Nup42 and IP₆ coordinate Gle1 stimulation of Dbp5/DDX19B for mRNA export in yeast and human cells

Abstract

The mRNA lifecycle is driven through spatiotemporal changes in the protein composition of mRNA particles (mRNPs) that are triggered by RNA-dependent DEAD-box protein (Dbp) ATPases. As mRNPs exit the nuclear pore complex (NPC) in Saccharomyces cerevisiae, this remodeling occurs through activation of Dbp5 by inositol hexakisphosphate (IP₆)-bound Gle1. At the NPC, Gle1 also binds Nup42, but Nup42’s molecular function is unclear. Here we employ the power of structure-function analysis in S. cerevisiae and human (h) cells, and find that the high-affinity Nup42-Gle1 interaction is integral to Dbp5 (hDDX19B) activation and efficient mRNA export. The Nup42 carboxy-terminal domain (CTD) binds Gle1/hGle1B at an interface distinct from the Gle1-Dbp5/hDDX19B interaction site. A nup42-CTD/gle1-CTD/Dbp5 trimeric complex forms in the presence of IP₆. Deletion of NUP42 abrogates Gle1-Dbp5 interaction, and disruption of the Nup42 or IP₆ binding interfaces on Gle1/hGle1B leads to defective mRNA export in S. cerevisiae and human cells. In vitro, Nup42-CTD and IP₆ stimulate Gle1/hGle1B activation of Dbp5 and DDX19B recombinant proteins in similar, non-additive manners, demonstrating complete functional conservation between humans and S. cerevisiae. Together, a highly conserved mechanism governs spatial coordination of mRNP remodeling during export. This has implications for understanding human disease mutations that perturb the Nup42-hGle1B interaction.

INTRODUCTION

Eukaryotic cellular function is dependent on the proper and efficient export of mRNAs from the nucleus through nuclear envelope (NE)-embedded NPCs to the cytoplasm. An essential rate-limiting step in this pathway occurs at the cytoplasmic face of the NPC on a structural platform over the NPC central channel^1,2. At this site, the ATPase cycle of the S. cerevisiae Dbp5 (DDX19B in human) mediates release of the export factor Mex67-Mtr2 (NXF1 in human) and other RNA-binding proteins (RBPs) from the translocating mRNPs^3,4. Several essential factors regulate the Dbp5 ATPase cycle at the NPC cytoplasmic face: the nucleoporin Nup159 (Nup214 in human) which facilitates ADP release⁵, and the protein Gle1 bound to IP₆ which enhances ATP loading for triggering RNA-dependent ATP hydrolysis^5–8. Control of the Dbp5 ATPase by Gle1-IP₆ and Nup159 at the NPC cytoplasmic face results in spatially coordinated mRNP remodeling of Mex67-Mtr2 and RBPs from exporting mRNPs. As such, directionality is conferred on the export mechanism, allowing release of the mRNP to the cytoplasm for protein synthesis.

Although first identified as an NPC-associated mRNA export factor in S. cerevisiae⁹, Gle1 is conserved in higher eukaryotes^10,11. In addition to regulating Dbp5 at the NPC, Gle1 is required for the function of distinct Dbps at non-NPC subcellular locales. In S. cerevisiae, Gle1-IP₆ regulates Dbp5 activity during translation termination, and Gle1 impacts the ATPase activity and RNA binding of Ded1, another DEAD-box protein, for efficient translation initiation independently of IP₆^12–15. In humans, hGle1 shuttles between the nucleus and cytoplasm with at least two isoforms, hGle1A and hGle1B, produced from a single gene^11,16. hGle1A and hGleB are identical except for the C-terminal region wherein differential splicing results in an extended 43 amino acids in hGle1B (Figure 1A)¹⁶. The hGle1B isoform localizes predominantly at the NPC and is required for mRNA export^16,17 whereas hGle1A is not functional at the NPC and has cytoplasmic roles at stress granules and in regulating the human Ded1 homologue DDX3 during translation¹⁸. Most recently, hGle1 has also been implicated in centrosome assembly and cilia function, potentially regulating an unidentified DEAD-box protein for localized mRNA metabolism¹⁹. Taken together, Gle1 and hGle1 play critical roles in regulating Dbps during multiple stages in gene expression at distinct subcellular locations.

Figure 1. Gle1-Nup42 interaction domains.

(A) Domain structure of Gle1, hGle1B, and hGle1A. For Gle1, K377 K378 indicates IP₆-coordinating residues identified in¹⁰, V513 A516 I520 indicates residues essential for Dbp5 ATPase regulation⁷, and Q491 K494 E501 E502 indicates residues essential for interaction with Nup42 identified in this paper. For hGle1B, K526 K527 are homologous to S. cerevisiae Gle1 K377 K378, with K479 K486 as additional positively-charged residues at the IP₆ interface (based on homologous residues in the Gle1 structure⁷, PDB 3RRM), and Q640 K643 E650 D651 indicates residues essential for interaction with hNup42 identified in this paper. Symbols denote changes from disease-associated mutations^25,27: asterisks are ALS-associated mutations, filled circle indicates LCCS1 Fin_major, open circles indicate other LCCS1 or LAAHD mutations. From amino to carboxy-terminus: S70X, ALS-associated nonsense mutation; T144_E145insPFQ, LCCS1 Fin_major; R569H, LCCS1; V617M, LAAHD; 654/IVS14-2A>C, ALS (alternate exon use results in novel C-terminal extension from this residue); I694T, LAAHD; R697C, ALS. (B) Domain structure of Nup42 and hNup42. For Nup42, 408–424 indicates minimal Gle1 interaction domain defined in this paper. (A & B) Colored regions indicate truncations used in biochemistry experiments. (C) S. cerevisiae Gle1 contains a conserved patch of residues opposite the Dbp5 binding site. From a Clustal Omega³⁵ sequence alignment between Gle1 and hGle1B, identical (red) and similar (orange) residues were mapped onto the Gle1-Dbp5 structure⁷, PDB 3RRM). (D) Q491, K494, E501, and E502 are surface accessible residues in the conserved patch on S. cerevisiae Gle1. Residues are indicated on zoomed region boxed in C.

It is speculated that Gle1’s multi-functional capabilities are dictated by unique protein interaction partners at different subcellular sites²⁰. Thus, to understand how spatial regulation of Dbp5 is controlled at the NPC cytoplasmic face, it is important to precisely pinpoint the roles of Gle1 interactions at the NPC. The amino terminal half of Gle1 contains a coiled-coil region that promotes Gle1 self-association and is required for NPC localization (Figure 1A)¹⁷. Intriguingly, this region of Gle1 crosslinks to several members of the Nup82 holo-complex¹, suggesting a potential mechanism for Gle1 localization at the NPC through direct interaction with this cytoplasmically-oriented complex. The region comprising the first 29 amino acids of the human Gle1 amino-terminus also binds to human Nup155, and is required for NPC localization (Figure 1A)²¹. Finally, the carboxy-terminal domain (CTD) of Gle1/hGle1B binds the CTD of Nup42/hNup42 (also known as Rip1 in S. cerevisiae and hCG1 or NUPL2 in human) (Figure 1B)^9,22,23. The recent structural model of the cytoplasmic NPC face proposes that Gle1, Dbp5, and Nup42 are possibly oriented toward the NPC central channel¹, positioning these factors for interaction with exporting mRNPs. However, the molecular details of the interactions that allow this Gle1 positioning are still undetermined. Revealing how Gle1 acts at the NPC is also needed to give insight into human disease mechanisms. Disease mutations linked to hGLE1 that alter hGle1 self-association, the hGle1-hNup42 interaction, and/or the respective pools of hGle1 at the NPC versus in the cytoplasm are associated with devastating pathologies including lethal congenital contracture syndrome 1 (LCCS1)^17,24,25 and amyotrophic lateral sclerosis (ALS)^26,27.

