Molecular and structural analysis of central transport channel in complex with Nup93 of nuclear pore complex

ABSTRACT

The central transport channel (CTC) of nuclear pore complexes (NPCs) is made up of three nucleoporins Nup62, Nup58 and Nup54. In which manner and capacity, these nucleoporins form the CTC, is not yet clear. We explored the CTC Nups from various species and observed that distinct biochemical characteristics of CTC Nups are evolutionarily conserved. Moreover, comparative biochemical analysis of CTC complexes showed various stoichiometric combinations of Nup62, Nup54 and Nup58 coexisting together. We observed the conserved amino‐terminal domain of mammalian Nup93 is crucial for the anchorage of CTC and its localization to NPCs. We could reconstitute and purify mammalian CTC·Nup93 quaternary complex by co‐expressing full length or N‐terminal domain of Nup93 along with CTC complex. Further, we characterized CTC·Nup93 complex using small angle X‐ray scattering and electron microscopy that revealed a “V” shape of CTC·Nup93 complex. Overall, this study demonstrated for the first time evolutionarily conserved plasticity and stoichiometric diversity in CTC Nups.

1. INTRODUCTION

Nuclear pore complexes (NPCs) are one of the macromolecular assemblies in eukaryotic cells and are made of up to ~30 different types of nucleoporins (Nups). The NPCs reside in the nuclear membrane and mediate the transport of various solutes across the nuclear envelope. ¹ , ² Emerging evidence indicates that NPC structure, in general, is likely to be diverse from the lower eukaryotes to the higher ones, ³ , ⁴ , ⁵ , ⁶ , ⁷ consistent with the observed differences between molecular weights of vertebrate (~120 MDa) and yeast (~60 MDa) NPCs. ⁸ , ⁹ Additionally, considerable evidence suggests that many nucleoporins are evolutionary divergent. ⁷ , ¹⁰ Three of the Nups (Nup62, Nup54, and Nup58), are known to form the central transport channel (CTC). Among them, Nup62 appears to be conserved across various species. However, the other two counterparts, Nup54 and Nup58, appear to have low sequence similarity in lower to higher eukaryotes. Moreover, the rate of evolution of CTC Nups also appears to be different among these species. ⁷ These three Nups are crucial for imparting the permeability barrier and mediating nucleocytoplasmic transport of NPC. ¹¹ , ¹² , ¹³ , ¹⁴ They have structured alpha‐helical regions and also harbor longer disordered regions containing repeats of Phenylalanine‐Glycine, also known as FG repeats. ¹³

Structured regions of CTC Nups have coiled‐coil motifs that can form various complexes such as Nup62·Nup54 and Nup54·Nup58, as shown using rat proteins. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Recombinant mammalian CTC complexes show distinct biochemical behavior by forming multimeric complexes of higher molecular weights, as well as different stoichiometries. ¹⁵ However, the equivalent CTC complexes of Saccharomyces origin did not show the existence of such heterogeneous complexes. ¹⁷ Moreover, the crystal structure of purified partial CTC complex from Xenopus species ¹⁹ shows the equimolar stoichiometry of each partner (Table S1) overruling the existence of diverse stoichiometry. All together, these studies project contrary views on assembly of CTC Nups at biochemical level, and fail to clarify why CTC complexes of mammalian origin behave differently from their equivalent yeast or Xenopus complexes.

The CTC complex is anchored to NPC scaffold via Nup93. ²⁰ It has been shown that knockdown of Nup93 in Caenorhabditis elegans alters the permeability of NPC. ²¹ Additionally, in absence of Nup93, in Xenopus, the recruitment of CTC to NPCs is affected and is mediated by the N‐terminal region (1–175) of Nup93. ²² Moreover, the structure of thermophilic yeast Nup62·Nup54·Nup58·Nic96 complex revealed the interaction of Nic96 (139–180) region with CTC complex. ²³ However, such interactions studies on mammalian CTC with Nup93 have not been established yet; mainly due to difficulties in isolating the recombinant Nup93.

In this study, we reconstituted CTC complex of various species such as Hydra and Danio and compared their behavior with equivalent complexes from rat, yeast, and Chaetomium species. We report that, despite various trends in the sequences properties and evolution, ⁷ the CTC complex demonstrates evolutionary preservation of dynamic nature. Our data shows that the various combinations of CTC components can co‐exist biochemically and such phenomenon seems to be evolutionarily conserved. Using the co‐immunoprecipitation experiments, we identified an amino‐terminal domain (1–150) of mammalian Nup93 required for interactions with CTC complex as well as to localize into NPCs. Further, we succeeded in biochemically assembling complexes of mammalian Nup93 with CTC, which was subjected to structural analysis using SAXS and electron microscopy, demonstrating the overall architecture of mammalian CTC·Nup93 quaternary complex for the first time as well as its in‐solution flexible behavior.

2. RESULTS

2.1. Distinct biochemical properties of CTC nucleoporins across the species

Biochemical reconstitution and isolation of CTC complexes have been reported from various species such as Saccharomyces, ² , ²⁰ Xenopus, ²⁴ Chaetomium ²³ , ²⁵ and rat. ¹² , ¹³ , ¹⁴ , ²⁶ We performed the phylogenetic analysis of all three CTC Nups viz. Nup62 (Figure 1a), Nup54 (Figure S1a) and Nup58 (Figure S1b) from a total of 73 different species (Table S2). We observed that CTC Nups from human, rat, Xenopus and Danio form a separate clade compared to distant clades of Chaetomium and Saccharomyces, whereas, the Hydra appears in‐between a separate clade for all three CTC Nups (Figures 1a and S1). Further, the sequence alignment of CTC Nups of these seven species using PROMALS3D ²⁷ (Figure S2) and secondary structure prediction analysis using PSIPRED ²⁸ , ²⁹ (Figures 1b–d and S3) suggested the presence of conserved secondary structure domains, despite the variations in the sequence identities among these species. Within the boundaries of the conserved helical regions the propensities for coiled‐coil motifs in the CTC Nups appear to be conserved (Figure S2). Overall, these analyses suggest that variations in CTC Nups sequences attribute to the evolutionary divergence of CTC nucleoporins, however, secondary structure domains and the coiled coil domains remain conserved.

FIGURE 1.

