Reversibility in nucleocytoplasmic transport

Abstract

Nucleocytoplasmic exchange of proteins and RNAs is mediated by receptors that usher their cargo through the nuclear pores. Peptide localization signals on each cargo determine the receptors with which it will interact. Those interactions are normally regulated by the small GTPase Ran. Hydrolysis of GTP provides the chemical energy required to create a bona fide thermodynamic pump that selectively and directionally accumulates its substrates across the nuclear envelope. A common perception is that cargo delivery is irreversible, e.g., a protein imported to the nucleus does not return to the cytoplasm except perhaps via a specific export receptor. Quantitative measurements using cell-free nuclei reconstituted in Xenopus egg extract show that nuclear accumulation follows first-order kinetics and reaches steady state at a level that follows a Michaelis–Menten function of the cytoplasmic cargo concentration. This saturation suggests that receptor-mediated translocation across the nuclear pore occurs bidirectionally. The reversibility of accumulation was demonstrated directly by exchange of the cytosolic medium and by fluorescence recovery after photobleaching. Based on our results, we offer a simple biophysical model that predicts the observed behavior. A far-reaching consequence is that the nuclear localization signal dictates the fate of a protein population rather than that of the individual molecules that bear it, which remain free to shuttle back and forth. This implies an open communication between the nucleus and cytoplasm and a ubiquitous mechanism for signaling in both directions.

Nuclear pores are large protein channels that traverse the lipid bilayers of the nuclear envelope (NE) (1–3). They accommodate a wide range of essential molecular traffic in the cell. The biochemical mechanism for nucleocytoplasmic transport depends on protein receptors, primarily of the importin/karyopherin family (4–6). These provide molecular specificity for nuclear import or export of proteins and RNA–protein complexes. Peptide signal sequences [nuclear localization signals (NLSs) and nuclear export signals (NESs)] on the cargo govern interaction with the appropriate receptors, which are classified as importins or exportins according to the direction of net transport. Because of specific interactions between a receptor and phenylalanine–glycine repeat motifs believed to line the nuclear pore complex channel (7–10), this receptor–cargo complex may pass through the pore whereas translocation of the bare cargo is strongly hindered. The regulating switch, Ran, is maintained in the nucleus primarily in the GTP form by the chromatin-bound guanosine exchange factor RCC1. For nuclear import, the receptor–cargo complex is disrupted by a competitive exchange with RanGTP (11), which leaves the cargo inside the nucleus. The newly formed receptor–RanGTP complex then returns to the cytoplasm, where the GTPase activating protein (RanGAP) accelerates hydrolysis of the GTP and so liberates the receptor to bind another cargo. Such cycles explain the capacity of the nuclear pore, as a system, to raise the nuclear concentration of a cargo above that in the cytoplasm, even when the cargo is more abundant than the receptor. Nuclear export is a related process, but in this case an exportin receptor binds a nuclear export signal-bearing protein cooperatively with RanGTP so that the direction of substrate accumulation is reversed.

The kinetics and energetics of transport have received considerable attention, with most of the focus on translocation per se of the receptor–cargo complex through the nuclear pore. Several studies show that translocation is an essentially passive process of facilitated diffusion, decoupled from GTP hydrolysis (12–15). In that case, vectorial transport (2, 16, 17) suggests a physically forbidden device that reduces entropy with no cost in energy, i.e., a Maxwell demon. At least one work shows that for large substrates hydrolyzable GTP is required for translocation (18); this raises the possibility of a rectifying gate at the pore. Irreversible passage suggests, however, that a nuclear concentration should rise indefinitely, as long as cargo and GTP are available. Although biochemical feedbacks might ensure that this does not occur in a given cell or for a given substrate, the universality of nucleocytoplasmic transport and its tolerance to artificial cross-species exchange of major factors suggest a more fundamental paradigm.

