The Nuclear Pore Complex

The nuclear pore complex is the exclusive gateway through the nuclear envelope, a double membrane continuous with the endoplasmic reticulum. At several thousand sites per mammalian nucleus, the outer and inner nuclear membranes curve and fuse into a single continuous membrane, and an NPC is threaded through each fusion pore. Structurally, the NPC is built with eightfold rotational symmetry: eight identical protomers arranged around a central channel, giving the whole assembly its characteristic symmetric ring appearance in electron micrographs. It is enormous by molecular standards — about 120 megadaltons in vertebrates (roughly 60 MDa in yeast, which has a smaller version), built from roughly 500-1000 individual protein subunits drawn from only about 30 distinct nucleoporin genes, each present in multiples of eight. Functionally, the NPC behaves as a dual-mode filter. For small, uncharged solutes and ions, and for proteins below roughly 40 kDa, it behaves like a passive channel: molecules diffuse through in either direction with a rate that falls off as size increases, with an effective pore diameter for passive diffusion of about 5 nm. For everything else — most functional proteins, all messenger RNAs, transfer RNAs, and pre-ribosomal particles — passage requires an active, signal-dependent process. This is not merely size exclusion; the pore's aqueous channel is wide enough (up to ~40 nm at its most dilated) to physically admit huge cargoes like 40S and 60S ribosomal subunits, provided those cargoes carry the correct molecular "passport." That passport takes the form of short peptide sequences: nuclear localization signals (NLS), first defined on the SV40 large T-antigen, direct import; nuclear export signals (NES), leucine-rich sequences first defined on HIV Rev protein, direct export. These signals are read by a family of soluble transport receptors called karyopherins (importins and exportins), which shuttle cargo through the pore by binding directly to the disordered FG-nucleoporin meshwork that fills the channel. Directionality — import receptors moving one way, export receptors the other — is not built into the pore itself but is imposed by a small GTPase, Ran, whose GTP-bound and GDP-bound forms are segregated to opposite sides of the nuclear envelope. This is the central conceptual payoff of the whole system: a symmetric, apparently non-directional pore achieves net vectorial transport purely because the regulatory protein Ran is asymmetrically distributed.

1. Architecture: the scaffold, the membrane ring, and the FG mesh. The NPC scaffold consists of concentric rings viewed from the nuclear-cytoplasmic axis. A cytoplasmic ring faces the cytoplasm and carries cytoplasmic filaments; a nuclear ring faces the nucleoplasm and anchors a basket-like structure, the nuclear basket, built from Nup153, Nup50, and TPR filaments that converge in a distal ring. Sandwiched between these are one or more inner rings that form the central spoke structure. Nucleoporins that build these rings — Nup107-160 complex members (the "Y-complex"), Nup93, Nup205, Nup188, and others — have folded, structured domains (β-propellers and α-solenoids) and are collectively called scaffold nucleoporins or structural Nups; they resemble, and are evolutionarily related to, coat proteins of vesicle coats like COPII and clathrin, consistent with the idea that the NPC and vesicular coats share a deep evolutionary origin (the "protocoatomer" hypothesis). Threaded through and anchored to this scaffold is a second class of nucleoporins, the FG-nucleoporins, so named because they contain multiple repeats of phenylalanine-glycine (FG) dipeptides embedded in long stretches of intrinsically disordered polypeptide. These FG repeats — in Nups like Nup62, Nup98, Nup214, Nup153 — project into and fill the central channel, forming a dynamic, cohesive meshwork often described (from two complementary models) as a "virtual gate" polymer brush or, given self-association of FG repeats via hydrophobic and cation-π interactions, a phase-separated hydrogel. Experiments reconstituting purified FG domains in vitro (notably from yeast Nsp1) showed they spontaneously form saturated hydrogels through which small proteins pass freely but through which folded, "inert" GFP-sized proteins are excluded — unless fused to a karyopherin-binding NLS.

2. Passive diffusion versus active, receptor-mediated transport. Ions, metabolites, and small proteins under about 40 kDa cross the FG mesh by simple diffusion, with permeability decreasing steeply as molecular mass approaches and exceeds this threshold — there is no hard cutoff, but a continuous decline, so that a 60 kDa protein diffuses through slowly and a 500 kDa complex essentially never does on its own. Cargo that must move efficiently and unidirectionally instead displays a signal peptide:

• A classical NLS is a short stretch rich in basic residues (e.g., the SV40 large T-antigen sequence PKKKRKV, a "monopartite" NLS; histones and nucleoplasmin carry "bipartite" NLS motifs with two basic clusters separated by a spacer).
• The NLS is recognized by importin-α, which in turn binds importin-β1; importin-β1 is the subunit that actually engages the FG-nucleoporins and ratchets the complex through the channel.
• A leucine-rich NES is recognized directly by the exportin CRM1 (exportin-1), which also engages FG repeats to cross in the opposite direction.