With respect to Nup42, some discrete functions at the NPC have been defined for its different domains. Our prior studies revealed a role for the FG domain in recruiting the mRNP to be in proximity to Gle1 and Dbp5 for remodeling²⁸, via the Nup42 FG domain interacting with Mex67-Mtr2 (Figure 1B)²⁹. In fact, a fusion between Gle1 and the FG domain of Nup42 (gle1-FG^nup42) bypasses the requirement for Nup42 in some genetic contexts²⁸. The Nup42 CTD is also required for specific functions. Nup42/hNup42 mediates the export of heat shock transcripts^22,30, and deletion of NUP42 together with IPK1, the kinase that produces IP₆, results in a temperature sensitive mRNA export defect³¹. Nup42 CTD expression is sufficient for the function in heat shock mRNA export³² and rescues nup42△ ipk1△ temperature sensitivity³¹; whereas, expression of the gle1-FG^nup42 fusion does not restore these defects in nup42△mutants²⁸. Since proper Gle1 and Dbp5 function are likewise required for heat shock mRNA export^30,33, and IP₆ is required for normal Gle1 regulation of Dbp5, we propose that the Nup42 CTD might also impact their interaction.

Nup42 was among the first identified NPC constituents³⁴, but molecular details of its role at the NPC remain unresolved. Current models propose that the Nup42 CTD serves as a docking site for Gle1 at the NPC, and hNup42 for hGle1B. However, hGle1A lacks efficient binding to hNup42 but localizes to the NPC upon hGLE1 knockdown^18,22. More importantly, cells with only hGle1A at the NPC are defective for mRNA export¹⁸. Thus, Gle1 interaction with Nup42 might be required for proper Gle1 function beyond any role in NPC localization. To further define the mRNA export mechanism mediated by the Gle1-Nup42 interaction at the NPC, we conducted a series of biochemical, genetic and cell biological studies with both the S. cerevisiae and human proteins and respective cells. We find here that Nup42/hNup42 and IP₆ have conserved functions in enhancing Gle1/hGle1B stimulation of Dbp5/DDX19B for mRNA export.

RESULTS

Gle1-Nup42 interaction is required for heat shock mRNA export

To further investigate the Gle1-Nup42 function, we used a structural homology approach to pinpoint residues critical for the interaction. Gle1/hGle1B binding to Nup42/hNup42 is so highly conserved that the hNup42 CTD interacts with S. cerevisiae Gle1 and rescues the heat shock mRNA export defect of nup42Δ mutants²³. Thus, we hypothesized that the amino acid residues involved in the Gle1/hGle1B-Nup42/hNup42 interaction are conserved. From a Clustal Omega³⁵ alignment between S. cerevisiae Gle1 and hGle1B, we mapped the conserved human residues onto the S. cerevisiae protein structure of gle1^H337R-IP₆-dbp5^L327V, PDB 3RRM⁷). A patch of conserved residues was apparent on the surface of Gle1 that was distinct from the Dbp5 binding interface (Figure 1C). Four polar residues were surface accessible on this patch: Q491, K494, E501, and E502 (Figure 1D). We altered these residues to alanine, first as pairs, then all together, and analyzed the interaction with nup42-CTD via the yeast two-hybrid assay (Y2H). Strains expressing each of the pairwise gle1 mutants (GBD-gle1-CTD^QK>AA and GBD-gle1-CTD^EE>AA) showed growth with GAD-nup42-CTD, but altering all four residues (GBD-gle1-CTD^QKEE>AAAA) resulted in no growth on quadruple dropout media, indicating defective interaction with GAD-nup42-CTD (Figure 2A).

Figure 2. Functional analysis of the proposed Nup42 binding site on Gle1.

(A) QKEE>AAAA disrupts the Gle1-Nup42 Y2H interaction. Indicated plasmids were transformed into the Y2H reporter strain, grown to early log phase at 23°C, and plated on the synthetic media lacking indicated amino acids for growth at 23°C. (B) gle1^QKEE>AAAA ipk1Δ phenocopies the growth defect of nup42Δ ipk1Δ. Indicated strains were grown to mid-log phase, serially diluted, and plated on YPD for growth at the indicated temperatures. (C) Gle1 is localized at NPCs when interaction with Nup42 is disrupted. Indicated strains containing carboxy-terminal GFP-tagged GLE1-GFP plasmids were grown to mid-log phase in YPD at 23°C and live cells were imaged by wide-field microscopy. Scale bar = 5μm. (D) gle1^QKEE>AAAA disrupts heat-shock mRNA export. Indicated strains were grown at 25°C to early-log phase, kept at 25°C or shifted to 42°C for 15 min, labeled with [³⁵S]methionine for an additional 15 min, and lysed. Lysates were separated by SDS-PAGE, and proteins were visualized by autoradiography. The positions of Hsp proteins, induced upon heat shock, are indicated by asterisks.

Next, we analyzed the function of gle1^QKEE>AAAA in vivo. The altered proteins were expressed similar to wild-type Gle1 levels (Figure S1A), rescued a gle1Δ mutant, and displayed no obvious growth defects at all temperatures analyzed, similar to the growth of nup42Δ (Figure 2B, top panel). When expressed as the only copy in cells (Figure S1B), GFP-tagged gle1^QKEE>AAAA localized at the NE rim to the same relative extent as wild-type Gle1-GFP. Similarly, Gle1-GFP was localized to the NE rim in nup42Δ cells (Figure 2C). However, gle1^QKEE>AAAA mutants demonstrate a temperature-sensitive growth defect when combined with ipk1Δ mutants, phenocopying the growth defect previously observed with nup42Δ ipk1Δ mutants (Figure 2B, bottom panel)³¹. Finally, we analyzed heat shock mRNA export by testing³⁵S methionine incorporation to protein synthesis after shifting to growth at 42°C. Under these conditions, poly (A)⁺ RNA is retained in the nucleus, but heat shock transcripts are permitted to export, resulting in preferential translation of heat shock proteins³⁶. Previous analysis had determined that nup42Δ mutants are defective in heat shock mRNA export and protein production^30,32, so we anticipated that loss of nup42 binding by the gle1^QKEE-AAAA protein would likewise impact the heat shock response. Indeed, the gle1^QKEE>AAAA mutant demonstrated a loss of heat shock protein production similar to nup42Δ mutants (Figure 2D, lanes 8, 10). Therefore, we concluded that the Gle1-Nup42 interaction is required for heat shock mRNA export, gle1^QKEE>AAAA disrupts this interaction, and association with Nup42 is not required for Gle1 localization at the NPC.

A minimal region of Nup42 that binds Gle1 is not sufficient for function

Prior studies defined the Nup42 CTD (residues 365–430) as sufficient for Y2H interaction with Gle1²³. Using XtalPred-RF³⁷, the Nup42 CTD was predicted to have significant regions of disorder (data not shown). To more narrowly define the required Gle1 interaction domain in the Nup42 CTD, we generated a series of nup42-CTD truncations in the Y2H bait construct (GAD-nup42-CTD) and assayed with GBD-gle1-CTD. Both the GAD-nup42^405–424 and GAD-nup42^408–425 strains showed growth with GBD-gle1-CTD, revealing a region of seventeen residues (408–424) as necessary for the Gle1-Nup42 interaction (Figure 3A). Moreover, GAD-nup42 truncations lacking residues 408–424 did not grow with GBD-gle1-CTD. From a Clustal Omega alignment of nup42-CTD and hnup42-CTD, residues 408–424 were in the most highly conserved region (Figure 3B). Observing conservation in the minimal Nup42 region required for Nup42-Gle1 binding is consistent with the fact that altering residues found on the conserved patch Gle1 disrupts interaction with Nup42 (Figure 2). BioLayer Interferometry analysis demonstrated a high affinity interaction between recombinant purified MPB-gle1-CTD and a biotinylated peptide corresponding to nup42^408–424, with a K_D of 0.0915 μM (Figure 3C). Purified recombinant MBP-gle1-CTD^QKEE>AAAA had significantly reduced association with the biotin-nup42^408–424 peptide with a K_D of 0.599 μM. Thus, the nup42^408–424 region was necessary and sufficient for physically binding Gle1.