Results

To approach the nuclear transport process from a thermodynamic perspective, we adopted cell-free nuclei reconstituted from Xenopus laevis egg extract. This is a classic functional in vitro model for nuclear assembly and transport studies (19–21). Reconstituted nuclei are suspended in a much larger cytosolic volume; this situation occurs in several natural cases including multinucleate cells, e.g., skeletal muscle, and in the syncytia of Drosophila larvae and plant endosperm. The cytosol thus serves as a thermodynamic reservoir for transport cargo and carriers. Precisely these conditions, where nuclear accumulation does not deplete the cytosol, enable unambiguous thermodynamic interpretation of transport measurements.

As a model transport cargo we used nucleoplasmin (NP), an abundant nuclear protein carrying classical bipartite NLSs (22). Its nuclear import depends on interaction with the importin α/β dimer (23). A GFP fusion (GFP-NP) was used in all experiments reported here. The molecular mass of the pentameric GFP-NP, 335 kDa, is far larger than the canonical cutoff of ≈40 kDa for passage across the nuclear pore by simple diffusion; hence, its translocation must be facilitated by a receptor.

Measurements were conducted by using a custom-built multimode microscope, sketched in Fig. 1A. It provides both a conventional wide-field epifluorescence mode and a fluorescence correlation spectroscopy (FCS) (24) mode with simultaneous differential interference contrast imaging. The latter was used to observe the precise location where local confocal fluorescence intensity was measured quantitatively. FCS confirmed the existence of a monodisperse population of GFP-NP with Stokes diameter 9.8 nm (Fig. 1B); for comparison, for GFP alone we measure a Stokes diameter of 2.7 nm (data not shown). Calibration of local GFP-NP concentration (by repeated dilution steps in an extract over four orders of magnitude 1 nM to 10 μM, up to ≈1 μM by FCS) (Fig. 1B Inset) with confocal fluorescence intensity verified a linear relation. Fig. 1C shows a typical import assay with a differential interference contrast image of a nucleus together with accumulation of fluorescent GFP-NP and exclusion of rhodamine-labeled dextran to verify integrity of the NE.

Fig. 1.

Experimental system. (A) Diagram of the multimode microscope, which incorporates transmission differential interference contrast imaging with simultaneous confocal epifluorescence intensity and FCS measurements, as well as wide-field epifluorescence imaging to a cooled CCD camera. A Zeiss ×40/N.A. 1.2 C-Apochromat water-immersion objective was used for all measurements. APD, avalanche photodiode; HID, metal-halide discharge lamp. See a detailed description in SI Text. (B) FCS was used to verify molecular mobility of the nuclear import substrate GFP-NP and to calibrate its local concentration in the nucleus (high concentration, 360 nM, in green) or cytosol (low concentration, 30 nM, in blue). Inflections at 1.00 and 0.96 msec, respectively, indicate similar diffusion constants in the two environments. Both fits are improved by an anomalous diffusion model, 〈X²〉 ∞ t^γ, where γ = 0.85 and 0.75, respectively (data not shown). (Inset) Linearity of confocal fluorescence intensity with GFP-NP concentration in Xenopus egg extract. Filled circles represent quantitative calibration by FCS, and stars show concentration estimated by dilution. (C) Nuclear accumulation of GFP-NP (green, Left), differential interference contrast imaging (Center), and exclusion of tetramethyl rhodamine-conjugated dextran (red, Right) to verify integrity of the NE.

To test the capabilities of the nuclear transport machinery, we probed the nuclear concentration of GFP-NP kinetically over a wide range of cytosolic concentrations. Fluorescent transport cargo (GFP-NP) and a fresh ATP regenerating system were added to an aliquot of reconstituted nuclei. The mixture was then introduced into an observation slide and mounted on the microscope, typically within 1 min. The precise concentration, [C]^C, of GFP-NP in the cytosolic reservoir was monitored and confirmed to remain constant. Accumulation curves of fluorescent cargo within the nucleus [C]^N were recorded, covering a range of [C]^C from 4 nM to 9 μM. Representative curves appear in Fig. 2. They show kinetics of a simple first-order process, i.e.

characterized by two parameters: [C]_SS^N is the nuclear cargo concentration reached in steady-state, and β is the initial rate of accumulation [see supporting information (SI) Movie 1]. Both parameters depend on the experimental variable [C]^C as depicted in Fig. 3.