Karyopherins move cargo through the mesh by transiently and weakly binding successive FG repeats, effectively "surfing" or hopping along the meshwork — a low-affinity, high-avidity mechanism that permits fast translocation (an NPC can transport roughly 100-1000 macromolecules per second per pore under demanding conditions) while still excluding non-receptor-bound proteins that lack any FG affinity.

3. The RanGTP gradient — why transport has a direction. The karyopherin/FG-Nup interaction by itself does not explain directionality; importin-β1 could in principle carry cargo either way. Directionality is imposed entirely by the small GTPase Ran, whose nucleotide state differs sharply between the two compartments:

• In the nucleus, chromatin-bound RCC1, Ran's guanine nucleotide exchange factor (GEF), continuously converts RanGDP to RanGTP.
• In the cytoplasm, RanGAP1 (aided by RanBP1/RanBP2) stimulates Ran's intrinsic GTPase activity, converting RanGTP to RanGDP.

The result is a steep, persistent gradient: RanGTP is high in the nucleus, low in the cytoplasm. This asymmetry, maintained at the cost of GTP hydrolysis, is the sole energetic input that makes nuclear transport vectorial (the FG-mesh crossing step itself does not require ATP or GTP hydrolysis directly).

• Import: Importin-β/α binds NLS-cargo in the cytoplasm (low RanGTP) and carries it through the pore. In the nucleus, RanGTP binds importin-β directly, displacing the cargo — releasing it inside the nucleus. Importin-β–RanGTP is then recycled back to the cytoplasm, where RanGAP-stimulated hydrolysis releases importin-β for another round.
• Export: CRM1 binds NES-cargo cooperatively together with RanGTP, forming a trimeric export complex — only assembling efficiently where RanGTP is abundant, i.e., in the nucleus. After crossing to the cytoplasm, RanGAP-driven hydrolysis of RanGTP to RanGDP disassembles the complex, releasing cargo into the cytoplasm.

Ran-GDP itself is returned to the nucleus by a dedicated carrier, NTF2, completing the cycle. The net effect: cargo-bound importins release cargo in the nucleus and cargo-bound exportins release cargo in the cytoplasm, purely because that is where the RanGTP concentration crosses the threshold that triggers complex assembly or disassembly.

4. Cargoes matched to the machine: an example transit. Consider a newly synthesized ribosomal protein, made on cytoplasmic ribosomes and destined for assembly into a pre-ribosomal particle in the nucleolus. It carries a classical NLS. Importin-α binds the NLS; importin-β1 binds importin-α and engages the cytoplasmic filaments' FG repeats, then "walks" the complex through the central channel via a chain of transient FG contacts. On the nuclear face, RanGTP (abundant because RCC1 continuously regenerates it on chromatin) binds importin-β1 and displaces importin-α/cargo, releasing the ribosomal protein into the nucleoplasm. Importin-β–RanGTP diffuses back through the pore; cytoplasmic RanGAP triggers GTP hydrolysis, releasing free importin-β for another import cycle and leaving RanGDP to be recycled back into the nucleus by NTF2. Conversely, an assembled 40S pre-ribosomal subunit exiting the nucleus after nucleolar maturation is escorted by the export receptor Crm1 (recognizing NES-bearing adaptor proteins on the particle) together with RanGTP; the trimeric export complex crosses to the cytoplasm, where RanGAP-driven hydrolysis disassembles it and releases the subunit for final cytoplasmic maturation — a transport step of direct relevance to The Nucleolus and Ribosome Assembly.

Feldherr's gold-particle microinjections (1960s-1970s) established size-dependent gating. Carl Feldherr injected colloidal gold particles of defined, calibrated diameters into the cytoplasm of amoeba and oocytes and tracked their fate by electron microscopy. Particles below roughly 9 nm diffused freely into the nucleus through structures visible as pores in the envelope; larger particles were excluded unless coated with proteins carrying nuclear-targeting information, in which case they were seen concentrated within the central channel of individual pores en route to the nucleus. This was the first direct physical demonstration that the nuclear envelope's pores are size-selective channels capable of actively conveying oversized cargo when the cargo is appropriately "labeled."

