Nuclear protein migration involves two steps: Rapid binding at the nuclear envelope followed by slower translocation through nuclear pores

Cell, Vol. 52, 655-664,

March

11, 1968, Copyright

0 1988 by Cell Press

Nuclear Protein Migration Involves Two Steps: Rapid Binding at the Nuclear Envelope Followed by Slower Translocation through Nuclear Pores W. D. Richardson,’ A. D. Mills,t S. M. Dilworth,t R. A. Laskey,t and C. Dingwallt l Department of Biology University College London Gower Street London WClE 6BT, England t Cancer Research Campaign Molecular Embryology Group Department of Zoology University of Cambridge Downing Street Cambridge CB2 3EJ, England

Summary When injected into the cytoplasm of Veto cells, nucleoplasmin rapidly concentrates in a narrow layer around the nuclear envelope and then accumulates within the nucleus. Transport into the nucleus can be reversibly arrested at the perinuclear stage by metabolic inhibitors or by chilling. Nucleoplasmin-coated colloidal gold particles concentrate around the nuclear envelope of Vero cells or Xenopus oocytes, and by electron microscopy of oocytes appear to be associated with fibrils attached to nuclear pore complexes. Perinuclear accumulation is not observed for the nonmigrating nucleoplasmin core fragment or nonnuclear proteins. We propose two steps in nuclear migration of proteins: rapid binding around the nuclear envelope, possibly to pore-associated fibrils, followed by slower, energy-dependent translocation through nuclear pores. j Introduction It is becoming clear that accumulation of large nuclear proteins in the cell nucleus is mediated by a selective transport mechanism rather than by selective binding to sites within the nucleus after entry by diffusion (Dingwall et al., 1982; reviewed by Dingwall and Laskey, 1986). The two main questions which arise are the following: What sequence features of nuclear proteins specify migration into the nucleus? What mechanisms respond to signal sequences to mediate selective entry into the nucleus? The first of these questions has received considerable attention recently (reviewed by Dingwall and Laskey, 1986). In this paper we describe experiments which address the second question. To study mechanisms mediating selective entry of nuclear proteins into the nucleus, we have used nucleoplasmin purified from Xenopus eggs. Nucleoplasmin is the most abundant nuclear protein in Xenopus oocytes, where it binds histones and transfers them to DNA during nucleosome assembly (Laskey et al., 1978; Earnshaw et al,, 1980; Kleinschmidt et al., 1985; Dilworth et al., 1987). The entry of nucleoplasmin into the nucleus has been studied

extensively. It provided the first evidence that information for accumulation in the nucleus is contained in a specific domain and that this domain specifies accumulation by selective entry through the nuclear envelope (Dingwall et al., 1982). Nucleoplasmin was also used to demonstrate that protein transport occurs through nuclear pores (Feldherr et al., 1984) and requires ATP (Newmeyer et al., 1986a, 1986b). We have investigated the mechanism of selective protein migration into nuclei by microinjecting nucleoplasmin or nucleoplasmin-coated colloidal gold into Xenopus oocytes and Vero (African green monkey kidney) cells. We subsequently visualized nucleoplasmin in microinjected Vero cells by immunofluorescence microscopy; coated gold particles were detected in Xenopus oocytes by thinsection electron microscopy, and in Vero cells by silver enhancement and light microscopy. We have found evidence that migration of nucleoplasmin into the nucleus occurs in two steps. Less than 1 min after injection into the cytoplasm of Vero cells, nucleoplasmin concentrates in a thin rim around the periphery of the nucleus. During the next few minutes (at 37%) nucleoplasmin enters and accumulates inside the nucleus. The distinction between these two steps is highlighted when Vero cells are maintained on ice, or when ATP production is reversibly inhibited by a combination of sodium azide and deoxyglucose. These treatments prevent accumulation of nucleoplasminor nucleoplasmin-coated gold particles within the nucleus, but still permit their stable accumulation around the nuclear envelope. Nucleoplasmin-coated gold particles microinjected into Xenopus oocytes also associate with the nuclear envelope, prior to and during accumulation within the nucleus. In the electron microscope they appear to be associated with fibrils attached to nuclear pore complexes on their cytoplasmic side. Fibrils associated with the nuclear surface of pore complexes have been described previously (Swift, 1956; Wischnitzer, 1958; Verhey and Moyer, 1967; Kessel, 1968; Franke and Scheer, 1970). Franke (1970) proposed a pore model with 3 nm fibrils extending into the cytoplasm as well as the nucleoplasm. Our data suggest that these fibrils on the cytoplasmic face of nuclear pores possess binding sites for nuclear proteins, and may be responsible for perinuclear accumulation of nucleoplasmin prior to transport through nuclear pores. Results Perinuclear Accumulation of Nucleoplasmin Injected into Cultured Cells Nucleoplasmin was first identified in the eggs and oocytes of Xenopus laevis, where it is the most abundant nuclear protein (for reviews of its properties, see Laskey et al., 1985; Dingwall et al., 1987). Purified nucleoplasmin was injected into the cytoplasm of cultured cells and subsequently visualized by a mixture of mouse anti-nucleoplasmin monoclonal antibodies followed by fluorescent second