Figure 3. Analysis of minimal Gle1-binding domain in Nup42.

(A) nup42^408–424 is required for interaction with Gle1. Indicated plasmids were transformed into the Y2H reporter strain, and struck to –Trp –Leu –His –Ade synthetic media for growth at 23°C. Growth on quadruple drop-out media is indicated as “+”. (B) nup42^408–424 lies in a conserved region of the protein. Clustal Omega³⁵ alignment between nup42-CTD and hnup42-CTD. Identical (black) and similar (grey) residues are indicated. The black line above the sequence indicates residues 408–424 of Nup42. (C) Affinity measurements between gle1-CTD and nup42^408–424. BioLayer Interferometry was performed with a biotin-nup42^408–424 peptide and recombinant gle1-CTD protein. Calculated K_D and R² correlation is indicated. (D) GFP-nup42^408–424 localization at the nuclear envelope is disrupted in gle1^QKEE>AAAA mutants. Strains containing the indicated plasmids were grown to mid-log phase at 23°C and live cells were imaged by wide-field microscopy. Scale bar = 5μm. (E) nup42^408–424 does not rescue the growth defect of nup42Δ ipk1Δ mutants. nup42Δ ipk1Δ was transformed with the indicated GFP-tagged constructs, grown to mid-log phase, and plated on –Leu synthetic media for growth at the indicated temperatures. (F) nup42^408–424 is not sufficient for heat shock mRNA export. nup42Δ mutants were transformed with the indicated GFP-tagged constructs, grown at 25°C to early-log phase in synthetic media, kept at 25°C or shifted to 42°C for 15 min, labeled with [³⁵S]methionine for an additional 15 min, and lysed. Lysates were separated by SDS-PAGE, and proteins were visualized by autoradiography. Hsp proteins are indicated by asterisks.

To determine whether this nup42^408–424 motif was sufficient for NE rim localization, GFP-tagged Nup42 proteins were expressed in nup42△ and gle1^QKEE>AAAA nup42△ cells. Direct fluorescence microscopy showed full-length GFP-Nup42 and GFP-nup42-CTD both localized comparably at the NE rim, indicative of NPC incorporation (Figure 3D). Although a higher background of pan-cellular signal was observed, GFP-nup42^408–424 also localized to the NE rim, and, importantly, GFP-nup42^408–424 localization was absent from the NE in gle1^QKEE>AAAA mutant cells (Figure 3D). This suggested that interaction with Gle1 is sufficient for localizing GFP-nup42^408–424 to the NPC. However, GFP-Nup42 and GFP-nup42-CTD were NE rim-localized in gle1^QKEE>AAAA mutants, indicating that additional NPC interactions besides Gle1 contribute to Nup42 localization. Interestingly, Nup42 proteins lacking the Gle1 interaction site (GFP-nup42Δ408–424 and GFP-nup42-CTDΔ408–424) were not detected by fluorescence microscopy or immunoblotting, suggesting that loss of interaction with Gle1 reduces Nup42 stability (Figure S2B).

Because nup42^408–424 bound Gle1 and localized to the NE rim, we speculated that the nup42^408–424 minimal domain might also be sufficient for functions attributed to the Nup42 CTD. Genetic tests were conducted to determine whether expression of nup42^408–424 alone could rescue defects associated with absence of the Nup42 CTD. Centromeric GFP-NUP42, GFP-nup42-CTD, and GFP-nup42^408–424 expression vectors were transformed into the temperature sensitive nup42Δ ipk1Δ mutant and growth was assessed at various temperatures. Similar to previously published results³¹, GFP-NUP42 and GFP-nup42-CTD rescued the temperature-sensitivity of nup42Δ ipk1Δ. However, GFP-nup42^408–424 did not rescue the growth defects (Figure 3E; land 6, 8). The Nup42 CTD is also required for export of heat-shock transcripts following a temperature shift to 42°C²³. Therefore, we tested whether GFP-nup42^408–424 cells produced heat shock proteins as an assay for heat-shock mRNA export. Although GFP-nup42-CTD cells produced heat-shock proteins, GFP-nup42^408–424 cells did not (Figure 3F). Untagged nup42^408–424 also did not rescue nup42△ growth and heat-shock mRNA export defects (data not shown). Thus, although nup42^408–424 was sufficient for interaction with Gle1 and localization to the NE, this region was not sufficient to rescue the function of the Nup42 CTD.

Nup42 interacts with IP₆-Gle1-Dbp5 complexes

The Gle1 CTD interacts with both Nup42 and Dbp5 and our mutation analysis indicated that Nup42 and Dbp5 interact at distinct Gle1 interfaces. To reveal the role of the Nup42 CTD, we tested whether Gle1 can simultaneously bind both Nup42 and Dbp5. Soluble binding assays were performed with purified recombinant proteins. We used the gain-of-function proteins MBP-gle1-CTD^H337R and/or dbp5^L327V from prior structural studies⁷ and GST or GST-nup42-CTD bound to resin. A trimeric GST-nup42-CTD/MPB-gle1-CTD^H337R/dbp5^L327V complex was observed only in the presence of IP₆ (Figure 4A; lane 13, 14). In this assay, binding of GST-nup42-CTD to dbp5^L327V was not observed in the absence of MBP-gle1⁻CTD^H337R (Figure 4A; lane 11, 14). Thus, Gle1 bound Nup42, Dbp5, and IP₆ at the same time forming an IP₆-trimeric protein complex.

Figure 4. Nup42 coordinates Gle1-Dbp5 interaction.

(A) Nup42, Gle1, and Dbp5 interact in a trimeric protein complex with IP₆. Soluble binding assays were conducted with purified recombinant proteins immobilized on glutathione coupled Sepharose. Input and bound fractions were analyzed by SDS-PAGE and Coomassie staining. Note: GST-nup42-CTD is proteolytically sensitive and resolves as several bands by SDS-PAGE (lane 2), as observed in²³. (B) GBD-gle1-CTD^QKEE>AAAA does not interact with GAD-Dbp5 via Y2H. Indicated plasmids were transformed into the Y2H reporter strain, grown to early log phase at 23°C, and plated on the synthetic media lacking indicated amino acids for growth at 23°C. (C) GBD-gle1-CTD does not interact with GAD-Dbp5 in nup42Δ mutants. Indicated plasmids were transformed into wild-type (wt) or nup42Δ Y2H reporter strains, grown to early log phase at 23°, and plated on the synthetic media lacking indicated amino acids for growth at 23°C. (D) Purified recombinant S. cerevisiae proteins used in ATPase assays. 1μg indicated proteins were resolved by SDS-PAGE and Coomassie stained. Asterisk indicates GST cleaved (not removed) from Dbp5 sample. (E) nup42-CTD enhances Gle1 stimulation of Dbp5 ATPase activity in the absence of IP₆. PK/LDH-coupled ATPase assays were performed with Dbp5 (500 nM) in the presence of 1 μM A₃₀ RNA and 2 mM ATP gle1-CTD or gle1-CTD^QKEE>AAAA (250 nM), IP₆ (1 μM), and varying amounts of nup42-CTD or nup42^408–424 peptide (250, 500, or 750 nM) were added as indicated. Reactions were incubated at 37°C for 40 min with the A₃₄₀ monitored every 40 seconds to calculate Kcat. Mean shown for n=3, and standard error is indicated by error bars.

