Nab3 Facilitates the Function of the TRAMP Complex in RNA Processing via Recruitment of Rrp6 Independent of Nrd1

Non-coding RNAs (ncRNAs) play critical roles in gene regulation. In eukaryotic cells, ncRNAs are processed and/or degraded by the nuclear exosome, a ribonuclease complex containing catalytic subunits Dis3 and Rrp6. The TRAMP (Trf4/5-Air1/2-Mtr4 polyadenylation) complex is a critical exosome cofactor in budding yeast that stimulates the exosome to process/degrade ncRNAs and human TRAMP components have recently been identified. Importantly, mutations in exosome and exosome cofactor genes cause neurodegenerative disease. How the TRAMP complex interacts with other exosome cofactors to orchestrate regulation of the exosome is an open question. To identify novel interactions of the TRAMP exosome cofactor, we performed a high copy suppressor screen of a thermosensitive air1/2 TRAMP mutant. Here, we report that the Nab3 RNA-binding protein of the Nrd1-Nab3-Sen1 (NNS) complex is a potent suppressor of TRAMP mutants. Unlike Nab3, Nrd1 and Sen1 do not suppress TRAMP mutants and Nrd1 binding is not required for Nab3-mediated suppression of TRAMP suggesting an independent role for Nab3. Critically, Nab3 decreases ncRNA levels in TRAMP mutants, Nab3-mediated suppression of air1/2 cells requires the nuclear exosome component, Rrp6, and Nab3 directly binds Rrp6. We extend this analysis to identify a human RNA binding protein, RALY, which shares identity with Nab3 and can suppress TRAMP mutants. These results suggest that Nab3 facilitates TRAMP function by recruiting Rrp6 to ncRNAs for processing/degradation independent of Nrd1. The data raise the intriguing possibility that Nab3 and Nrd1 can function independently to recruit Rrp6 to ncRNA targets, providing combinatorial flexibility in RNA processing.


Introduction
Non-coding RNAs (ncRNAs) are fast emerging as key regulators of gene expression in eukaryotic cells [1,2]. Dysregulation of ncRNAs is linked to cancer and neurodegeneration [3,4]. Understanding how ncRNAs are produced, processed, and destroyed is therefore critical to elucidating gene regulation and has become an area of major research focus in recent years. Transcriptional termination and 3'-end processing of ncRNAs are functionally linked but how they are coupled is poorly characterized [5]. In Saccharomyces cerevisiae, unlike most mRNAs, ncRNAs are not terminated and 3'-end processed by the conventional cleavage and polyadenylation machinery, but by a trio of evolutionarily conserved complexes: the RNA exosome, the TRAMP complex, and the NNS complex [5]. Critically, the interactions and mechanisms employed by these complexes to terminate and process/degrade ncRNAs are not well understood.
The exosome is recruited to RNAs by exosome cofactors that recognize the RNA and regulate exosome function [9,10]. Certain exosome cofactors also facilitate transcriptional termination of ncRNAs [5,10]. Exosome cofactors are usually single RNA binding proteins or protein complexes containing RNA binding proteins that recognize RNAs in a sequence or structurespecific manner and interact with core exosome and/or Rrp6 [5,10]. Three examples of exosome cofactors in Saccharomyces cerevisiae are the Rrp47 protein, the TRAMP complex, and the Nrd1-Nab3-Sen1 (NNS) complex [9,10].

Results
NAB3 suppresses the thermosensitive growth of air1/2 cells To study the function of TRAMP, we exploited thermosensitive air1-C178R air2ΔTRAMP mutant cells, hereafter referred to as air1/2 cells, containing an integrated C178R substitution in AIR1 zinc knuckle 5 combined with deletion of AIR2 [16]. The air1/2 cells are thermosensitive, show impaired growth at 30°C, and can be suppressed by TRAMP components [16] (Fig. 1A).
To identify additional factors that facilitate TRAMP function, we performed a high-copy suppressor screen with air1/2 cells. From 60 suppressor (SUP) plasmids that improved air1/2 growth at 30°C, 44 plasmids contained either AIR1 (e.g. SUP3-2) or AIR2 (Fig. 1A). Four other plasmids, including SUP11-3, all contained a large insert including the NAB3 gene and potently suppress air1/2 growth at 30°C (Fig. 1A). A plasmid clone of NAB3 suppresses air1/2 growth at 30°C to a similar degree as suppressor SUP11-3 (Fig. 1B), confirming that suppression is conferred by NAB3. NAB3 suppression of air1/2 growth could mean that Nab3 specifically facilitates Air1 function or more generally facilitates TRAMP function. To this point, NAB3 can suppress the slow growth of another TRAMP mutant, trf4Δ, at 25°C (S1A Fig), suggesting that Nab3 facilitates TRAMP function.
As TRAMP is an exosome cofactor and Air1/2 facilitate RNA recognition, the Nab3 RNA binding protein could simply replace RNA binding function of the Air proteins in air1/2 cells and target TRAMP/exosome to RNA substrates. This suppression mechanism predicts that NAB3 should bypass the requirement for AIR1/2. However, NAB3 cannot improve air1Δair2Δgrowth (Fig. 1C) and thus is not a bypass suppressor of AIR1/2 function.
