Conserved residues at the Mtr4 C-terminus coordinate helicase activity and exosome interactions

Abstract

Mtr4 is an essential RNA helicase involved in nuclear RNA processing and degradation and is a member of the Ski2-like helicase family. Ski2-like helicases share a common core architecture that includes two RecA-like domains, a winged helix, and a helical bundle (HB) domain. In Mtr4, a short C-terminal tail immediately follows the HB domain and is positioned at the interface of the RecA-like domains. The tail ends with a SLYΦ sequence motif that is highly conserved in a subset of Ski2-like helicases. Here, we show that this sequence is critical for Mtr4 function. Mutations in the C-terminus result in decreased RNA unwinding activity. Mtr4 is a key activator of the RNA exosome complex, and mutations in the SLYΦ motif produce a slow growth phenotype when combined with a partial exosome defect in S. cerevisiae, suggesting an important role of the C-terminus of Mtr4 and the RNA exosome. We further demonstrate that C-terminal mutations impair RNA degradation activity by the major RNA exosome nuclease, Rrp44 in vitro. These data demonstrate a role for the Mtr4 C-terminus in regulating helicase activity and coordinating Mtr4-exosome interactions.

INTRODUCTION

RNA processing and surveillance pathways are critical for proper cellular function. Disruption of these pathways can lead to a variety of disease states, including cancer, autoimmunity, and neurodegenerative diseases.^1–5 The essential Ski2-like helicase, Mtr4, plays a central role in nuclear RNA decay, providing a physical link between upstream RNA substrate targeting events and downstream RNA degradation events.⁶ Mtr4 has been linked to liver tumorgenesis in humans⁷ and studies have shown that mutations or depletion of Mtr4 cause various RNA defects in vivo.^8–10 In addition, mutations in the cytoplasmic homolog of Mtr4, Ski2, cause the intestinal and immune disorder, trichohepatoenteric syndrome (THES).^5,11,12 Mtr4 engages with a variety of adaptor proteins/complexes to target RNA substrates. For example, the Mtr4-mediated TRAMP, NEXT, and PAXT complexes target substrates for degradation,^13–15 while Mtr4 interactions with ribosomal processing factors such as Nop53 and Utp18 promote processing of rRNA for ribosome maturation.¹⁶ Once an RNA is targeted, Mtr4 delivers the substrate to the RNA exosome for processing or complete degradation.^17–19

The eukaryotic RNA exosome is found both in the cytoplasm and nucleus, and is activated by Ski2 (in the cytoplasm) and Mtr4 (in the nucleus). The cytoplasmic exosome is primarily involved in mRNA surveillance and turnover,²⁰ whereas the nuclear exosome function includes RNA biogenesis, maturation, and surveillance.²¹ The exosome is comprised of nine catalytically inert proteins that form a barrel-like core (Exo9). This core structure includes six PH-like proteins that form the hexameric base and three S1/KH domain “cap” proteins that sit on top and serve as the interface with exosome cofactors.^22,23 In S. cerevisiae, the Exo9 core associates with ribonuclease Rrp44 (or Dis3, DIS3 in human) in both the cytoplasm and the nucleus. Rrp44 interacts with the PH-like ring and generally accesses RNA substrates after they traverse the Exo9 core.^24–26 The nuclear exosome contains an additional ribonuclease, Rrp6, which interacts with the cap proteins and engages RNA substrates directly.^27–29 In addition to Rrp6, the nuclear exosome utilizes two additional co-factors, Rrp47 (RRP47 or C1D in human) and Mpp6 (MPHOSPH6 in human). Rrp6, Rrp47 and Mpp6 each interact with Mtr4 and facilitate interaction with the exosome near the cap ³⁰. Mtr4 association with the exosome is thought to activate RNA degradation by directing unwound, single-stranded RNA through the exosome core to Rrp44.^18,19

Ski2-like helicases play important roles in DNA repair, recombination, transcription, translation, splicing, and RNA turnover processes.³¹ A distinguishing feature of this family is their relatively large size (typically 100-200 kDa) and their ability to provide a platform for larger macromolecular assemblies.³¹ These helicases contain a common core architecture composed of two RecA-like domains, a winged-helix domain, and a helical bundle (HB) domain.^9,19,32 As with other helicases, conserved sequence motifs in the RecA-like domains are associated with ATP binding/hydrolysis, RNA binding/translocation, and communication between ATP hydrolysis and RNA unwinding. Much of the functional variability between Ski2-like helicases appears to be achieved outside of the RecA-like domains through “accessory” domains that flank the helicase core.^32–37

Multiple studies from our lab and others have identified features of Mtr4 accessory domains that are important for helicase function and interactions. Mtr4 contains an unstructured N-terminal region that interacts with the N-termini of exosome cofactors Rrp6 and Rrp47.³⁰ A large insertion in the winged-helix termed the arch domain serves as a recruiting platform for several adaptor proteins⁶ and directly engages with RNA substrates.^30,38–40 Ribosome biogenesis and processing factors (e.g. Nop53, Utp18, NVL) and exosome adaptor complexes (e.g. TRAMP and NEXT) contain an arch interacting motif (AIM) that promotes binding and complex formation with Mtr4.^{14,16,35,41–43}

An unexplored accessory region of Mtr4 is a short tetrapeptide, which we refer to as the C-terminal tail, that extends off the HB domain. This region of Mtr4 is highly conserved⁴⁴ with a consensus SLYΦ sequence. Interestingly, the C-terminal tail is positioned at the interface of the RecA-like domains, suggesting that it may influence interactions in this region and potentially affect Mtr4 function. It is also located near binding sites for ZCCHC8 (human NEXT complex)^14,45 and Mpp6 (exosome cofactor),^18,19 suggesting a potential role for the C-terminal tail in Mtr4-complex interactions and exosome activity. Here, through a combination of biochemical and genetic analyses, we show that mutations in the C-terminal SLYΦ motif alter Mtr4 and TRAMP helicase activity. We also show that mutations of the SLYΦ motif alter the ability of Mtr4 to activate exosome degradation by Rrp44. Genetic analyses reveal a common function between the C-termini of Mtr4 and Rrp6. Our data demonstrate that the Mtr4 C-terminal tail is critical for helicase function and may facilitate productive interactions with the exosome.