To investigate whether Nup42 impacts the interaction between wild-type Gle1 and Dbp5 in cells, we used the Y2H assay. Notably, although GAD-DBP5 and GBD-gle1-CTD cells grew on selective media, GAD-DBP5 and GBD-gle1-CTD^QKEE>AAAA cells did not (Figure 4B). Thus, in the Y2H assay, Nup42 interaction was required for Gle1 to bind Dbp5. As a further test of this, we compared Y2H results in a reporter strain with NUP42 deleted. The nup42△ reporter strain expressing GAD-DBP5 and GBD-gle1-CTD did not grow (Figure 4C). Notably, the GBD-gle1-CTD^VAI>DDD mutant (shown previously to disrupt Gle1 stimulation of Dbp5⁷) also did not interact with GAD-DBP5 by Y2H (Figure 4B), even though GBD-gle1-CTD^VAI>DDD did grow with GAD-nup42-CTD (Figure 2A). These results provided further support that Gle1 binds Nup42 and Dbp5 at different interfaces and that Nup42 might impact the interaction between Gle1 and Dbp5.

Gle1 stimulation of Dbp5 ATPase activity is enhanced by Nup42 or IP₆

To directly determine whether Nup42 CTD has an effect on Gle1 stimulation of the RNA-dependent Dbp5 ATPase activity, in vitro assays were conducted using our established methods and bacterially-expressed recombinant proteins (Figure 4D). Upon addition of IP₆, Gle1 stimulates the ATPase activity of Dbp5 ~6-fold in a colorimetric enzyme coupled ATPase rate assay (Figure 4E; lane 2, 7, 8), as previously determined^6,8. Under these conditions, when recombinant purified H₆-nup42-CTD was added at 750nM into the assay, no obvious change in ATPase activity was observed (Figure 4E, lane 12). Because our hypothesis is that Nup42 enhances the interaction between Gle1 and Dbp5, and IP₆ performs a similar function^6,8, we then tested whether H₆-nup42-CTD affected the Dbp5 ATPase activity in the absence of IP₆. Indeed, dose-dependent enhancement of gle1-CTD -stimulated Dbp5 ATPase activity was observed when H₆-nup42-CTD was added to this assay without IP₆ present (Figure 4E, lanes 9–11). Of note, the full Nup42 CTD (residues 365–430) was required for this stimulation as the nup42^408–424 peptide did not enhance ATPase activity when titrated at the same molar concentrations (Figure 4E, lanes 19–21). H₆-nup42-CTD did not stimulate Dbp5 ATPase activity in the absence of gle1-CTD (data not shown). Finally, although gle1-CTD^QKEE>AAAA showed somewhat reduced stimulation of Dbp5 in the presence of IP₆ (Figure 4E; lane 8, 14), H₆-nup42-CTD demonstrated no dose-dependent enhancement of Dbp5 ATPase activity in the presence of gle1-CTD^QKEE>AAAA. Therefore, through interaction with Gle1, Nup42 and IP₆ each augmented Gle1-mediated stimulation of Dbp5 in manners that were not additive.

The hNup42-hGle1B interaction is conserved in human cells and required for mRNA export

Because the QKEE residues for Nup42 interaction in S. cerevisiae Gle1 are conserved in hGle1B (Figure 1D), we investigated whether hGle1B requires hNup42 interaction in a similar manner. We first examined the localization of mCherry-hGle1B and mCherry-hgle1B^QKED>AAAA in HeLa cells. Although mCherry-hgle1B^QKED>AAAA was expressed at lower levels than mCherry-hGle1B (Figure S3A), mCherry-hgle1B^QKED>AAAA was detected at the NE rim in a subset of transfected cells (Figure 5A). More importantly, the proportion of cells exhibiting NE-localized mCherry-hgle1B^QKED>AAAA increased upon siRNA-mediated hGLE1 knockdown (Figure 5B), indicating that endogenous hGle1 outcompetes hgle1B^QKED>AAAA for NE rim localization. This is identical to hGle1A behavior wherein it only localizes at the NE rim in the absence of hGle1B¹⁸. To analyze whether mCherry-hgle1B^QKED>AAAA was functional for mRNA export, we tested for complementation of the mRNA export defect upon hGLE1 knock down (Figure S3B). As previously demonstrated¹⁷, knockdown of hGLE1 (A and B) with specific siRNAs resulted in accumulation of poly(A)⁺ RNA in the nucleus, and expression of siRNA-resistant mCherry-hGLE1B^R rescued the mRNA export defect (Figure 5C, 5D). However, expression of siRNA-resistant mCherry-hgle1B^{R QKED>AAAA} did not rescue of mRNA export. No change in the nuclear:cytoplasmic ratio of poly(A)⁺ RNA was observed in mCherry-hgle1B^R-QKED>AAAA transfected cells, compared to vehicle and mCherry (Figure 5D). Indeed, cells with rim-localized mCherry-hgle1B^QKED>AAAA maintained a significant mRNA export defect (Figure 5C). These results indicated that the conserved QKED residues are required for hGle1B function, suggesting that interaction with hNup42 is required for mRNA export in human cells.

Figure 5. hGle1B interaction with hNup42 is required for mRNA export.

(A–B) mCherry-hgle1B^QKED>AAAA is localized to the nuclear rim in a portion of cells. (A) HeLa cells expressing exogenous siRNA-resistant mCherry-hGLE1B^R and mCherry-hgle1B^R-QKED>AAAA after treatment with CTRL or hGLE1 siRNA were imaged by confocal microscopy. Scale bar=10μm. (B) Quantification of cells with rim-localized mCherry signal from the experiment in A. Approximately sixty cells were quantified from three independent experiments, and the mean is indicated with standard error indicated by error bars. (C-D) mCherry-hgle1B^QKED>AAAA does not rescue the mRNA export defect upon hGLE1 knockdown. (C) HeLa cells were transfected with mCherry, mCherry-hGLE1B^R or mCherry-hgle1B^R-QKED>AAAA plasmids following CTRL or hGLE1 knockdown. After fixation, cells were processed for in situ hybridization using 488-labeled oligo d(T) probe followed by mCherry immunofluoresence. Scale bar=10μm. (D) Quantification of nuclear:cytoplasmic mean fluorescence intensity of oligo d(T) signal. Approximately 100 cells were quantified per condition from three independent experiments with nuclear:cytoplasmic intensity ratio depicted in box and whisker plots (whiskers representing 10%–90%). Statistical analysis includes ANOVA with a post-hoc t-test. (E-F) hNUP42 depletion results in a poly(A)⁺ mRNA export defect. (E) HeLa cells were transfected with CTRL or hNUP42 siRNA for 72h. After fixation, cells were processed for in situ hybridization using Cy3-labeled oligo d(T) probe followed by hNup42 immunofluorescence. Scale bar=10μm. (F) Quantification of nuclear:cytoplasmic mean fluorescence intensity of oligo d(T) signal. Approximately 200 cells were quantified per condition from three independent experiments with nuclear:cytoplasmic intensity ratio depicted in box and whisker plots (whiskers representing 10%–90%). A t-test was performed to analyze statistical significance.

hNup42 is required for human mRNA export

Prior studies by our laboratory and others tested for roles of the human homologues of S. cerevisiae Gle1, Dbp5, and Ipk1 in mRNA export through NPCs. Knockdown of hGLE1B, DDX19B, and hIPK1 (also known as IPPK) or dominant negative functional perturbation results in nuclear accumulation of poly(A)⁺ RNA^16,17,38,39, reflecting a block to mRNA export. Although we previously showed that hNup42 is required for heat-shock mRNA export²², mRNA export under normal growth conditions was not assessed. To test this directly, siRNA knockdown of hNUP42 in HeLa cells was conducted and poly(A)⁺ RNA localization analyzed at 37°C. A robust nuclear accumulation of poly(A)⁺ RNA signal, with loss of cytoplasmic signal, was observed when hNup42 was depleted (Figures 5E, S3C). Quantification demonstrated a statistically significant increase in the nuclear:cytoplasmic signal for poly(A)⁺ RNA upon hNUP42 knockdown (Figure 5F). Therefore, hNup42 was essential for mRNA export in human cells, corroborating the mRNA export defect observed in mCherry-hgle1B^R-QKED>AAAA cells. Further, as depletion of each of the factors all result in mRNA export defects, we concluded that the molecular mechanism for human hGle1B, DDX19B, hNup42, and IP₆ function in mRNA export is similar to that of their S. cerevisiae homologues.