Overview of the diversity among the CTC Nups. (a) A representative unrooted phylogenetic tree for the protein sequences of Nup62 from 73 species. The numbers on each branch represent bootstrap value and “*” indicates the location of all the species discussed in this study (Sc—Saccharomyces cerevisiae, Ct—Chaetomium thermophilum, Hv—Hydra vulgaris, Xl—Xenopus laevis, Hs—Homo sapiens, Rn—Rattus norvegicus, Dr—Danio rerio). Organization of secondary structure domains for Nup62 (b), Nup54 (c), and Nup58 (d) of various species, based on the predictions using PSIPRED and visualized as a drawing with IBS illustrator. Predicted α‐helical domains are represented by cylinder and β‐sheets by arrow. FG‐repeat in the sequences are represented by the red bars

Since, we observed Danio and Hydra in separate clades in our phylogenetic analysis, we decided to study the biochemical behavior of CTC complexes from these species. We cloned and purified the complex consisting of structured regions of three CTC Nups Nup62 (311–507), Nup58 (258–478), and Nup54 (368–536) from Danio (Figure 2a), referred to hereafter as Danio CTC. Similarly, we purified the equivalent CTC complex consisting of Nup62 (341–534), Nup58 (144–322), and Nup54 (373–547) from Hydra (Figure 2e), referred to hereafter as Hydra CTC. The rat CTC complex consisting of Nup62 (322–525), Nup58 (239–415), and Nup54 (332–510) was purified as reported previously ¹⁶ , ¹⁸ (Figure 2c). Size exclusion chromatography followed by SDS‐PAGE analysis of purified complexes shows the presence of all three CTC proteins of Danio (Figure 2b), Hydra (Figure 2f), and Rat (Figure 2d). Resolving these complexes on the native‐PAGE show a single band for the Danio complex whereas Hydra CTC complex shows multiple species of complexes (Figure 2b,f; indicated by arrow). Similarly, as reported previously ¹⁵ rat CTC‐complex shows the presence of multiple bands with smudged lanes (Figure 2d). In SEC‐MALS analysis of Danio complexes, we observed a homogenous distribution of averaged mass, equivalent to expected mass of ~69 kDa at equimolar ratio (Figure 3a, Table S3). We observed the apparent increase in molecular weights of rat CTC complex (from ~72 to ~92 kDa) with increased protein concentration (Figure 3b, Table S3) as reported previously. ¹⁵ The CTC complex of Hydra showed lower molar mass (~47 kDa) compared to the expected mass (~64 kDa; Figure 3c; Table S3). Appearance of a small merged peak proximal to the main peak, and distributed range of molecular weights at higher concentration (3 mg/ml), suggest the possibility of multiple molecular species in this complex (~85–90 kDa, orange line Figure 3c). Consistent with this observation, Hydra CTC complex shows multiple distinct bands on native gel (Figure 2f). Thus, CTC complex of Danio appears to be equimolar, with minimal tendency of formation of oligomeric assemblies. Similar biochemical behavior of CTC complex is reported for yeast and Chaetomium CTC complexes. ¹⁷ , ²³ Biochemical behavior of Hydra CTC complex is similar to the previously reported rat CTC complexes, including the propensity to form higher oligomeric complexes with varied stoichiometry at increasing concentration. ¹⁵ , ¹⁷

FIGURE 2.

Purified CTC complexes from various species. The purified CTC complexes from (a,b) Danio, (c,d) Rat, and (e,f) Hydra were analyzed using SEC on Superdex200 10/30 GL. The diagrammatic representation of the boundaries of various nucleoporins used for the cloning and expression of Danio (a), Rat (c), and Hydra (e) is shown; FG repeats are indicated with vertical blue lines. SEC profiles Danio (b), Rat (d), and Hydra (f) are plotted against the absorbance (A_280nm). The fractions were analyzed on reducing SDS‐PAGE (12%) or native non‐reducing‐PAGE (10%). The orange bar represents the position of loaded fractions on the chromatogram. The various oligomeric species of CTC complexes are indicated by arrow. Peak position of each chromatogram peak over PAGE analysis is indicated by “**” asterisks

FIGURE 3.

SEC‐MALS. The size exclusion chromatography coupled with multi‐angle light scattering analysis for (a) Danio (b) Rat, and (c) Hydra CTC complexes. In each chromatogram, the Rayleigh scattering for the various concentrations is shown by the blue (3 mg/ml), grey (1 mg/ml), and black (0.5 mg/ml) lines. Molecular weight distribution across the peak are represented as orange (3 mg/ml), yellow (1 mg/ml), and green (0.5 mg/ml) lines

2.2. Role of alpha‐beta domain of Nup54 in CTC dynamics

Nup54 contains a region consisting of alpha‐beta secondary structures from residues 181 to 326 additional to the C‐terminal coiled coil domains. Most of the studies on the rat CTC complexes are performed using the Nup54 devoid of such region owing to solubility of complexes while purification. ¹⁵ , ¹⁶ , ¹⁸ To test the role of alpha‐beta domain of Nup54 on the observed plasticity of the CTC complexes, we cloned and purified the mammalian CTC complex consisting of alpha‐beta domain of Nup54 (181–507; Figure 4a). Since the rat CTC complex comprising of longer Nup54 showed solubility issues, we replaced it with corresponding human Nup54 (181–507), having 99.4% sequence similarity with each other (Figure S3d). The longer chimeric CTC complex was purified using similar approach to CTC complex (Figure 2c,d) and subjected to comparative characterization. The fractions of SEC analysis of this CTC complex (Figure 4a) resolved on the SDS‐PAGE showed the presence of three bands. This CTC complex obtained indeed shows the presence of higher molecular weight species observed on the non‐reducing PAGE (NR‐PAGE; Figure 4b). Corroboratively the SEC‐MALS analysis consistently showed the higher order oligomeric states in the alpha‐beta domain containing CTC complex (Figure 4c). The observed plasticity of this longer CTC complex is consistent with the previously reported rat CTC complexes. ¹⁵ , ¹⁶ , ¹⁸

FIGURE 4.