Cell 656

Figure

1. Accumulation

of Nucleoplasmin

at the Nuclear

Envelope

Precedes

Nuclear

Translocation

Nucleoplasmin was injected into the cytoplasm of Vero cells, which were incubated as described below, then fixed and stained with mouse antinucleoplasmin antibodies followed by rhodamine-tagged rabbit anti-mouse and sheep anti-rabbit antibodies. Uninjected cells by comparison showed negligible fluorescence. (a) Incubation for 20 min at 37%. Nucleoplasmin has accumulated within the nucleus. (b) Nucleoplasmin was microinjected as in (a), but at 20°C and the coverslip was flooded with fixative immediately after microinjection of the last cell. The mean incubation time was less than 1 min at 20%. A sharp perinuclear ring is evident (arrows). (c) The pentameric trypsin-resistant core of nucleoplasmin was microinjected and incubated as in (a). The nucleoplasmin core remains in the cytoplasm and appears evenly distributed without a sharp perinuclear fluorescence. (d) A control monoclonal IgG was microinjected and subsequently visualized with a rhodamine-conjugated anti-mouse IgG antibody (Amersham International). It shows a random cytoplasmic distribution.

antibodies (see Experimental Procedures). Twenty minutes (at 37%) after injection, all of the nucleoplasmin had accumulated in the cell nucleus (Figure la). No detectable fluorescence, and hence no nucleoplasmin, remained in the cytoplasm. Incubation for as little as 2 min at 37% gave identical results, showing that nuclear accumulation of nucleoplasmin is extremely rapid. This figure clearly illustrates that unmodified nucleoplasmin in Vero cells achieves a nuclear-to-cytoplasmic concentration ratio greater than the value of 3 reported for fluorescently modi-

fied nucleoplasmin in hepatoma cell polykaryons (Schulz and Peters, 1986). Uninjected cells showed no significant fluorescence and so are presumed not to contain a large pool of nucleoplasmin or its analogs. When the cells were fixed less than 1 min (at 20%) after injection of nucleoplasmin, sharp perinuclear fluorescence was seen (Figure lb). The capacity of this perinuclear region for nucleoplasmin appears easily saturated, since increasing the mass of microinjected nucleoplasmin only results in increased fluorescence in the bulk of the cytoplasm rather

Nuclear 657

Protein

Migration

Involves

Two Steps

shows a gradation from the cell perimeter increasing to the nuclear envelope. This effect can be seen with various other nonnuclear proteins, e.g., immunoglobulins (IgG) (Figure Id), bovine serum albumin (BSA), or polyvinylpyrrolidone (PVP) (data not shown), or with core nucleoplasmin incubated for less than 1 min (data not shown), and simply reflects the increased depth of cytoplasm around the nucleus. This contrasts with the sharp perinuclear fluorescence at early times after injection of complete nucleoplasmin.

Figure 2. Chilling Allows Association

Cells Reversibly Inhibits Nuclear with the Nuclear Envelope

Accumulation

but

Precooled Vero cells were maintained on ice following cytoplasmic microinjection of proteins. Fixation and probing with fluorescent antibodies were as in Figure 1. (a) Nucleoplasmin was injected into the cytoplasm of cells that were precooled on ice and subsequently incubated for a further 1.5 hr on ice before being flooded with cold fixative. Low temperature has inhibited nuclear accumulation, but there is a strong punctate perinuclear fluorescence. (b) Cells were treated exactly as for (a) except that the pentameric core of nucleoplasmin was substituted for complete nucleoplasmin. These cells show a random cytoplasmic distribution. (c) Cells, microinjected with nucleoplasmin and incubated exactly as in (a), were then shifted up to ?Z“‘C for 10 min prior to fixation. The arrest of nuclear transport at the perinuclear stage is reversible.

than the perinuclear region. Thus binding to the envelope is seen most clearly when low concentrations are injected. Figure lc shows a 20 min incubation after cytoplasmic microinjection of the pentameric core of nucleoplasmin, which has lost the peptide that specifies nuclear entry (Dingwall et al., 1982). The distribution of fluorescence

Pore Translocation but Not Perinuclear Accumulation Requires Active Cell Metabolism By inhibiting metabolic activity in Vero cells, we were able to block nuclear translocation preferentially without visibly affecting perinuclear accumulation. Figure 2a shows Vero cells injected with nucleoplasmin, like those in Figure la, but precooled on ice and then maintained on ice for 1.5 hr after injection. Nuclear accumulation is inhibited and there is fluorescent staining outside the nucleus, particularly as a narrow perinuclear band similar to that seen in Figure lb. In contrast, injection of the pentameric core of nucleoplasmin into Vero cells maintained on ice (Figure 2b) revealed general cytoplasmic fluorescence similar in appearance to that in Figure lc. To test if the effect of low temperature is reversible, chilled cells were injected with nucleoplasmin, maintained on ice for 1.5 hr to prevent nuclear entry (parallelling the experiment of Figure 2a), and shifted up to 37°C for 10 min before fixation. Figure 2c shows that the effect of chilling is fully reversible. These results confirm that perinuclear accumulation can occur without translocation, and indicate that nuclear translocation but not perinuclear accumulation requires metabolic activity. This conclusion was confirmed and extended by treating cells with a mixture of deoxyglucose (6 mM) and sodium azide (10 mM) in the absence of glucose to deplete ATP This treatment blocks nuclear translocation (Figure 3a and 3b), again revealing perinuclear accumulation at the nuclear envelope. The high-power micrograph (Figure 3b) further illustrates that the perinuclear fluorescence has a punctate appearance, reminiscent of the pattern described for nuclear pore complex-specific staining (Davis and Slobel, 1986). Deoxyglucose used alone retarded but did not abolish nuclear accumulation, judged by the observation that some cells still displayed cytoplasmic fluorescence after 30 min at 37%. Sodium azide alone had very little apparent effect on nuclear accumulation. These observations suggest that the nuclear migration of proteins can be divided biochemically into at least two stages. The first step involves the recognition of a nuclear protein by a component of the nuclear periphery. The recognition and binding are very rapid and do not depend on cell metabolism. The second step is translocation through the nuclear envelope and accumulation within the nucleus; this step requires energy, possibly ATP hydrolysis, To determine whether protein-coated gold particles would behave like the free proteins, we first injected nucleoplasmin-coated gold into Vero cells and studied its subcellular distribution by silver enhancement (Danscher,