We then tested whether the nab3-ΔNBD and nab3-11 RRM mutants can suppress air1/2 growth. The nab3-ΔNBD mutant suppresses air1/2 growth similarly but not identically to NAB3 at 30°C (Fig. 2C). The nab3-ΔNBD mutant protein is expressed at a similar level to wildtype Nab3 protein (Fig. 2D). This result suggests that the Nab3 Nrd1-binding domain is not essential for air1/2 suppression. In contrast, the nab3-11 mutant cannot suppress air1/2 growth air2Δ growth at 30°C. The air1-C178R air2Δ cells containing vector, TRAMP component or suppressor 2μ URA3 plasmid were spotted and grown at indicated temperatures. (B) NAB3 suppresses the air1-C178R air2Δ thermosensitive growth at 30°C to the same degree as NAB3 suppressor plasmid, SUP11-3. The air1-C178R air2Δ cells containing vector, AIR1, TRF4, SUP11-3 or NAB3 μ URA3 plasmids were spotted and grown at indicated temperatures. See also S1 Fig. (C) NAB3 is not a bypass suppressor of AIR1/2. NAB3 cannot improve air1Δ air2Δ growth. Two air1Δ air2Δ strains (ACY1095, ACY2036) containing AIR2 URA3 maintenance plasmid and vector, AIR1-GFP, TRF4, TRF5 or NAB3 2μ HIS3 plasmid were spotted on control/ 5-FOA and grown at indicated temperatures. (D) NRD1 and SEN1 do not suppress the air1-C178R air2Δ thermosensitive growth at 30°C. The air1-C178R air2Δ cells containing vector, TRF4, NAB3, NRD1 or SEN1 2μ URA3 plasmid were spotted and grown at indicated temperatures. See also S1 Fig. (  The nab3-ΔNBD mutant protein shows greatly reduced binding to Nrd1. TAP-tagged Nrd1 was precipitated from lysates of NRD1-TAP cells expressing Myc-tagged Nab3, nab3-11 or nab3-ΔNBD and bound (B), unbound (U), and input fractions were analyzed by immunoblotting to detect Nab3-Myc proteins and 3-phosphoglycerate kinase (Pgk1) as a loading control. Samples were also probed with anti-Nrd1 antibody to detect Nrd1-TAP proteins. The percentage of bound Nab3 protein relative to input protein and bound wild-type Nab3 (% Bound) is shown below the bound lanes. The percentage of input Nab3 protein relative to input wild-type Nab3 protein (% Input) is shown below the input lanes. The percentages of protein were calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. at 30°C (Fig. 2C). Notably, nab3-ΔNBD, but not nab3-11, also suppresses trf4Δ growth (S1A Fig).
As the Nab3 Nrd1-binding domain is not essential for air1/2 suppression, we assessed whether this domain is essential for Nab3 function. We tested whether the nab3-ΔNBD mutant can function as the sole copy of the essential NAB3 gene. The nab3-ΔNBD mutant cells are viable, but show a moderate growth defect, while nab3-11 mutant cells exhibit a severe growth defect ( Fig. 3E), indicating that the Nab3 Nrd1-binding domain and thus Nab3 interaction with Nrd1 is not essential for the cellular function of Nab3.
Critically, all the nab3 RRM mutants, except nab3-11, are expressed at similar levels to Nab3 (Fig. 3C). Expression of nab3 RRM mutants in wild-type cells does not impair growth, showing that the inability of most nab3 RRM mutants to suppress air1/2 growth is not due to toxicity (S3B Fig).
To determine if the nab3 RRM mutants are functional in vivo, we tested whether the nab3 RRM mutants can function as the sole copy of the essential NAB3 gene. The nab3-R331A, nab3-F333A, and nab3-S399A mutants are not viable, whereas nab3-S400A mutant has growth comparable to NAB3 (Fig. 3E). Thus, Nab3 RNA binding is essential for Nab3 function in vivo.
NAB3 reduces IMD2 CUT terminator readthrough product from a reporter in air1/2 cells Nab3 and Nrd1 recognize elements in the terminators of ncRNAs/CUTs and mRNAs to facilitate termination and processing/degradation [5,20,21]. Notably, nrd1/nab3 mutants exhibit readthrough of NNS terminators in ncRNAs and Nrd1/Nab3 binding site mutations within these NNS terminators also cause readthrough [5,8,20,21]. Moreover, trf4 Δ cells show readthrough of RNA terminators [42,43]. To begin to assess termination of CUTs in air1/2 cells and test the impact of NAB3 on CUT termination, we employed a pREF-GFP reporter for CUT terminator readthrough [38]. The pREF-GFP reporter contains a galactose-inducible promoter and the IMD2 CUT intergenic terminator (IT) upstream of GFP [38] (Fig. 4A). In wild-type cells, upon galactose induction, little GFP is observed as transcription is efficiently terminated at the terminator before GFP. In termination defective cells, GFP is produced as transcription is not terminated at the terminator and readthrough to GFP occurs. We galactose-induced the pREF-GFP reporter in wild-type, air1/2, and other TRAMP mutants and examined the GFP level in the cells by immunoblotting. GFP expression is increased in air1/2 and TRAMP mutants containing pREF-GFP (Fig. 4B).