CTC complex with alpha‐beta domain of Nup54 exhibits higher order oligomers. (a) The purified CTC complex containing alpha‐beta domain of Nup54 (181–507), Nup62 (322–525), and Nup58 (239–415) was analyzed using SEC on Superdex200 10/30 GL. (b) The fractions were analyzed on reducing SDS‐PAGE (12%) as well as native non‐reducing‐PAGE (NR‐PAGE; 12%). The orange bar above the gel represents the loaded fractions corresponding to the chromatogram. Partial‐proteolysis of Nup54 is indicated by “*” asterisk. The various oligomeric species of CTC complex analyzed using NR‐PAGE are indicated by arrow. (c) The SEC‐MALS analysis of CTC complex containing alpha‐beta domain of Nup54 using 1 mg/ml (black) or 1.5 mg/ml (blue)

2.3. Danio CTC complex suggests coexistence of various stoichiometric complexes of CTC

The recombinant purification of CTC complexes ¹⁵ , ¹⁸ , ¹⁹ , ²³ is achieved generally in three steps—such as affinity chromatography, ion exchange followed by size, and exclusion chromatography. We hypothesized that multi‐step purification approach may rule out the co‐purification of varied stoichiometric complexes. Therefore, to identify co‐existing CTC complexes, Danio CTC purification was performed using only affinity chromatography (using His_6x‐Nup54) and was immediately analyzed using SEC, avoiding the intermediate steps. This approach clearly showed the coexistence of Danio CTC and Nup58(257–478)·Nup54(368–536) complexes, termed as Nup58·Nup54 complex (Figure 5a,b). Native gel analysis corroborates this data, with distinct bands for both complexes (indicated by arrow, Figure 5b). Similarly, we observed the co‐existence of Danio CTC and Nup58·Nup54 complexes when GST‐tagged Nup58 (Figures S4a,b) was used for GST affinity‐based purification. To analyze the integrity of these complexes, Ni‐NTA affinity purified fractions corresponding to the peaks of Nup58·Nup54 complex (Peak‐1) or Danio CTC (Peak‐2) complex were pooled and re‐analyzed. We observed distinct bands for Nup58·Nup54 (Figure 5c) and Danio CTC complexes (Figure 5d). We also confirmed Nup58·Nup54 complex formation by expressing only these two proteins (Figures S4c,d) that indicated the corresponding complexes observed in Figure 5a,b. These results demonstrate that the Danio CTC, which appeared to be homogenous in 1:1:1 stoichiometry (when purified using affinity followed by ion exchange chromatography steps), can actually form co‐existing complexes of various stoichiometries, as reported for the rat CTC complex, ¹⁵ when purification strategies are altered. By eliminating the ion exchange chromatography step, we could establish the coexistence of Danio CTC complex with Nup54·Nup58 complex. Based on this data, we suggest that the CTC complexes reported earlier from the yeast, ¹⁷ Chaetomium, ²³ or Xenopus ¹⁹ could also contain similar co‐existing complexes in varying amounts, which are eliminated by additional stages of purification.

FIGURE 5.

Co‐existing CTC Nup‐complexes in Danio. (a) Ni‐NTA affinity purified Danio CTC complex was directly loaded on the analytical SEC column (Superdex‐200 10/30 GL) and the absorbance (A_280nm) is plotted against elution volume. (b) Eluted fractions were analyzed on SDS‐PAGE (12%) and Non‐reducing native PAGE (NR‐PAGE, 10%), corresponding to fractions, indicated by orange bar, in the chromatogram. “*” represents the fraction with Nup62(311–507)·Nup54(368–536) bands and negligible amounts of Nup58(257–478). Roughly, the spread of Peaks #1 and #2 is indicated by colored bars. (c) Two major peaks in panel‐A named #1 (c) and #2 (d) were pooled, run on Superdex‐200 10/30 GL column and eluted fractions were analyzed same as mentioned above. The arrow indicates the position of multiple complexes on the NR‐PAGE. Peak position of chromatogram peak corresponding on PAGE analysis is indicated by “†” dagger symbol

2.4. Amino‐terminal region of Nup93 (1–150) is essential for CTC interaction and localization in NPCs

To identify the regions of mammalian Nup93 essential for interactions with CTC Nups, we made several deletion constructs of Nup93 tagged with GFP (Figure 6a). Domain boundaries were selected based on PSIPRED ²⁸ , ²⁹ analysis of Nup93 (Figure 6a). Transfection of these GFP tagged constructs in HEK‐293F cells, followed by pull‐down experiments, suggest that the N‐terminal region (1–150) of Nup93 interacts with CTC complex (Figure 6b). We observed that the C‐terminal region of Nup93 (176–819) fails to interact with the CTC complex (Figure 6b). Further, to identify the precise boundaries of the domain, we made few more deletions in N‐terminal region of namely Nup93 (1–82) and Nup93 (96–150). Pull down assays with these constructs show detectable interaction of Nup93 (1–82) but not Nup93 (96–150), with CTC (Figure 6b). This interaction of Nup93 (1–82) domain with CTC corroborates previous data, where a short sequence from an equivalent domain in CtNic96 (139–180) interacts with Chaetomium CTC. ²³ Interestingly, the inability of Nup93 (96–150) to bind CTC also correlates with the lack of interaction of CTC with the equivalent region of CtNic96 (262–301), ²³ suggesting conserved modes of interaction between Nup93 and CTC complex. Further, to test the localization of these Nup93 regions to NPCs, we performed fluorescence microscopy. We observed that amino terminal region of Nup93 (1–150) localizes to NPC (Figure 7a,b). Further deletions of the N‐terminal region, Nup93 (1–82) and Nup93 (96–150), showed diminished localization to the nuclear rim (Figure 7a,b). Taken together, these results suggest the critical role of mammalian Nup93 (1–150) domain for localization of Nup93 to CTC·Nup93 complex in NPCs.

FIGURE 6.

Amino‐terminal region of Nup93 is needed for the CTC interaction. (a) A schematic representation of the various domains of mammalian Nup93, generated based on PSIPRED analysis and were cloned in pEGFP‐C1 vector. (b) A pull‐down assay with either anti‐GFP‐nanobody (GFPNB) or GST (as a control) was performed using various Nup93 constructs transfected to HEK293F cells. Pulled complexes were resolved on SDS‐PAGE (10%) and blotted. Western blot analysis of pulled‐down complexes probed with anti‐GFP antibody (1:5,000) or mAB414 (1:5,000), along with the cleared lysate as Input. The Nup93 domains are indicated by the numbers on right. Arrow indicates the location of expected bands appearing in the pulled samples. The interactions of Nup93 domains with CTC are summarized in the right panel, “+” denotes the observed interaction while “−” indicates the absence of interaction by our study

FIGURE 7.