hNup42 and IP₆ each enhance hGle1B stimulation of DDX19B ATPase activity

To further examine the conservation of hNup42 and IP₆ functions in the human system, direct in vitro ATPase assays were conducted with DDX19B. Initial experiments were performed with purified bacterially-expressed DDX19B (denoted DDX19B*) (Figure S4); however, the bacterially-expressed DDX19B* with or without hgle1B-CTD had ~8.3 fold lower enzymatic activity compared to that of S. cerevisiae Dbp5, gle1-CTD and IP₆ (Figure S4). We then tested His-tagged DDX19B (H₆-DDX19B) purified from a Baculoviral expression system in insect cells based on reported successful DDX19B in vitro protein interaction studies and reports that DDX19B is post-translationally modified in mammalian cells^40,41 (Figure 6A), The H₆-DDX19B purified from insect cells displayed significant ATPase activity which was stimulated ~4-fold by hgle1B-CTD in the presence of IP₆ (Figure 6B; lane 2, 6, 7). Strikingly, titration of hnup42-CTD resulted in dose-dependent enhancement of the hGle1B-stimulated H₆-DDX19B ATPase activity in the absence of IP₆ (Figure 6B, lanes 8–10). hNup42 and IP₆ did not have an additive effect (Figure 6B; lanes 7, 10, 11). Finally, although IP₆ further stimulated the hgle1B^QKED>AAAA mediated activation of H₆-DDX19B (Figure 6B; lane 12, 13), hnup42-CTD did not enhance hgle1B^QKED>AAAA stimulation of H₆-DDX19B ATPase activity (Figure 6B, lanes 14–16). In sum, in vitro human DDX19B activation was dependent on IP₆, hGle1B, and hNup42, with the human and S. cerevisiae proteins showing fully conserved in vitro functions.

Figure 6. hGle1B interactions with hNup42 and IP₆ enhances stimulation of DDX19B activity and are required for mRNA export.

(A) Purified recombinant human proteins used in ATPase assays. 1μg indicated purified proteins were resolved by SDS-PAGE and Coomassie stained. (B) hGle1B and hNup42 stimulate DDX19B ATPase activity in absence of IP₆. PK/LDH-coupled ATPase assays were performed with H₆-DDX19B (500 nM, expressed in insect cells) in the presence of 1 μM A₃₀ RNA and 2 mM ATP. hgle1B-CTD or hgle1B-CTD^QKED>AAAA (250 nM), IP₆ (1 μM), and varying amounts of hnup42-CTD (250, 500, or 750 nM) were added as indicated. Reactions preformed as in Figure 3D. Mean shown for n=3, and standard error is indicated by error bars. (C) hGle1B-IP₆ interaction is conserved in human proteins. PK/LDH-coupled ATPase assays were performed with H₆-DDX19B (500 nM) in the presence of 1 μM A₃₀ RNA and 2 mM ATP, hgle1B-CTD, hgle1B-CTD^KK>QQ, or hgle1B-CTD^KKKK>QQQQ (250 nM), and hnup42-CTD (750 nM) or IP₆ (1 μM) were added as indicated. Purified proteins are shown in 6A. Reactions performed as in Figure 3D. Mean shown for n=3, and standard error is indicated by error bars. (D) hGle1B-IP₆ interaction is required for mRNA export. HeLa cells were transfected with vehicle or GFP-hGLE1B^R, GFP-hgle1B^R-KK>QQ, or GFP-hgle1B^R-KKKK>QQQQ plasmids following CTRL or hGLE1 knockdown. Nuclear:cytoplasmic mean fluorescence intensity of oligo d(T) signal was quantified as in Figure 5F.

hGle1B requires conserved IP₆ binding residues for DDX19B stimulation and mRNA export

To further investigate the IP₆ dependence of hGle1B-DDX19B activity in human cells, we tested whether hGle1B coordinates IP₆ through the same surface accessible protein interface as the S. cerevisiae Gle1. We previously defined the residues required for IP₆ binding to S. cerevisiae Gle1 (K377 and K378) by identification of positive residues that were conserved across fungal and metazoan species¹⁰. In prior structural modeling of fungal, plant and human Gle1, it was observed that the IP₆ binding pocket of hGle1 and Arabidopsis thaliana Gle1 is less positively-charged than that of S. cerevisiae⁴². However, the homologous IP₆-coordinating lysine (K) residues of hGle1B (K526, K527) are conserved. Altering these lysine (K) residues of hGle1B (K526, K527) to glutamine (Q) (hgle1B-CTD^KK>QQ) abolished the enhanced stimulation of H₆-DDX19B in the presence of IP₆ (Figure 6C, comparing lanes 7, 10). We also tested a hGle1B variant with changes in two additional positively-charged residues (K479, K486) whose homologous residues are found near the IP₆ binding pocket of S. cerevisiae Gle1. The hgle1B-CTD^KKKK>QQQQ also did not show IP₆ dependent stimulation of H₆-DDX19B. Importantly, the hgle1B-CTD^KK>QQ protein still stimulated H₆-DDX19B at a basal level (Figure 6C, comparing lane 1 with 10 and 11), and with the addition of hnup42-CTD, hgle1B-CTD^KK>QQ enhanced H₆-DDX19B ATPase activity levels comparable to that with wild-type hgle1B-CTD and hnup42-CTD (Figure 6C, lanes 12–14). Thus, the IP₆ interface changes on hgle1B-CTD^KK>QQ did not perturb the hNup42 interface or overall folding.

Further supporting a conserved role for an IP₆ interaction with hGle1B during mRNA export, we found that expression of the siRNA-resistant GFP-hgle1B^KK>QQ did not rescue the mRNA export defect seen with siRNA hGLE1-knockdown in human cells, whereas wild-type GFP-hGle1B did rescue (Figure 6D). As such, mRNA export was inhibited in human cells when the IP₆-dependent hGle1B function was perturbed by altering the conserved K526 and K527 residues. Thus, although the plant and human IP₆ interactions surfaces are less positively charged compared to S. cerevisiae and not fully homologous in terms of protein sequence, these results with the prior studies⁴² documented a conserved requirement for IP₆ during hGle1B function in mRNA export in all tested eukaryotic cells.

DISCUSSION

Here we reveal important insights into the functional organization of the mRNA export machinery at the NPC cytoplasmic face. Our biochemical and physiological assays demonstrate a unique role for Nup42 in mediating Gle1 regulation of the Dbp5 ATPase activity. In both S. cerevisiae and human systems, Nup42 is not critical for localizing Gle1 at the NPC. Rather, Nup42/hNup42 enhances the Gle1/hGle1B stimulation of the RNA-dependent Dbp5/DDX19B ATPase (Figure 4E and 6B). We find that Gle1 utilizes distinct surface accessible binding sites for interaction with IP₆, Nup42 and Dbp5 (Figure 1C), and a nup42-CTD/gle1-CTD/Dbp5 trimeric complex forms in the presence of IP₆. As some of the hGLE1 disease-associated mutations alter the carboxy-terminal region of hGle1B required for hNup42 binding (Figure 1A)^25,27, we predict that perturbed DDX19B activation and mRNP remodeling dynamics at the NPC underlie some of these human pathologies.