Having established that air1/2 cells express GFP from pREF-GFP, suggesting readthrough of the IMD2 CUT terminator and/or impaired degradation of the readthrough RNA, we wished to determine if NAB3 impacts the GFP level in air1/2 cells containing pREF-GFP. NAB3 greatly reduces the GFP level (24% GFP), NRD1 increases the GFP level (152% GFP), and SEN1 does not affect the GFP level (91% GFP) in air1/2 cells (Fig. 4C). These data suggest that NAB3 reduces the readthrough of the IMD2 CUT terminator and/or increases the Ser399 (orange) also contacts the U 3 nucleotide. The Nab3 RRM residues, Phe371 (blue) and Pro374 (brown), mutated in the nab3-11 (nab3-F371L-P374L) RRM mutant, are also highlighted. The Nab3 RRM-RNA NMR structure was reproduced from the PDB file 2L41 [41] using MacPyMOL software [62] and altered and annotated using Adobe Photoshop and Illustrator CS4 (Adobe). (B) nab3 RRM mutants, nab3-R331A, nab3-F333A, nab3-S399A do not suppress the air1-C178R air2Δ thermosensitive growth at 30°C. nab3 RRM mutant, nab3-S400A, suppresses air1-C178R air2Δ thermosensitive growth at 30°C. The air1-C178R air2Δ cells containing vector, NAB3 or nab3 RRM mutant 2μURA3 plasmid were spotted and grown at indicated temperatures. See also S1 and S2 Figs. (C) All nab3 RRM mutant proteins, except nab3-11, are expressed in air1-C178R air2Δ cells to similar levels as Nab3. Lysates of air1-C178R air2Δ cells expressing Myc-tagged Nab3 or nab3 RRM mutants at 30°C were analyzed by immunoblotting to detect Myc-tagged proteins and 3phosphoglycerate kinase (Pgk1) as a loading control. (D) The nab3-R331A and nab3-S399A RRM mutant proteins show binding to Nrd1 similar to wild-type Nab3. TAP-tagged Nrd1 was precipitated from lysates of NRD1-TAP cells expressing Myc-tagged Nab3, nab3-R331A or nab3-S399A and bound (B), unbound (U), and input fractions were analyzed by immunoblotting to detect Nab3-Myc proteins, Nrd1-TAP proteins and 3-phosphoglycerate kinase (Pgk1) as a loading control. The percentage of bound Nab3 relative to input protein and bound wild-type Nab3 (% Bound) is shown below the bound lanes. The percentage of input Nab3 protein relative to input wild-type Nab3 protein (% Input) is shown below the input lanes. The percentages of protein were calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. The original immunoblot is shown in S3A Fig. (E) nab3Δcells expressing nab3 RRM mutants, nab3-R331A, nab3-F333A or nab3-S399A, are not viable, but nab3Δ cells expressing nab3-S400A RRM mutant or nab3-Δ NBD mutant are viable. nab3Δ cells expressing nab3-11 RRM mutant show a severe growth defect. nab3Δ cells containing NAB3 URA3 maintenance plasmid and vector, NAB3, nab3 RRM mutant or nab3-Δ NBD mutant 2μ HIS3 plasmid were spotted on control/5-FOA and grown at indicated temperatures.  NAB3 reduces the level of IMD2 CUT terminator readthrough product from a reporter and native IMD2 CUT and readthrough RNA, but NAB3 does not significantly affect the termination of native IMD2 CUT in air1/2 cells. (A) Schematic of pREF-GFP IMD2 CUT terminator reporter plasmid [38]. The pREF-GFP reporter contains the IMD2 CUT intergenic terminator (IT) upstream of GFP open reading frame (GFP) under the control of a galactoseinducible promoter (pGAL). (B) Level of GFP readthrough product from pREF-GFP reporter is increased in air1-C178R air2Δ cells and other TRAMP mutants. Wild-type, air1Δ, air2 Δ, trf4Δ, and air1-C178R air2Δ cells containing pREF-GFP reporter were grown in the presence of galactose (Gal) to induce to GFP expression for 3hr at 30°C and lysates were analyzed by immunoblotting to detect GFP and 3-phosphoglycerate kinase (Pgk1) as a loading control. Samples of cells just after addition of galactose (0 hour time point) were also analyzed. Different immunoblots are outlined by black boxes. (C) NAB3, but not NRD1 or SEN1, decreases level of GFP readthrough product from pREF-GFP reporter in air1-C178R air2 Δ cells. air1-C178R air2 Δ cells containing pREF-GFP reporter plasmid and vector, NAB3, NRD1 or SEN1 were grown in the presence of galactose (Gal) to induce to GFP expression for 3hr at 30°C and lysates of cells were analyzed by immunoblotting as described previously. Nonadjacent lanes in the same immunoblot are separated by white space. (D) nab3-Δ NBD mutant, like NAB3, decreases level of GFP readthrough product from pREF-GFP reporter in air1-C178R air2Δ cells. nab3-R331A, nab3-F333A, and nab3-degradation of the IMD2 readthrough RNA in air1/2 cells, suggesting that NAB3 can improve the termination/degradation of CUTs in air1/2 cells.
To determine whether the NAB3 effect on IMD2 terminator readthrough and/or IMD2 readthrough RNA degradation in air1/2 cells requires Nab3 interaction with RNA or Nrd1, we tested if the nab3 RRM mutants and nab3-204-248 NBD mutant affect the GFP level. Most nab3 RRM mutants do not reduce the GFP level (126-288% GFP), but the nab3-ΔNBD mutant greatly decreases the GFP level (8% GFP) in air1/2 cells (Fig. 4D). These data suggest that the NAB3-mediated decrease in IMD2 CUT terminator readthrough and/or increase in IMD2 readthrough RNA degradation in air1/2 cells requires the Nab3 RRM, but does not require Nrd1 interaction.