Fig. 2.

Nuclear accumulation kinetics. (A) Representative curves showing the time course of nuclear concentration for dilute cargo reservoirs (36 nM to 180 nM). Dotted lines are fits to simple first-order kinetics. (B) Representative curves showing the nuclear concentration, [C]^N, normalized by the cytosolic concentration, [C]^C. Up to 1.3 μM the kinetics were similar (shaded circles, 36 nM; others as marked). At higher concentrations both initial rates and steady-state ratios decreased. RanQ69L inhibited nuclear accumulation, which settled at a concentration ≈40% of the cytosolic concentration ([C]^C = 400 nM).

Fig. 3.

M-M behavior of GFP-NP accumulation. (A) The initial accumulation rates β are plotted here as a function of the cytosolic reservoir concentration, [C]^C, after normalization to account for the nuclear volume. The dotted line shows the M-M curve fit: β = β_max([C]^C/K_m + [C]^C), with β_max = 12.7 nM/sec and K_m = 2.7 μM. (B) Steady-state nuclear cargo concentration is plotted as a function of the experimentally fixed cytoplasmic concentration, [C]^C. Again the data follow a M-M form: [C]_SS^N = [C]_SS^max([C]^C/([K]_m^SS + [C]^C) with [K]_m^SS = 1.7 μM and [C]_SS^max = 27.6 μM. Vertical bars in both plots represent standard deviations from the number of measurements indicated. Horizontal bars show the concentration range over which the measurements were pooled for clarity.

The nonhydrolyzing RanQ69L mutant is a well known inhibitor of nuclear import. On addition of RanQ69L-GTP to an assay with [C]^C = 400 nM, [C]^N reached steady-state at ≈40% of [C]^C; i.e., GFP-NP was effectively excluded (Fig. 2B).

The initial rates β depend on [C]^C in a Michaelis–Menten (M-M) form (Fig. 3A). Two regimes appear: at low concentration the initial uptake rate is simply proportional to [C]^C, whereas at higher concentrations β becomes independent of [C]^C. This behavior is expected in a receptor-mediated process in analogy with enzyme–substrate kinetics (25). The M-M behavior of β, together with the inhibition of import by RanQ69L, confirm that nucleocytoplasmic transport in our model system acts as an active molecular pump that works against a concentration gradient.

Fig. 3B shows the intranuclear concentrations [C]_SS^N reached in steady state as a function of [C]^C. These also display a M-M relation, with the transition between low and high concentration behavior occurring at 1.7 μM. For lower [C]^C, the nuclear cargo concentration [C]_SS^N is ≈12 times the cytosolic. This ratio drops to 3 at the highest working concentrations accessible in our preparation. By extrapolation of the curve, the data predict that by [C]^C = 27.6 μM, [C]_SS^N will reach unity and further accumulation will no longer occur. Thus, the pump becomes less effective as its load is increased, as is reasonable for any physical machine.

Our data suggest an alternate paradigm for nucleocytoplasmic transport: steady state occurs when forward and reverse reactions reach detailed balance. This corresponds to saturation of the accumulation curves in Fig. 2 and implies that inward and outward fluxes of the receptor–cargo complex are equal. Thus, we should expect a bidirectional exchange of cargo mediated by the importin receptor itself; NLS cargoes may leave the nucleus via equilibration of receptor–cargo complexes across the NE. To probe this counterflow transport pathway more directly, we performed two further sorts of experiments.