Cdl 658

e

Figure 3. Inhibitors Coated Gold

of ATP Production

Reversibly

Arrest

Nuclear

Transport

but Not Perinuclear

Localization

of Nucleoplasmin

or Nucleoplasmin-

(a-c) represent fluorescent micrographs similar to those in Figures 1 and 2; (d-f) show the distributions of microinjected colloidal gold revealed by silver enhancement. (a) Cells were preincubated for 30 min in medium without glucose, supplemented with 6 mM deoxyglucose and 10 mM sodium azide. Cells were then microinjected with nucleoplasmin and incubated at 37% for 1 hr before fixation. Most of the fluorescence is cytoplasmic, but a perinuclear ring is also visible. (b) Higher power micrograph of cells treated as in (a). The punctate appearance of the perinuclear ring is more clearly seen. (c) Cells were treated as in (a), and then prior to fixation the cells were washed with normal medium (containing 6 mM glucose) and incubated in that medium for a further 30 min. Reinstating ATP production in this way allows accumulation within the nucleus. (d) Nucleoplasmin-coated colloidal gold was injected into the cytoplasm, and incubation was at 37% for 2 hr. After fixation the colloidal gold distribution was visualized by silver enhancement (see Experimental Procedures). A significant proportion of the nucleoplasmin-coated gold has accumulated in the nucleus. (e) Nucleoplasmincoated gold was injected into cells pretreated with deoxyglucose and sodium azide, and incubation was at 37% for 1 hr. Silver enhancement shows that nuclear accumulation has been prevented, but a perinuclear ring is visible. (f) Gold particles coated with nucleoplasmin core were injected, and the cells were incubated at 37oC for 2 hr. Neither nuclear accumulation nor perinuclear association is evident.

Nuclear 659

Protein

Migration

Involves

Two Steps

1981) and light microscopy. Figure 3d shows that a significant fraction of the nucleoplasmin-coated gold accumulates in the nucleus within 2 hr at 37%. Prior treatment of cells with deoxyglucose and sodium azide resulted in a cytoplasmic distribution of injected nucleoplasmin-coated gold, with a perinuclear ring (Figure 3e). The perinuclear ring was absent when gold coated with the pentameric, nonmigrating nucleoplasmin core was injected (Figure 3f). Thus the qualitative behavior of protein-coated colloidal gold is similar to that of free proteins (although it migrates more slowly; data not shown). To study the perinuclear accumulation step in more detail, thin-section electron microscopy of Xenopus oocytes injected with nucleoplasmin-coated colloidal gold was carried out. Perinuclear Accumulation of Nucleoplasmin-Coated Gold Particles in Xenopus Oocytes Feldherr et al. (1984) have shown that colloidal gold particles coated with X. laevis nucleoplasmin enter the nucleus through nuclear pores and accumulate in the nucleus after injection into the cytoplasm of Xenopus oocytes. Figures 4a and 4b show our results obtained using the same approach. Oocytes were microinjected near the center of the vegetal hemisphere with 15 nm colloidal gold which had been coated with nucleoplasmin. Figure 4a shows, in addition to colloidal gold particles in the nuclear pore complex and within the nucleus, a striking accumulation of gold particles at the highly convoluted nuclear envelope. Figuree4b shows that after 24 hr the perinuclear accumulation is more marked and that there is very little gold in the adjacent cytoplasm, suggesting that the nucleoplasmin-coated gold does not readily exchange with the cytoplasm. A small group of nuclear pores is visualized where the envelope is cut tangentially; most contain centrally placed colloidal gold particles (arrows in Figure 4b). Colloidal gold coated with the nucleoplasmin core (F? ure 4c) or BSA (Figure 4e) does not show perinuclear localization but does show a general cytoplasmic distribution. This is also seen with other nonnuclear macromolecules (e.g., IgG and PVP; data not shown). The amount of gold injected into the cytoplasm is approximately the same in each case, and therefore the perinuclear accumulation cannot simply reflect the amount of gold injected. Apparently contradictory results from early experiments with isolated nuclei of amphibian oocytes are considered in the Discussion. Perinuclear Localization of Colloidal Gold Coated with Nucleoplasmin Tail Peptide Pepsin cleavage of nucleoplasmin produces a carboxyterminal tail peptide of about 60 amino acids containing four regions of homology to postulated nuclear localization signals (Dingwall et al., 1987). This purified tail peptide accumulates rapidly in the nucleus after microinjection into the cytoplasm of oocytes (Dingwall et al., 1982). In contrast, colloidal gold coated with this peptide does not migrate efficiently into the nucleus; however, it does accumulate around the nuclear envelope (Figure 4d). The lack of significant accumulation in the nucleus may reflect