MATERIAL AND METHODS

Sequence and structural analysis of Ski2-like helicases

Sequence conservation was determined by multiple sequence alignment of model organisms and conservation scoring with the ConSurf server.⁴⁶ ConSurf output scores of 7-9 (on a scale of 1-9, with 9 indicating highest conservation) are highlighted in Figure 1. S. cerevisiae and H. sapiens Mtr4 structures were aligned to the RecA1 domain of the helicase core using PyMOL ⁴⁷. All available structures were examined and categorized into open and closed core conformations. The PDB accession codes for the open core conformation are: 4QU4 and 5OOQ. Closed core conformations: 2XGJ, 4U4C, 6FSZ, 6FT6, 6IEG, 6IEH, 6C90, 6RO1, 6D6Q, 7S7B, 7Z52. Molecular graphics were rendered using PyMOL. The rotation and translation of the RecA2 and HB domains compared to RecA1 was quantified in PyMOL using the angle_between_domains command downloaded from pymolwiki.org.

Figure 1. Sequence and structural analysis of Ski2-like helicases.

(A) Structures of Ski2-like helicases: Mtr4 (2XGJ), Ski2 (4A4Z), Hel308 (2P6U), and Brr2 (5DCA). Ski2-like helicases share a conserved helicase core, which consists of RecA1 (blue), RecA2 (orange), winged helix (green), and HB (pink) domains. For clarity, accessory domains flanking the Ski2-like helicase core are not shown. The loop structure (purple) that extends off the C-termini of the HB domains of Mtr4 and Ski2 lies at the interface of the RecA-like domains, whereas the corresponding region extends in a different direction for the other family members (dotted circle). (B) Sequence alignment and conservation of Ski2-like helicases show high conservation at the C-termini of Mtr4 and Ski2. While the HB domain is at the C-terminus of Mtr4 and Ski2 (arrow), additional domains extend beyond the HB domain in other Ski2-like helicases (e.g. Hel308 and Brr2). Sequence conservation of each Ski2-like helicase was determined by conservation scoring with the ConSurf server. ConSurf output scores of 7-9 (on a scale of 1-9, where 1 is the least conserved and 9 is the most conserved) are highlighted and colored from light yellow (7) to orange (9), respectively. (C) The C-terminal motif (SLYL) interacts with RecA1 and RecA2 domains. Mtr4 and Ski2 share a conserved SLYΦ sequence, where Φ represents a hydrophobic residue (L, M, T, V, F). Available Mtr4 structures show that S1070 and L1071 interact with the RecA2 domain, while L1073 interacts with the RecA1 domain. Y1072 interacts with both RecA1 and RecA2 domains. Hydrogen bonding, hydrophobic, and pi-stacking interactions are indicated. Mtr4 C-terminal and reciprocal RecA mutations used in this study are listed.

Mutagenesis, protein expression and purification

C-terminal mutations of Mtr4 were made using complementary primers containing the targeted mutations using a Q5 Site-Directed Mutagenesis kit (NEB). Mutations were targeted to disrupt interactions in observed binding pockets (L1071, Y1072, and L1073) and carboxylate of terminal amino acid. Mutations were confirmed by Sanger sequencing (ACGT, Inc.). Full-length wild-type Mtr4 (Mtr4^WT) and Mtr4 mutants were recombinantly expressed on a pET151/D-TOPO vector with an N-terminal hexahistidine tag and induced using an autoinduction protocol.^48,49 Cell lysis was performed by sonication. Protein purification was performed using Ni affinity, Heparin, DEAE, and gel filtration chromatography as described previously for Mtr4^WT.⁴⁸ While most of the Mtr4 mutants purified essentially as wild-type, the C-terminal truncation (Mtr4^ΔSLYL) mutant eluted as a broad peak off gel filtration (Fig. S4). Trf4 and Air2 were co-expressed on a pETDuet vector with an N-terminal hexahistidine tag and purified as described previously. Trf4 contains an active site mutation (D236A/D238A) and Air2 contains a 121 residue C-terminal deletion (final construct contains residues 1-223). Protein concentration was determined using a NanoDrop spectrophotometer (ThermoFisher) with an extinction coefficient of 90190 M⁻¹ cm⁻¹ for Mtr4 and 78730 M⁻¹ cm⁻¹ for Trf4-Air2 (calculated using ExPASy ProtParam⁵¹). For Mtr4^Y1072F, Mtr4^Y1072A and Mtr4^ΔSLYL mutants, an extinction coefficient of 88700 M⁻¹ cm⁻¹ was used. Since the Mtr4^ΔSLYL mutant eluted in a broad peak off gel filtration (suggesting potential instability), protein concentration was verified with in-gel quantification and a BSA standard curve.

All wild-type exosome constructs and plasmids were obtained from the Lima lab. Expression and purification of exosome subunits and reconstitution were carried out as described previously.⁵² Briefly, eight exosome proteins were co-expressed in pRSFDuet vectors (Rrp42/Mtr3, Rrp43/Rrp46, Rrp41/Rrp45, Rrp6/Rrp47) with N-terminal Smt3 tags on Rrp42, Rrp46, Rrp41, and Rrp47. Each of the remaining exosome proteins (Csl4, Rrp4, Rrp40, Rrp44, Mpp6) were independently expressed in pSMT3 TOPO vectors with N-terminal Smt3 tags. A catalytically inactive Rrp6 construct (D238N) was made using the Q5 Site-Directed Mutagenesis kit (NEB). All proteins were recombinantly expressed in an Escherichia coli BL21-CodonPlus (DE3)-RIL competent cell line (Novagen) using IPTG induction. All expression was performed using baffled flasks except for Rrp44, which only produced satisfactory levels of protein expression in a 10 L fermenter growth. Exo9 was reconstituted by combining the core and cap proteins (Rrp42, Mtr3, Rrp43, Rrp46, Rrp41, Rrp45, Csl4, Rrp4, Rrp40). Additional exosome proteins were added to reconstitute Exo11^Rrp44/Mpp6 (Exo9+Rrp44+Mpp6), Exo12^{Rrp44/Rrp6/Rrp47} (Exo9+Rrp44+Rrp6+Rrp47), and Exo13^{Rrp44/Rrp6/Rrp47/Mpp6} (Exo9+Rrp44+Rrp6+Rrp47+Mpp6) as described previously.⁵²

RNA substrate design and purification

Oligoribonucleotides were synthesized and HPLC-purified by Integrated DNA Technologies (Coralville, IA). Duplex substrates were prepared by heating an equimolar mixture of each strand in annealing buffer (20 mM Tris-Cl pH 8.0, 100 mM KCl, 0.5 mM EDTA) to 95°C for 5 min after which samples were allowed to slowly cool to room temperature. Annealed substrates were purified by gel filtration using Superdex 200 increase 10/300 GL column (GE Healthcare) pre-equilibrated with annealing buffer.

