We tested the hypothesis that gp210, an integral membrane protein of nuclear pore complexes (NPCs), mediates nuclear pore formation. Gp210 has a large lumenal domain and small COOH-terminal tail exposed to the cytoplasm. We studied the exposed tail. We added recombinant tail polypeptides to Xenopus nuclear assembly extracts, or inhibited endogenous gp210 tails using anti-tail antibodies. Both strategies had no effect on the formation of fused flattened nuclear membranes, but blocked NPC assembly and nuclear growth. Inhibited nuclei accumulated gp210 and some nucleoporin p62, but failed to incorporate nup214/CAN, nup153, or nup98 and were defective for nuclear import of lamin B3. Scanning and transmission EM revealed a lack of “closely apposed” inner and outer membranes, and the accumulation of novel arrested structures including “mini-pores.” We conclude that gp210 has early roles in nuclear pore formation, and that pore dilation is mediated by gp210 and its tail-binding partner(s). We propose that membrane fusion and pore dilation are coupled, acting as a mechanism to control nuclear pore size.
The eukaryotic genome is enclosed by two nuclear membranes. A mechanism for fusion between the inner and outer membranes to generate pores is essential for the genome to communicate with the cytoplasm; indeed, the evolution of eukaryotic organisms probably depended on a “porogenic” fusion mechanism. Assembling pores have diameters that range from 6–40 nm (Goldberg et al., 1997). In vertebrates, mature pores have a diameter of ∼50–70 nm and are occupied by nuclear pore complexes (NPCs),* which regulate molecular traffic between the nucleoplasm and cytoplasm (Bayliss et al., 2000; Wente, 2000; Yoneda, 2000). Vertebrate NPCs have a maximum mass of 125 MD (Reichelt et al., 1990; Panté and Aebi, 1994) and consist of multiple copies of ∼40 distinct proteins (Miller and Forbes, 2000), termed nucleoporins. NPCs are anchored at the pore membrane domain, where the inner and outer membranes merge. Soluble nucleoporins are recruited to the pore membrane during NPC assembly. Pore formation was proposed to be triggered by the binding of soluble proteins to membranes (Fabergé, 1974) or by chromatin-induced indentations of the inner nuclear membrane, as seen by transmission EM (TEM; Maul et al., 1971). A role for chromatin in the formation of mature functional NPCs is likely, even though chromatin is not essential for porogenic membrane fusion per se (Maul, 1977; Vasu and Forbes, 2001). Unidentified soluble nucleoporins are required to stimulate pore formation in regions of flattened nuclear membranes (Macaulay and Forbes, 1996), and also in ER-like membranes known as annulate lamellae, which lack chromatin (Dabauvalle et al., 1991; Meier et al., 1995). Along with most aspects of NPC assembly, the mechanism of porogenic membrane fusion is an important open question in biology.
Membrane fusion is central to secretion, endocytosis, and the biogenesis of the ER, Golgi apparatus, and mitochondria (Bennet and Scheller, 1993). These fusions are mediated by cytosolic proteins that first disrupt the cytosolic leaflet of each bilayer (Robinson and Martin, 1998). In contrast, nuclear pore fusion involves the lumenal leaflets of the nuclear inner and outer membranes, and therefore probably involves proteins within the lumenal space. Viral fusogens, such as hemagglutinin (HA) protein of influenza virus (Skehel and Wiley, 2000), are viewed as possible models for lumenal membrane fusion events in normal cells. At low pH, membrane-embedded HA trimers undergo a conformational change that exposes their fusogenic peptides, allowing them to destabilize the opposing lipid bilayer and sequentially trigger membrane hemifusion, pore formation, and pore dilation (Hernandez et al., 1996; Kozerski et al., 2000). A conceptually different possibility is that soluble nucleoporins might assemble on the chromatin surface and then recruit surrounding membranes laterally as membranes attach to chromatin during nuclear assembly. However, this hypothetical mechanism is restricted; it could function only during the few minutes in telophase before chromatin is enclosed by membranes, it cannot explain pore formation during G1, S, or G2 phases of the vertebrate cell cycle, nor explain pore formation in eukaryotes (e.g., Saccharomyces cerevisiae) whose nuclei remain intact during mitosis.
Wozniak et al. (1989) and Greber et al. (1990) proposed that nuclear pore formation might be mediated by gp210, an integral membrane protein found at pores (Gerace et al., 1982). Gp210 is linked structurally to NPCs, as antibodies that target lumenal epitopes of gp210 decrease both active and passive transport through NPCs (Greber and Gerace, 1992). The majority of gp210 (95% of its mass) lies within the nuclear envelope lumen, but its short COOH-terminal tail is exposed to the cytosol (Greber et al., 1990). This cytoplasmic tail is specifically phosphorylated during mitosis, presumably to facilitate NPC disassembly (Favreau et al., 1996). Circumstantial evidence for gp210 as a fusogen came from topological studies. Rat gp210 was predicted by hydropathy analysis to have two membrane spanning domains (Wozniak and Blobel, 1992). However, topological studies showed that only one, the COOH-terminal hydrophobic domain, actually spans the membrane, whereas the other sits in the lumenal domain (Greber et al., 1990). Wozniak and Blobel (1992) suggested that this “extra” hydrophobic domain might destabilize lipid bilayers, triggering fusion. Also analogous to HA, which forms trimers, gp210 was calculated to form dimers or trimers (Gerace et al., 1982). Recent cross-linking studies show that gp210 forms stable dimers (Favreau et al., 2001).