the instability of the tail peptide in the oocyte cytoplasm (Dingwall et al., 1982). Alternatively, coating onto colloidal gold may reduce signal recognition. Whatever the reason, this result further emphasises that accumulation at the envelope can occur in the absence of transport into the nucleus. Fibrillar Components of the Nuclear Pore Complex To investigate the structural basis of perinuclear accumulation, we have studied the nuclear envelopes of isolated oocyte nuclei by electron microscopy using a procedure described by Franke et al. (1981). The nuclear pore when tangentially sectioned (in the plane of the nuclear envelope) appears as a relatively densely staining ring, consistent with the eightfold globular symmetry described previously (Franke et al., 1981; Unwin and Milligan, 1982). Tangential sections immediately above and below the plane of the nuclear envelope show less densely staining circular arrays without polygonal symmetry (data not shown). In cross section (to the nuclear envelope) (Figure 5b) these arrays appear as fibrils extending from the nuclear pores in either direction. We cannot, however, conclude that these fibrils are continuous through the pore complex. Although we do not present direct evidence here that these fibrils are essential for perinuclear accumulation, we do show below that they bind colloidal gold coated with nucleoplasmin, and we regard them as serious candidates for a role in this mechanism. The pore-associated fibrils extend further into the nucleus than into the cytoplasm and appear as a hollow cylindrical array, as suggested previously (Franke, 1970). The diameters of a number of nucleoplasmic pore-associated fibrils were estimated from electron micrographs of sections stained with uranyl acetate and lead citrate and printed at magnifications ranging from 100,000x to 275,000x; the mean was calculated to be 3.3 nm (en-, = 0.95, n = 50). Estimating the diameter of the poreassociated fibrils on the cytoplasmic side proved more difficult because they are much shorter and form a dense mass; however, measurements that have been made indicate a similar diameter to those on the nucleoplasmic face (mean = 3.6 nm, onml = 1.6, n = 7). These sizes are within the range of those described previously (Kessel, 1968; Franke, 1970). To visualize these structures more clearly, oocyte nuclei were rapidly isolated (cl min) and fixed. It is clear that during this procedure the nucleus has swollen, allowing the fibrils to be seen clearly. This expansion could conceivably cause artifactual binding of nuclear contents to the pores. However, fibrillar arrays can also be seen in the nucleus of some fixed whole oocytes (arrowheads in Figure 5a), suggesting that the nucleoplasmic fibrils do not originate from artifactual binding of nuclear contents during isolation. Colloidal Gold Coated with Nucleoplasmin Decorates the Pore-Associated Fibrils When nuclei are isolated from oocytes after injection of colloidal gold coated with nucleoplasmin, particles of gold

Cell 660

.

.

t

.

Nuclear 661

Protein

Migration

Involves

Two Steps

can be seen centrally within the pore and also inside the cylindrical space delimited by the pore-associated fibrils on the nuclear side of the nuclear envelope (Figure 5~). Colloidal gold is only rarely found associated with these nuclear fibrils; however, the fibrils extending into the cytoplasm are decorated with colloidal gold (Figure 5c), and this can be visualized as a ring of gold in sections tangential to the nuclear envelope (data not shown). To visualize fibrils clearly, we have isolated nuclei after injecting gold into intact oocytes; this introduces a potential artifact, because the envelopes of isolated nuclei can bind coated colloidal gold particles nonspecifically in vitro (Feldherr, 1962, 1964, 1965, 1974). However, we have minimized nonspecific binding in two ways. First, we have included BSA (at 5 mglml) in the isolation buffer; this decreases nonspecific binding. Second, we have concentrated our attention on infoldings of the highly convoluted nuclear envelope. We do not observe nonspecific binding of colloidal gold within these infoldings, possibly because they retain a high concentration of cytoplasmic protein during nuclear isolation. When artifactual binding was minimized in these ways, nucleoplasmin-coated gold remained bound to the fibrils on the cytoplasmic surface of the nuclear pore (Figure 5c), whereas binding was not observed for gold coated with the nucleoplasmin core or BSA. The results we present are internally consistent when either intact Xenopus oocytes or Vero cells are compared with the cytoplasm-filled infoldings of isolated Xenopus oocyte nuclei. They suggest that nuclear-migrating species like nucleoplasmincoated colloidal gold bind to the pore-associated fibrils. Although we have minimized nonspecific binding, the existence of this potential artifact prevents us from saying to what extent binding to the fibrils accounts for the perinuclear accumulation seen by both light and electron microscopy. Nevertheless, the fibrils are obviously interesting candidates for this role, and their involvement deserves further study. Discussion Evidence for Two Steps in Nuclear Accumulation of Proteins There is growing evidence that nuclear proteins accumulate in the nucleus as the result of a selective, receptor-

