 1 and 60 minutes after injection into control cells and cells treated with 50 μg/ml cycloheximide for 3 hours, are shown in Fig. 1A (left). The N/C fluorescence ratios were quantified at the time points indicated, and are plotted in Fig. 1B. There was no significant difference in nuclear import kinetics in the control versus the experimental cells, demonstrating that cycloheximide had no effect on the signal-mediated transport of a smaller substrate (approximately 70 Å in diameter), which is appreciably below the exclusion limit of the nuclear pores. Similarly, a 3 hour treatment with cycloheximide had no significant effect on the passive diffusion of either FITC-labeled ovalbumin or BSA (Fig. 1A, right; Fig. 1B). The high N/C ratios obtained after 1 minute in the diffusion experiments were due to background fluorescence from overlapping and underlying cytoplasm. This was demonstrated by injecting FITC-labeled, 260 kDa dextran, which is too large to diffuse across the envelope, but still had an initial N/C ratio of approximately 0.6.

Diffusion of proteins through the transport channel is impeded by both steric factors and drag. The steric factor is a measure of the difficulty for a particle to enter a channel of comparable dimensions. Assuming these impediments, an effective channel diameter of 90 Å has been previously calculated (Paine et al., 1975). If the channel diameter increased by 60 Å, as indicated by the gold experiments, drag and steric interactions would be much reduced, and the diffusion of BSA would be expected to increase by about 100-fold and ovalbumin by 17-fold. No such increase was observed, demonstrating that the increase in the signal-mediated transport of large gold does not reflect an overall increase in functional pore diameter.

As a further test for the possible involvement of p53 in the regulation of transport capacity, experiments were performed using the inhibitor pifithrin-α (Komarov et al., 1999). The nuclear uptake of colloidal gold was examined in control fibroblasts, and cells that were incubated in 30 μM pifithrin-α for 3 hours. The results (Table 3) show that exposure to pifithrin-α significantly increased functional pore size (both the percentage of large particles that entered the nucleus and the N/C ratio) to approximately the same degree as cycloheximide treatment. However, the same pifithrin-α treatment (30 μM for 3 hours) had no significant effect on the nuclear import rate of FITC-BSA-NLS, compared with controls (Fig. 2).

It was initially found that treatment of fibroblasts with cycloheximide produced an increase in the functional size of the nuclear transport channel. Further studies, in which cycloheximide was used in conjunction with a proteasome inhibitor, demonstrated that the increase in functional pore diameter, estimated to be approximately 60 Å, was initiated by a short-lived protein and, in addition to the initiating protein, at least one other factor is required to modulate transport capacity. Although cycloheximide significantly increased the nuclear uptake of large gold particles, it had no effect on either the import rates of smaller NLS-containing substrates, or passive diffusion, indicating that the drug treatment did not alter the activity of the soluble nuclear transport factors, or the overall diameter of the transport channel.

Although conclusive evidence is not available, a possible candidate for the initiating factor is the short-lived tumor suppressor, p53. As discussed in the Introduction, antibodies to p53 as well as transfection with mutant forms of p53 cause a functional increase in pore diameter, similar in magnitude to those reported here. The results obtained with pifithrin-α, which blocks p53-dependent transcription activation, provide further support for the involvement of p53 and indicate that a factor downstream of p53 is required to enhance transport.

Our previous results (see Introduction) that extracts from SV40-transformed BALB/c 3T3 cells can increase functional pore size when microinjected into non-transformed cells, and that the increase can be blocked by PKC inhibitors, raises the possibility that cycloheximide-dependent transport effects might also require kinase activity. Unfortunately, experiments designed to test this hypothesis were inconclusive. Cells were simultaneously treated with cycloheximide and various kinase inhibitors, a procedure that prevented an increase in the transport of large gold particles; however, 3 hour exposures to kinase inhibitors alone also altered transport capacity, making it difficult to evaluate their specific effects on cycloheximide activity.

It is currently believed that signal-mediated translocation through the nuclear pore complex occurs by sequential steps of adsorption and desorption of the receptor-substrate complex to FG repeat nucleoporins that are associated with the central transport channel (Bayliss et al., 2000). However, the mechanism that restricts the passive diffusion of molecules approximately 25 Å in diameter and above (Paine et al., 1975), but allows NLS-containing substrates as large as 250 Å in diameter to rapidly enter the nucleus, remains controversial. Theoretically, this could be accomplished in several ways. One possibility is a mechanical gating system, which could involve either two separate gates, located at either end of the transport channel (Akey, 1990; Kiseleva et al., 2000), or a single gate, positioned in the middle of the channel (Feldherr and Akin, 1997). Activation of a gating system by the receptor-substrate complex could allow selective translocation of large, signal-containing substrates, while restricting diffusion of molecules that are not targeted for exchange. Rout et al. introduced the concept of a virtual gate (Rout et al., 2000). This model is based on a detailed study of yeast nuclear pore complexes, which indicated that both ends of the transport channel are encircled by filamentous FG-containing nucleoporins. It was suggested that these filamentous proteins undergo active Brownian movement and present a formidable barrier to the diffusion of large molecules that lack FG-binding domains; however, substrates that contain FG-binding sequences (e.g. transport receptors) would be transiently retained, thus facilitating diffusion through the central channel. Most recently, a ‘selective phase model’ has been proposed for exchange through the pores (Ribbeck and Gorlich, 2001). According to this model, the central transport channel is occupied by a low density meshwork of FG-repeat nucleoporins. It is suggested that this meshwork would function as a semi-liquid phase, in which transport complexes containing FG-binding sites could partition and, thus, translocate across the pores. The finding, in this investigation, that changes in functional channel size are not accompanied by overall changes in the physical characteristics of the transport channel (as determined from diffusion studies), does not appear to distinguish among the above possibilities. However, it does imply that, as a result of cycloheximide treatment, there is an increased capacity of the gate to dilate, whether it be mechanical or virtual. Presumably, this would involve conformational changes of the nucleoporins, which are initiated by their interaction with the receptor-substrate complex. It is also possible that the degree of dilation is regulated by the phosphorylation state of the nucleoporins.

Although all models of the pore complex incorporate a centrally located pathway for the translocation of macromolecules, the specific nature of the pathway is still in question. Several investigators have proposed that transport occurs through a cylindrical transporter element. In the model developed by Akey and Radermacher, for example, the transporter element is represented as a hourglass-shaped structure with minimum and maximum outer diameters of 320 and 420 Å, respectively (Akey and Radermacher, 1993). The wall, at its thinnest point, is approximately 75-90 Å thick. Alternatively, it is possible that the central pathway is simply an open channel through the pore complex, rather than a distinct structural element (Stoffler et al., 1999). We have calculated that channel diameters of approximately 310 Å and 370 Å in normal and cycloheximide-treated cells, respectively, would be necessary to account for the observed transport of large gold particles. It would appear that the open channel model would be more compatible with these size requirements than a more restricted pathway associated with a transporter element.

Future investigations directed toward understanding the overall significance, and the molecular events that regulate the functional dimensions of the central transport channel should include the following: (1) establishing whether variations in pore size can occur in all cells, or are restricted to specific cell types; (2) identifying the factor(s) downstream of p53 that activates the subsequent events responsible for increasing pore size (possible candidates for this factor include cyclin G, p21 and mdm2); and (3) determining whether cycloheximide and pifithrin-α treatments are accompanied by changes in the phosphorylation patterns of nucleoporins. This data might help identify specific pore components that are directly involved in modulating pore size.

This work was supported by grant MCB-9723015 from the National Science Foundation.