Amino‐terminal region of Nup93 is needed for the localization in nucleus. (a) HEK293 cells transfected with various deletion constructs of Nup93 were treated to remove cytoplasm, fixed and then imaged using SP8 confocal microscope (Leica). Cells were stained with mAB414 (1:300), recognizing various Nups of NPCs; followed by anti‐mouse IgG‐AF594 (1:1,000) and GFP‐Booster‐ATTO488 (1:1,000). Nuclei were stained with DAPI. Representative regions for each image, enlarged from the same image (indicated by a dotted box), are shown in right panel with “selected region” label for each channel. From clockwise 1. GFP, 2. mAB414, 3. GFP + mAB414, and 4. DAPI. Scale‐bar—25 μm. (b) Intensity profiles for the various deletion constructs of Nup93 quantified using the ImageJ for the small subset of the cells (4–6 in numbers), in independent experiment. Fluorescence intensity (in arbitrary units) is plotted against the length in pixels

2.5. Reconstitution of mammalian CTC·Nup93(1–150) and CTC·Nup93(1–819) complexes

To test the in vitro ability of Nup93 (1–150) domain to recognize CTC complex, we co‐expressed the full‐length Nup93 (1–819) and Nup93 (1–150) as GST fusion protein with the rat CTC complex (as His‐6x tagged to Nup58) using the Escherichia coli expression system. Tandem Ni‐NTA and GST affinity purification, followed by SDS‐PAGE analysis of purified complexes, revealed the isolation of quaternary complexes namely CTC·Nup93(1–819) (Figure 8b) as well as CTC·Nup93(1–150) (Figure 8d). Further, elution profiles of these complexes after size exclusion chromatography showed that both full length Nup93 (1–819) or Nup93 (1–150) domains were able to form stable complex with the CTC Nups (Figure 8a–d) under given biochemical conditions. SEC‐MALS analysis of these quaternary complexes showed molecular weight of ~82 kDa and ~155 kDa for CTC·Nup93(1–150) and CTC·Nup93(1–819), respectively (Figure S5), indicating the possibility of 1:1:1:1 stoichiometry for each nucleoporin (Nup93, Nup62, Nup54, and Nup58). These observations, taken together, indicate that amino‐terminal region of mammalian Nup93 (1–150) forms the minimal quaternary complex with CTC. It should be noted that full length human Nup93 is notoriously difficult to purify alone, however, when co‐expressed with CTC complex, it could be purified to homogeneity as a stable complex.

FIGURE 8.

In vitro interaction of amino‐terminal region of mammalian Nup93 with CTC complex. Schematic representation of boundaries various CTC Nups (a,c,e) used for co‐expression and purification with full length Nup93(1–819) (a,b) and amino‐terminal Nup93 (1–150) (c,d). SEC analysis using Superdex‐200 10/30 GL of purified CTC·Nup93(1–819) (b); CTC·Nup93 (1–150) (d) and CTC alone (f). Elution fractions (represented by blue bar) containing the complexes were resolved on SDS‐PAGE (10 or 12%) and stained with Coomassie‐blue. Expected bands of various proteins are indicated in right side of gel. Peak position of each chromatogram corresponding on PAGE analysis is indicated by “*” asterisk

2.6. Solution structure of Danio CTC and rat CTC·Nup93(1–150) complexes

Since the structure of mammalian CTC in complex with Nup93 is not known, we employed, small angle X‐ray scattering (SAXS), an established method for structural analysis of dynamic proteins, to study the solution structure of both Danio CTC and rat CTC·Nup93 complexes. We estimated the radius of gyration (R _g) using low Q‐region and radius of cross section (R _c), assuming globular and rod‐like scattering shape of the predominant scattering molecules in solution. The length of persistence (L) of the respective scattering entities were calculated by employing L = [12(R _g ² − R _c ²)]^1/2 (Table 1). Guinier analysis and double log plots revealed that both CTC and CTC·Nup93 complexes are monodisperse, as their Guinier plots are linear in the region with Q*R _g < 1.5, with Q being the momentum transfer (Figures 9a,b,f,g and S6k). The residual plot of the Guinier analysis is randomly distributed along zero suggestive of a good monodisperse sample quality (Figures S6e,j). Values of R _g calculated from the slope of the Guinier plot (R _g = 47.3 Å for CTC and 51.7 for CTC·Nup93 Å) suggest that the both complexes adopt an extended state (Table 1). The distance distribution functions P(r) calculated from the scattering data using the program GNOM ³⁰ are shown in Figure 9c,h for CTC and CTC·Nup93 complex, respectively. The P(r) functions along with total quality estimate (TQE) are similarly shaped for both complexes and show peaks around 47 Å for CTC complex and 52 Å for CTC·Nup93 complex (Figure 9c,h). The maximum distance D _max 216 and 220 Å is observed for both complexes. The elongated and highly asymmetric shape of the P(r) functions corroborates the extended state of the CTC, as well as CTC·Nup93 complex. The Kratky plot I(Q)·Q ² versus Q, which reflects the degree of compactness of the scattering particles, has a shape typical of a partially flexible protein with a flat peak at low Q and an increase in I(Q)·Q ² in the larger Q range (Figure S6). The comparison of the Kratky plot of CTC and CTC·Nup93 complex (Figure S6a,f,l) also revealed that both the tertiary and Quaternary complexes have flexible structures, along with ordered regions in similar manner as indicated by Porod exponents of 3.4 for both CTC (Figure S6a,b) and CTC·Nup93 (Figure S6f,g,l) complexes. Additionally, molecular weight of the sample was calculated based on a SAXS invariant Vc (volume of correlation). This method of molecular weight estimation is concentration‐independent and exclusive to the structural state of the scattering molecule in solution. ³¹

TABLE 1.

SAXS data reduction and model statistics

Structural parameters		Rat (Nup62322‐525·Nup54332‐510·Nup58239‐415·Nup931‐150)	Danio (Nup62311‐507·Nup54368‐536·Nup58258‐478)
Guinier analysis	R g	5.17 nm	4.73 nm
	R c	2.31 nm	2.00 nm
	L	16.02	14.85 nm
Indirect Fourier transformation	D max	21.67 nm	22 nm
	R g	5 nm	5 nm
Porod's exponent		3.2	3.4
χ 2 for models using
GASBOR		5.29	25.82
Molecular weight by partial volume (Vc)	kDa	113.130	80.697

FIGURE 9.