Figure

4. Association

of Nucleoplasmin-Coated

Gold with the Nuclear

mediated transport mechanism (reviewed by Dingwall and Laskey, 1986). Although we confirm that the nuclear pore complex is the route of entry (Feldherr et al., 1984), data presented here show that translocation through the nuclear pore does not provide a complete explanation of nuclear accumulation. Feldherr et al. (1984) further suggested that nuclear protein migration appeared to involve multiple steps. We have directly confirmed this suggestion and have resolved, both temporally and metabolically, two distinct steps. Figures lb, 2a, 3a, 3b, and 3e show that nucleoplasmin binds rapidly to the nuclear envelope even when accumulation within the nucleus is blocked. Thus we conclude that rapid binding to the nuclear envelope is a preliminary step in nuclear migration and that this is followed by slower transit through the nuclear pore. Newmeyer et al. (1986a, 1986b) had previously shown that ATP is required for protein accumulation in the nucleus. The evidence from the chilling and metabolic blocking experiments described here indicates that ATP is only required for the translocation step and not for the rapid association with the nuclear envelope. Specificity of Binding to the Nuclear Envelope Binding to the nuclear envelope is specific for nuclear proteins or peptides in both injected Vero cells (Figures lb, 2a, 3a, 3b, and 3e) and injected oocytes (Figures 4a, 4b, 4d, and 5~). Thus gold coated with IgG (Figure Id), BSA (Figure 4e), PVP, porcine 6-melanocyte stimulating hormone, gastrin (data not shown), or the core pentamer of nucleoplasmin (Figures lc, 2b, 3f, and 4c) showed neither preferential perinuclear association nor nuclear accumulation in intact Xenopus oocytes, Vero tissue culture cells, or isolated Xenopus oocyte nuclei under the controlled conditions described above. At first sight this appears to conflict with early data showing binding of polylysine-, polyproline-, or PVPcoated colloidal gold to the envelope of isolated amphibian oocyte nuclei (Feldherr, 1964, 1974). In those studies the absence of competing macromolecules at the nuclear envelope during fixation may account for the apparent nuclear pore association seen in isolated nuclei. We, too, have observed binding of gold coated with nonnuclear molecules to isolated nuclei in vitro in the absence of BSA. However this nonspecific effect was not observed when coated gold particles were injected into intact oocytes

Envelope

of Xenopus

Oocytes

Colloidal gold (15 nm) coated with different proteins was microinjected into the cytoplasm of Xenopus oocytes. After incubation at 22OC, the oocytes were fixed, thin-sectioned, and stained for electron microscopy(see Experimental Procedures). Abbreviations: C, cytoplasm; N, nucleus. Bar = 1 hm. (a) Colloidal gold was stabilized with nucleoplasmin, and injected oocytes were incubated for 4 hr. The electron micrograph shows gold around the nuclear envelope, in the nuclear pore orifices (arrowheads), and evenly distributed within the nucleoplasm. (b) Similar to (a), but after a 24 hr incubation at 22%. Note first that the gold has still not all accumulated within the nucleus but that some remains outside, where it is associated with the nuclear envelope; there is very little gold beyond a 200 nm perinuclear layer. The arrowheads indicate a part of the nuclear envelope that is cut tangentially, revealing the circular pores; most pores contain a gold particle. (c) Oocytes were injected with colloidal gold coated with the trypsin-resistant core of nucleoplasmin. The amount of colloidal gold injected was similar to that in (a) and (d), but it is distributed randomly through the cytoplasm rather than concentrated locally. (d) Tail fragment cleaved from nucleoplasmin by a brief digestion with pepsin was stabilized onto colloidal gold and microinjected as in (a). The electron micrograph shows that, after a 2 hr incubation, colloidal gold is distributed within the cytoplasm and at higher density around the nuclear envelope, but that it is virtually absent from the nucleus. (e) Colloidal gold coated with the control protein BSA shows only a random cytoplasmic distribution of gold particles without any preferential pore or nuclear envelope association.

Cell 662

Figure

5. Nucleoplasmin-Coated

Gold

Particles

Associate

with Fibrils

Attached

to the Cytoplasmic

Side of Nuclear

Pore Complexes

A sectioned whole Xenopus oocyte occasionally reveals fibrillar arrays extending away from the pore complex (a). However, when Xenopus oocyte nuclei are isolated and rapidly fixed, swelling occurs and reveals this structure more reliably and more clearly. Abbreviations: C, cytoplasm (a) or cytoplasmic face (b and c); N, nucleus (a) or nuclear face (b and c). Bar = 0.2 urn. (a) A cross section of a nuclear envelope from a whole fixed oocyte showing fibrillar arrays associated with the nuclear pore (between double arrowheads). (b) Fibrils extend from both the cytoplasmic and nuclear faces of pore complexes. The fibrillar arrays on the nuclear face of the pore are much longer than those on the cytoplasmic face. It is suggested that associated with pore complexes are cylindrical fibrillar arrays composed of 3-4 nm fibrils that extend a short distance into the cytoplasm and much further into the nucleus. (c) Nucleoplasmin-stabilized colloidal gold was prepared, microinjected, and incubated as described for Figures 4a and 4b. After a 3 hr incubation the oocyte nuclei were isolated in 5:l isolation buffer containing 5 mglml SSA, briefly fixed, and processed for electron microscopy. This section shows part of an infolding of nuclear envelope containing cytoplasm (note ribosomes) to minimize artifactual binding (see text). Colloidal gold can be seen in the center of the pore complex at the level of the nuclear envelope and decorating the pore-associated fibrils on the cytoplasmic face.