NAB3 decreases the levels of native IMD2 CUT and readthrough RNA in air1/2 and trf4Δ cells NAB3 reduction of the GFP level from the pREF-GFP reporter in air1/2 cells suggests that Nab3 decreases IMD2 CUT readthrough, but the decrease in GFP level could also result from downstream effects of Nab3 on RNA processing/degradation or translation. To test if NAB3 affects the native IMD2 CUT in air1/2 or trf4Δ cells, we probed a Northern blot of total RNA from air1/2, trf4Δ or wild-type cells containing vector, AIR1, TRF4, NAB3 or NRD1, with an IMD2 CUT-specific probe. The level of the short IMD2 CUT is greatly increased in air1/2 and trf4Δ cells with vector alone, compared to wild-type cells with vector alone, as expected for TRAMP mutants that impair exosome degradation of CUTs (Fig. 4E). Importantly, the level of a long IMD2 CUT readthrough RNA is also increased in air1/2 and trf4Δ cells with vector alone, relative to wild-type cells, consistent with the increased GFP from pREF-GFP in these cells and supporting the notion that TRAMP mutants also have termination defects (Fig. 4E). AIR1 greatly reduces to 28-29% and TRF4 partially reduces to 60-68% the levels of the IMD2 CUT and readthrough RNA in air1/2 cells, relative to cells containing vector alone, consistent with AIR1 and TRF4 reactivating TRAMP degradation/termination (Fig. 4E). NAB3 reduces to 85% the level of the IMD2 CUT, whereas NRD1 reduces to 95% the level of the IMD2 CUT, in air1/2 cells (Fig. 4E). Moreover, NAB3 decreases to 70% the IMD2 CUT readthrough RNA, whereas NRD1 decreases to 89% the readthrough RNA, in air1/2 cells (Fig. 4E). In addition, NAB3 reduces to 72-84% the level of IMD2 CUT and readthrough RNA, but NRD1 does not S399A RRM mutants do not decrease level of GFP from pREF-GFP reporter in air1/2 cells. air1-C178R air2Δ cells containing pREF-GFP reporter plasmid and vector, NAB3, nab3-ΔNBD mutant or nab3 RRM mutants were grown in the presence of galactose (Gal) to induce to GFP expression for 3hr at 30°C and lysates were analyzed by immunoblotting as described previously. The percentage of GFP relative to Pgk1 loading control and GFP in cells containing vector alone (% GFP Rel Vector) is shown below the lanes and was calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. (E) NAB3 reduces the levels of native IMD2 CUT and readthrough RNA in air1-C178R air2 Δ and trf4Δ mutant cells. Northern blot of total RNA from wild-type, air1-C178R air2Δ, and trf4Δ cells containing vector, AIR1, TRF4, NAB3 or NRD1 grown at 30°C were probed with an IMD2 CUT-specific probe. Northern blot was probed with ACT1 probe as a loading control. The IMD2 CUT (IMD2 Short/CUT) and IMD2 CUT readthrough RNA (IMD2 Long/RT) are labeled. A longer exposure of IMD2 CUT readthrough RNA is shown above. The percentage of IMD2 CUT and readthrough RNA relative to ACT1 loading control and cells containing vector alone (% IMD2 Short Rel Vector; % IMD2 Long Rel Vector) is shown below lanes and was calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. See also S4A Fig. (F) NAB3 does not significantly affect Pol II occupancy downstream of the IMD2 CUT at Primer Pair 2-5 positions in air1-C178R air2Δ cells relative to air1/2 cells containing vector alone (p-value 0.4), suggesting that Nab3 overexpression does not significantly affect IMD2 CUT termination in air1/2 cells. As a control, AIR1 significantly decreases Pol II occupancy downstream of the IMD2 CUT in air1-C178R air2Δ cells at Primer Pair 4 and 5 positions compared to air1/2 cells containing vector alone (p-value 0.05), indicating that Air1 significantly affects termination and suggesting that air1/2 cells have a termination defect. Anti-Pol II ChIP was performed on air1-C178R air2Δ cells containing vector, AIR1 or NAB3 and relative Pol II occupancy was measured across IMD2 gene by qPCR with IMD2 Primer Pair 1-5 as described in Material and Methods. Mean RNA Pol II occupancy values from three independent experiments normalized to Primer Pair 1 within IMD2 CUT are shown with error bars that represent standard error of the mean. Statistical significance of differences in mean Pol II occupancy values was determined using unpaired t test and significant differences in mean values (pvalue 0.05) are denoted with asterisks. Schematic of IMD2 CUT and downstream IMD2 gene is shown with positions of IMD2 qPCR Primer Pairs 1-5 above and base pair distances between primer pairs below. See also S4B Fig decrease the level of the IMD2 CUT and readthrough RNA in trf4Δ cells (Fig. 4E). These data indicate that NAB3 can reduce both the IMD2 CUT and readthrough RNA in air1/2 cells and does so to a greater extent than NRD1, suggesting that Nab3 affects the degradation/termination of the IMD2 CUT and readthrough RNA to a larger degree than Nrd1. NAB3 suppression of TRAMP mutant growth therefore correlates with NAB3 reduction of IMD2 CUT and readthrough RNA in TRAMP mutant cells.