Small angle X‐ray scattering (SAXS) of CTC and CTC·Nup93(1–150) complexes. Double log plot Log₁₀ Q vs. Log₁₀ I(Q) for CTC complex (a) and CTC·Nup93 complex (f); A ln I·(Q) vs. Q ² plot for the CTC complex (b) and CTC·Nup93 complex (G); pair‐wise distribution analysis P(r) for CTC complex (C) and CTC·Nup93 (1–150) complex (H) were plotted. The calculated models for CTC and CTC·Nup93 obtained after GASBOR analysis (red dots; Panel d and h) were fitted with the experimental data for CTC (blue dots, panel D) and CTC·Nup93 (Green dots, Panel i), respectively. Molecular envelope obtained with GASBOR analysis for CTC complex was fitted with the model of Danio CTC (e). Similarly, the molecular envelope obtained by GASBOR analysis for CTC·Nup93 complex was fitted with the model for the CTC·Nup93 (j). TQE—total quality estimate of the P(r) plot

Based on SAXS experimental data, 3D models for CTC and CTC·Nup93(1–150) were generated using ab initio modeling in GASBOR. ³² The lowest χ ² model and its fitting with P(r) plot for each is shown in Figure S6c,d and Figure S6h,i, respectively. A good fitting model consistent with lower χ ² (5.29) was obtained for rat CTC·Nup93 (Figure 9d,e,i,j; Table 1). However, the χ ² value of about 25.82 for the CTC complex model is relatively higher, indicating that the CTC complex in absence of Nup93 forms a loose and flexible structure (Figure 9d,e,i,j; Table 1). Further, we used a homology model of CTC using CTC·Nic96 complex structure (PDB:5IJO) as reference. ⁴ For CTC homology model, the Nic96 chain was removed and residues were mutated to that of Danio. A similar model for CTC·Nup93(1–39) was generated using the same reference (PDB:5IJO), and by mutating the residues as per the rat CTC domains used in experiments. Both these homology models (CTC and CTC·Nup93(1–150) complex) were then used to fit in SAXS data‐derived envelope and their fit is shown in Figure 9e,j, respectively. The envelops created by calculating the SAXS data fitted well with the equimolar models of CTC and CTC·Nup93(1–150) complexes. Overall, our SAXS data clearly showed that CTC complex is flexible in nature and has an elongated structure. Although the CTC·Nup93(1–150) complex is of a similar shape, it has relatively reduced flexibility. The Nup93(1–150) region binds in the pocket of CTC, as evident from the Chaetomium CTC·Nic96 structure. ²³

2.7. EM based structure of mammalian CTC·Nup93(1–819) complex

Till now, no structural information for entire Nup93 or its homolog is available, mainly due to difficulties in the purification of recombinant Nup93. However, a partial structure of yeast Nic96 (200–820) has been reported, ³³ , ³⁴ where the CTC‐interacting N‐terminal domain is missing. Also, in the structure of Chaetomium CTC·Nic96 quaternary complex, only a 40 amino acid long region (139–180) is shown. We could purify full length mammalian Nup93 in complex with CTC using an altered protocol as described before. However, we could not perform SAXS analysis on mammalian CTC·Nup93(1–819) due to very low solubility of the complex. Therefore, we analyzed the structure of CTC·Nup93(1–819) using negative stain electron microscopy. Purified mammalian CTC·Nup93(1–819) complex was adsorbed on EM grids and stained with uranyl acetate (Figure 10a). A dataset of approximately 4,000 particles was collected and utilized for single particle analysis using RELION suite. ³⁵ 2D class averages of these particles showed the distinct “V” shape feature of the complex (Figure 10b). These classes were further subjected to 3D reconstruction (Figure 10c), using a 3D model of human Nup93 generated by homology modeling, using the yeast Nic96 (200–820 amino acid) crystal structure as a template. The resolution of the structure is 27.4 Å, based on a FSC cut‐off of 0.143 (Figure S7a,b) ³⁶ . One copy of modeled CTC·Nup93(1–39) and one copy of Nup93(171–819) were fitted into the 3D density map and both structures were found to fit well with cross‐correlation values of 0.86 (Figure S7c). The overall architecture of mammalian CTC·Nup93 appears to be a distinct “V” in shape, with some expected flexibility in the middle and C‐terminal domains of Nup93. The amino terminal domain of Nup93 (1–150) is embedded in the pocket formed by CTC near its C termini, while the rest of Nup93 is separated from the helical domains of CTC by the three‐helix bundle region of the former.

FIGURE 10.

EM analysis of CTC·Nup93(Full length) complex. (a) A representative micrograph of purified CTC·Nup93(1–819) complex stained with 2% uranyl acetate. Scale bar—100 nm. Representative particles used for the analysis are shown in inset with the arrow. (b) 2D classes obtained using 4,156 particles in Relion 2.1 and (c) shows the 3D density map (at ~27.4 Å) obtained

3. DISCUSSION

We analyzed sequences of CTC Nups from various species and observed that the predicted domain organization is conserved across various species, despite significant variations in the protein sequences (Figures 1 and S1–S3). In general, many of the nucleoporin sequences have diverged from the lower eukaryotes to the higher ones. ⁷ , ¹⁰ Similar to the conserved architectural features, we observed that the biochemical behavior of CTC complexes is also comparable (Figures 2, 3, 4, 5). In case of the rat CTC complexes, it has been observed that these complexes, at higher concentration, can form higher order oligomers, with diverse stoichiometric combinations. ¹⁵ We reconstituted the equivalent CTC complexes of Danio and Hydra and analyzed them for their ability to form diverse oligomers. We observed that, with increasing concentrations, Hydra CTC complex can form higher order complexes. On the other hand, Danio CTC complex does not form such higher order complexes at similarly higher concentrations, and appears to have equimolar stoichiometry (Figures 2 and 3). At first, such behavior of Danio CTC complex appeared similar to the Saccharomyces CTC complex. ¹⁷ However, we could alter the homogenous state of Danio CTC complex by altering the purification strategy, which clearly demonstrated that the pool of Danio CTC complex is as dynamic as reported for Hydra CTC in this study and rat CTC. ¹⁵ We previously reported that CTC Nups are species specific, ⁷ thus, it is probable that physiochemical properties of CTC complexes will also vary in species specific manner. Due to differences in such intrinsic biochemical reactivity based on purification strategies, the heterogeneous CTC complex from Danio could be separated easily. However, in case of rat and Hydra CTC complexes, such separation was not feasible using the same purification strategy. Based on our analysis, we suggest that biochemically dynamic behavior of CTC complexes is conserved across the species and evidence for such dynamicity can be obtained by following unbiased purification strategies.