Oligoribonucleotides used to prepare binding, helicase (unwinding), and ATPase substrates correspond to the ds₁₆UA₅ substrate used in previous Mtr4 studies^36,50 and include the following 5’ to 3’ sequences with duplex regions underlined: R16 (displacement strand, with 5’ fluorescein label), 5′ FAM-AGCACCGUAAAGACGC-3′; R16 (displacement strand, without fluorescein label), 5′-AGCACCGUAAAGACGC-3′; R22A (loading strand, UAAAAA 3’ single-stranded overhang), 5′-GCGUCUUUACGGUGCUUAAAAA-3′.

Oligoribonucleotides used to prepare decay (degradation) substrates correspond to the ds₁₇A₁₀ substrates used in previous Mtr4-exosome studies^30,53 and include the following 5’ to 3’ sequences with duplex regions underlined: R17 (displacement strand), 5’-AAG UGA UGG UGG UGG GG-3’; R27 (loading strand, with 5’ fluorescein label and 10 nt poly(A) 3’ single-stranded overhang), 5’ FAM-CCC CAC CAC CAU CAC UUA AAA AAA AAA-3’.

RNA Binding assay

Dissociation constants (K_D) between Mtr4 and RNA (ds₁₆UA₅) were measured using an electrophoretic mobility shift assay (EMSA). Increasing concentrations of Mtr4^WT and Mtr4 mutants were mixed with 60 nM RNA in binding buffer (40 mM MOPS pH 6.5, 100 mM NaCl, 0.5 mM MgCl₂, 5% [v/v] glycerol, 0.01% [v/v] nonidet-P40 substitute, 0.5 U/μl of RiboLock, 2 mM dithiothreitol). Samples were incubated on ice for 45–60 min and diluted 1:4 with loading buffer (20% [v/v] glycerol). Samples were separated on a native 4–20% gradient polyacrylamide TBE gel for 100-210 min at 100 V in the cold room (4°C). The gel was imaged using fluorescein mode (530/28 Emission filter/Bandpass, nm) with a ChemiDoc MP Imaging system (Bio-Rad Laboratories). The fraction of RNA bound was quantified by densitometry using Image Lab (Bio-Rad Laboratories) and fit as described previously.³⁸ Data were fit using KaleidaGraph (Synergy Software). Statistical significance comparing Mtr4^WT to each mutant from at least three independent assays were determined using a one-way ANOVA followed by a Dunnett’s post hoc test.

RNA unwinding assay

Pre-steady state unwinding assays were performed as described previously.⁴⁸ Briefly, unwinding activity was determined by monitoring the displacement of a 16 bp RNA duplex with a UAAAAA 3’ single-stranded overhang (ds₁₆UA₅). Unwinding reactions were performed in 1 X helicase buffer (40 mM MOPS (pH 6.5), 100 mM NaCl, 0.5 mM magnesium chloride, 5% glycerol, 0.01% nonidet-P40 substitute (Amresco), 2 mM dithiothreitol and 0.5 U/μl of RiboLock (ThermoFisher Scientific)) at 30°C. Reactions were pre-incubated for 5 min with 1.2 μM of Mtr4^WT or Mtr4 mutant protein and 10 nM RNA (final concentration). For assays with TRAMP, Mtr4 was mixed with inactive Trf4-Air2 and incubated for at least 10 minutes prior to adding the RNA. 100 nM DNA-trap oligonucleotide that is complementary to the fluorescently labeled displacement strand was then added into the reaction and incubated for an additional 5 min. Reactions were initiated with equimolar ATP and MgCl₂ at saturating concentrations of 1.6 mM. At specified time points, 10 μL aliquots of the reaction were removed and quenched in a 1:1 ratio with quench buffer containing a final concentration of 0.5% sodium dodecyl sulfate, 5 mM ethylenediaminetetraacetate (EDTA), and 10% glycerol. Quenched reactions were then run on a native gel (15% polyacrylamide TBE-PAGE) at 120 V for 90-120 min in the cold room (4°C). Fluorescein-labeled RNA was visualized using the fluorescein mode (530/28 Emission filter/Bandpass, nm) with a ChemiDoc MP Imaging system (Bio-Rad Laboratories). Statistical analysis comparing Mtr4^WT or TRAMP^WT to each mutant from at least three independent assays were determined using a one-way ANOVA followed by a Dunnett’s post hoc test.

ATPase assay

RNA-stimulated ATPase activity was measured using a malachite green phosphate kit according to manufacturer guidelines (BioAssay Systems). Absorbance was monitored at 650 nm with a VERSA max tunable plate reader (Molecular Devices). An increase in absorbance at 650 nm correlates to free inorganic phosphate production and indirectly measures ATP hydrolysis. ATPase assays were performed with 0.2 μM of protein and 5 μM of RNA in 1 X helicase buffer (40 mM MOPS (pH 6.5), 100 mM NaCl, 0.5 mM magnesium chloride, 5% glycerol, 0.01% nonidet-P40 substitute (Amresco), 2 mM dithiothreitol and 0.5 U/μl of RiboLock (ThermoFisher Scientific)) at 30°C. A control reaction (without RNA) was run for each reaction to obtain intrinsic background ATPase activity of Mtr4 and Mtr4 mutants. Mtr4 has no detectable intrinsic ATPase activity without RNA. Mtr4 and RNA (ds₁₆UA₅) were pre-incubated at 30°C for 10-15 minutes in the assay buffer prior to initiating the reaction with saturating concentrations of ATP and MgCl₂ (1.33 mM each). For TRAMP experiments, 4-5 μM of Mtr4 or Mtr4 mutants were incubated on ice for at least 30 minutes with inactive Trf4-Air2 in a 1:1 ratio. TRAMP (0.2 μM) was then added to reaction buffer with RNA and pre-incubated at 30°C for 10-15 minutes prior to initiating with ATP and MgCl₂. At specified time points (0, 1, 2, 3, 4, 5, 6 minutes), aliquots of the reaction were removed and quenched with working reagent (provided in the kit). Initial rates ((μM Pi min⁻¹) / μM Mtr4) were calculated by fitting a linear trend line to the inorganic phosphates produced over time (Fig. S7) using GraphPad Prism. Statistical significance comparing Mtr4^WT or TRAMP^WT to each mutant from at least three independent assays were determined using a one-way ANOVA followed by a Dunnett’s post hoc test.