By indirect immunofluorescence, many soluble nucleoporins reaccumulate rapidly at assembling nuclear envelopes (Chaudhary and Courvalin, 1993; Bodoor et al., 1999; Haraguchi et al., 2000). In contrast, gp210 is detectable at early stages of nuclear assembly, but most gp210 does not reaccumulate until late telophase or early G1 (Chaudhary and Courvalin, 1993; Bodoor et al., 1999). Note that gp210 is not excluded from nuclei at early stages of assembly; rather, it simply reaccumulates slowly, raising questions about its theoretical early role in pore formation. The only other known integral membrane nucleoporin in vertebrates, POM121, accumulates rapidly during nuclear assembly (Bodoor et al., 1999; Daigle et al., 2001). POM121 has a large exposed domain and a small lumenal domain (Soderqvist and Hallberg, 1994). Overexpression of POM121 in mammalian cells causes NPCs to form and accumulate in ER membranes, suggesting direct or indirect roles in triggering NPC formation (Imreh and Hallberg, 2000). In yeast, overexpression of a soluble nucleoporin named nup53, which has an amphipathic α-helix, induces the formation of nuclear inner membrane tubules with empty pores (Marelli et al., 2001).
We used Xenopus extracts to directly test the role of gp210, if any, in pore formation. Xenopus egg extracts are a powerful and well-characterized system for studying nuclear pore formation (Miller and Forbes, 2000; Vasu and Forbes, 2001). We focused on the small exposed tail of gp210, because it is freely accessible to reagents added to cell-free extracts. We found that a recombinant gp210 tail polypeptide and antibodies against the exposed COOH-terminal tail of Xenopus gp210 both inhibited pore formation. The arrest morphologies suggest an unanticipated direct function for the gp210 tail and its binding partners in the dilation of nascent nuclear pores.
To study gp210 function in the Xenopus system, we first had to identify Xenopus gp210. We identified a partial cDNA corresponding to the COOH terminus of Xenopus gp210 by using the rat gp210 sequence to search the Xenopus EST database (see Materials and methods; this EST clone had a mutation). We then generated the corresponding wild-type PCR product from a Xenopus oocyte cDNA library. This partial cDNA encoded a polypeptide of 251 residues that was 39% identical and 47% similar to the COOH-terminal 251 residues of rat gp210 (Fig. 1 A). As shown below, antibodies against the COOH terminus confirmed that this cDNA encoded gp210.
Recombinant gp210 tail polypeptide inhibits nuclear growth
We hypothesized that the function of gp210 may require interactions between its cytoplasmic tail and other nucleoporins. To test this hypothesis, we first expressed the 61-residue gp210 tail as a 32.8-kD fusion protein with glutathione-S-transferase (GST), from which the 6.8-kD tail was purified after cleavage by biotinylated thrombin (Fig. 1 B). The purified gp210 tail was added, at time zero, to nuclear assembly reactions consisting of reconstituted Xenopus egg cytosol and membranes plus demembranated sperm chromatin (Lohka and Masui, 1984; Newmeyer and Wilson, 1991). Purified gp210 tail was added to final concentrations ranging from 1 nM to 1 mM. Nuclei were allowed to assemble for 90 min and then examined by light microscopy (Fig. 1 C). Nuclear growth was fully inhibited by tail concentrations of 10 μM or more. The growth-inhibited nuclei had phase-dense membranes at the chromatin surface, but failed to expand or decondense the chromatin. Inhibition by exogenous gp210 tails was specific, because control nuclei assembled in the presence of an unrelated polypeptide (thrombin-cleaved, purified 10 μM GST) assembled and grew normally (unpublished data). Further experiments were done using the tail polypeptide at 10 μM, the lowest concentration tested that terminally arrested nuclear envelope growth. This represented a fivefold excess over endogenous gp210 (∼2 μM), as estimated from densitometry analysis of immunoblots of Xenopus membrane proteins and known concentrations of the recombinant Xenopus gp210 tail (unpublished data).
To determine the time course of inhibition, reactions were supplemented at time zero with 10 μM gp210 tail (or buffer control) and imaged after 5, 10, 20, 30, 60, and 90 min of assembly (Fig. 2 A). Membrane recruitment to the chromatin surface was neither delayed nor inhibited by exogenous gp210 tails. However, defects in nuclear growth and chromatin decondensation were obvious within 20–30 min, when buffer addition control nuclei became fully enclosed by a nuclear envelope and began enlarging. Nuclear growth requires functional NPCs and active nucleocytoplasmic transport (Macaulay and Forbes, 1996; Wiese et al., 1997). The growth arrest caused by exogenous gp210 tails suggested defects in NPC assembly or function.