An active role for Nup42 and hNup42 during mRNA export is distinct from the historically attributed function of these proteins in localizing Gle1/hGle1 to the NPC. Here, in nup42Δ mutants, we find Gle1-GFP localizes to the NPC as robustly as in wild-type cells, and GFP-gle1^QKEE>AAAA is also localized to the NE rim (Figure 2C). In fact, Nup42 is at least partly reliant on Gle1 for its localization since GFP-nup42^408–424 is lost from the rim in gle1^QKEE>AAAA mutants (Figure 3D). Moreover, the hGle1A isoform which lacks hNup42 binding localizes to the NPC in the absence of hGle1B¹⁸, and hGle1 remains localized to the NE rim upon hNUP42 knockdown²². A recent structural analysis of the Nup82 complex (Nup159, Nup82, Nsp1) revealed the interactions that anchor this complex to the S. cerevisiae NPC cytoplasmic side¹, with the FG domains for Nup159 and Nsp1 situated across the NPC channel for their function in mRNA export^28,43. The Fernandez-Martinez study also reports substantial interactions between the Gle1 amino terminal region (NTD) and the Nup82 complex, which corroborates our prior finding that the hGle1 NTD is required for NPC localization (both its coiled-coil region and the Nup155 interaction region (Figure 1A,^17,21;. This also explains why overexpression of GLE1 rescues growth defects of nup82 mutants⁴⁴. In contrast, Nup42’s only reported crosslinking partner is Gle1¹. Thus, the Gle1 NTD is the primary determinant of NPC localization whereas the Gle1 CTD interaction with Nup42 and IP₆ is required for mRNP remodeling.

Importantly, although a minimal conserved peptide of seventeen amino acids in Nup42 (nup42^408–424) is sufficient for interaction with Gle1, additional sequence in the Nup42 CTD is required for function (Figures 3, 4). A conserved patch on Gle1 is essential for interaction with the minimal nup42^408–424 peptide (Figure 1D, 2, 3), placing this interface at the opposite face of where Dbp5 binds Gle1 (Figure 1C,⁷). However, Nup42 is required for the Y2H interaction between Gle1 and Dbp5 in cells (Figure 4C). We predict that the sequence in the Nup42 CTD flanking the minimal Gle1 binding peptide (residues 408–424) is responsible for mediating the Gle1-Dbp5 interaction and the nup42-CTD/gle1-CTD activation of Dbp5. To date, we have not been able to determine whether this occurs directly (with Nup42 interacting with both Gle1 and Dbp5) or indirectly (with Nup42 binding Gle1 and altering its conformation). We did not observe a Nup42-Dbp5 interaction in our soluble binding assay (Figure 4A lane 11), but it is possible that low affinity interactions are not detected by this method. We also analyzed the IP₆-trimeric GST-nup42-CTD/MPB- gle1^H337R/dbp5^L327V protein complex formation in the presence of different nucleotides; however, no obvious differences in gle1^H337R-dbp5^L327V interaction were observed, potentially due to the use of gain-of-function proteins required for this assay (Figure 4A).

The hypothesis that Nup42 mediates an interaction between Gle1 and Dbp5 suggests that the Nup42 CTD role is similar to that of IP₆, explaining why no additive effect is seen with IP₆ and nup42-CTD during in vitro ATPase assays (Figure 4E and 6B). The redundant functions also explain why nup42Δ ipk1Δ double mutants are temperature-sensitive with no growth defects observed for single mutants in S. cerevisiae³¹. IP₆ is required for GST-nup42-CTD/MBP-gle1-CTD^H337R/dbp5^L327V trimeric protein complex formation in soluble binding assays (Figure 4A). We speculate that IP₆ is required in vitro for trimeric complex formation but does not enhance Gle1 stimulation of Dbp5 in the presence of Nup42 because the former experiment requires capture of a stable interaction, whereas observation of ATPase activity only relies on transient stimulation. Having two parallel ways to support Gle1-Dbp5 interaction through either Nup42 or IP₆ potentially allows combinatorial control over the NPC-localized activity of Dbp5 in cells. Such spatial control is likely especially critical for modulating Gle1 through its multiple subcellular locations and functions.

We find that hGle1B, like S. cerevisiae Gle1, demonstrates enhanced stimulation of DDX19B in the presence of IP₆ or hNup42-CTD. However, in contrast to S. cerevisiae, hGle1B interactions with hNup42 and IP₆ are each individually required for mRNA export in human cells, indicating stricter control over this process in vivo (Figure 5 and 6D). The molecular requirement for hNup42 during nuclear mRNA export might also be due to the fact that the IP₆ binding pocket of hGle1B is less positively-charged⁴² and potentially binds to IP₆ more weakly than S. cerevisiae Gle1 does to IP₆. However, as others have reported, the IP₆ binding pocket of Arabidopsis thaliana Gle1 has even lower positive charge than hGle1B but demonstrates enhanced stimulation of LOS4 (A. thaliana Dbp5 orthologue) in the presence of IP₆⁴². Therefore, a requirement for IP₆ in Gle1 stimulation of DEAD-box ATPases during mRNA export is conserved between S. cerevisiae, plants and human cells.

Based on our proposed mechanism for Nup42 CTD function, we posit that Nup42/hNup42 acts at an early step in the ATPase cycle of Dbp5/DDX19B at the NPC, where Gle1 enhances ATP loading onto Dbp5. This model is diagrammed in Figure 7. With the Nup42 FG domain recruiting mRNPs for remodeling by Dbp5²⁸, the Nup42 CTD ensures activation of Dbp5 (via the Gle1-Nup42 CTD interaction) when Mex67-bound mRNPs are present (via the Nup42 FG-Mex67 interaction). We propose that Nup42 serves as an important sensor that couples these steps in the mechanism. We anticipate that these interactions might direct specific interaction between Dbp5/DDX19B and mRNA to enable remodeling of specific proteins. As other DEAD-box proteins are activated by interaction partners in a similar manner to Gle1 activation of Dbp5 (eIF4G for eIF4A and THO for Sub2^45,46), similar mechanisms might exist to spatially regulate these activities. This work also illustrates how Nup42/hNup42 as a specific binding partner at the NPC cytoplasmic face allows specificity to Gle1 and hGle1B function differentially from cytoplasmic Gle1/hGle1A function. With S. cerevisiae Nup42 being specifically required for heat shock mRNA export, we propose that Nup42 binding might also preferentially stabilize Gle1 and confer thermostability on the mRNA export mechanism. As such, this might explain how the Nup42/Gle1-mediated mRNA export mechanism is crucial under heat shock conditions.

Figure 7. Schematic for Nup42 function at the NPC cytoplasmic face.

(A) Dbp5 dynamically associates with the NPC. (B) The Nup42 CTD stimulates Gle1-IP₆ interaction with Dbp5, with Gle1-IP₆ promoting ATP loading onto Dbp5. Mature mRNPs interact with the Mex67-Mtr2 heterodimer via adaptors such as the poly(A)⁺ binding protein Nab2, permitting export through the NPC. (C) The FG domain of Nup42 recruits the mRNP in close proximity to Dbp5, and Dbp5-ATP binds RNA. Dbp5 might also be a constituent of the exporting mRNP. (D) ATP hydrolysis induces a conformational change in Dbp5 and the bound RNA for release of Nab2, and Mex67-Mtr2, which are recycled into the nucleus for additional rounds of mRNP export. The remodeled mRNP is released into the cytoplasm. Dbp5 interacts with the Nup159 amino terminal domain (NTD) to facilitate ADP release.

One important distinction between the Dbp5 and DDX19B activities might be linked to a role for potential post-translational modifications or small effectors in the human system. A recent study reported roles for DDX19B phosphorylation in altering its nuclear activity during the DNA damage response⁴⁰. We show here that DDX19B* purified from bacteria is not activated by hgle1-CTD in the presence of IP₆ or hnup42-CTD, and demonstrates relatively low enzymatic activity (Figure S4). In contrast, DDX19B purified from insect cells shows full, conserved functionality, including gle1-CTD-mediated stimulation that is enhanced in the presence of hnup42-CTD or IP₆ (Figure 6). Our future studies will be focused on pinpointing potential DDX19B modifications needed for activity and regulation during mRNA export.