RNA degradation assays

Exoribonuclease assays were performed as described by Zinder and Lima.⁵² Briefly, Rrp6 and/or Rrp44 (Dis3) activity was determined by monitoring the degradation of a 17 bp RNA duplex with a 10 nt 3’ poly(A) overhang (ds₁₇A₁₀). All assays include wild-type Exo13^WT (Exo9+Rrp44+Rrp6+Rrp47+Mpp6) or inactive Rrp6 (Exo13^Rrp6exo-) and were performed in triplicate. Exoribonuclease assays were performed in 1 X RNA decay buffer (20 mM HEPES-KOH pH 7.5, 50 mM potassium acetate, 1.1 mM magnesium acetate, 2.5 mM DTT) at 30°C. Final concentrations were 60 nM exosome, 65 nM Mtr4, 1 mM ATP, 0.5 U/μl of RiboLock (ThermoFisher Scientific), and 5 nM RNA substrate. Exosome complexes were pre-incubated with or without Mtr4 at room temperature in 1 X RNA decay buffer for 45-60 minutes prior to initiating the reaction. At specified time points, 10 μL aliquots of the reaction were removed and quenched with 5 μL of quench buffer, followed by proteinase K digestion at 37°C for 1 hr. The final concentration of quench buffer contains 5% (v/v) glycerol, 0.1% (w/v) SDS, 10 mM EDTA, 1 U/mL proteinase K (Thermo Scientific). Quenched reactions were run on a denaturing gel (7 M urea, 20% TBE-PAGE). Fluorescein-labeled RNA was visualized using the fluorescein mode (530/28 Emission filter/Bandpass, nm) with a ChemiDoc MP Imaging system (Bio-Rad Laboratories). Rrp44 product was determined for Exo13^Rrp6exo- reactions by quantifying the intensity of remaining substrate over time and dividing it by the intensity of substrate before initiating the reaction (t= 0 minutes). The percent of total RNA was then averaged for each time point and plotted against time.

Yeast methods

To analyze Mtr4 functions in yeast, the MTR4 gene, from 288 bp upstream of the start codon to 265 bp downstream of the stop codon was inserted into the centromeric plasmid pRS415. Mutations were introduced by cloning a synthetic gene fragment (Twist bioscience or Genscript) containing the last 34 codons of MTR4 and its 3’UTR and replacing the wild-type fragment (as a ZraI SacI restriction fragment).

To analyze Rrp6 mutants, the RRP6 gene, from 111 bp upstream of the start codon to 349 bp downstream of the stop codon was inserted into the centromeric plasmid pRS413. The Rrp6^{ΔN (129-733)}, Rrp6^D238N, and Rrp6^{ΔC (1-520)} constructs were generated by standard molecular biology techniques. For Rrp6^ΔN, codons 1 to 128 were replaced with GFP to maintain stable expression. For Rrp6^ΔC, codons 522 to 733 were deleted.

To analyze Mtr4 function, yeast plasmids with mutated MTR4 genes were transformed into yav1150 (MatA ura3-Δ0 leu2-Δ0 his3-Δ1 mtr4Δ::NEO [MTR4, URA3]). Multiple transformants were selected on media lacking uracil and leucine (SC-Leu-Ura, Sunrise Science), serially diluted and plated on 5-FOA (5-fluoro-orotic acid) containing plates. 5-FOA is converted to the toxic 5-fluoro-uracil by the Ura3 enzyme, and therefore growth on 5-FOA media indicates that the plasmid containing wild-type MTR4 on the URA3 plasmid has been lost, and the mutant MTR4 encoded on the LEU2 plasmid is functional. Each transformant gave similar results and a representative one is shown.

To analyze Mtr4 function in the absence of Rrp6, yeast plasmids with mutated MTR4 genes were transformed into yav1233 (Mat alpha ura3-Δ0 leu2-Δ0 his3-Δ1 met15-Δ0 mtr4Δ::HYG rrp6Δ::NEO [MTR4, URA3]). Multiple transformants were selected on media lacking uracil and leucine (SC-Leu-Ura, Sunrise Science), serially diluted and plated on 5-FOA (5-fluoro-orotic acid) containing plates. Each transformant gave similar results and a representative one is shown.

To analyze Mtr4 function combined with specific mutations in rrp6, each of the strains that contained rrp6Δ and an Mtr4 point mutant (e.g. Mat alpha ura3-Δ0 leu2-Δ0 his3-Δ1 met15-Δ0 mtr4Δ::HYG rrp6Δ::NEO [mtr4-S1070A, LEU2] was transformed with the Rrp6 plasmids described above. Two transformants were selected on media lacking histidine and leucine (SC-His-Leu, Sunrise Science), serially diluted and plated on SC-His-Leu plates. Both transformant gave similar results and a representative one is shown.

Western blots were performed as previously described.⁹ Deletion of MTR4 is lethal. In order to analyze MTR4 (wild-type or mutants) in an mtr4 deletion background, yeast were supplemented with Mtr4 mutants expressed exogenously from a low copy number plasmid.

RESULTS

Sequence and structural analysis of the C-terminus of Mtr4

A characteristic feature of Ski2-like helicases is a four-domain structure composed of two RecA-like domains, a winged helix domain, and a HB domain that form a ring-like core structure around which additional “accessory” domains may be attached.³² In the case of Mtr4 and Ski2, the HB domain is the C-terminal domain of the protein. In contrast, other Ski2-like helicases (e.g. Suv3, Brr2, Slh1, Hel308, and Mer3) contain additional domains C-terminal to the HB domain. Consequently, the C-termini of Mtr4 and Ski2 are positioned differently from what is observed for other Ski2-like family members. Specifically, the C-termini of Mtr4 and Ski2 are positioned at the interface of the RecA1 and RecA2 domains, whereas the C-termini of other Ski2-like helicases do not traverse the RecA1-RecA2 interface, but rather extend in the opposite direction into additional domains (Fig. 1A).