To determine if the NAB3 decrease of the IMD2 CUT and readthrough RNA in air1/2 cells is dependent on Nab3 interaction with RNA or Nrd1, we probed a Northern blot of total RNA from air1/2 cells containing vector, NAB3, nab3 RRM mutants or nab3-Δ NBD mutant with an IMD2 CUT-specific probe. nab3-Δ NBD decreases the level of IMD2 CUT and readthrough RNA similar to NAB3 in air1/2 cells (S4A Fig). In contrast, the nab3 RRM mutants, nab3-11, nab3-R331A, and nab3-S399A, decrease the level of the IMD2 CUT and readthrough RNA to a lesser degree than NAB3 (S4A Fig). These data support the idea that Nab3 requires RNA interaction, but does not require Nrd1 binding, to terminate/degrade and decrease the IMD2 CUT readthrough RNA.
NAB3 does not significantly affect termination of native IMD2 CUT in air1/2 cells The Nab3-mediated decrease in the level of the IMD2 CUT readthrough RNA detected in air1/ 2 cells could be due to increased termination and/or increased degradation. To assess whether Nab3 affects the termination of the native IMD2 CUT in air1/2 cells, we examined RNA Pol II occupancy on the IMD2 gene in air1/2 cells containing vector, AIR1 or NAB3 by Pol II ChIP. We employed five primer pairs to examine Pol II occupancy and normalized all data back to Primer Pair 1 located within the IMD2 CUT (Fig. 4F). Given that the chromatin was sheared to a size ranging from 300-500 base pairs and the distances between the primer pairs ( Fig. 4F), we expected to detect differences in Pol II occupancy most readily with Primer Pair 4 and 5. The air1/2 cells expressing AIR1 show a statistically significant decrease in Pol II occupancy downstream of the IMD2 CUT at Primer Pair 4 and 5 positions relative to the air1/2 cells containing vector alone (p-value 0.05; Fig. 4F). In contrast, the air1/2 cells expressing NAB3 show no statistically significant change in Pol II occupancy downstream of the IMD2 CUT at Primer Pair 2-5 positions relative to air1/2 cells containing vector alone (p-value 0.4; Fig. 4F). These data suggest that overexpression of Nab3 does not significantly affect termination of the IMD2 CUT in air1/2 cells. To extend this analysis, we also performed Pol II ChIP on the native snR13 snoRNA gene, which contains a well-characterized NNS-dependent terminator [20,21,44], in air1/2 cells containing vector or NAB3. As described for the IMD2 CUT, air1/2 cells expressing NAB3 do not exhibit a statistically significant change in Pol II occupancy downstream of snR13 compared to air1/2 cells containing vector alone (p-value 0.3; S4B Fig), suggesting overexpression of Nab3 does not significantly affect termination of the snR13 gene in air1/2 cells. These results indicate that Nab3 overexpression does not have a significant effect on the termination of the IMD2 CUT, suggesting that Nab3 suppression of air1/2 cells and reduction of IMD2 CUT readthrough RNA predominantly involves Nab3 rescue of degradation.
These results indicate that NAB3 suppression of air1/2 growth requires RRP6.
To ascertain if the catalytic activity of Rrp6 is required for NAB3 suppression of the air1/2 growth, we tested whether NAB3 can suppress air1/2 rrp6Δcells expressing the catalytically inactive mutant of Rrp6, rrp6-D238A [47]. NAB3 does not suppress air1/2 rrp6Δ cells containing rrp6-D238A indicating that NAB3 suppression is dependent on the catalytic activity of Rrp6 (S5A Fig). As NAB3 reduces the levels of IMD2 CUT and CUT readthrough RNA in air1/2 cells, we determined if NAB3 reduction of these RNA levels is dependent on RRP6. We probed a Northern blot of total RNA from air1/2 rrp6Δ cells containing vector, NAB3, AIR1, TRF4 or NRD1 with an IMD2 CUT-specific probe. AIR1 decreases to 48-78% the levels of the IMD2 CUT and readthrough RNA in air1/2 rrp6Δ cells (Fig. 5B). In contrast, NAB3 does not decrease (107%) the level of the IMD2 CUT and only weakly decreases to 94% the IMD2 readthrough RNA in air1/2 rrp6Δ cells (Fig. 5B). These data indicate that NAB3 reduction of the IMD2 CUT and readthrough RNA depends on RRP6, suggesting Nab3 enhanced degradation of the IMD2 CUT requires Rrp6.

Nab3 interacts with Rrp6 independent of Nrd1
The requirement of RRP6 for NAB3 suppression of air1/2 growth and reduction of the IMD2 CUT and readthrough RNA suggests that Nab3 may physically interact with Rrp6. Nrd1 coprecipitates with Rrp6 [19], raising the possibility, given the Nrd1-Nab3 interaction, that Nab3 could interact with Rrp6. To test if Nab3 directly interacts with Rrp6, we examined the binding of recombinant GST-tagged full-length Nab3 to His-tagged full-length Rrp6 in an in vitro protein binding assay. As controls, we also tested the binding of GST and GST-tagged Rrp47, an exosome cofactor known to directly interact with Rrp6 [48], to His-Rrp6. GST-Nab3 binds to His-Rrp6 but not as strongly as GST-Rrp47 binds to His-Rrp6 (Fig. 5C). This result shows that Nab3 directly interacts with Rrp6 and indicates that Nab3 can bind Rrp6 independent of Nrd1.