(Figures 4c and 4e) or intact cultured Vero cells (Figures lc, Id, 2b, and 3f). Furthermore, nonspecific binding to isolated oocyte nuclei was decreased by including BSA in the isolation buffer. BSA did not decrease specific nuclear

envelope binding of gold coated with nuclear proteins (Figure 5%). Feldherr also reported a weaker nuclear envelope association of gold coated with PVP after injection into intact

Nuclear 663

Protein

Migration

Involves

Two

Steps

amoebae (Feldherr, 1962, 1965). The cause of this is unclear, but neither we nor Feldherr (personal communication) have observed this nonspecific binding in intact microinjected Xenopus oocytes. The binding we have observed to the nuclear envelopes of microinjected oocytes or Vero cells has been specific for nuclear-migrating proteins. Evidence that Binding to the Nuclear Envelope Is an Intermediate Step in Nuclear Entry We have confirmed and extended the suggestion by Feldherr et al. (1984) that binding to the nuclear envelope is an intermediate stage in nuclear migration. First, it is specific for nuclear proteins, as explained above. Second, it is observed immediately after injection of nucleoplasmin into the cytoplasm of Vero cells, before nuclear accumulation is obvious. Third, interruption of nuclear migration by chilling cells or by inhibiting their ATP production allows the formation of stable perinuclear rings of nucleoplasmin (Figures 2a, 3a, 3b, and 3e). This nucleoplasmin subsequently enters the nuclei when inhibition is removed by raising the temperature or transferring the cells into inhibitor-free medium containing glucose (Figures 2c and 3~). Therefore we conclude that the first step in nuclear migration is selective binding to the outside of the nuclear envelope. Are the Pore-Associated Fibrils Involved in the First Step of Nuclear Accumulation? Figure 5c shows an association of nucleoplasmin-coated colloidal gold particles with the fibrils that extend into the cytoplasm from the nuclear pore complexes. We have not yet been able to identify the components of the fibrils, and it remains to be seen if they can account fully for the perinuclear accumulation seen in Figures l-4. However, we feel that their structure and possible involvement in the nuclear transport process deserve further study. In view of growing evidence for a selective transport mechanism, the simplest tenable model of nuclear transport envisages that a single signal sequence interacts with a specific translocation mechanism at the pore orifice. However, the results we present here indicate that this is an oversimplification, and that this step is preceded by rapid accumulation around the nuclear envelope. Experimental

Procedures

Reagents Nucleoplasmin was isolated from homogenates of Xenopus eggs using monoclonal anti-nucleoplasmin antibody columns. Eggs were collected as described before (Laskey et al., 1977) except that the current modification of Barth saline excluded CaClz and replaced HEPES with a Tris buffer system. Eggs were then completely dejellied (Laskey et al., 1977) washed twice with distilled water and twice with buffer S (0.25 M sucrose, 70 mM potassium acetate, 20 mM HEPES, 5 mM MgCIz, 5 mM EGTA, 0.5 mM PMSF [pH 7.41) and allowed to pack in a centrifuge tube on ice. The supernatant was removed, and the packed egg volume was estimated and then overlaid with 0.5 vols of fresh buffer S for centrifugation at 12,000 rpm, 4OC, for 20 min in a Sorvall SS34 rotor. The supernatant was transferred to a fresh tube and centrifuged at 30,000 rpm (145,000 x g) for 30 min at 4OC in a Beckman 50Ti rotor. The clear supernatant was collected and stored in liquid nitrogen. Anti-nucleoplasmin antibody columns were prepared from two