A study with CTC complex of Chaetomium lacking alpha‐beta domain of Nup54 suggested the absence of this domain as a trigger to the higher order oligomer formation of CTC complex. ²³ It is interesting to note that Chaetomium has very short length Nup54 with shorter alpha‐beta domain (Figures S2c and S3c). In our study, we observed that the mammalian CTC complex with additional alpha‐beta domain of Nup54 (187–507) exhibit dynamic behavior similar to the CTC complex. Therefore, we conclude that the plasticity of the mammalian CTC complex is not dependent on the Nup54 alpha‐beta region and such dynamic nature of CTC complex is conserved (Figure 4).

In yeast (Saccharomyces or Chaetomium), the amino‐terminal region of Nic96 has been shown to be crucial for the anchorage and interaction with Nup57·Nup49·Nsp1 complex. ²⁰ , ²³ , ²⁵ , ³⁷ The R1 domain of Chaetomium Nic96(139–180) is known to be important for the interaction with CTC Nups. ²³ An equivalent domain containing region of Xenopus Nup93 (1–189) is known to interact with Nup62 complex, and the absence of this region affects the transport function of NPCs. Lack of this domain, however, did not affect the completion of the nuclear envelope. ²² In our study, we observed that the amino‐terminal region of mammalian Nup93 (1–150) can form a stable complex with CTC and is crucial for its localization to the NPC (Figure 6, 7, 8). The smaller domains from the amino‐terminal region, namely Nup93 (1–82) and Nup93 (96–150), fail to localize to the nuclear rim, suggesting that the combination of more than one stretch might be needed for the localization of Nup93 to NPCs. The R2 domain of Chaetomium Nic96 (262–361), which is contained within the amino‐terminal region of Nic96, is known to interact exclusively Nup192/Nup188, as opposed to R1 which exclusively interacts with CTC complex. ²³ , ²⁵ Since, small domain of Nic96 consisting of R1 and R2 domains are involved in the interaction with several scaffold proteins of NPCs, such interactions plausibly could be similar with the mammalian Nup93. Therefore, we speculate that the cumulative interactions of these two domains of mammalian Nup93 (Nup93 [1–82] and Nup93 [96–150]) might be needed for the appropriate localization of Nup93 to NPCs.

It is interesting to note that the dwelling time for the components of CTC and Nup93 on the NPCs vary significantly. ³⁸ Nup62 and Nup58 have dwelling times in the range of ~11 and ~8.5 hr respectively, whereas Nup93 has a dwelling time of ~70 hr on NPCs. ³⁸ Longer dwelling times of Nup93 are consistent with the observation that Nup93 is one of the long living proteins synthesized in cells. ³⁹ With such differential rates of CTC components, we can speculate that the formation of heteromeric CTC assemblies with either two or three protein partners (Figure 5) is supported by a temporary pool of proteins maintaining CTC stoichiometry on NPCs. Since, CTC Nups harbor coiled coil domains and interact with each other through these domains, ¹⁸ , ⁴⁰ we proposed the chain replacement mechanism, ⁴⁰ as the basis for the formation of CTC complexes of various stoichiometries. However, in vivo existence and the plausible functions of various CTC complexes needs to be investigated further.

Our biochemical data on reconstitution also indicates that mammalian Nup93 interacts with CTC complex in 1:1:1:1 stoichiometry. Our SAXS analysis of CTC and CTC·Nup93(1–150) clearly showed that CTC complex has intrinsic plasticity which is relatively stabilized when Nup93 (1–150) is in bound form. The binding pocket of Nup93 (1–150) with CTC is similar to that reported in quaternary complex structure (CTC·Nic96) from Chaetomium species. ²³ Here for the first time, we have shown the approach for reconstitution and isolation of full‐length mammalian Nup93 with CTC complex, which was structurally analyzed using electron microscopy. The low‐resolution structure obtained upon single particle reconstruction from negatively stained micrographs revealed a distinct “V” shape of the mammalian CTC·Nup93 quaternary complex. Our analysis thus forms the basis for reconstitution of mammalian Nups complexes for atomic resolution structural analysis.

In conclusion, the comparative study of CTC complex revealed evolutionary conservation of its dynamic nature and coexistence of diverse stoichiometric combinations. We also showed that Nup93 anchors the CTC complex in 1:1:1:1 stoichiometry only and this quaternary complex, in its purified form, is a “V” shaped structure with dynamic nature. Due to differences in the dwelling times of Nup93 and CTC Nups, it is important to understand the physiological role and relevance of CTC complex plasticity in NPC mediated transport functions.

4. MATERIALS AND METHODS

4.1. Phylogenetic analysis of CTC Nups

All possible homologs of human Nup54, Nup58, and Nup62 were identified using HMMER server using human sequence as a query sequence. ⁴¹ The dataset created using phmmer searches were aligned (structure guided) using PROMALS3D. ²⁷ Common representative species sequences were selected for three proteins (Nup54, Nup58, and Nup62) representing the total sequence divergence of CTC Nups. Final dataset for each protein consisted of 73 species (Table S2). These alignments were then edited in Jalview ⁴² to remove the low complexity FG repeat regions and Jpred ⁴³ to predict the secondary structures considering Rattus norvegicus as template. These modified aligned sequences were subjected to phylogenetic analysis using Phylip (http://evolution.genetics.washington.edu/phylip/). As a test of phylogeny, 1,000 bootstraps were used to generate the final consensus tree. The final consensus trees were build using SumTrees method from DendroPy ⁴⁴ at a threshold of 75%. Secondary structure prediction for the various CTC Nups was performed using the PSIPRED server ²⁸ , ²⁹ and the sequences were visualized using the IBS illustrator. ⁴⁵