To determine if Nab3 can bind to Rrp6 independent of interactions with Nrd1 or RNA in yeast, we precipitated TAP-tagged Rrp6 from lysates of RRP6-TAP yeast cells expressing Myctagged Nab3 or nab3-ΔNBD mutant in the absence or presence of RNase A, and analyzed the bound fractions by immunoblotting. In the absence of RNase, binding of nab3-Δ NBD to Rrp6 is reduced but not abolished compared to Nab3 (Fig. 5D). In the presence of RNase, binding of nab3-Δ NBD to Rrp6 is similar to Nab3 (Fig. 5D). Importantly, the Nab3-Rrp6 interaction is reduced but not abolished by RNase-treatment and the nab3-NBD-Rrp6 interaction is not decreased by RNase-treatment (Fig. 5D). These results indicate that a proportion of Nab3 can interact with Rrp6 in the absence of the Nrd1-binding domain and RNA, suggesting that Nab3 can bind to Rrp6 independent of Nrd1. In further support, we performed a reverse Nab3-Rrp6 coprecipitation and found that Rrp6 binds to both Nab3 and a nab3-Δ1-248 NBD mutant (S5B Fig).
As Nab3 interacts with the catalytic exosome subunit, Rrp6, we determined if Nab3 can also interact with the other main catalytic subunit of the core exosome, Dis3/Rrp44. Notably, Nrd1 coprecipitates with Dis3 [19]. To assess interaction between Nab3 and Dis3, we precipitated TAP-tagged Dis3 from cells expressing Myc-tagged Nab3 and examined the bound fraction by immunoblotting. Nab3 co-purifies with Dis3 (Fig. 5E). To assess whether Nab3 can interact with Dis3 independent of Nrd1, we precipitated TAP-tagged Dis3 from cells expressing the nab3-ΔNBD mutant and assayed for co-purification of Nab3. Strikingly, unlike Nab3, the nab3-ΔNBD mutant does not bind to Dis3 (Fig. 5E). These results indicate that the Nab3 interaction with Dis3 is dependent upon the Nab3 Nrd1-binding domain and therefore likely interaction with Nrd1.
Human RALY protein suppresses the thermosensitive growth of air1/2 cells To address the question of functional conservation of Nab3, we performed a BLAST search with the Nab3 RRM and identified the human RALY (hRALY) protein that contains an RRM and C-terminal domain with homology to Nab3 (Figs. 6A and S6). The hRALY RRM shares 31% identity (23/74 residues) and a similar predicted β1α1β2β3α2β4 secondary structure with the Nab3 RRM (Fig. 6B). Moreover, the Nab3 RRM RNA-interacting residues Arg331 and Phe333 are conserved in the hRALY RRM as Arg22 and Phe24 (Fig. 6B). Importantly, hRALY can suppress the growth defect of air1/2 TRAMP mutant cells (Fig. 6C). In contrast, hRALY RRM mutants, hRALY-R22A and hRALY F24A, cannot suppress the growth of air1/2 cells (Fig. 6D). However, the hRALY RRM mutants are expressed to lower levels than the wild-type hRALY protein (Fig. 6E), leaving open the caveat that the inability of hRALY RRM mutants to suppress air1/2 cells could be due to low expression levels and not to impairment of the RRM/ RNA binding function of hRALY. At this time, it is therefore not possible to definitively conclude whether RNA binding by hRALY is required for suppression. Although Nab3 and hRALY share similar RRMs, the Nab3 RRM alone is not sufficient to suppress air1/2 cells. A Nab3 truncation mutant nab3-1-448, which lacks the C-terminal 354 amino acids, but retains the intact RRM, also does not suppress air1/2 cells (S7A Fig), even though the truncated protein is expressed (S7B Fig) and properly localized to the nucleus (S7C Fig). The C-terminal region of hRALY with homology to the C-terminal domain of Nab3 could therefore contribute to hRALY suppression of air1/2 cells.