In one test we imposed a sudden change in the reservoir concentration of cargo by mixing with a second extract. Nuclei that had reached their steady-state accumulation in the cytosol with a certain cargo concentration were mixed into a new cytosol with lower cargo concentration. Experiments were conducted for [C]^C < [K]_m^SS so that [C]_SS^N was simply proportional to [C]^C. After a kinetic delay of variable duration, the nuclear cargo concentration dropped and then resettled at a new, lower value (Fig. 4A and see SI Movie 2). The ratios of both nuclear and cytosolic cargo concentrations, before and after dilution, were equal (Fig. 4B). Thus, the very same proportionality of nuclear to cytoplasmic concentration was restored after the dilution. Dextran exclusion verified NE integrity throughout the experiments.

Fig. 4.

Counterflow of GFP-NP on dilution. (A) Nuclei that had reached steady state in accumulation of GFP-NP ([C]^C = 600 nM) were transferred to a new extract with [C]^C reduced by a factor of 5. GFP-NP left the nucleus, and its concentration resettled at a new steady state. (See also SI Movie 2.) (B) The counterflow assay was performed for dilutions factors of 2 (circles), 2.5 (squares), and 5 (triangles), all beginning from 600 nM original cytosolic concentration of GFP-NP. In each case the nuclear concentration dropped by the same factor, so their ratios follow a line of slope unity (dashed line).

In a second test we studied the flux balance with minimal intervention by following kinetics of fluorescence recovery in the nucleus after photobleaching (Fig. 5A; see also SI Movie 3). Although this technique is normally applied to measure diffusivity in a single medium, here we bleached an isolated chamber, the nucleus, and detected the influx of fresh fluorescent proteins from the cytosolic reservoir. We allowed an import assay to proceed to a concentration approaching saturation and then photobleached the accumulated cargo at a single point. The nucleus darkened throughout, confirming the mobility of the tracer cargo. Photobleaching rendered the accumulated nuclear GFP-NP invisible, causing a downward shift of the measured fluorescence intensity in the nucleus (bleach step, Fig. 5A). The darkened proteins remain otherwise unchanged, however. If cargo accumulation were irreversible, then on recovery the accumulation curve would continue to follow its original exponential approach to saturation according to Eq. 1, minus the shift (dashed line in Fig. 5A). Strikingly, we observed that reaccumulation of fluorescent GFP-NP instead retraced the original curve closely, as if from time 0. A similar recovery was observed in >30 assays with [C]^C ranging from 4 nM to 1.5 μM. The total concentration of dark and bright GFP-NP is not sensitive to the bleaching manipulations, so their sum must continue to follow the first-order kinetics appropriate to the relevant [C]^C. Therefore, the fast influx of fluorescent cargo requires a concomitant efflux of the darkened molecules. The photobleaching assay thus provides noninvasive, unambiguous evidence for bidirectional exchange across the nuclear pores.

Fig. 5.

Photobleaching demonstrates counterflow kinetics noninvasively. Nuclear accumulation of transport substrates [GFP-NP (A), 1.3 μM; NLS-2×GFP (B), 1.5 μM; GFP-hnRNP A1 (C), 1.5 μM] approached saturation with first-order kinetics. After photobleaching, the reaccumulation of fresh fluorescent substrate followed the original uptake kinetics closely, as opposed to continuing the saturation curve that would be expected for irreversible, strictly vectorial import (dashed line in A). (See also SI Movie 3.)

To test whether the counterflow might be peculiar to GFP-NP, we repeated the photobleaching assay using a NLS-2×GFP hybrid (molecular mass ≈ 65 kDa), and a GFP fusion to heterogeneous nuclear ribonucleoprotein (hnRNP) A1 protein (molecular mass ≈ 65 kDa). The former addresses the same importin α/β dimeric receptor pathway as NP, but as a specifically engineered protein it is unlikely to contain cryptic localization signals in addition to the NLS. The latter is a canonical substrate of the monomeric transportin (karyopherin β2) receptor, binding via the M9 NLS (26). Both of these behaved similarly to the GFP-NP: recovery followed the original accumulation curve (Fig. 5 B and C). This provides further confirmation that the machinery of nucleocytoplasmic transport achieves its accumulation in the presence of a constant bidirectional exchange.