4.2. Preparation of expression plasmids

To clone CTC Nups, the RNA was isolated from the embryos of Danio rerio using the Trizol reagent (Invitrogen). The isolated RNA was used for cDNA synthesis using the Superscript‐II cDNA synthesis kit (Invitrogen) as per manufacturer's instructions. CTC Nups of Danio were PCR amplified using Phusion® polymerase (NEB) and gene specific primers (Table S4), denoted henceforth with prefix Dr. DrNup62 (311–507) was cloned between the EcoRV and XhoI sites and DrNup54 (368–536) was cloned between EcoRI and NotI sites of pRSF‐Duet1 vector. The region encompassing these two ORFs were digested with NcoI and XhoI and sub‐cloned into pET‐28a plasmid. Structured region of DrNup58 (258–478) was cloned between EcoRI and NotI sites of pGEX‐4 T‐1 (GE healthcare) as well as NdeI and NotI sites of pET‐21a. For obtaining CTC Nups from Hydra, RNA was extracted from 2 days starved Hydra magnipappilatta polyps (about ~50) and cDNA synthesis was carried as per manufacturer's instructions. Gene specific primers were used for amplification of Hydra CTC Nups (Table S4), denoted henceforth with Hm. HmNup62 (341–534) was cloned between NcoI and NotI and HmNup54 (373–547) between NdeI and BamHI of pETDuet‐1 vector. Nup58 (144–322) was cloned between NdeI and KpnI sites of pET‐28a vector. For purification of rat CTC complex, a modified pET‐28a plasmid expressing rat Nup62 (322–525), rat Nup54 (332–510), and Nup58 (239–415) as reported previously ¹⁶ , ¹⁸ was used in this study. For purification of CTC complex containing alpha‐beta domain (181–507) of Nup54, rat CTC expression plasmid was modified by inverse PCR to express rNup62(322–525) and rNup58 (239–415). For alpha‐beta domain containing Nup54 (181–507), RNA was extracted from HEK293 cells and cDNA was synthesized using the Superscript‐II cDNA synthesis kit (Invitrogen) as per manufacturer's instructions. Human Nup54 specific primers (Table S4) were used to amplify Nup54 and cloned between NcoI and HindIII in pETduet1 vector. To obtain various deletion constructs of Nup93, desired regions (Figure 6a, Table S4) were PCR amplified using cDNA prepared from RNA of HEK‐293 cells and cloned between KpnI and XhoI sites of pEGFP‐C1 vector. For obtaining GST‐fusion constructs of Nup93, Full‐length Nup93 (1–819) and N‐terminal region Nup93 (1–150) were PCR amplified and cloned between BamHI and XhoI sites of pGEX‐4T‐1 vector (Table S4). Human Nup93 (1–150) region has identical sequence with the rat Nup93 (1–150) region. All plasmids were sequenced for the confirmation of intended regions.

4.3. Protein purification

Purifications of Danio and Hydra complexes were performed using the protocol used for the rat CTC complex. ¹⁶ , ¹⁸ To study the co‐existence of the various CTC complexes of Danio, Ni‐NTA purified complexes were subjected to size exclusion chromatography using Superdex200 10/30 GL (GE healthcare) and analyzed on SDS‐PAGE. CTC complex containing Nup54 alpha‐beta domain (181–507) was purified by co‐transforming pET28a vector, expressing rNup58 (239–415) and rNup62 (322–525), with pETduet1 expressing Nup54 (181–507) in E. coli BL21 (RIL) cells. The transformed cells were grown in 37°C and induced with 0.5 mM IPTG at OD₆₀₀ about ~0.45 and incubated at 22°C post induction for 12–14 hr. Harvested cells were subjected to Ni‐NTA affinity purification as mentioned above. Eluted samples were pooled and digested with thrombin (Merk) followed by size exclusion chromatography using Superdex20010/30 GL (GE healthcare). Rat CTC·Nup93 complexes with Nup93 (1–819) or Nup93 (1–150) were purified using following steps. The pGEX‐4T1‐Nup93 (1–819) or pGEX‐4T1‐Nup93 (1–150) vectors were co‐transformed with Nup62 expression vector in E. coli BL21 (RIL) or E. coli BL21 (C43) cells. Cells were cultured in 4–8 L of LB broth containing 0.1% glucose, 10 mM MgCl₂ and corresponding antibiotics. Cells grown at 37°C were induced with 0.2 mM IPTG at OD_600nm between 0.4 and 0.8 and grown at 22°C for 12–14 hr. Cells pellets were lysed in lysis buffer (10 mM Tris pH 8.0, 250 mM NaCl, 20 mM Imidazole, 5 mM β‐ME, 1 mM PMSF) by homogenization (Avestin) and cleared lysate was bound to Ni‐NTA agarose beads (Qaigen), washed with 20× CV and eluted with elution buffer (lysis buffer with 300 mM Imidazole). Eluted samples were subjected to GSH‐Sepharose affinity purification (Pierce) and eluted with lysis buffer containing reduced glutathione (10 mM). Eluted samples were concentrated using 10 kDa concentrator (AmiconUltra, Merk) and digested with 5–10 U thrombin per mg of protein while dialysis for 20–24 hr. Dialysed samples were analyzed using superdex200 16/60PG (GE healthcare) column in running buffer (10 mM Tris pH 8.0, 150 mM NaCl, 0.5 mM EDTA, 5 mM β‐ME, and 1 mM PMSF). Integrity of CTC·Nup93 (1–819) or CTC·Nup93 (1–150) complexes was confirmed with analytical SEC using Superdex20010/30 GL (GE healthcare) and SDS‐PAGE analysis.

4.4. SEC‐MALS

The purified protein complexes analyzed with SEC using Superdex200 10/30 GL column coupled with multi‐angle light scattering (MALS) instrument (Wyatt technology). The Wyatt Dawn Helios‐II light scattering detector and Wyatt Optilab T‐rex refractive index detectors were connected in‐line with the chromatography system for simultaneous MALS measurements. The BSA was used as standard (2 mg/ml; Pierce) and data was analyzed with Astra software (Wyatt Technologies). All experiments were performed at room temperature.

4.5. Pull‐down assay

HEK293F cells (ThermoFisher) were cultured in Freestyle‐293 (Gibco) medium, and about 0.6–1 μg of GFP‐fusion plasmids (per 1 × 10⁶ cells/ml) were transfected to HEK‐293F using PEI. Cells were harvested after 48 hr and stored in −20°C. Cells were lysed in 1X DPBS (Gibco) supplemented with 0.2% Tween20 (Sigma) and protease inhibitor cocktail (Sigma). Followed by sonication, lysates were cleared at 14,000 rpm at 4°C. Each cleared lysate (~ 1–1.5 mg) was mixed with 50 μg of GST‐tagged anti‐GFP nanobody (referred as GFP‐NB henceforth). ⁴⁶ As a control, equal amount of each lysate was mixed with 50 μg of purified GST. Lysates were incubated with 50 μl of GSH‐sepharose (Pierce) beads for 4 hr with rotation in cold cabinet. Beads were washed four times with 1xDPBS and complexes were eluted in 80 μl of elution buffer. The pulled‐out samples were resolved on denaturing SDS‐PAGE. Western blot analysis was carried out using Anti‐GFP antibody (1:5,000, Sigma #G1546), anti mAB414 (1:5,000, Abcam #ab24609). HRP‐conjugated mouse IgG (Sigma #A3673) was used at 1:7,000 dilution for developing signal. Images were recorded using Imager600 (GE healthcare) or BioImager (BioRad). Pull‐down experiments were performed thrice independently.