Sequence analysis also reveals strong conservation in the C-termini of Mtr4 and Ski2 that is not observed in other Ski2-like helicases (Fig. 1B). Specifically, the last four residues include a strictly conserved SLYΦ sequence, where Φ represents a hydrophobic residue (most commonly leucine) (Fig. S1). The local environment around the C-terminal tail (SLYL) consists of conserved motifs responsible for RNA binding in RecA1 and RecA2 (motif Ia, Ib, and IVa, respectively), and communication between ATP hydrolysis and RNA unwinding in RecA2 (motif Va) (Fig. S2A).⁵⁴ However, the C-terminal tail does not appear to directly interact with these motifs, except for S1070, which interacts with a single residue in motif IVa (E485) in all available structures.^{14,16,18,35,39,41,42,45} The E485 interaction appears to stabilize the C-terminal tail but is removed from the RNA binding site by ~10 Å. While direct interactions with conserved motifs are limited, extensive interactions are observed between C-terminal residues and RecA residues that flank these motifs (Fig. 1C). The precise positioning of the C-terminal tail residues correlates with the conformation of the helicase core domains. A comparison of existing S. cerevisiae and H. sapiens Mtr4 structures reveals two distinct orientations of the RecA-like domains in the helicase core, which we describe as “open core” and “closed core” conformations (Fig. 2A and Table S1). Most of the Mtr4 structures adopt a closed core conformation where the C-terminal tail interacts with residues at the interface of RecA1 and RecA2 (Fig. 1B). However, two structures (4QU4 and 5OOQ) adopt a conformation where the local interface of the RecA-like domains opens by approximately 4 Å compared to the closed core structures, and the C-terminal tail is completely disengaged from RecA1. The C-terminal tail appears somewhat disordered in the open conformation as evidenced by a lack of electron density that led to incomplete modeling of the amino acid sidechains in this region. Notably, for all closed conformation structures, sufficient electron density is observed to model all C-terminal residues. Interestingly, there appears to be a correlation between substrate binding to the helicase core and the core conformation. All closed core structures are either bound to a substrate (nucleotide and/or RNA) or engaged in protein-protein interactions with core domains. In contrast, the core domains of both open core structures are in an apo (unbound) state. We further note that the open core structures were solved in different space groups, suggesting that the open conformation is not a crystallization artifact.

Figure 2. Comparison of open and closed Mtr4 helicase core structures.

(A) An electrostatic surface of the RecA-like domains for an open (PDB ID 4QU4) and closed (PDB ID 2XGJ) structure are shown. The conserved C-terminal tail is shown in sticks at the interface of the RecA-like domains. The open conformation shows minimal contact between the C-terminal motif (SLYL) and RecA-like domains. In the closed conformation, the RecA-like domains move ~4 Å closer together and multiple interactions are observed between the C-terminus and the RecA-like domains. L1071, Y1072 and L1073 each occupy separate binding pockets. (B) Structural alignment of the RecA1 domain in the open and closed structures shown in (A) reveals a concerted conformational rearrangement of RecA2 and HB domains. Vectors (pink and orange lines) were generated in PyMOL from c-alpha positions between open and closed structures and range from 3-14 angstroms (Å) in length.

The transition between open and closed conformations is accomplished by repositioning of RecA2 and HB domains with respect to RecA1 (Fig. 2B and movie S1). (No significant structural changes are observed within individual domains.) Comparing open and closed conformations, RecA2 is rotated on average 10 degrees and shifted 9 Å. This is accompanied by a repositioning of the HB domain (rotated on average 14 degrees and displaced 9 Å) with RecA2 (Fig. 2B and Fig. S3). In the closed conformation, the last 3 C-terminal residues (L1071, Y1072, and L1073) interact with both RecA-like domains where each side chain is stabilized in a separate binding pocket (Fig. 2A). The C-terminal tail follows the same path in all closed structures, although some variation in the local positioning of sidechains is observed depending on the structure (Fig. S2B). L1071 binds in a hydrophobic pocket in RecA2. Y1072 is involved in a hydrogen bonding network with E488, which is just upstream of motif IVa. Notably, E488 seems to be the center of the hydrogen bond network as it interacts with motif Ia (Q207) and motif IVa (H475 and K484), suggesting that the Y1072 interaction with E488 could influence these motifs. Furthermore, Y1072 is near R210 and forms a cation-π interaction in some structures, which is just upstream of motif Ia. The backbone of Y1072 and the carboxyl group of the terminal amino acid (L1073) also interact with R210. Interestingly, upon closure of the RecA-like domains, the L1071 backbone flips ~180° to position the sidechain into the hydrophobic pocket of RecA2. This repositioning significantly reorients Y1072 and L1073, allowing them to interact with their respective binding pockets. In summary, the C-terminal tail of Mtr4 is physically positioned at the interface of conserved motifs and poised to affect helicase activity. In addition, it has been reported that fusing GFP to the C-terminus of Mtr4 is synthetically lethal when combined with an Rrp6 deletion, and that fusing an epitope to the C-terminus of Ski2 inactivates its function, suggesting that the C-terminus of Mtr4 has important functions in vivo.^30,55,56

C-terminal residues (SLYL) affect Mtr4 helicase activity

Given the positioning of the conserved C-terminal residues at the RecA1-RecA2 interface, we reasoned that they may play a role in Mtr4 helicase activity. To test this possibility, we constructed a series of point and deletion mutants targeting the conserved residues in S. cerevisiae Mtr4 (Fig. 1C). The mutations were designed to disrupt interactions of L1071, Y1072, and L1073 binding pockets. We also probed the length requirement of the C-terminal tail by constructing a deletion (Mtr4^ΔL1073) and extension (Mtr4^+6Ala) mutant, which we predicted would disrupt the positioning of the C-terminus. The mutants were recombinantly expressed in E. coli and purified similar to Mtr4^WT, except for Mtr4^ΔSLYL. The Mtr4^ΔSLYL mutant eluted off gel filtration in a broad peak, suggesting that the C-terminal deletion is destabilizing (Fig. S4). All other mutants purified and eluted off gel filtration at the expected volume similar to Mtr4^WT.