Gp210 tail-inhibited nuclei fail to incorporate nucleoporins and lack lamin B3
To determine if tail-inhibited nuclei were defective for NPC assembly or nuclear import, we tested the incorporation of FG-repeat nucleoporins using mAb 414, and tested the accumulation of lamin B3, which depends on signal-mediated nuclear import (Loewinger and McKeon, 1988; Meier et al., 1991; Firmbach-Kraft and Stick, 1995). Aliquots were removed from assembly reactions after incubating for 5, 10, 20, 30, 60, and 90 min, and the nuclei were purified (see Materials and methods), resolved by SDS-PAGE, blotted, and probed for either nucleoporins (Fig. 2 B) or lamin B3 (Fig. 2 C) using mAb 414 and mAb S49F, respectively. The corresponding control and inhibited samples were processed for Western blotting in pairs so that the signals were directly comparable in qualitative terms. As a further control, blots were stripped and reprobed with an antibody that recognizes nuclear membrane LAP2 proteins (Fig. 2, B and C, bottom), or an antibody against Xenopus gp210 (Fig. 2 C, bottom; this antibody is described below). These experiments confirmed that nuclear membranes were recruited to chromatin. Several isoforms of LAP2 were detected, as expected (Gant et al., 1999). The reduced signals for both LAP2 and gp210 in inhibited nuclei were consistent with their smaller surface areas. Interestingly, the gp210 blots showed that gp210 accumulated over time in both the control and inhibited nuclei (Fig. 2 C), demonstrating that gp210 can accumulate in nuclei that lack at least a subset of soluble nucleoporins.
mAb 414 recognizes five nucleoporins (nup358, nup214/CAN, nup153, nup98, and p62), all of which carry the O-linked N-acetylglucosamine modification (Davis and Blobel, 1987; Aris and Blobel, 1989; Shah and Forbes, 1998). Nup358 was not detected by mAb 414 in our Western blots, presumably because it failed to transfer under our conditions. The other four nucleoporins accumulated rapidly over time in the positive control nuclei (Fig. 2 B, buffer), as expected. The tail-inhibited nuclei acquired p62, although not to control levels (Fig. 2 B, gp210 tail). Tail-inhibited nuclei appeared to be depleted of nup214, nup153, and nup98, despite near-normal accumulation of gp210 in these nuclei (Fig. 2 C).
These results suggested that gp210 tail polypeptides selectively blocked the recruitment of several soluble nucleoporins, but not gp210 itself. Nup214 normally localizes to NPC cytoplasmic filaments (Kraemer et al., 1994), whereas nup153 and nup98 localize to the NPC basket and proposed intranuclear filaments (Sukegawa and Blobel, 1993; Radu et al., 1995; Fontoura et al., 2001). Our conclusion that NPC assembly was blocked in tail-inhibited nuclei was independently supported by the greatly reduced import of lamin B (Fig. 2 C). To explain these defects, we hypothesized that our recombinant tail polypeptides competed for putative tail-binding nucleoporins, which could be either soluble or membrane associated. Our results thus implicate a putative gp210 tail–binding nucleoporin(s) as having an early role in NPC assembly.
Antibodies to the gp210 tail also arrest nuclear growth
We raised rabbit polyclonal antibodies against the predicted COOH-terminal 16 residues of Xenopus gp210 (Fig. 1 A, underlined). This serum (no. 3860) was used to probe immunoblots of Xenopus egg membrane and cytosol fractions. The immune antibodies specifically recognized a membrane protein that migrated on SDS gels at ∼195 kD (Fig. 3 A), similar to the antigen recognized by a proposed anti-Xenopus gp210 monoclonal antibody (Gajewski et al., 1996). Our 195-kD antigen was undetectable in the cytosolic fraction, as expected for an integral membrane protein. Recognition of the 195-kD membrane protein was specifically competed by pretreating serum 3860 with the peptide antigen (Fig. 3 A, +). When used for indirect immunofluorescence of cultured Xenopus epithelial (A6) cells, immune (but not preimmune) serum 3860 specifically stained the nuclear envelope in a punctate pattern (Fig. 3 B, Xgp210) that colocalized with the family of FG repeat nucleoporins recognized by monoclonal antibody 414 (Fig. 3 B, nups). Punctate nuclear envelope staining was also seen in nuclei assembled in Xenopus extracts (unpublished data). The homology, appropriate antigen size on gels, membrane cofractionation, and colocalization with NPCs at the nuclear envelope all verified that clone AW642061 represents Xenopus gp210, and further showed that our antibodies specifically recognized Xenopus gp210.
To independently inhibit gp210 tail function, we assembled nuclei in the presence of serum 3860 antibodies, which were predicted to bind the COOH-terminal tip of the endogenous gp210 tail. Immune or preimmune antibodies were added at time zero to nuclear assembly reactions, and assembly was monitored at different times by light microscopy (Fig. 4 A). Nuclei assembled in the presence of immune antibodies successfully recruited membranes, but the chromatin failed to decondense and a block to nuclear envelope growth was obvious by the 20-min time point (Fig. 4 A, Im). By light microscopy, the arrested phenotype was indistinguishable from that caused by recombinant gp210 tails.