hybridoma clones, PA3C5 (an anti-tail nucleoplasmin antibody) or PAlC2 (an anti-core nucleoplasmin antibody). Tissue culture fluids in which the hybridomas had been allowed to grow to confluence were clarified by centrifugation at 10,000 x g and precipitated with ammonium sulfate (60% saturated solution). The precipitate was redissolved in 2 M NaCI, 0.1 M sodium phosphate (pH 8.5) and dialyzed against the same buffer. The resultant dialysate was cross-linked to protein A-Sepharose CL46 (Bioprocessing Ltd.) at a ratio of 5 mg IgG to 1 ml packed matrix using 20 mM dimethyl pimelimidate dihydrochloride (Pierce) in 0.2 M triethanolamine (pH 8.2) (Schneider et al., 1982). Excess cross-linker was inactivated using 0.2 M ethanolamine (pH 8.2) and the matrix was washed into 100 mM NaCI, 20 mM HEPES (pH 7.4). Egg homogenates were incubated batchwise with the antibody matrix for 3-4 hr at 4°C. The matrix was then isolated, washed, and eluted with either 0.02 M citrate (pH 2.8) or 0.1 M glycine (pH 2.8). The eluate was immediately neutralized. After identification of the nucleoplasmin peak, fractions were pooled, dialyzed, and concentrated to l-10 mglml. Tail fragments of nucleoplasmin were prepared from pepsindigested purified nucleoplasmin (Dingwall et al., 1982); they were separated from core fragments and partially digested pentamers by collecting the flow-through of an anti-core nucleoplasmin antibody column. Pentameric core fragments of nucleoplasmin were prepared and purified from a trypsin digest of purified nucleoplasmin (Dingwall et al., 1982). Trypsin was used because it yields more completely digested nucleoplasmin core than pepsin. Colloidal gold was made by modification of the citrate method of Slot and Geuze (1981). Twenty milligrams of chloroauric acid (BDH) was dissolved in 200 ml of ultrapure water; 4.6 ml of 1% trisodium citrate (BDH) in ultrapure water was added to give a mean particle size of about 15 nm. The mixture was gently swirled, left for 5 min, and then gently refluxed (Slot and Geuze, 1981). Stabilization of colloidal gold was achieved using protein or peptide solutions (>l mglml) (Slot and Geuze, 1981). Stabilized colloidal gold was washed and concentrated by centrifugation before microinjection. Microinjection of Tissue Culture Cells Vero cells, grown on glass coverslips, were microinjected (Richardson and Westphal. 1983) with nucleoplasmin or other proteins at 1 mglml. The injected volume was estimated to be 1O-10-1O-1r ml (Graessmann et al., 1980). After appropriate incubation, the coverslips were flooded with cold fixative (4% formaldehyde in phosphate-buffered saline). Formaldehyde fixation minimizes loss of nucleoplasmin during fixation (Krohne and Franke, 1980). Nucleoplasmin was located using a mixture of monoclonal anti-core nucleoplasmin antibodies (PAlC2 and PB4D3), then rhodamine isothiocyanate-conjugated rabbit anti-mouse F(ab’)z fragment followed by rhodamine isothiocyanate-conjugated sheep anti-rabbit F(ab)z fragment (kindly donated by Roger Morris, NIMR). Alternatively, Vero cells were injected with colloidal gold stabilized with either complete nucleoplasmin or core nucleoplasmin. These cells were fixed as above and subjected to silver enhancement (see below). To test the requirements for ATP, cells were also injected following a 0.5 hr preincubation with the inhibitors sodium azide and deoxyglucase. These were added to glucose-minus medium at the final concentrations of 10 mM and 6 mM, respectively. Following microinjection the cells were further incubated for 1 hr at 3PC before fixation as above. Microinjection of Xenopus Oocytes Oocytes were removed from anesthetised X. laevis and maintained in a modified Barth saline solution (88 mM NaCI, 2 mM KCI, 1 mM MgClp, 0.5 mM CaClz, 15 mM Tris-HCI [pH 7.61). They were separated into single oocytes just prior to the injection into the center of the vegetal hemisphere of about 150 nl of protein-stabilized colloidal gold. Oocytes were incubated at 20°C-220C in modified Barth saline. Preparation for Electron Microscopy Injected oocytes were processed either as whole oocytes, by placing the oocyte directly into fixative (2.5% gluteraldehyde, 2% formaldehyde, 0.1 M sodium cacodylate, 2% dimethyl sulfoxide. 1 mM CaClz [pH 7.41) or by isolating the oocyte nucleus in 5:l isolation buffer (83 mM KCI, 17 mM NaCI, 10 mM Tris-HCI [pH 7.41, with 5 mglml BSA to reduce nonspecific binding) before transferring it to the same fixative. Isolation of nuclei and transfer to fixative routinely took less than 30

Cell 664

sec. Material was fixed for 4 hr at 20°C then overnight at 4%; subsequently it was washed in cacodylate buffer, postfixed (1% OsOl in cacodylate buffer [pH 7.41) overnight at 4OC, dehydrated through ethanol; and embedded in either epoxy resin (TAAB Laboratories) or LR White acrylic resin (London Resin Co. Ltd.). Silver to pale-gold sections were taken onto collodion- and carbon-coated 600 mesh copper grids, stained with uranyl acetate and Reynolds lead citrate (5 min each), and visualized using a Philips EM300 at 60 kV, or a JEOL 200 CX at 80 kV. Silver Staining After fixation in 4% formaldehyde in phosphate-buffered saline, colloidal gold in microinjected Vero cells was visualized by the silver lactate amplification method of Danscher (1981). In brief, freshly prepared solutions of silver lactate (7.3 mglml) and hydroquinone (57 mglml) were mixed in the dark with sodium citrate buffer (2 M, pH 3.5) and a protective colloid of gum arabic (250 mglml) in the ratio 3:3:2:12 by volume. After incubation at room temperature for 30 min in the dark, the cells were washed extensively in warm (40%) water and mounted for brightfield light microscopy.

Franke, W. W., Scheer, U., Krohne, G., and Jarasch, nuclear envelope and the architecture of the nuclear Biol. 97, 39s-50s.

E.-D. (1981). The periphery. J. Cell

Graessmann, A., Graessmann, M., and Mueller, C. (1980). Microinjection of early SV40 DNA fragments and T antigen. Meth. Enzymol. 65, 816-825. Kessel, R. G. (1968). Fine structure 36, 658-664.

of annulate

lamellae.

J. Cell Biol.

Kleinschmidt, J. A., Fortkamp, E., Krohne, G., Zentgraf, H., and Franke, W. W. (1985). Co-existence of two different types of soluble histone complexes in nuclei of Xenopus laevis oocytes. J. Biol. Chem. 260, 1166-1176. Krohne, G., and Franke, W. W. (1980). located in nuclei of diverse vertebrate 167-189.

A major soluble acidic protein species. Exp. Cell Res. 129,

Laskey, R. A., Mills, A. D., and Morris, N. R. (1977). Assembly of SV40 chromatin in a cell-free system from Xenopus eggs. Cell 70, 237-243.