A final probe of this exchange was performed by combining the photobleaching and dilution assays with RanQ69L. With this we explored the hypothetical possibility that the balanced steady-state counterflow of GFP-NP might be mediated by distinct export receptors having identical kinetics to the import receptors recognizing the NLS. Because Ran-GTP generically binds exportin-type receptors cooperatively with nuclear export signal cargo, the mutant Ran does not block exportin-mediated translocation of the nuclear pore (27). Indeed, after photobleaching we observed reaccumulation only to 40% of the cytoplasmic concentration (SI Fig. 6A), similar to the case where RanQ69L was introduced before the initial accumulation (Fig. 2). Thus, the recovery was receptor-dependent. When nuclei that had reached steady-state were supplemented with RanQ69L and then diluted 10-fold into cargo-free extract, so that [C]^N/[C]^C stood at 130, no outward leak was observed over 10 min. On photobleaching, the same 40% recovery in [C]^N/[C]^C occurred (SI Fig. 6B). These results are inconsistent with a role of exportin in mediating the counterflow of GFP-NP.

Discussion

Cell-free reconstituted nuclei provided an ideal experimental model with which to clarify the underlying biophysical mechanism of nuclear accumulation. First-order kinetics (Fig. 2) clearly showed that the nuclear concentration reaches a steady-state value, rather than continuously increasing, even though the total quantity of GFP-NP in the large cytosolic volume greatly exceeded the quantity in the nuclei. Preservation of the same functional form also indicated that accumulation was mediated by a single mechanism over the tested range spanning three orders of magnitude. The prior stabilization of the import system, followed by sudden addition of cargo to a constant cytosolic concentration, allowed us to saturate the uptake kinetics and observe the M-M behavior. This was anticipated but not possible to achieve in cultured cells (28), perhaps because the need to stop and “restart” transport with metabolic poisons dictated the use of small substrates that were able to pass the pore autonomously as well as by receptor-mediated transport.

The measured ratio of nuclear to cytoplasmic concentrations also obeyed M-M behavior in the steady-state concentration (Fig. 3B). This had not been anticipated previously and cannot be reconciled with the perception of nuclear import as a unidirectional process. Recalling that the cytosolic pool serves as a reservoir of receptors and cargoes, accumulation would not be expected to reach steady state (except perhaps by energy depletion), and only the kinetics might depend on the cytosolic cargo concentration, [C]^C. Reversibility of the accumulation process was demonstrated directly by dilution of transport cargo in the cytosolic medium (Fig. 4). Most notably, a new steady state was established at just the same nuclear concentration, [C]_SS^N, that would have been reached on accumulation with the final cargo concentration. The transport system thus settled at a true thermodynamic endpoint, which might be reached by a variety of kinetic paths.

At steady state the net molecular flux across the pores is zero by definition. This might occur for any rate of transport events in either single direction, provided that the rate in the opposite direction is equal. By photobleaching the nucleus (Fig. 5) we could distinguish the two directional fluxes. The already-accumulated cargo darkened, so that the robust recovery of fluorescence in the nucleus was due to fresh protein that entered from the cytosol. The bleached molecules are biochemically indistinguishable from the unbleached, however, so the fast recovery of fluorescence is possible only in the presence of an equally fast counterflow to remove the darkened cargo. Moreover, the rate of passage in either direction was found to be similar to the initial rate of accumulation (β in Fig. 3) when the transport substrate was first introduced, i.e., just the maximum rate suited to the relevant cytosolic concentration.