Nuclei assembled in the presence of preimmune or immune antibodies were isolated, processed for SDS-PAGE, and immunoblotted in pairs (control and inhibited) for the incorporation of mAb 414–reactive nucleoporins and lamin B3 (Fig. 4, B and C). Similar to tail-inhibited nuclei, the antibody-inhibited nuclei incorporated detectable levels of only one FG repeat nucleoporin, p62 (Fig. 4 B), and failed to accumulate lamin B3 (Fig. 4 C), consistent with an early arrest of NPC assembly and lack of nuclear transport activity.
Tail- and antibody-inhibited nuclei are enclosed by flattened nuclear membranes
The above results suggest that exogenous tail peptides and gp210 antibodies both arrested nuclear assembly at a very early stage, with membranes attached to chromatin but no functional pores. Nuclear pore formation and NPC assembly can be assayed by scanning EM (SEM) as soon as patches of nuclear membranes flatten onto the chromatin surface (Wiese et al., 1997). In wild-type reactions, pores form asynchronously and rapidly, and many mature NPCs are already present by ∼30 min, when chromatin becomes enclosed by nuclear membranes (Wiese et al., 1997). To determine if our inhibitors disrupted these cytosol-dependent membrane fusion events, we imaged the inhibited and control nuclei (assembled for 2 h) by SEM (see Materials and methods). SEM visualization of control nuclei, assembled in the presence of buffer or preimmune serum, showed full-size nuclei with fused and flattened nuclear envelopes and mature NPCs, as expected (Fig. 5). Nuclei assembled with either gp210 tail polypeptides or anti-tail antibodies were growth arrested (Fig. 5, top), consistent with our light microscopy results, and were all enclosed by flattened and fused membranes, showing that nuclear vesicles fused normally. Importantly, these arrested nuclei lacked NPCs (Fig. 5; center and bottom). However, tiny dimples or holes with diameters of 3–7 nm were visible on the surface of tail-inhibited nuclei (Fig. 5, + tail, arrows in bottom panel). These tiny holes were scarce in the antibody-arrested nuclei. We concluded that our inhibitors arrested nuclear pore formation at two distinct stages, with tail polypeptides arresting after the formation of mini-pores. Antibodies against the gp210 tail appeared to arrest pore formation at an earlier stage.
TEM reveals novel arrested intermediates in pore formation
We next used TEM to visualize the arrested nuclear membranes in cross section (see Materials and methods). Positive control nuclei assembled in the presence of buffer (Fig. 6 A) or preimmune serum (Fig. 6 E) formed nuclear envelopes with mature NPCs, as expected. The membranes surrounding a mature NPC are shown (tangential section) and traced in Fig. 6, I and J. In contrast, nuclei arrested by the competing tail polypeptides (Fig. 6, B–D) or anti-gp210 antibodies (Fig. 6, F–H) had striking, unprecedented morphologies: the inner and outer membranes appeared to be fused at frequent intervals, forming discrete electron-dense foci (Fig. 6, B and F, arrows). In the mini-pores produced by exogenous tail polypeptides, the two membranes were almost parallel on an axis perpendicular to the nuclear envelope, and were separated by a gap of 1–5 nm (Fig. 6, C and D). Tracings of the membrane edge suggested that in two of three arrested structures shown, membrane fusion might have been completed to yield a narrow (1–5 nm) pore (Fig. 6 D, white dotted lines). Holes in this size range were barely detectable in the surface images shown in Fig. 5 (+ tail). We estimated that the frequency of mini-pores was lower in the SEM images, probably due to masking of smaller pores by the 2.5-nm layer of chromium.
Membranes arrested by anti-tail antibodies were pointy rather than parallel, and we were unable to trace membranes within the electron-dense gap (Fig. 6 H, white dotted lines). Few, if any, mini-pores were detected in antibody-arrested membranes by surface imaging (Fig. 5). Thus, exogenous tails and anti-tail antibodies produced subtly different arrest morphologies when viewed by TEM, independently supporting our surface imaging results. Antibodies appeared to inhibit either during or immediately after pore formation, whereas tail polypeptides appeared to inhibit later, during pore enlargement (dilation).
We next quantitated four definable morphological features in the TEM images (Fig. 7 A). These features were dimples (D), defined as short angular electron-dense indentations of the outer membrane; twinned (T) membranes, defined as ∼40-nm paired indentations of the inner and outer membranes (originally described by Maul et al., 1971); mature NPCs (M); and both types of unusual arrested structures (A) (only the pointy structure is shown in Fig. 7 A). We counted these features, measured the total length (microns) of nuclear envelope examined (Fig. 7 B), and graphed their frequencies per micrometer of nuclear envelope (Fig. 7 C). This quantitation showed that mature NPCs were largely absent from the inhibited nuclei, strongly suggesting that gp210 is required for NPC formation. Few, if any, (0–5%) arrested structures were present in control nuclei, but comprised the majority (70%) of structures seen in tail- or antibody-arrested nuclei (Fig. 7 B). The slightly higher frequency of total pore-related structures in the inhibited nuclei (Fig. 7 B, structures/micrometer) suggested that these structures accumulated stably. Twinned membranes were relatively abundant in control nuclei (34–36% of total structures), but significantly reduced (to 2%) in tail-inhibited nuclei and reduced by half (to 15%) in antibody-inhibited nuclei. Similar structures, with closely-apposed inner and outer membranes, were shown by Maul et al. (1971) to be highly correlated with forming (or disassembling) pores. Because twinned membranes were abundant in our control nuclei, but missing or reduced in arrested nuclei, we suggest that the gp210 tail and its partners might trigger the close apposition of the inner and outer nuclear membranes in addition to mediating pore dilation.