The ability of the C-terminal mutants to bind a 16 bp RNA duplex with a UAAAAA 3’ single-stranded overhang (ds₁₆UA₅) was measured using an electrophoretic mobility shift assay (EMSA). A fluorescein label was placed on the 5’ end of the shorter duplex strand. Apparent dissociation constants (K_D) were determined as shown in Table S2 and Fig. S5. None of the mutants tested showed significant changes in binding to the RNA substrate (compared to Mtr4^WT), except for Mtr4^ΔSLYL, which showed a >5-fold increased K_D. These results suggest the C-terminal mutants do not fundamentally impair the ability of Mtr4 to bind RNA.

We then measured the effect of the C-terminal mutations on ATP hydrolysis. RNA-stimulated ATPase activity was measured using a malachite green phosphate assay with the same RNA substrate used in the binding experiments (without a fluorescein label). Interestingly, most C-terminal mutations do not exhibit significant changes in RNA-stimulated ATPase activity, except for Mtr4^L1071D (Fig. 3A (gray bars) and Table S3). We also examined C-terminal reciprocal mutants (Mtr4^R210A and Mtr4^E488A) to test the interactions of the SLYL motif with the RecA-like domains. Mtr4^R210A (RecA1) and Mtr4^E488A (RecA2) mutants disrupt interactions with Y1072 and L1073, respectively. Both mutants significantly reduced RNA-stimulated ATPase activity, suggesting that interaction of the SLYL motif with the RecA-like domains is an important modulator of ATPase activity. However, since E488 is also involved in an extensive hydrogen bonding network with residues in motif Ia and motif IVa, we cannot clearly distinguish between effects due to the loss of interactions with Y1072 and disruption of conserved helicase motifs.

Figure 3. Effect of C-terminal mutants on Mtr4 and TRAMP unwinding and ATPase activities.

ATPase (A) and unwinding (B) activity rates of Mtr4^WT and Mtr4 mutants alone are shown as gray bars, while TRAMP ^WT and TRAMP mutants are shown as green bars. The ATPase and unwinding activities correspond to Mtr4 C-terminal mutants (purple label) and complementary RecA1 (blue) and RecA2 (orange) mutants. Numerical values of the data are presented in Table S3 and S4. Data presented here represent averages from three independent experiments. Error bars represent one standard deviation (SD). Statistical analysis of Mtr4 mutants and TRAMP mutants was performed against Mtr4^WT and TRAMP^WT controls, respectively. Asterisks represent statistical significance where (*) is p <0.05 and (**) is p <0.01 (one-way ANOVA followed by a Dunnett’s post-hoc test). Statistical analysis of TRAMP enhancement was also performed comparing Mtr4 and TRAMP activities (data not shown for clarity). For unwinding activity, TRAMP enhancement was found to be statistically significant for all C-terminal mutants.

To characterize unwinding activities for C-terminal mutants, we measured unwinding rate constants on the same RNA substrate using a strand displacement assay. Most C-terminal mutations exhibit significantly reduced unwinding activity, and detectable activity is abolished with the Mtr4^R210A and Mtr4^E488A mutants (Fig. 3B (gray bars), Table S4 and Fig. S6). Mtr4^Y1072F unwinding activity is the same as Mtr4^WT. Interestingly, Mtr4^ΔSLYL shows almost no unwinding activity (Fig. S6), whereas the 6-alanine extension (Mtr4^+6Ala) shows a 2-fold increase in unwinding activity.

TRAMP (Trf4-Air2-Mtr4) complex formation enhances the unwinding activity of Mtr4^WT by approximately 4-fold.⁵⁰ We therefore asked if the C-terminal unwinding defects could be overcome by addition of Trf4 and Air2 (to form TRAMP). Trf4 is a poly(A) polymerase that competes with Mtr4 for the RNA substrate and also uses ATP as the nucleotide substrate. To avoid competition with polymerase activity, we used a Trf4 construct where active site catalytic residues have been mutated to alanine (Trf4^D236A/D238A).^50,57 Except for Mtr4^ΔSLYL, TRAMP formation enhances unwinding activity for all C-terminal mutants (Fig. 3B (compare gray vs. green bars) and Table S4). However, TRAMP unwinding activity is still significantly reduced with several Mtr4 C-terminal mutants (TRAMP^S1070A, TRAMP^L1071D, TRAMP^Y1072A, and TRAMP^ΔL1073) compared to the wild-type Mtr4 TRAMP complex (TRAMP^WT).

Genetic interaction suggests parallel functions for the C-termini of Mtr4 and Rrp6

After demonstrating that C-terminal residues influence Mtr4 unwinding activity in vitro, we then asked how mutations at the C-terminus affect Mtr4 function in vivo. MTR4 mutant alleles (mtr4^S1070A, mtr4^L1071D, mtr4^Y1072F, mtr4^Y1072A, mtr4^L1071D, mtr4^ΔL1073, mtr4^ΔSLYL, and mtr4^+6Ala) were placed in a low copy LEU2 yeast plasmid. Because Mtr4 is essential, we introduced the mutant (or wild-type control) plasmids into an mtr4Δ strain that contains MTR4 exogenously expressed on a URA3 plasmid. We then grew the strains on media containing 5-FOA to select for cells that have lost the URA3 plasmid and thus only express MTR4 from the LEU2 plasmid. Western blot analysis showed expression of the single amino acid mutants and the Mtr4^+6Ala mutant were similar to the unmutated Mtr4 (Fig. 4A). The Mtr4^ΔSLYL construct expressed at somewhat reduced levels compared to the unmutated Mtr4, consistent with the instability observed with the recombinant protein.

Figure 4. Growth complementation defects by Mtr4 C-terminal mutants in rrp6Δ and rrp6-ΔC.