To further delineate the mRNA export mechanism, additional structural details regarding hNup42 and hGle1B molecular interactions in the context of the NPC are needed, and further analysis is required to fully understand how the mRNP remodeling platform is arranged to interact with exporting mRNPs. Overall, we demonstrate that the NPC protein Nup42 interacts with Gle1/hGle1B to locally regulate Dbp5/DDX19B activity. Importantly, this work demonstrates a remarkable conservation of function for the mRNP remodeling factors Gle1/hGle1B, Nup42/hNup42, Dbp5/DDX19B, and IP₆ during mRNA export in human and S. cerevisiae cells. These results also reveal novel unexpected steps in the molecular mechanism for mRNA export at the NPC cytoplasmic face that impact normal and disease cell function.

METHODS

Yeast Strains, Growth, and Y2H Analysis

Table S1 lists the yeast strains used in this study. Yeast genetic methods including mating, sporulation, dissection, and transformations were conducted according to standard procedures⁴⁷. Yeast strains were grown at indicated temperatures in either YPD (2% peptone, 2% dextrose, 1% yeast extract) or selective minimal media lacking appropriate amino acids and supplemented with 2% dextrose and 5-fluoroorotic acid (5-FOA; United States Biological) as needed at 1.0 mg/mL. For growth analysis on plates, indicated strains were grown in YPD liquid media to mid-log phase (OD₆₀₀ ~0.5), serially diluted, and equal numbers of cells were spotted onto YPD media for growth at the indicated temperatures. For Y2H analysis, vectors expressing GBD and GAD fusions were transformed into the reporter strain (PJ69-4A) or nup42Δ reporter strain (SWY6432) and selected on –Trp –Leu synthetic media. The resulting strains were then grown to mid-log phase, serially diluted, and equal numbers of cells were spotted onto –Trp –Leu or –Trp –Leu –His –Ade media. The plates were then incubated at 23°C for the indicated number of days.

Vector Construction

Table S2 lists the vectors used in this study. Vectors were cloned using standard molecular biology strategies or by Gibson Assembly (New England Biolabs, Ipswich, MA), and site-directed mutagenesis was performed using QuikChange (Agilent Genomics, Santa Clara, CA), and sequencing confirmed all vectors generated.

Live Cell Microscopy

Yeast strains were grown to mid-log phase in YPD or synthetic media lacking appropriate amino acids at the indicated temperatures. Cultures were collected, re-suspended in synthetic complete media at room temperature, and imaged. Wide-field images were acquired using a microscope (BX50; Olympus) equipped with a motorized stage (Model 999000, Ludl), Olympus 100× NA1.3 UPlanF1 objective, and digital charge coupled device camera (Orca-R2; Hamamatsu). Images were processed with ImageJ (NIH) or Adobe Photoshop CS6.

Heat Shock Protein Production Assay

The [³⁵S]methionine incorporation assay of heat shock proteins was performed as described^28,48. Nascently produced radiolabeled proteins from whole-cell lysate were separated by SDS-PAGE, and the resulting gel was dried and exposed to autoradiography film.

Immunoblotting

For immunoblotting of yeast S. cerevisiae whole-cell lysates, cells were cultured and lysates were prepared as previously described in SDS loading buffer²⁸. Proteins were resolved by SDS-PAGE and blotted using affinity-purified guinea pig anti-Gle1 (ASW 43¹⁰), rabbit anti-GFP (Thermo Fisher Scientific, Waltham, MA), or anti-yPgk1 (Thermo Fisher Scientific). IRDye 800CW-conjugated goat anti-mouse (LI-COR, Lincoln, NE) or Alexa Fluor 700 goat anti-rabbit (Thermo Fisher Scientific) antibodies (1:5000) were visualized with the Li-Cor Odyssey scanner (Lincoln, NE).

For HeLa cell lysate immunoblotting, cells were grown in 60 mm dishes (Fisher Scientific, Pittsburg, PA) and lysed in RIPA buffer (Sigma, St. Louis, MO) supplemented with EDTA-free cOmplete protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). Proteins were resolved by SDS-PAGE and blotted with rat anti-mCherry (Sigma), anti-hNup42/NUPL2 (Sigma), mouse anti-beta-actin (Sigma), or anti-hGle1¹⁷ antibodies. Infrared 700- or 800-conjugated secondary antibodies (Thermo Fisher Scientific) were visualized with the Li-Cor Odyssey scanner.

HeLa cell culture and transfection

HeLa cells were cultured in complete DMEM media (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS, Atlanta Biologicals, Norcross, GA) at 37°C in 5% CO₂. Knockdown add-back of hGLE1 was performed using the previously validated protocol¹⁷. Negative control siRNA and hGLE1 siRNAs were purchased from Qiagen (Valencia, CA). hNUP42 SMARTpool siRNA was purchased from Dharmacon (Lafayette, CO). Cells were reverse-transfected with indicated 25 nM siRNAs using HiPerFect (Qiagen) and then transfected 24 h later with relevant constructs (siRNA-resistant mCherry-hGLE1B^R or mCherry-hgle1B^R-QKED>AAAA for Figure 4; siRNA-resistant GFP-hGLE1B^R or GFP-hgle1B^R-KK>QQ for Figure 5) using Fugene 6 (Promega, Madison, WI) per manufacturer’s instructions.

HeLa immunofluorescence

HeLa cells expressing mCherry-hGLE1B^R and mCherry-hgle1B^R-QKED>AAAA were grown on 1.5 mm round coverslips in a 24-well plate (Fisher Scientific) and immunofluorescence was performed as previously described¹⁸ using mAb414 (Biolegend, San Diego, CA) and anti-mCherry (Sigma). Cells were imaged using a 63x/1.4 NA oil-immersion objective on a confocal microscope (Leica TCS Sp5). Images were processed with ImageJ (NIH) or Adobe Photoshop CS6.

HeLa mRNA export assay

The previously validated knockdown add back approach (for hGLE1¹⁷ or described hNUP42 knockdown was employed followed by in situ hybridization of 488-oligo d(T) to poly(A)⁺ RNA. Cells were imaged using a 63x/1.4 NA oil-immersion objective on a confocal microscope (Leica TCS Sp5). The nuclear:cytoplasmic ratio was determined by measuring mean intensity of these compartments in Image J (NIH) and dividing nuclear by cytoplasmic signal. Images were processed with ImageJ (NIH) or Adobe Photoshop CS6.

Protein Expression and Purification

Proteins were expressed (except DDX19B* and H₆-DDX19B) in E. coli Rosetta cells (Novagen) cultured in Terrific Broth under chloramphenicol and corresponding plasmid antibiotic resistance (kanamycin or ampicillin) selection. Protein expression was induced with IPTG (0.2 mM) at an OD₆₀₀ of 0.5–0.8 for 18 hours at 18°C. Bacteria were lysed by sonication in Buffer B (20 mM HEPES (pH 7.5), 150 mM NaCl, 0.5 mM DTT, and 20% glycerol) supplemented with EDTA-free cOmplete protease inhibitor cocktail (Sigma) and 2 mM PMSF. All purified proteins were snap-frozen in an ethanol-dry ice bath and stored at −80°C.