Microinjection into Xenopus oocytes defined the NLS and showed transport is signal-mediated, not size-mediated alone. Robert Laskey, Colin Dingwall, and colleagues studied nucleoplasmin, an abundant nuclear protein of Xenopus oocytes, and showed that its nuclear accumulation depended on a specific basic tail domain; removing it abolished nuclear uptake, and fusing it to otherwise cytoplasmic reporter proteins was sufficient to drag them into the nucleus. Around the same time, Dingwall and Laskey, and independently Kalderon, Roberts, Richardson, and Alan Smith working with SV40 large T-antigen, mapped the essential signal to a short stretch of basic amino acids (PKKKRKV in T-antigen) — the first molecularly defined nuclear localization signal. Critically, these experiments used microinjection directly into either the cytoplasm or nucleus of live oocytes: proteins injected into the cytoplasm with an intact NLS accumulated in the nucleus over minutes, while NLS-deleted mutants stayed cytoplasmic, and injecting the same protein directly into the nucleus (bypassing the envelope) trivially resulted in nuclear localization regardless of signal — controlling for the possibility that "nuclear proteins" were simply retained wherever placed rather than actively imported.

In vitro reconstitution in digitonin-permeabilized cells dissected the biochemical requirements for transport. Using cultured cells whose plasma membrane, but not nuclear envelope, is selectively permeabilized with digitonin (which extracts cholesterol from the plasma membrane preferentially), researchers including Dirk Görlich and Larry Gerace could add back purified, defined cytosolic components to fluorescently labeled NLS-cargo and directly observe which factors were necessary and sufficient for import. This approach identified and validated the requirement for importin-α, importin-β, Ran, and NTF2, and showed that import could be reconstituted from purified components, that it required GTP hydrolysis by Ran (non-hydrolyzable GTP analogs blocked cargo release in the nucleus), and that RanGAP-driven GTP hydrolysis in the cytoplasm was needed to recycle transport factors.

FG-domain hydrogel reconstitution demonstrated the biophysical basis of the selectivity filter. Dirk Görlich's group (Frey and Görlich, 2007) expressed and purified the isolated FG-repeat domain of the yeast nucleoporin Nsp1 and showed it spontaneously self-assembles in vitro into a saturated hydrogel. Small, "inert" proteins (green fluorescent protein alone) failed to enter this hydrogel efficiently, mirroring their exclusion from real nuclear pores, while the same GFP fused to importin-β or bearing an NLS-importin complex entered and partitioned into the gel readily. This directly demonstrated, outside any cellular context, that FG-repeat sequences alone are sufficient to reconstitute a selective sieve with properties matching the intact NPC — strong evidence that FG-Nup self-association, not scaffold geometry alone, creates the size/binding-based gate.

Cryo-electron tomography and cryo-EM resolved the scaffold architecture. Beginning in the 2000s and maturing through composite structures published in the 2010s-2020s (work from groups including Beck, Hoelz, Rout, Chait, and Schwartz), cryo-electron tomography of intact nuclear envelopes combined with high-resolution crystal and cryo-EM structures of purified nucleoporin subcomplexes (notably the Y-complex/Nup107-160 complex) allowed near-atomic composite models of the entire NPC scaffold to be built. These structures confirmed the eightfold symmetric ring architecture, quantified copy numbers of individual Nups (many present in multiples of 8, 16, or 32 per pore), and revealed the structural homology between coat nucleoporins and vesicle coat proteins (COPII, clathrin), supporting a shared evolutionary ancestry for the NPC and vesicle-budding machinery.

Example 1 — Estimating transport flux through a single pore. A rapidly growing HeLa-like cell has roughly 3,000-5,000 nuclear pore complexes per nucleus, and the nucleus must import on the order of 10^6 histone and ribosomal-protein molecules per generation time of ~24 hours to keep pace with DNA replication and ribosome biogenesis, alongside constant housekeeping traffic. Take a simplified estimate: suppose a cell needs to import 2 × 10^6 protein molecules in a 20-minute (1,200 s) window during S-phase histone synthesis, and it has 4,000 NPCs. Flux per pore = (2 × 10^6 molecules) / (4,000 pores × 1,200 s) = (2 × 10^6) / (4.8 × 10^6) ≈ 0.42 molecules/second/pore This is well within experimentally measured single-pore capacities (estimates from live-cell imaging and biochemical flux measurements put maximal per-pore transport rates at roughly 100-1,000 molecules/second under saturating conditions), so a real cell easily has the capacity to meet histone demand through active NLS-mediated import — with enormous headroom to also handle constitutive housekeeping transport, mRNA export, and ribosomal subunit export simultaneously. This calculation illustrates why the "one-gate, many-cargoes" system does not create a bottleneck: the FG-mesh's per-pore throughput vastly exceeds typical physiological demand, and cells modulate total capacity mainly by changing pore number (nuclear pore density roughly doubles during Xenopus oocyte maturation, for instance) rather than needing each pore to work harder.