Model

Based on these experimental observations we offer a minimalistic analytical model of nucleocytoplasmic transport, described in detail below and in SI Fig. 7. The universality of nucleocytoplasmic transport suggests that the fundamental mechanism may be understood more broadly than the details of its specific biochemical implementation in a particular species. We consider the nucleus as a chemical reactor, within which an exchange reaction takes place among the transport receptor (T), RanGTP (R), and cargo (C): [TC]^N + [R]^N ⇌ [C]^N +[TR]^N. (The relevant kinetic parameters incorporate the subsidiary interactions [T]^N + [C]^N ⇌ [TC]^N and [T]^N + [R]^N ⇌ [TR]^N implicitly.) The nuclear level of RanGTP, [R]^N, is maintained by RanGEF and the RanGDP-specific transport receptor NTF2. RanGTP is required for all nucleocytoplasmic transport, not only the experimental probe, so its level is buffered and constant. Reactants pass in and out of the vessel in the company of transport receptors. Namely, [TC]^N and [TR]^N are the only chemical species that can communicate with the outside environment via the pores. In the cytoplasm, RanGAP activity catalyzes the reaction [TR]^C → [T]^C + [R′]^C + Pi, where [R′] represents the RanGDP form and Pi is the dephosphorylation product. This is the essential energy-consuming, irreversible step that holds the exchange reaction out of equilibrium. For simplicity we assume that all cytoplasmic Ran is RanGDP. [TC]^C can then reach a simple equilibrium in relation to [C]^C:

with [K]_m^TC the affinity constant for the reaction [T] + [C] ⇌ [TC]. For low [C]^N (initial accumulation stages), influx takes place into an environment that is nearly empty of [TC]^N because of the exchange reaction in favor of [TR]^N. With the experimental insights of Fig. 3 now in hand, we note the second implication of the exchange reaction: as accumulation progresses and the nuclear concentration, [C]^N, rises, this species will begin to compete with [R]^N for binding to the transport receptor. A balance is reached when the higher affinity of RanGTP for the receptor is compensated by the higher concentration of soluble nuclear cargo. The restored [TC]^N is then indistinguishable from a freshly arrived complex that has not yet been disrupted by Ran. It is therefore free to return to the cytoplasm, where it will partition according to Eq. 2. A similar suggestion was offered in the context of computer simulation, although its consequences were not fully explored (29).

The analysis can predict the dependence of [C]^N on [C]^C if we focus on the receptor–cargo complex. The total rate of change of [TC]^N is the sum of its net flux across the NE and the kinetics of the exchange reaction. At steady state both of these terms must be zero, so that

where p_in and p_out are kinetic permeabilities of the nuclear pore in the inward and outward directions, and k_on and k_off refer to the exchange reaction. The final nuclear cargo concentration [C]_SS^N is determined, in the model as in the experiment, by the external cytosolic concentration, [C]^C. Combining with Eq. 2, we obtain

and recognize the M-M form in the bracketed terms. The unbracketed terms in Eq. 4 describe [C]_SS^max. These have a numerical influence on the accumulation ratio but do not introduce a dependence on [C]^C. For a simple symmetric channel p_in = p_out so the first factor would be unity; this may or may not be the case for the nuclear pore. K^TCR in the second term is the relative affinity of the transporter for RanGTP versus the NLS of interest, i.e., it is the “chemical strength” of the NLS. The pump is driven energetically by a steady resupply of nuclear RanGTP, [R]^N, and hindered to the extent that transporter–RanGTP complexes [TR]^N accumulate in the nucleus. Finally, the nuclear accumulation of NLS cargo depends on the cytoplasmic concentration of available transport receptors [T]^C. Note that although increasing [T]^C may enhance import kinetics, as confirmed recently using permeabilized cells (30), in a complete transport model its effect on steady-state accumulation may equally be inhibitory because of nuclear partitioning as [TR]^N. This possibility was identified by simulation and confirmed by microinjection to tissue culture cells (31).

The model distills the requirements for Ran-regulated transport to an essential minimum. No assumption of vectorial translocation through the pore was made, supporting a similar ansatz made in computer simulations (29, 31, 32). However, the proliferation there of adjustable parameters and coupled equations, combined with the technical complexity of their solution, makes it difficult to prove this as a specific hypothesis. The simpler analytical derivation shows straightforwardly that directional accumulation is achieved without requiring directional passage of receptor–cargo or receptor–RanGTP complexes. In this light the observation of “abortive” import translocations, where cargo–receptor complexes enter the nuclear pore but return to the cytoplasmic side, is most relevant (30); after their arrival in the pore the memory of the compartment of origin would be lost (34). For small proteins whose rate of diffusive equilibration may approach that of receptor-mediated translocation, a simple leak would reduce the nuclear to cytoplasmic ratio (28).