The H₆-DDX19B plasmid was transformed into DH10Bac™ (Thermo Fisher Scientific) E. coli cells and plated on Luria agar plates containing tetracycline (10 μg/mL), gentamycin (7 μg/mL), kanamycin (50 μg/mL), IPTG (40 μg/mL), and X-Gal (100 μg/mL). White colonies were re-struck three times before the recombinant bacmid was purified from 1.5 mL of an overnight culture in LB containing tetracycline, gentamycin, and kanamycin. Transposition was verified by PCR with M13/pUC forward and reverse sequencing primers. Sf9 cells were cultured in Sf-900(TM) III SFM (Thermo Fisher Scientific) at 27°C. P0 virus stocks were generated by transfecting Sf9 cells (2 mL at 0.5×10⁶ cells/mL) in a 6-well plate using Cellfectin(R) II Reagent (Thermo Fisher Scientific) and harvested three days post transfection. P1 virus was generated by adding 0.25 mL of P0 stock to 100-mm Petri dishes containing 10 mL of cells at 1×10⁶ cells/mL and harvested three days post infection. P2 virus stocks were generated by adding 0.2 mL of P1 virus stocks to 20 mL of 1×10⁶ cells/mL Sf9 cells in 125 mL shake flasks. P2 virus stocks were harvested after three days post infection with shaking at 140 rpm. Protein expression was achieved using High Five™ cells (Thermo Fisher Scientific), which were cultured in Express Five(R) SFM (Thermo Fisher Scientific) at 27°C with shaking at 140 rpm. P2 virus stocks (10 mL) were used to infect 450 mL of High Five™ cells (1×10⁶ cells/mL) in 1 L shake flasks. High Five™ cells were harvested 48 hrs post infection and lysed using a Dounce homogenizer in the presence of Buffer B and 2 mM PMSF.

For MBP, MBP-gle1-CTD^H337R, H₆-MBP-TEV-gle1-CTD, H₆-MBP-TEV-gle1-CTD^QKEE>AAAA, H₆-MBP-TEV-hgle1B-CTD, H₆-MBP-PPS-hgle1B-CTD^QKED>AAAA, H₆-MBP-PPS-hgle1B-CTD^KK>QQ, and H₆-MBP-PPS-hgle1B-CTD^KKKK>QQQQ proteins, the soluble fractions were purified with amylose resin (New England Biolabs) according to manufacturer recommendations. The amino-terminal tags on H₆-MBP-TEV-gle1-CTD, H₆-MBP-TEV-gle1-CTD^QKEE>AAAA, H₆-MBP-TEV-hgle1B-CTD, H₆-MBP-PPS-hgle1B-CTD^QKED>AAAA, H₆-MBP-PPS-hgle1B-CTD^KK>QQ, and H₆-MBP-PPS-hgle1B-CTD^KKKK>QQQQ were cleaved by addition of H₆-TEV or H₆-PPS to the amylose elution fractions and incubation overnight at 4°C. His-MBP and protease were removed by collecting flow-through after Ni Immobilized Metal Affinity Chromatography (Ni IMAC) (New England Biolabs). Flow-through was dialyzed overnight in Buffer B.

H₆-DDX19B and H₆-nup42-CTD were purified by Ni IMAC. Elution fractions were pooled and dialyzed overnight in Buffer B (H₆-nup42-CTD) or Buffer B with 400 mM NaCl (H₆-DDX19B).

GST, GST-Dbp5, GST-dbp5^L327V, GST-nup42-CTD, and GST-hnup42-CTD were purified by glutathione-coupled Sepharose chromatography. For GST-Dbp5, GST-dbp5^L327V, and GST-hnup42-CTD, the GST tag was cleaved by Factor Xa (New England Biolabs) digestion and inactivated by the addition of PMSF. The GST tag was removed from dbp5^L327V by ion-exchange chromatography as in⁴. All proteins were dialyzed overnight in Buffer B.

GST-DDX19B* was expressed in E. coli Rosetta cells cultured in Terrific Broth under ampicillin and chloramphenicol antibiotic selection. Protein expression was induced with IPTG (0.1 mM) at an OD₆₀₀ of 0.8 at 18°C for 18 hours. The harvested cell pellet was suspended in 30 mM HEPES pH 7.5, 400 mM NaCl, 10% Glycerol, 1 mM DTT, 2 mM MgCl₂ supplemented with ETDA-free cOmplete protease inhibitor cocktail and 2 mM PMSF and cells were lysed by sonication. Cleared lysate was incubated with buffer equilibrated glutathione-coupled sepharose 4B resin for 1.5 hours at 4°C with rotation. The resin was washed with 12 column volumes of buffer before elution with 10 mM reduced glutathione. The GST tag was removed by Factor Xa incubation for 36 hours at 4°C after the NaCl concentration was reduced to 200 mM by dilution with buffer containing no NaCl. GST was removed by passing the protein over buffer-equilibrated glutathione-coupled sepharose 4B resin and collecting the flow through. Protein homogeneity was verified by SDS-PAGE.

ATPase Assay

Colorimetric enzyme-coupled ATPase rate assays were performed as described⁵. Briefly, reaction mixtures (100μL), containing 10 mM HEPES (pH 7.5), 45 mM NaCl, 2 mM MgCl₂, 1 mM DTT, 20U SUPERase·In™ RNase Inhibitior (Thermo Fisher Scientific), 6 mM PEP (Sigma), 1.2 mM NADH (Sigma), 2.5 mM ATP (Sigma), 2 μL PK/LDH (Sigma), 1 μM polyA RNA, 500 nM Dbp5 or H₆-DDX19B, 1 μM IP₆ (Sigma, as indicated), 250 nM gle1-CTD or hgle1B-CTD (or mutants), and titration of H₆-nup42-CTD or hnup42-CTD as indicated. A₃₄₀ was monitored every 40 sec for 40 min at 37°C in a BioTek Synergy HT microplate reader. Kcat/sec was calculated as (((OD₃₄₀/min × 2.5)/6.22×10⁻³)/μM protein).

BioLayer Interferometry

BioLayer interferometry assays were performed utilizing a ForteBIO Octet-Red96 at the Vanderbilt Antibody and Protein Resource Core. Biotinylated nup42^408–424 peptide (GenScript, Piscataway, NJ) was immobilized on a streptavidin conjugated biosensor at a well concentration of 0.5 μM in buffer containing 20 mM HEPES (pH 7.4), 200 mM NaCl, 1 mM DTT, 10% glycerol, 0.25 mM IP₆, and 0.01% Nonidet P-40. Immobilization was carried out over 5 minutes, followed by a 5-minute incubation in 100 μg/mL biocytin to cap any free streptavidin sites on the biosensor surface. Following a 5-minute wash and 1-minute baseline, gle1-CTD or gle1-CTD^QKEE>AAAA was allowed to associate for 5 minutes. Dissociation was followed for an addition 5 minutes. A single biosensor was used to measure binding at each concentration in parallel and responses in buffer alone were used to control for signal drift. The data analysis software provided by ForteBio was used to normalize all responses to an appropriate baseline, and subtract the buffer only control. The K_D was calculated from the determined k_on and k_off rates for the interaction.

In vitro binding assay

400 pmol GST or GST-nup42-CTD, 250 pmol MBP or MBP-gle1^H337R-CTD, and 500 pmol dbp5^L327V were incubated with equilibrated glutathione-coupled Sepharose in Buffer B. IP₆, ADP, or AMP-PNP, were added at 50 μM. Samples were incubated at 4°C for 2 hours and washed three times with Buffer B (IP₆, ADP, or AMP-PNP added in wash buffer where appropriate). Bound proteins were eluted in SDS loading buffer, resolved on a 12% (bottom) 7.5% (top) Tris-Glycine gel, and Coomassie stained.

SYNOPSIS.

Activation of the DEAD-box ATPase Dbp5 by Gle1-IP₆ is required for mRNA export through nuclear pore complexes (NPCs). In this paper, Adams, et al. demonstrate that a constituent of the NPC, Nup42, impacts the ATPase activity of Dbp5 through interaction with Gle1. This function, as well as IP₆-hGle1B interaction, is required for mRNA export in human cells and altered in human diseases. This work reveals how Gle1-Dbp5 activity is locally coordinated for function at the proper place and time.