Finally, we considered the possibility that our inhibitors might destabilize or revert mature NPCs, rather than (or in addition to) blocking de novo pore assembly. To test this model, we first preassembled nuclei for 90 min and then added 10 μM tail polypeptides, gp210 antibodies, or the corresponding controls. We imaged these reactions by SEM immediately and 1 h later. The control and experimental nuclei all remained intact with mature NPCs and were indistinguishable by SEM (see Fig. S1, available at http://www.jcb.org/cgi/content/full/jcb.200108145/DC1), suggesting that our reagents did not grossly disrupt the structure or function of mature NPCs.
We report the effects of two independent reagents that targeted the exposed cytosolic tail of gp210. Neither reagent affected NPC morphology when added to assembled nuclei. Importantly, neither reagent disrupted the membrane fusion events required to generate flattened nuclear membranes. However, when added to assembling nuclei, both reagents efficiently arrested nuclear pore formation at an early stage. To explain the mechanisms of arrest, we propose that exogenous tail polypeptides compete for tail-binding nucleoporins and block their binding to endogenous gp210. We propose that the anti-tail antibodies either cross-link endogenous gp210 tails and block their function sterically, or block their interactions with tail-binding partners, or both. The antibodies might arrest pore formation at a slightly earlier stage than exogenous/competing tail polypeptides. The antibody-arrested structures were electron dense, and individual membranes could not be traced, whereas polypeptide-inhibited membranes appeared to form narrow mini-pores 1–5 nm in diameter, which might permit limited diffusion. These tail-inhibited nuclei accumulated lamin B3 to low levels, consistent with (but not proving) the presence of narrow diffusion-competent pores. Structures consistent with tiny pores (∼3–7-nm diameter) were also seen by surface imaging of tail-inhibited nuclear membranes. Importantly, our inhibitors also reduced the formation of closely-apposed twinned membranes, proposed by Maul et al. (1971) to be precursors to pore formation. These results strongly support our major conclusion that gp210 has early roles in nuclear pore formation.
Antibodies inhibited pore formation more severely than competing tails. When gp210 bound, these antibodies might interfere sterically with neighboring proteins that function independently of gp210. Alternatively, bound antibodies might cluster gp210 or induce other conformational changes that disrupt its lumenal domain. Antibodies that directly bind the lumenal domain of gp210 in vivo inhibit both active transport and passive diffusion (Greber and Gerace, 1992). We therefore speculate that our anti-tail antibodies might also functionally disrupt the lumenal domain of gp210.
Inhibitors have been useful to dissect the NPC assembly pathway. Two inhibitors, GTPγS and NEM, block the formation of fused, flattened membrane cisternae (Boman et al., 1992; Newport and Dunphy, 1992; Macaulay and Forbes, 1996). To study pore formation in vitro, one must start with fused flattened membranes. The fusion of nuclear/ER membranes involves the GTP-binding protein Ran (Hetzer et al., 2000; Zhang and Clarke, 2000) and p97 ATPase complexes (Hetzer et al., 2001). Nuclear membranes can self-assemble as stacks of flattened cisternae in the absence of chromatin (Dabauvalle et al., 1991; Meier et al., 1995). Membrane fusion is delayed by a Ca2+ buffer, BAPTA (Sullivan et al., 1993; Shumaker et al., 1998). BAPTA also completely blocks pore formation when added at time zero (Macaulay and Forbes, 1996), or causes “star ring” intermediates to accumulate if added later in assembly (Goldberg et al., 1997). Wheat germ agglutinin, a lectin that binds O-linked N-acetylglucosamine–modified nucleoporins, also blocks pore formation (Meier et al., 1995; Goldberg et al., 1997). Our present findings add two novel intermediates to the biochemically-defined pathway of pore formation. BAPTA-arrested nuclei have flattened parallel poreless membranes that lack FxFG repeat nucleoporins. In contrast, reagents that disrupt gp210 tail function appear to slow a proposed membrane apposition event before fusion, but then allow porogenic membrane fusion to initiate or progress to the mini-pore stage. We propose that the BAPTA-arrested and gp210 tail–arrested intermediates bracket the porogenic fusion event, providing direct experimental access to the mechanisms of pore formation and dilation.
