Abstract
The approximately 120 MDa mammalian nuclear pore complex (NPC) acts as a gatekeeper for the transport between the nucleus and cytosol1. The central channel of the NPC is filled with hundreds of intrinsically disordered proteins (IDPs) called FG-nucleoporins (FG-NUPs)2,3. Although the structure of the NPC scaffold has been resolved in remarkable detail, the actual transport machinery built up by FG-NUPs—about 50 MDa—is depicted as an approximately 60-nm hole in even highly resolved tomograms and/or structures computed with artificial intelligence4,5,6,7,8,9,10,11. Here we directly probed conformations of the vital FG-NUP98 inside NPCs in live cells and in permeabilized cells with an intact transport machinery by using a synthetic biology-enabled site-specific small-molecule labelling approach paired with highly time-resolved fluorescence microscopy. Single permeabilized cell measurements of the distance distribution of FG-NUP98 segments combined with coarse-grained molecular simulations of the NPC allowed us to map the uncharted molecular environment inside the nanosized transport channel. We determined that the channel provides—in the terminology of the Flory polymer theory12—a ‘good solvent’ environment. This enables the FG domain to adopt expanded conformations and thus control transport between the nucleus and cytoplasm. With more than 30% of the proteome being formed from IDPs, our study opens a window into resolving disorder–function relationships of IDPs in situ, which are important in various processes, such as cellular signalling, phase separation, ageing and viral entry.
Main
IDPs are flexible, dynamic macromolecules that lack a fixed tertiary structure and can adopt a range of conformations to perform various functions across the cell. IDPs are highly relevant for human physiology and have central roles, among others, in neurodegenerative ageing diseases and cancer. IDPs are also key players in phase separation and are involved in the formation of biomolecular condensates13,14,15,16,17,18,19,20,21. In the nanosized NPC, which has a total molecular weight of approximately 120 MDa in mammals, there are hundreds of IDPs enriched in phenylalanine (F) and glycine (G) residues, known as FG-NUPs1. The FG-NUPs form a permeability barrier in the central channel of the NPC, which regulates nucleocytoplasmic transport by restricting the passage of large cargo unless it presents a nuclear localization sequence or a nuclear export sequence2,3. Nuclear transport receptors can specifically recognize these sequences and efficiently shuttle the cargo through the barrier. With recent advances in cryo-electron tomography, crystallography, proteomics and artificial intelligence (AI)-based structure prediction, approximately 70 MDa of the NPC scaffold enclosing the central channel has been resolved with near-atomic resolution4,5,6,7,8,9,10,11. However, signals from the highly dynamic FG-NUPs are by and large not accessible to those structural biology techniques, and the actual transport machinery inside the central channel—another approximately 50 MDa—is not captured, leaving an approximately 60-nm hole in the centre of the scaffold structure. Consequently, the protein conformational state inside the NPC remains elusive, which has led to several partially conflicting hypotheses for the morphologies of the FG domains in their functional state22,23,24,25,26,27,28. With approximately 30% of the entire eukaryotic proteome being intrinsically disordered, the problem that the conformational state is not easily studied in cells extends far beyond NPC biology. Besides magnetic resonance and scattering techniques13,14, single-molecule fluorescence of purified and labelled proteins has become a powerful tool for probing the conformations of proteins in solution; advanced studies have even shown that this is possible in cells if such probes are microinjected29,30,31. However, the NPC is assembled only in late mitosis and during nuclear growth in interphase32, and its labelling thus requires genetic encoding. Established fluorescent protein-based technologies such as GFP or self-labelling protein tags such as SNAP-tag33, however, do not readily enable the extraction of multiple distance distributions for the same protein, owing to the sheer size of the fluorescent label and the inherently limited freedom of labelling.
In this study, we developed a method to probe distance distributions of FG-NUPs inside the NPCs by combining fluorescence lifetime imaging of fluorescence resonance energy transfer (FLIM–FRET) with a site-specific synthetic biology approach. We show that the methods deliver quantitative results when using permeabilized cells with functional transport machinery, and offer sound agreement with qualitative measurements from live cells. We focused on NUP98 because it is the essential constituent of the NPC permeability barrier and is accessible to this technology34,35,36. By measuring the distance distribution for 18 labelled chain segments of NUP98 in the NPC using FLIM–FRET, we showed that the FG domain is exposed to—in the terminology of the Flory polymer model12—‘good solvent’ conditions inside the NPC. This enables the protein to adopt much more extended conformations in the functional state than in the highly collapsed state of single chains in solution at ‘poor solvent’ conditions. We combined our residue-specific measurements with coarse-grained molecular dynamics (MD) at residue resolution. This enabled us to integrate the distance distribution data and the recently solved scaffold structure7 into a molecular picture of FG-NUP distribution and motion in the central channel of a functional NPC.
In-cell site-specific labelling of NUP98
High-precision fluorescence measurements of the conformation of FG-NUPs in their functional state require the introduction of labelling tags with minimal linkage errors and minimal disruption of the structures and functions of the labelled proteins. To this end, we performed site-specific labelling with a non-canonical amino acid (ncAA) using genetic code expansion (GCE)37. We used the pyrrolysine orthogonal tRNA–synthetase suppressor pair to reassign the amber stop codon (TAG) to incorporate the ncAA trans-cyclooct-2-en-L-lysine (TCO*A) at that site. The chemical functionality of this ncAA residue can then be reacted with an organic fluorophore containing a tetrazine moiety to undergo an inverse-electron-demand Diels–Alder reaction (click chemistry38). Thus, the dye is stably attached to the protein via a small chemical linker, causing minimal disruption to the protein structure and its function. One potential downside of this technique is that it is not mRNA-specific, leading to background labelling of untargeted proteins with their naturally occurring stop codons being suppressed. To circumvent this problem, we utilized our recently developed film-like, synthetic orthogonally translating organelles (OTOs) to form a distinct protein translational machinery on the outer mitochondrial membrane surface39. These organelles exclusively reassigned two amber codons for the target FG-NUP and incorporated TCO*A at the two specified sites with high selectivity, ensuring minimal interference with endogenous protein translation and negligible background staining (see Fig. 1a and Extended Data Fig. 1 for the improved contrast when comparing OTO technology with conventional GCE). The incorporated ncAAs were reacted with a mixture of donor and acceptor dyes for FRET measurements. This results in FRET species mixed with donor-only and acceptor-only species. Therefore, the chosen measurement method must be able to distinguish the FRET species from the other two. Furthermore, quantitative high-precision FRET measurements have high requirements on the properties of the FRET dye pair, such as photostability, Förster radius, monoexponential decay of fluorescence lifetime and fluorescence emission clearly distinguishable from background….