The model may assist in interpreting the present experiment if we note that the Michaelis constant [K]_m^SS measured in steady state (Fig. 3B) appears as [K]_m^TC in Eq. 4. This suggests that the major parameter determining nuclear accumulation of a given cargo is its affinity to transport receptors in the cytoplasm. The numerical value from the experiment, 1.7 μM, indicates a much weaker binding than suggested by in vitro assays using purified components [e.g., 12 nM for SV40 NLS (35), 48 nM for the isolated NP NLS (36), or ≈100 nM for yeast Kap121 or Kap123 and cognate NLS (28)]. The egg extract represents an in situ, if not in vivo, cellular environment where both specific and nonspecific competitors may be present. Thus, the weaker affinity is entirely consistent with indications that only a fraction of the receptors are actually available for transport (28, 37). Interestingly, the kinetic Michaelis constant (Fig. 3A) is larger than that for steady state (Fig. 3B). This suggests that the step limiting nuclear accumulation in steady state is distinct from that which limits the import or translocation kinetics. If the former is cargo–receptor affinity, the latter may be pore permeability.

Conclusion

A widespread assumption is that, once delivered, a transport cargo remains in the nucleus (or cytoplasm) until and unless it is picked up by a receptor of the opposite directionality. In this sense, the localization signal would determine the fate of the individual molecule. In fact many proteins shuttle back and forth by virtue of their possession of both signal types or because of complex formation with other cellular factors. The data we present here show that shuttling is an intrinsic feature of the accumulation mechanism. Counterflow alone has been observed previously by artificial reversal of the RanGTP gradient (27). In the case of exportin-mediated translocation, a counterflux to the nucleus was observed by microinjection into the cytoplasm of Xenopus oocytes. Rates comparable to the export direction were observed, but only for substrates with substantial rates of receptor-free translocation (38). Most elegantly, tRNA was recently shown to shuttle in vivo (39). The present study adds the observation of a receptor-mediated molecular exchange of imported cargo between the nucleus and cytoplasm, taking place under conditions of net directional accumulation. A logical conclusion is that the localization (or export) signal determines the fate of a protein population, while individual cargo molecules may move back and forth repeatedly through the nuclear pores provided that they remain in solution and accessible to transport receptors.

This subtlety has far-reaching biological significance. It provides naturally for a basal level of exchange for cytoplasmic degradation or protein quality control. Moreover, it implies a much more open communication between the nucleus and cytoplasm than hitherto appreciated. For example, protein modifications or associations taking place far from the NLS region would be communicated back to the cytoplasm because the transport receptor recognizes only the signal peptide. Thus, the localization mechanism provides an intrinsic, ever-present mechanism for signaling in both directions. Strict confinement to one cellular compartment could be achieved by binding to insoluble structures such as chromatin, lamin, or cytoskeleton, or by chemical modification of the localization signal itself.

Materials and Methods

X. laevis Clarified Crude Egg Extract.

A gray-amber crude extract of X. laevis eggs was prepared following a standard protocol (33). The extract was then clarified by centrifugation at 26,500 × g for 15 min at 4°C in the Optima TL table-top ultracentrifuge (Beckman Coulter, Fullerton, CA) with TLS-55 swing-up rotor. Sucrose was added to 0.2 M final concentration. Aliquots were frozen in LN2 and stored at −80°C.

In Vitro Reconstitution of Nuclei from Clarified Crude Extract.

Clarified crude extract was diluted 3-fold with extract buffer (10 mM Hepes-NaOH, pH 7.4/50 mM KCl/2.5 mM MgCl₂/250 mM sucrose/20 mg/ml BSA) and supplemented with an ATP regenerating system (10× stock; final concentrations: 1 mM ATP, 10 mM phosphocreatine, and 50 μg/ml creatine phosphokinase). Nuclei were reconstituted at 20°C for 90 min by addition of demembranated sperm chromatin (2,000 units/μl) prepared as described in ref. 19. Chemical reagents were obtained from Sigma Israel Chemical Co. (Rehovot, Israel).