Nucleoporin p62 may be recruited early during porogenesis
Gp210-arrested nuclei accumulated p62, but failed to accumulate nup214/CAN, nup98, or nup153. The p62 result was probably not an artifact, because we gradient purified our nuclei before running gels for Western blotting. There is currently no evidence that p62 binds directly to the gp210 tail. Thus, p62 might be recruited to nascent pores independent of gp210. The possible incorporation of p62 into gp210 tail–arrested nuclei is intriguing, because p62 and its binding partners form a ring structure (Macaulay and Forbes, 1995; Hu et al., 1996) and are critical for transport activity (Finlay et al., 1991).
Gp210 functions early in pore formation and its tail is involved in nuclear pore dilation
Our results suggest that interactions between the exposed tail of gp210 and its putative binding partners might trigger an important early event, namely the close apposition of inner and outer nuclear membranes before fusion. Although membrane fusion per se was not blocked, reagents that interfered with the gp210 tail or its putative partners also caused pore formation to arrest at two subsequent stages. For the antibody-arrested stage, we could not determine if membrane fusion was complete, due to the electron density of these pointy structures. However the mini-pores that accumulated in the presence of exogenous tail polypeptides appear to represent a normal intermediate in which the pore has formed, but cannot dilate. Based on previous SEM studies, nascent pores are thought to progressively increase in diameter from ∼1–5 nm (dimples) to ∼30–40 nm (stabilizing pores) and ∼50–70 nm (mature pores; Maul, 1977; Goldberg et al., 1997). This model is supported by mini-pores, which have diameters of 3–7 nm when visualized by SEM and 1–5 nm when visualized by TEM. (Note that the 2.5-nm chromium coat on SEM samples would obscure pores <3 nm in diameter). Thus, our findings suggest direct roles for gp210 and its partners in triggering pore formation and dilating nascent pores. We speculate that soluble tail-binding nucleoporins play a positive role by forming spokes or struts between gp210 tails, which insertionally expand the size of the pore. Note that if tail-binding partners had inhibitory roles, pores would expand uncontrollably in the presence of competing gp210 tail polypeptides. We can therefore rule out inhibitory roles for tail-binding nucleoporins.
Pore dilation: a regulated event during nuclear pore formation?
Kozerski et al. (2000) proposed that the hydrophobicity of the tail domain of influenza HA is crucial for pore dilation, because dilation was blocked by a hydrophilic tail. The only known function of the HA tail is passive: to exist, and not interfere with fusion or pore dilation. Thus, the final diameter of the pore may be irrelevant to a virus. Gp210 tails from Xenopus, rat, and Caenorhabditis elegans also contain 30–45% hydrophobic residues, although their positions are not conserved. However, we conclude that the tail of gp210 is not passive, because our results suggest that nuclear pore dilation is coupled to the assembly of soluble nucleoporins. Without this link, NPCs might assemble incorrectly, or worse, pores themselves might enlarge and fenestrate the nuclear envelope, destroying its integrity as a boundary. Our results strongly suggest that the tail of gp210 plus one or more unidentified tail-binding nucleoporins mediate the close apposition of nuclear membranes, and nuclear pore dilation. Possible mechanisms to control pore dilation include the oligomerization of gp210 lumenal domains into a grommet that encircles the pore membrane domain, or the assembly of tail-binding struts, or both.
Nuclear membrane fusion and pore dilation could be coupled efficiently if the same protein (or two interacting proteins) mediated both events. Our results thus do not rule out, and moreover support, the hypothesis that gp210 directly mediates porogenic membrane fusion (Wozniak et al., 1989; Greber et al., 1990; Wozniak and Blobel, 1992). However, our results restrict the still hypothetical, fusogenic activity of gp210 to its highly-conserved lumenal domain (Cohen et al., 2001). Testing these models for gp210 function will require determining the gp210-null phenotype and identifying nucleoporins that bind its tail.
Materials And Methods
Partial cDNA cloning of Xenopus gp210
We identified an 807-bp cDNA clone (AW642061) in the Blackshear/Soares normalized database for Xenopus egg library ESTs and obtained the EST clone from Research Genetics Inc. After sequencing this DNA, we compared its predicted ORF for the gp210 tail with rat gp210. The predicted Xgp210 tail had 23 fewer amino acids, due to an extra nucleotide that created a stop codon. To determine if this stop codon was a PCR artifact in the EST clone, we did a PCR reaction from a stage I/II Xenopus oocyte cDNA library (from A. Wolffe, National Institutes of Health, Bethesda, MD) using primers 5′-TGTAAAACGACGGCCAGT-3′ and 5′-CAGGAAACAGCTATGACC-3′. This partial cDNA was subcloned into the pCRT7 TOPO TA vector (Invitrogen) to generate clone Xp-gp210, which was DNA sequenced (unpublished data) and shown to lack the truncating mutation. The longer amino acid sequence predicted by our clone, Xp-gp210, is shown in Fig. 1.