weakly decreases the IMD2 CUT readthrough RNA in air1-C178R air2Δ rrp6Δ cells. Northern blot of total RNA from air1-C178R air2Δ rrp6Δ cells expressing vector, AIR1, TRF4, NAB3 or NRD1 grown at 30°C were probed with an IMD2 CUT-specific probe. Northern blot was probed with scR1 probe as a loading control. The IMD2 CUT (IMD2 Short/CUT) and IMD2 CUT readthrough RNA (IMD2 Long/RT) are labeled. A longer exposure of IMD2 CUT readthrough RNA is shown above. The percentage of IMD2 CUT and readthrough RNA relative to sCR1 loading control and cells containing vector alone (% IMD2 Short Rel Vector; % IMD2 Long Rel Vector) is shown below lanes and was calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. (C) Nab3 directly interacts with Rrp6 in vitro. Recombinant GST-Nab3, GST-Rrp47 or GST was incubated with recombinant His-tagged Rrp6 and glutathione Sepharose beads and bound and input fractions were analyzed by SDS-PAGE and immunoblotting to detect His-Rrp6 and GST fusion proteins. Nonadjacent lanes in the same immunoblot are separated by white space. (D) nab3-ΔNBD mutant protein binds to Rrp6 in an RNA-independent manner. TAP-tagged Rrp6 was precipitated from lysates of RRP6-TAP or untagged control cells expressing Myc-tagged Nab3 or nab3-ΔNBD in the absence or presence of RNase A. Bound and input fractions were analyzed by immunoblotting to detect Nab3-Myc proteins and membrane was stained with Ponceau S (Pon S) stain to detect total protein as a loading control. Nonadjacent lanes in the same immunoblot are separated by white space. Different immunoblots are outlined by black boxes. See also S5B Fig. (E) nab3-ΔNBD mutant protein, unlike wildtype Nab3, does not bind to Dis3. TAP-tagged Dis3 was precipitated from lysates of DIS3-TAP cells expressing Myc-tagged Nab3 or nab3-ΔNBD and bound (B), unbound (U), and input fractions were analyzed by immunoblotting to detect Nab3-Myc proteins and 3-phosphoglycerate kinase (Pgk1) as a loading control. Different immunoblots are outlined by black boxes. (F) NAB3 and nab3-ΔNBD mutant suppress the thermosensitive growth of air1-C178R air2Δdis3Δcells expressing a catalytically inactive dis3 mutant, dis3-D551N. air1-C178R air2Δdis3Δcells containing dis3-D551N and vector, AIR1, TRF4, NAB3, nab3-11, nab3-ΔNBD, NRD1 or SEN1 2 μ URA3 plasmid were spotted and grown at indicated temperatures.   Fig. (B) Alignment of hRALY and Nab3 RRMs showing that the hRALY RRM has 31% identity (identical residues shaded in gray) and a similar predicted β1α1β2β3 α2β4 secondary structure (99.8% confidence; α-helices in green; β-sheets in blue) to the Nab3 RRM. Nab3 RRM RNA-binding residues R331 and F333 that are conserved in the hRALY RRM as R22 and F24 are boxed. RRM consensus motifs RNP1 and RNP2 are underlined. Alignment and secondary structure prediction of hRALY RRM based on Nab3 RRM crystal structure (PDB ID: 2XNQ [63]) was generated by Phyre2 server [64] (C) hRALY suppresses the air1-C178R air2Δ thermosensitive growth at 30°C. The air1-C178R air2Δ cells containing vector, NAB3, nab3-11 or hRALY 2μ URA3 plasmid were spotted and grown at indicated temperatures. (D) hRALY RRM mutants hRALY-R22A and hRALY-F24A do not suppress the air1-C178R air2Δ thermosensitive growth at 30°C. The air1-C178R air2Δ cells containing vector, AIR1, NAB3, nab3-11, hRALY, hRALY-R22A or hRALY-F24A 2μ URA3 plasmid were spotted and grown at indicated temperatures. (E) hRALY-R22A and hRALY-F24A RRM mutant proteins are expressed in air1-C178R air2 Δ cells but not to the same level as hRALY. Lysates of air1-C178R air2Δ cells expressing

Discussion
Non-coding RNAs (ncRNAs) play key roles in gene regulation and disease [1][2][3][4]. Understanding how the TRAMP and NNS complex exosome cofactors and the exosome coordinate the processing of ncRNAs is therefore critically important. Here, we find that the Nab3 RNA-binding protein but not Nrd1 of the NNS complex facilitates the function of TRAMP in ncRNA processing/degradation, suggesting a key functional difference between Nab3 and Nrd1. Nab3 suppresses the thermosensitive growth and reduces ncRNA levels of TRAMP mutants independent of Nrd1 interaction. Moreover, Nab3 improvement of TRAMP mutant growth is dependent on the catalytic activity of the nuclear exosome subunit Rrp6 but not the core exosome subunit Dis3. In addition, Nab3 directly binds to Rrp6 and Nab3 coprecipitates with Rrp6 independent of Nrd1 interaction. In the established Nrd1-dependent model for Nab3 function, Nab3 works together with Nrd1 in the NNS complex to recognize ncRNA terminators, interact with TRAMP, and recruit Rrp6/core exosome to terminate and process/degrade ncRNAs (Fig. 7A) [5,19,29,30]. The data presented here suggest a Nrd1-independent model for Nab3 function in which Nab3 facilitates TRAMP by recruitment of Rrp6 to ncRNAs for processing/degradation independent of Nrd1 (Fig. 7B). This model raises the possibility that Nrd1 and Nab3 can function independently to recruit the exosome to ncRNA targets, allowing combinatorial flexibility in processing of RNAs. We also find that the human RALY protein that shares homology with Nab3 improves the function of TRAMP mutants.
The NNS complex and TRAMP exosome cofactors facilitate termination of ncRNAs and recruit/stimulate the exosome to process/degrade these ncRNAs (Fig. 7A). However, the mechanisms and interactions employed by the NNS and TRAMP complexes to recruit Rrp6 and the core exosome are not fully characterized. Nrd1 coprecipitates with Rrp6, the core exosome, and TRAMP and stimulates the activity of purified exosome in vitro [19]. Moreover, Nrd1 directly interacts weakly with Rrp6 and strongly with Trf4 [32]. Putative human TRAMP components (hTRF4-2, ZCCHC7, hMTR4) also coprecipitate strongly with hRRP6 [18]. In addition, Trf4 also directly interacts strongly with Rrp6 and TRAMP stimulates the activity of Rrp6 in vitro [32,45]. Both Nrd1 and TRAMP therefore interact with Rrp6. We now find that Nab3 directly interacts with Rrp6 and a nab3 mutant that lacks the Nrd1-binding domain coprecipitates with Rrp6 but not with Dis3/core exosome. Nab3 and Nrd1 of the NNS complex and TRAMP may thus independently recruit Rrp6 to ncRNA targets, providing flexibility in processing/degradation. The NNS complex and TRAMP may also interact together via Nrd1-Trf4 interaction to enhance recruitment of Rrp6 and the core exosome to ncRNA targets for rapid processing/degradation.