Acknowledgments The authors would like to thank Dr. Brij Gupta, Mr. Mick Day, and Mrs. Maggie Bray for help with electron microscopy; Dr. Durward Lawson for advice on inhibitors of ATP synthesis; Mr. Michael Mosley for technical assistance; and Barbara Rodbard for typing the manuscript. W. D. R. thanks the Nuffield Foundation for a grant to buy microinjection equipment. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received

Franke, W. W., and Scheer, U. (1970). The ultrastructure of the nuclear envelope of amphibian oocytes: a reinvestigation. 1. The mature oocyte. J. Ultrastruct. Res. 30, 288-316.

May 26, 1987; revised

January

11, 1988.

Laskey, R. A., Honda, 6. M., Mills, A. D., and Finch, J. T. (1978). Nucleosomes are assembled by an acidic protein which binds histones and transfers them to DNA. Nature 275, 416-420. Laskey, R. A., Dingwall, C.. Mills, A. D., and Dilworth, S. M. (1985). Transport and assembly of nuclear proteins, In Nuclear Envelope Structure and RNA Maturation, UCLA Symp. Mol. Cell. Biol. New Ser. 26, E. A. Smuckler and G. A. Clawson, eds. (New York: Alan R. Liss, Inc.), pp. 193-205. Newmeyer, D. D., Lucocq, J. M., Burglin, T. R., and De Robertis, E. M. (1986a). Assembly in vitro of nuclei active in nuclear protein transport: ATP is required for nucleoplasmin accumulation. EMBO J. 5, 501-510.

References

Newmeyer, D. D., Finlay, D. R., and Forbes, D. J. (1986b). In vitro transport of a fluorescent nuclear protein and exclusion of non-nuclear proteins. J. Cell Biol. 703, 2091-2102.

Davis, L. I., and Blobel, G. (1986). Identification and characterization of a nuclear pore complex protein. Cell 45, 699-709.

Richardson, W. D., and Westphal, regulation and the adeno-associated Microbial. Immunol. 709, 147-165.

Danscher, chemistry

G. (1981). 77, 81-88.

Localization

of gold in biological

tissue.

Histo-

Dilworth, S. M., Black, S. J., and Laskey, R. A. (1987). Two complexes that contain histones are required for nucleosome assembly in vitro: role of nucleoplasmin and Nl in Xenopus egg extracts, Cell 51, 1009-1018. Dingwall, C., and Laskey, R. A. (1986). Protein cleus Ann. Rev. Cell Biol. 2, 367-390.

import

into the cell nu-

Dingwall, C., Sharnick, S. V., and Laskey, R. A. (1982). A polypeptide domain that specifies migration of nucleoplasmin into the nucleus. Cell 30, 449-458. Dingwall, C., Dilworth, S. M., Black, S. J., Kearsey, S. E., Cox, L. S., and Laskey, R. A. (1987). Nucleoplasmin cDNA sequence reveals poly glutamic acid tracts and a cluster of sequences homologous to putative nuclear localization signals. EMBO J. 6, 69-74. Earnshaw, W. C., Honda, B. M., Laskey, R. A., and Thomas, J. 0. (1980). Assembly of nucleosomes: the reaction involving X. laevis nucleoplasmin. Cell 27, 373-383. Feldherr, C. M. (1962). The nuclear annuli as pathways plasmic exchanges. J. Cell Biol. 74, 65-72. Feldherr, C. M. (1964). Binding ble effect on nucleocytoplasmic

for nucleocyto-

within the nuclear annuli and its possiexchanges, J. Cell Biol. 20, 188-192.

Feldherr, C. M. (1965). The effect of the electron-opaque pore material on exchanges through the nuclear annuli. J. Cell Biol. 25, 43-53. Feldherr, C. M. (1974). The binding characteristics Exp. Cell Res. 85, 271-277. Feldherr, C. M., Kallenbach, a karyophilic protein through 99, 2216-2222.

of the nuclear

annuli.

E., and Schultz, N. (1984). Movement of the nuclear pores of oocytes. J. Cell Biol.

Franke, W. W. (1970). On the universality of nuclear ture. Z. Zellforsch. Mikrosk. Anat. 705, 405-429.

pore complex

struc-

H. (1983). Adenovirus early gene virus helper effect. Curr. Top.

Schneider, C., Newman, R. A., Sutherland, D. R., Asser, U., and Greaves, M. F. (1982). A one-step purification of membrane proteins using a high efficiency immuno matrix. J. Biol. Chem. 257; 1076610769. Schulz, B., and Peters, R. (1986). Intracellular transport of a karyophilic protein. In Nucleocytoolasmic Transport, R. Peters and M. Trendelenburg, eds. (Berlin: Springer-Verlag), pp. 171-184. Slot, J. W., and Geuze, H. J. (1981). Sizing of protein A-colloidal gold probes for immunoelectron microscopy. J. Cell Biol. 90, 533-536. Swift, H. (1956). The fine structure Biochem. Cytol. 2, 415-418.

of annulate

lamellae.

J. Biophys.

Unwin, P. N. T., and Milligan, R. (1982). A large particle associated with the perimeter of the nuclear pore complex. J. Cell Biol. 93, 63-75. Verhey, C. A., and Moyer, F. H. (1967). Fine structural sea urchin oogenesis. J. Exp. Zool. 764, 195-226. Wischnitzer, S. (1958). envelope of amphibian

An electron microscopy oocytes. J. Ultrastruct.

changes

during

study of the nuclear Res. 7, 201-222.