Expression construct for GST–gp210 tail fusion
Different primers were used to generate a cDNA encoding only the cytoplasmic tail (COOH-terminal 61 residues) of Xenopus gp210 from the Xenopus stage I/II oocyte DNA library noted above. These new primers added BamH1 and Sma1 sites to the 5′ and 3′ regions of the gp210 tail, respectively (sense primer, 5′-GGGGCAGCAATAACCGGATCCATTATGTCA-3′; antisense primer, 5′-ACCCCAAATCCCGGGACAACATCGTTGTAC-3′). The PCR product and pGEX-2T vector (Amersham Pharmacia Biotech) were both sequentially digested with BamH1 and Sma1, gel purified, and ligated together. Correct insertion of the Gp210 tail fragment into pGEX-2T was verified by double-stranded DNA sequencing (unpublished data). The resulting construct, named Xenopus tail (Xt)-gp210, was expressed in bacteria as a recombinant protein consisting of the gp210 tail fused at its NH2 terminus to GST.
Protein expression and thrombin cleavage
The pGEX-2T vector containing the gp210 tail insert was transformed into competent Escherichia coli cells (strain BL21 DE3 pLysS; Novagen, Inc.) and grown in LB media containing 40 μg/ml ampicillin and 36 μg/ml chloramphenicol to an OD600 of 0.5. Protein expression was induced for 3 h using 1 mM IPTG. Cells were pelleted and the fusion protein purified by chromatography on glutathione-Sepharose according to manufacturer specifications (Amersham Pharmacia Biotech). The purified fusion protein was incubated with biotin-conjugated thrombin (Novagen, Inc.) for 4 h at 22–24°C. Biotinylated thrombin was removed by binding to streptavidin-Sepharose (Novagen, Inc.). Isolated gp210 tail polypeptide was either used fresh or stored in aliquots at –20°C.
A peptide corresponding to the predicted COOH-terminal 16 residues of Xenopus gp210 (GSFYFQRLLYPNCPLV) was synthesized, purified by reverse phase HPLC with the use of a C18 analytical column, and conjugated to keyhole limpet hemocyanin by Boston Biomolecules. This peptide was used to immunize rabbit 3860. mAb 414, which is specific for FG repeat nucleoporins, was from Covance Research Products. The mAb against Xenopus lamin B3, 46F7, was the gift of Reimer Stick (Göttingen University, Göttingen, Germany). Rabbit serum 2806 was raised against recombinant Xenopus LAP2 residues 1–165, comprising the “constant” region of Xenopus LAP2 (Gant et al., 1999; Dechat et al., 2000). Rabbit serum production was done by Covance Research Products.
Nuclear assembly reactions
Demembranated Xenopus sperm chromatin (used at 1,000 sperm/μl extract) was prepared as previously described (Newmeyer and Wilson, 1991) and incubated at 22–24°C in reconstituted cell-free extracts of Xenopus eggs supplemented with 2 mM ATP, 10 mM creatine phosphate, and 50 μg/ml creatine kinase (Hutchison et al., 1988). Assembly reactions were then supplied with purified recombinant Xenopus gp210 tail, immune or preimmune serum 3860, or an equal volume of wash buffer (50 mM Hepes, pH 7.6, 50 mM KCl, 2.5 mM MgCl2, 1 μg/ml leupeptin, 1 μg/ml aprotinin) as a control. These reagents contributed no more than 10% to the final volume.
Isolation of nuclei assembled in vitro
To isolate in vitro–assembled nuclei, 50-μl assembly reactions were diluted with 1 ml of ice-cold nuclear isolation buffer (NIB) (20 mM Tris-HCl, pH 7.4, 70 mM KCl, 5 mM MgCl2, 2 mM DTT, 2% polyvinylpyrolidone). Diluted reactions were pooled and layered over 60% Percoll (in NIB) and centrifuged at 3,000 g for 10 min at 4°C. Material was collected from the NIB–Percoll interface, supplemented with ice-cold NIB to a volume of 1 ml, layered over 0.5 ml of 25% sucrose (wt/vol in NIB), and centrifuged at 2,500 rpm in a benchtop microfuge to yield a nuclear pellet. The nuclear pellet was resuspended in SDS sample buffer and boiled for 3 min. For analysis by SDS-PAGE and Western blotting, the equivalent of 2 × 105 nuclei was loaded per lane, as confirmed by removing an aliquot and counting the number of pelleted nuclei.
SDS-PAGE, immunoblotting, and densitometry
Proteins were resolved on precast 4–12% Bis-Tris polyacrylamide gels (Novex) by electrophoresis at 200 V for 1 h, and then either stained with Coomassie blue or transferred to a polyvinylidene difluoride membrane. Immunoblots were probed with a 1:500 dilution of anti-gp210 tail serum 3860 in PBS-T (PBS containing 0.1% Tween-20). Alternatively, blots were probed for 1 h at 22–24°C with (a) mAb 414 diluted to 1:5,000 with PBS-T or (b) monoclonal antibody 46F7 at a 1:500 dilution to detect Xenopus lamin B3. After three washes in PBS-T, blots were incubated for 1 h at 22–24°C with 1:25,000 dilution of HRP-conjugated secondary antibodies (anti–rabbit or anti–mouse; Pierce Chemical Co.). Membranes were washed as before and visualized by ECL (Amersham Pharmacia Biotech). To probe for LAP2, each blot was stripped by incubation with 15% H2O2 (in PBS) for 30 min at 22–24°C, washed for 15 min in PBS-T, incubated with serum 2806 diluted 1:50,000, and processed as described above. LAP2-probed blots were then restripped and blotted for gp210 as described above.