4.6. Confocal microscopy

About 1–5 μg of various Nup93‐GFP plasmid constructs were transfected in HEK‐293 cells cultured in 12 well dishes containing poly‐l‐lysine (Sigma) coated coverslips. After 36 hr of transfection, cells were washed three times with 1xPBS followed by the treatment with CSK buffer ⁴⁷ to remove the cytoplasm. Cells were then fixed with 4% para‐formaldehyde for 10 min followed by washing with 1× PBS. Cells were blocked with 10% BSA and then incubated with mAB‐414 (1:300; Abcam #ab24609) for 12–16 hr at 4°C. Coverslips were washed thrice with 1× DPBS and incubated for 1 hr with GFP‐Chromobody‐Atto488 (1:1,000, Chromtech #GBA488) and anti‐mouse IgG‐AF594 (1:1,000, ThermoFisher #A21203). After washing with 1x DPBS cells were stained with DAPI and mounted in the Vectashield. Cells were imaged with Leica SP8 confocal microscope coupled with 480 nm and 543 nm laser. Recorded images were analyzed using the ImageJ to obtain the plot profiles (using 4–6 cells for each construct) across the nuclear rim. Imaging experiments were repeated twice.

4.7. SAXS analysis

The purified Danio CTC (6.6 mg/ml) or rat CTC·93 (0.9 mg/ml) complexes were used for small angle X‐ray scattering (SAXS) at the European Synchrotron Radiation Facility (ESRF), Grenoble, France on beamline BM29. Scattering datasets were collected in the flow mode using in a standard setup with sample to detector distance of 2.87 mm and X‐ray beam (λ = 0.991 Å) on PILATUS 1 M detector using 0.5 s exposure per frame for total 10 frames at 277 K. The samples buffer (10 mM Tris–HCl, 150 mM NaCl, 1 mM DTT, and 0.5 mM EDTA) was used for the buffer scattering subtraction automatically via EDNA. ⁴⁸ The SAXS datasets were analyzed using ATSAS 2.8. ⁴⁹ DAMMIF or GASBOR programs, were used to build the initial molecular envelope. For GASBOR analysis, total 600 dummy residues for CTC complex and 710 dummy residues for CTC·Nup93 complex were used to reconstruct the low‐resolution structure within the shape constraints derived during distance distribution function P(r) analysis with five iterations independently. Theoretical I(Q) profile of these models obtained in CRYSOL was compared with I(Q) of the raw data; model having lowest Chi‐square (χ ²) value was fitted and visualized in PyMOL. SUPCOM was used to fit SAXS model with the homology models based on PDB ID 5IJO. ²³ The five models obtained using GASBOR analysis for Danio CTC complex had model discrepancy (Normalized spatial discrepancy, NSD) value of 2.657 with standard deviation (SD) of 0.188 calculated using Damsel program. ⁵⁰ For GASBOR analysis of rat CTC‐Nup93 complex, five models had model discrepancy (NSD) value of 2.549 with SD of 0.019. Molecular weight was estimated using DATVC program based on the volume of correlation (V c) in ATSAS 2.8 suite.

4.8. Negative staining and electron microscopy

Carbon coated 300 mesh copper grids (#CF300‐Cu; EMS) were plasma cleaned for 60 seconds (Model 1020, Fischione Instruments) and 4 μl of untagged complex (at a concentration of 0.05 mg/ml) was adsorbed onto the grids for 2 min. After blotting the sample using Whatman™ filter paper‐1, 4 μl of 2% Uranyl acetate (pH ~7.4) was applied to the grid for 1 min. The staining was repeated once more. Afterwards, the grids were blotted and air dried for 2 min, and imaged in a FEI Tecnai microscope (Thermofisher) operating at 200 kV. Micrographs were recorded at 3,600 × 3,600 pixels with a pixel size of 1.38 Å. Single particle analysis, with 89 micrographs, was performed using Relion 2.1. ³⁵ CTF correction was done using CTFFIND3. ⁵¹ 4,156 particles were utilized to obtain 2D classes, and a 3D reconstruction was generated using a modeled structure of human Nup93. The human Nup93 model was generated in pGenthreader, ²⁹ using the crystal structure of yeast Nic96 (PDB: 2QX5) ³³ as a template. The generated models were fitted in the 3D density using UCSF Chimera.

AUTHOR CONTRIBUTIONS

Parshuram Sonawane: Conceptualization; data curation; formal analysis; methodology; writing‐original draft; writing‐review and editing. Pravin Dewangan: Conceptualization; data curation; formal analysis; methodology; writing‐original draft; writing‐review and editing. Pankaj Kumar Madheshiya: Data curation; formal analysis; methodology. Kriti Chopra: Data curation; formal analysis; methodology; writing‐review and editing. Mohit Kumar: Data curation; formal analysis; methodology; writing‐review and editing. Sangeeta Niranjan: Data curation; formal analysis; methodology. Mohammed Yousuf Ansari: Data curation; formal analysis; methodology; writing‐review and editing. Jyotsna Singh: Data curation; formal analysis; methodology. Shrankhla Bawaria: Data curation; formal analysis; methodology. Manidipa Banerjee: Formal analysis; methodology; project administration; resources; writing‐review and editing. Radha Chauhan: Conceptualization; funding acquisition; project administration; supervision; writing‐original draft; writing‐review and editing.

CONFLICT OF INTEREST

The authors declare no conflict of interest or financial interests.

Supporting information

Appendix S1: Supplementary Information

XXX

Sonawane PJ, S. Dewangan P, Madheshiya PK, et al. Molecular and structural analysis of central transport channel in complex with Nup93 of nuclear pore complex. Protein Science. 2020;29:2510–2527. 10.1002/pro.3983