Protein Expression.

6×His-tagged recombinant human GFP-NP was expressed in BL21 cells (0.5 mM IPTG, 3 h in a shaking incubator at 36°C) and purified by FPLC over Ni²⁺-NTA resin columns under native conditions (binding buffer: 20 mM Tris, pH 8/500 mM NaCl/20 mM imidazole; elution buffer: 20 mM Tris, pH 8/500 mM NaCl/400 mM imidazole). The Tris buffer was exchanged by dialysis to PBS (pH 7.4). Glycerol was added to 5%, and aliquots were frozen in LN2. GFP-hnRNP A1 protein was expressed similarly.

The GFP-NP plasmid was a kind gift of M. Michael (Harvard University, Cambridge, MA). NLS-2×GFP was obtained from O. Medalia (Ben Gurion University of the Negev, Beer Sheva, Israel), expressed from the plasmid of S. Musser (Texas A&M University, College Station, TX). RanQ69L was a gift from D. Forbes (University of California, San Diego, CA).

Import Assay.

GFP-NP was diluted in advance with clarified crude extract to reach the desired concentration. A total of 0.5 μl of this mixture was added to 9 μl of fully assembled nuclei, supplemented with a fresh ATP regenerating system as above. A total of 0.5 μl of 150-kDa TRITC-dextran (Sigma, St. Louis, MO) was added as an exclusion marker for integrity of the NE. RanQ69L was added ≈10 min before measurement (final concentration of 5 μM). For observation, slides were prepared from standard 1- × 3-inch microscope slides and 24- × 40-mm no. 1.5 glass coverslips mounted cross-wise with parallel strips of stretched Parafilm. This forms a chamber ≈30 μm thick. Samples were introduced by pipette, and the chambers were sealed immediately with paraffin wax.

Fluorescence intensity measurements within the nucleus were recorded in time lapse, normally with 1-sec exposure and 29-sec pause. Excitation laser power was kept low. FCS confirmed the absence of photobleaching at the same power and established the radius and half-height of the illuminated volume as 0.23 μm and 0.75 μm. The precise cytosolic concentration and the average fluorescence per GFP-NP molecule were also measured by FCS. To evaluate and compare accumulation rates quantitatively, the data were normalized by the nuclear volume and by the brightness per molecule.

Photobleaching Assay.

A standard import assay was prepared, and accumulation kinetics were recorded. Before reaching saturation, the Ar laser intensity was increased 50-fold by rotating a polarizer. Photobleaching proceeded for 60 sec, until the emission intensity was reduced to a new saturation at low value. The laser power was then reduced to the original value, and reaccumulation of fluorescent GFP-NP was recorded.

Counterflow Transport Assay by Dilution.

Aliquots of fully assembled nuclei were complemented as above with GFP-NP at final concentration of 0.2 mg/ml (≈600 nM) and TRITC-dextran and incubated for at least 30 min at room temperature. The sample was then diluted in equal parts to a new preparation with GFP-NP at concentrations of 0.1 mg/ml (≈300 nM), 0.08 mg/ml (≈240 nM), and 0.04 mg/ml (≈120 nM), resulting in final dilution of cytosolic GFP-NP by 2, 2.5, and 5, respectively. Bright nuclei originating from the high-concentration GFP-NP sample were surrounded by dimmer nuclei originating from the low-concentration sample. Fluorescence images were collected in time lapse with 1-sec exposure and 29-sec pause. Integrity of the NE was verified by dextran exclusion throughout the experiment.

Abbreviations

FCS

fluorescence correlation spectroscopy

NLS

nuclear localization signal

M-M

Michaelis–Menten

NP

nucleoplasmin

NE

nuclear envelope.