Aliquots of assembly reactions were diluted fourfold with Fix-Stain buffer (15 mM Pipes, pH 7.2, 80 mM KCl, 5 mM EDTA, 15 mM NaCl, 3.3% formaldehyde, and 10 μg/ml Hoescht 33258). Samples were imaged on a Nikon Microphot using a cooled CCD camera with UV filters to detect DNA and phase contrast to visualize nuclear membranes. Xenopus A6 cells were cultured in a 60% solution of Leibovitz L15 media containing 10% FBS, MEM nonessential amino acid solution (1×), penicillin, and streptomycin (each at 0.1 mg/ml). Cells were fixed in PBS with 4% paraformaldehyde, and stained for indirect immunofluorescence using mAb 414 at 1:5,000, or serum 3860 at 1:100. Goat ant–rabbit secondary antibodies were conjugated with either Cy3 or FITC and used at 1:500. DNA was stained with Hoescht. Images were collected and processed using IPLab Spectrum software.
Field emission SEM
A 4-μl aliquot of each assembly reaction was diluted in 500 μl buffer (150 mM sucrose, 50 mM KCl, 2.5 mM MgCl2, 50 mM Hepes, pH 8, 1 mM DTT, 1 μg/ml each of aprotinin and leupeptin) and sedimented at 3,000 g for 10 min at 4°C onto acetone-washed silicon chips (Fred Pella). Chips were incubated for at least 10 min in SEM fix buffer (150 mM sucrose, 50 mM KCl, 80 mM Pipes-KOH, pH 6.8, 1 mM MgCl2, 0.25% gluteraldehyde (EM grade), 2% paraformaldehyde [Sigma-Aldrich]). Chips were then washed in 0.2 M sodium cacodylate, pH 7.2, post-fixed for 10 min in 1% OsO4 (in 0.2 M Na cacodylate, pH 7.2), washed in water, stained for 10 min in 1% uranyl acetate (in water), and dehydrated in a graded ethanol series (30%, 50%, 70%, 95%, and two times at 100% ethanol, followed by two times at 100% Arklone). Samples were then critical point dried from high purity CO2 using Arklone (trichlorotrifluoroethane; ICI) as a transitional solvent. Dried samples were sputter coated with a 2.5-nm coat of chromium in an Edwards Auto 306 coating unit. Samples were viewed on the top stage of an ABT dual stage SEM (DS-130F) (Topcon Corp.) at a 30-kV accelerating voltage.
Nuclear assembly reactions (50 μl) were diluted in 3% gluteraldehyde, 1.5% paraformaldehyde, 0.1 M sodium cacodylate, and 3 mM MgCl2 to a final volume of 1 ml, and fixed for 1 h on ice. Nuclei were then pelleted in a Beckman Coulter Microfuge E horizontal benchtop centrifuge for 1 min at high speed at 4°C. Pellets were incubated on ice four times (10 min each) with 0.1 M sodium cacodylate and 3 mM MgCl2, repelleted, and then incubated with 2% osmium tetroxide (in 0.1 M sodium cacodylate with 3 mM MgCl2) for 1 h on ice. Pellets were washed twice with water (5 min each), incubated in 2% aqueous uranyl acetate at 22–24°C, and then dehydrated through a graded ethanol series (30, 50, 70, 95, 100%). After three 10-min incubations in 100% ethanol, samples were embedded in Spurrs, sectioned, and viewed on a Philips microscope at 60 kV.
Online supplemental material
Fig. S1 (available online at http://www.jcb.org/cgi/content/full/jcb.200108145/DC1) shows that exogenous gp210 tail polypeptides and anti-tail antibodies have no detectable effects on preformed NPCs. Nuclei were preassembled for 90 min, treated for 1 h with buffer or 10 μM tail polypeptide, or immune or preimmune serum 3860, and imaged by SEM.
We thank R. Stick for the antilamin antibody, and M. Matunis, C. Machamer, K.K. Lee, J. Cronshaw (Johns Hopkins University), Y. Gruenbaum, M. Cohen (Hebrew University of Jerusalem, Jerusalem, Israel), M. Zastrow (Johns Hopkins University), and an anonymous reviewer for interesting discussions or comments on the manuscript. We are grateful to S. Rutherford, M. Goldberg, and T. Allen (Paterson Institute, Manchester, England) for their help with SEM sample preparation and image collection.
This work was funded by a postdoctoral fellowship from the Wellcome Trust, UK (to S.P. Drummond), and a research grant from the National Institutes of Health (RO1-GM48646 to K.L. Wilson).
The online version of this article includes supplemental material.
S.P. Drummond's present address is Department of Structural Cell Biology, Paterson Institute for Cancer Research, Christie Hospital, Wilmslow Road, Manchester M20 9BX, UK.
Abbreviations used in this paper: GST, glutathione-S-transferase; HA, hemagglutinin; NIB, nuclear isolation buffer; NPC, nuclear pore complex; SEM, scanning EM; TEM, transmission EM.