(A) Western blot of C-terminal mutants. The second lane (Mtr4 control) represents Mtr4 expressed endogenously in a wild-type background. The remaining lanes (3-12) represent different alleles of MTR4 (WT or mutant) expressed exogenously from a plasmid in an mtr4 deletion background. (B) Most Mtr4 C-terminal mutants show normal growth phenotypes in an mtr4Δ strain. The indicated plasmids were introduced into a yeast strain that had the MTR4 gene deleted from the chromosome and contained a plasmid encoding Mtr4^WT with a URA3 selectable marker. Growth on 5-FOA plates selects for cells that have lost the URA3 plasmid, and thus shows the growth phenotype associated with each mutant. Only the Mtr4^ΔSLYL mutant shows a slightly slower growth phenotype. (C) Select Mtr4 C-terminal mutants (S1070A, L1071D, Y1072A, ΔSLYL, and +6Ala extension) show slower growth phenotype in an rrp6Δ strain on 5-FOA plates. (D) Growth assays of select C-terminal mutants (S1070A, Y1072A, and +6Ala extension) on Sc-His-Leu plates to select for the LEU2 plasmid encoding Mtr4 and the HIS3 plasmid encoding Rrp6. C-terminal mutants: S1070A, Y1072A, and +6Ala show slower growth phenotype in an rrp6-ΔC (deletion of amino acids 520-733) strain, suggesting that the C-termini of Mtr4 and Rrp6 have a shared function.

To test for functional defects in Mtr4, we serially diluted the mtr4Δ strain containing the MTR4 URA3 plasmid and a LEU2 plasmid expressing either wild-type MTR4 (Mtr4^WT), mutant mtr4, or empty vector, and plated the dilutions on 5-FOA media. The resulting plates were incubated for 4 days at either 30°C or 37°C to test the viability of Mtr4 mutants (Fig. 4B). As expected, Mtr4^WT was functional, whereas an empty vector was inviable at both temperatures. None of the single amino acid changes had an obvious effect on growth, but the Mtr4^ΔSLYL mutant grew slightly slower, presumably due to the instability of the protein.

Because previous studies have shown functional interactions between Mtr4 and Rrp6,^9,30,58 we then tested the effect of the C-terminal mutants in an rrp6Δ strain. To achieve this, we introduced the plasmids containing wild-type or mutant MTR4 into an mtr4Δ rrp6Δ double deletion strain that contained the MTR4 URA3 plasmid, serially diluted transformants and plated them on 5-FOA media (Fig. 4C). Interestingly, growth defects are observed with select Mtr4 C-terminal mutant alleles (mtr4^S1070A, mtr4^L1071D, mtr4^Y1072A, mtr4^ΔSLYL, and mtr4^+6Ala). Since the C-terminal mutants alone have no observable defect in vivo, this suggests that the C-terminus of Mtr4 shares a similar role with Rrp6.

We next sought to determine which feature of Rrp6 is responsible for the observed growth defects. To test this, we introduced constructs containing different alleles of RRP6 into three slow-growing rrp6Δ strains expressing mutant alleles of MTR4 (mtr4^S1070A, mtr4^Y1072A and mtr4^+6Ala) and grown for 3 days. Growth assays showed that wild-type RRP6 but not an empty vector restored growth. Mutant versions of Rrp6 that lacked either the Rrp47 interacting region or catalytic activity similarly restored growth (Rrp6^{ΔN (129-733)} and Rrp6^D238N respectively), but Rrp6 with a C-terminal truncation (Rrp6^{ΔC (1-520)}) did not (Fig. 4D). Amino acids 520 to 733 of Rrp6 are required for interaction with the RNA exosome. These results indicate that the Mtr4 SLYL motif is important for function when the Rrp6 C-terminus (residues 520-733) is disrupted.

C-terminal residues affect RNA degradation by Rrp44

To determine what effect the C-terminal tail of Mtr4 plays on exosome activity, we tested the ability of Mtr4 mutants to stimulate exosome degradation of a 17 bp RNA duplex with a 10 nt 3’ poly(A) overhang (ds₁₇A₁₀). We specifically examined Mtr4 mutants that showed in vivo growth defects (Mtr4^S1070A, Mtr4^L1071D, Mtr4^Y1072A, and Mtr4^+6Ala) in combination with Exo13^WT (Exo9+Rrp44+Rrp6+Rrp47+Mpp6). Two distinct RNA intermediates were observed in all decay assays that are attributable to the Rrp6 degradation product (higher band) and the Rrp44 degradation product (lower band) (Fig. S8). In the absence of Mtr4, Exo13^WT trims the RNA by ~10 nucleotides. This activity is due to Rrp6, as inactivation of Rrp6 (Exo13^Rrp6exo-) produces no degradation products (Fig. 5). Upon addition of Mtr4, Exo13^WT produces a smaller Rrp44 degradation product (~4-6 nts). However, when Mtr4^WT is replaced by Mtr4^S1070A, Mtr4^L1071D, or Mtr4^Y1072A, Rrp44 products are only weakly observed, indicating that the RNA is no longer accessible to Rrp44. This observation is consistent with the reduced unwinding activity of these mutants.

Figure 5. Effect of Mtr4 C-terminal mutants on RNA degradation by the exosome.

RNA decay assays were performed in triplicate using a fluorescently-labeled 17 bp RNA duplex with a 10 nt 3’ poly(A) overhang on the translocation strand (ds₁₇A₁₀). RNA degradation was initiated by adding Exo13^Rrp6exo- in combination with Mtr4 or C-terminal mutants (purple). (A) Representative RNA decay assay gels. The location of the Rrp44 degradation product is indicated. Rrp44 degradation of ds₁₇A₁₀ requires Mtr4. (B) Rrp44 product calculated by quantifying the substrate consumption of Exo13^Rrp6exo- and C-terminal mutants from gels, with error bars representing plus or minus one standard deviation. Mtr4^S1070A and Mtr4^Y1072A enhance Rrp44 degradation activity, albeit slower than Mtr4^WT. Mtr4^L1071D is unable to enhance Rrp44 activity.