Example 2 — Diagnosing a transport-signal mutant. A researcher studies a fluorescently tagged transcription factor, "Protein X," normally found concentrated in the nucleus of transfected cultured cells. Three engineered constructs are tested by fluorescence microscopy after transfection:

Reasoning through this: wild-type localization to the nucleus, despite Protein X (65 kDa, well above the ~40 kDa passive diffusion ceiling) being too large for unaided diffusion, already implies active signal-dependent import. Deleting the basic cluster abolishes concentrated nuclear accumulation — the protein now distributes according to its passive-diffusion equilibrium, appearing in both compartments at similar levels because at 65 kDa some slow passive leakage still occurs across the ~40 nm dilated channel and because nuclear envelope breakdown during mitosis re-equilibrates soluble proteins each cell cycle in dividing cultured cells. Re-supplying any competent NLS (even a heterologous one from an unrelated viral protein, SV40 T-antigen) restores active nuclear concentration, demonstrating that (a) the deleted basic cluster was a genuine, portable NLS — sufficient in isolation to redirect a passenger protein — and (b) the import machinery (importin-α/β, Ran) recognizes the sequence motif itself rather than anything specific to Protein X's native context. This is precisely the logic Kalderon and colleagues used originally to map the SV40 large T-antigen NLS, and it remains the standard operational test for "is this a bona fide localization signal."

Günter Blobel, working at Rockefeller University from the late 1960s through his death in 2018, built the conceptual framework — the "signal hypothesis" — that a protein's destination within the cell is encoded in its own amino acid sequence, read by dedicated cellular machinery. Although Blobel's Nobel-winning work (1999, Physiology or Medicine) centered on signal sequences directing proteins to the endoplasmic reticulum, the same logic he championed — that topogenic signals within a protein sequence direct it through a channel to a specific compartment — was directly extended by his own lab and by others to explain nuclear import, culminating in identification of NLS motifs and the karyopherin transport receptors. Blobel's laboratory also produced foundational structural and biochemical characterization of the nuclear pore complex itself, including early identification of individual nucleoporins and models of NPC architecture using electron microscopy of Xenopus oocyte nuclear envelopes — chosen because a single oocyte nucleus, the germinal vesicle, contains tens of millions of nuclear pores and can be manually dissected free of cytoplasm, making it an unusually generous source of intact nuclear envelope for both microscopy and biochemistry.

Nuclear transport is disrupted in disease and exploited therapeutically. Mutations or mislocalization of nucleoporins (such as NUP98 gene fusions) drive subsets of acute myeloid and lymphoblastic leukemia by generating aberrant fusion transcription factors. In amyotrophic lateral sclerosis and frontotemporal dementia linked to the C9orf72 repeat expansion, nucleocytoplasmic transport itself is disrupted — repeat RNA and dipeptide-repeat proteins sequester nucleoporins and Ran-cycle components, impairing global nuclear import/export and contributing to neurodegeneration, a finding that has made restoring nuclear transport an active drug-discovery target. The exportin CRM1 is clinically targeted directly: selinexor, an oral CRM1 inhibitor, is FDA-approved for multiple myeloma and diffuse large B-cell lymphoma, working by trapping tumor-suppressor proteins (like p53) in the nucleus where they can act, rather than being exported and degraded. Many viruses subvert nuclear transport machinery for their own replication — HIV-1 uses karyopherin-related pathways to import its pre-integration complex through nuclear pores even in non-dividing cells, and influenza and herpesviruses manipulate nucleoporins to redirect host transport toward viral gene expression, making the NPC an active target for antiviral strategies. In biotechnology, engineered NLS/NES tags are now routine tools — appending an SV40 NLS to a CRISPR-Cas9 construct, for instance, is what allows the nuclease to reach genomic DNA efficiently in gene-editing applications, directly descended from the original Kalderon/Dingwall experiments described above.

Ask the tutor: How does the nuclear pore complex distinguish a folded, correctly-assembled cargo from a misfolded protein carrying the same NLS?