Previous studies have shown that inactivation of Rrp6 does not impact Rrp44 degradation activity in vitro.^53,59 To focus on Rrp44 activity and remove competition with Rrp6 activity, we, therefore, performed the same degradation experiments described above using an inactive form of Rrp6 (Exo13^Rrp6exo-). When Rrp6 is inactivated, Rrp44 activity is enhanced by Mtr4^WT and to a reduced extent by Mtr4^S1070A and Mtr4^Y1072A. In contrast, Mtr4^L1071D is unable to activate Rrp44 activity. These differences in degradation correlate with differences in unwinding activity for each of these mutants. Unwinding activity is nearly abolished (approximately 8-fold reduction) with Mtr4^L1071D, whereas Mtr4^S1070A and Mtr4^Y1072A only result in partial (approximately two-fold) reduction of unwinding activity (Fig. 3).

No difference in degradation is observed when Mtr4^WT is replaced by Mtr4^+6Ala, using Exo13^WT or Exo13^Rrp6exo-, indicating that the 6-alanine extension has no effect on exosome degradation activity using this substrate.

DISCUSSION

Here, we show that the conserved C-terminal tail of Mtr4 is important for Mtr4 function. Although none of the residues in the SLYL motif directly interact with the ATP or RNA binding sites, every point mutation in that region resulted in reduced RNA unwinding activity, except for the conservative Y1072F mutation (Fig. 3). Notably, deletion of the SLYL motif (Mtr4^ΔSLYL) promotes protein aggregation (Fig. S4), indicating that the C-terminal tail has a stabilizing effect on the Mtr4 structure. Placement of the tail and the RecA1-RecA2 interface suggests that it may play an important role in coordinating RecA domain interactions. Our structural analysis further suggests that the C-terminal tail may also help anchor the HB domain to RecA2, providing a mechanism for concerted movement of the HB and RecA2 domains observed between the open and closed core conformations (Fig. 2B).

Previous studies have shown that TRAMP assembly enhances Mtr4 unwinding activity several-fold.^36,50 We observe a similar TRAMP-induced enhancement of unwinding activity for each of the SLYL mutants, except Mtr4^ΔSLYL (Fig. 3, Fig. S6, and Table S4). This suggests that C-terminal Mtr4 mutants can productively bind to Trf4-Air2 to form TRAMP. It is possible that the C-terminus directly impacts the function of other upstream RNA targeting complexes, such as ZCCHC8 from the human NEXT complex, which binds to Mtr4 near the C-terminal tail.^14,45 However, the function of the tail appears to be largely independent of upstream RNA targeting complexes because the conservation is also observed in systems that don’t form a TRAMP complex. For example, fission yeast and plants have duplicated Mtr4 genes, both of which retain the SLYL sequence (Fig. S1), with one gene encoding an Mtr4 that forms a TRAMP complex and the other Mtr4 forming a NEXT-like complex^60–63. Similar conservation is observed in Ski2, which forms an unrelated SKI complex.⁶⁴

The conservation of the SLYΦ motif in RNA exosome-associated helicases, Ski2 and Mtr4, but not in other Ski2-like helicases suggests that the C-terminus is important for regulating RNA exosome activity. The RNA degradation patterns observed with the Mtr4 C-terminal point mutants support this view. Mtr4^S1070A and Mtr4^Y1072A only partially activate Rrp44 degradation on the ds₁₇A₁₀ substrate, while Mtr4^L1071D was unable to activate Rrp44 (Fig. 5). Reduced Rrp44 activation by C-terminal mutants seem to be consistent with their decreased unwinding activity (Fig. 3).

Notably, none of these mutants exhibit growth defects in the presence of active Rrp6 (Fig. 4B). Rather, growth defects are only observed in an rrp6Δ background (Fig. 4C). Similarly, Mtr4^+6Ala exhibits a growth defect in an rrp6Δ background. However, in contrast to the Mtr4 C-terminal point mutants, Mtr4^+6Ala shows no defects in unwinding or on exosome activation under the conditions tested (Fig. 3B, Fig. 5). Interestingly, Schuch et al. showed that fusion of a GFP tag to the C-terminus of Mtr4 is completely inviable in an rrp6Δ background.³⁰ It appears that the addition of 6 alanines is largely sufficient to reproduce the GFP effect. The C-terminal extension may physically impede interactions with the exosome or other co-factors.

We further isolated the growth defects to the C-terminal domain (CTD) of Rrp6 (Fig. 4D), indicating that the C-termini of Mtr4 and Rrp6 have a shared function. Although Mtr4 interacts directly with the N-terminus of Rrp6 through the Mtr4 N-terminus and the arch domain, no direct interactions have been observed between Mtr4 and the Rrp6 CTD. We suspect that interactions between the C-termini of Mtr4 and Rrp6 are mediated through other factors, possibly including the exosome co-factor Mpp6. Cryo-EM and crystal structures show Mpp6 interactions with Mtr4 near the Mtr4 C-terminus.^18,19 Additionally, a double knockout of Rrp6 and Mpp6 is synthetically lethal.⁶⁵ While not visible in existing structures, we envisage physical overlap or close proximity of the Rrp6 CTD and the N-terminal portion of Mpp6. Our model suggests that RNA delivery to the exosome may be accomplished when the C-termini of Mtr4 or Rrp6 are disrupted, but simultaneous disruptions significantly impede RNA processing by the exosome, giving rise to the observed growth defects (Fig. 6).

Figure 6. Model for Coordination of C-termini of Mtr4 and Rrp6 in Rrp44 activation.

(Left) Mtr4 interactions with the exosome include direct interactions with the N-terminus of Rrp6 (and Rrp47) and interactions between Mpp6 and the Mtr4 C-terminus. Proper association leads to degradation of RNA substrates by Rrp44. (Center) Mtr4 association with the exosome could be maintained despite disruption of Rrp6 interactions with Exo9 (Rrp6^ΔC mutant) or disruption of Mtr4 interactions with Mpp6 (Mtr4 C-terminal mutations), explaining why neither mutation has an observable in vivo growth defect. (Right) Simultaneous deletion of the C-termini of Mtr4 and Rrp6 may disrupt formation of a productive Mtr4-exosome complex, consistent with the observed slow growth defects.

CONCLUSIONS

The C-terminus of Mtr4 contains a highly conserved SLYL motif that is positioned at the interface of the helicase RecA-like domains. Here, we show that mutations to this motif generally disrupt helicase activity and exosome decay by Rrp44. We further demonstrate that the C-termini of Mtr4 and Rrp6 share a common function (possibly mediated by Mpp6), suggesting an important role for the Mtr4 C-terminus in exosome interactions.