To better understand TRAMP function, we utilized air1/2 TRAMP mutant cells [16]. In air1/2 cells, the air1-ZnK5 protein shows reduced stability and decreased binding to Trf4 that leads to reduced integrity of TRAMP [16]. The primary defect in air1/2 cells, as in other TRAMP mutants, is reduced TRAMP integrity/function that leads to decreased recruitment/ stimulation of Rrp6/core exosome and thus defective processing/degradation of ncRNAs and some mRNAs [12,13,16]. Impaired TRAMP function in trf4Δ mutant cells also causes terminator readthrough of snoRNA, CUT, and some mRNA genes [13,42,43], suggesting TRAMP plays a role in transcription termination. In support, we find that Air1 significantly reduces RNA Pol II occupancy downstream of the IMD2 CUT in air1/2 cells, indicating that Air1 and TRAMP affect termination and suggesting that air1/2 cells have a termination defect. Exactly how TRAMP affects transcription termination is an important question for future study. Impaired growth of TRAMP mutant cells is thus correlated with undegraded and/or non-terminated RNAs. Notably, polyadenylation defective TRAMP mutant cells are viable [13], indicating that polyadenylation is not the essential function of TRAMP and suggesting that exosome recruitment and other activities, such as termination, are the more vital functions of TRAMP.
In this study, we identified NAB3 as a potent suppressor of the impaired growth of air1/2 and trf4Δ TRAMP mutant cells, linking Nab3 of the NNS complex to TRAMP function. Surprisingly, NRD1 does not suppress air1/2 cells and a nab3 mutant that lacks the Nrd1-binding domain still suppresses air1/2 cells, suggesting Nab3 harbors a Nrd1-independent function that it shares with TRAMP. As NAB3 suppression and decrease of IMD2 CUT RNA in air1/2 cells is dependent on Rrp6, and Nab3 directly interacts with Rrp6, we suggest that the Nab3 mechanism of suppression of air1/2 cells involves Nab3 recruitment of Rrp6 to target ncRNAs for processing/degradation (Fig. 7B). Given that TRAMP recruits/stimulates Rrp6, this Nab3 suppression mechanism would seem logical. In support of a Nab3 role in RNA degradation, mutation of a Nab3 binding site in the IMD2 CUT terminator leads to an increase in IMD2 CUT RNA [28]. Importantly, NAB3 does not significantly affect RNA Pol II occupancy downstream of the IMD2 CUT or the snR13 gene in air1/2 cells, suggesting that the Nab3 suppression mechanism does not involve Nab3 improvement of termination. Conceivably, the Nab3 suppression mechanism could also involve an as yet uncharacterized function of Nab3 in TRAMP activity or the ability of Nab3 to reduce the cellular requirement for TRAMP.
The finding that NAB3, but not NRD1, suppresses the growth of TRAMP mutants is surprising, given that Nrd1 directly interacts with Nab3, Trf4, and Rrp6 [29,32]. As the Nab3 RRM is essential for suppression of TRAMP mutants and Nab3 and Nrd1 RRMs recognize different RNA sequences, one intriguing explanation is that Nab3 recognizes important ncRNAs with NNS terminators composed predominantly or exclusively of Nab3 binding sites that Nrd1 cannot recognize in TRAMP mutant cells. On this note, AU-rich sequences (e.g. UAAA; AAAU) and extended Nrd1/Nab3 binding sites have recently been identified in artificial NNSdependent terminators that are critical for Nrd1-Nab3 interaction, present in native ncRNAs, In the established Nrd1-dependent model, Nab3 works in partnership with Nrd1 in the NNS complex to bind to ncRNA terminators, interact with TRAMP and recruit Rrp6 and the core exosome to terminate and process/degrade ncRNAs [5,19,29,30]  and could serve as supermotifs for NNS recognition [50]. As Air2 binds to adenosine RNA [15], TRAMP could help Nrd1-Nab3 to cooperatively recognize these AU-rich sequences. If critical ncRNAs contained terminators with AU-rich-Nab3 site supermotifs, this could explain why Nab3, but not Nrd1, specifically suppresses TRAMP mutants. This possibility implies that Nab3 and TRAMP process/degrade a common set of ncRNAs that are critical for cell growth, but are not regulated by Nrd1.
In support of the idea of Nab3-and Nrd1-specific RNA targets, two RNA cross-linking and RNA-Seq studies that mapped Nab3 and Nrd1 sites transcriptome-wide found that a greater percentage of reads for Nab3 RNAs map to RNA Pol II transcripts than that for Nrd1 RNAs (Nab3-73% Pol II vs Nrd1-59% Pol II [23]; Nab3-42% Pol II vs Nrd1-36% Pol II [24]). As yeast cells express more Nrd1 (*20,000 mols/cell) than Nab3 (*6,000 mols/cell) [51], this result suggests that Nab3 may bind to more RNA Pol II transcripts than Nrd1. Notably, comparison of the top 100 Nab3 and Nrd1 cross-linked sites from one study reveals that Nab3 crosslinks more efficiently to snoRNAs than Nrd1 (Nab3-59/100 sites are snoRNAs vs Nrd1-33/ 100 sites are snoRNAs [52]). Studies have also reported that the processing/degradation of certain ncRNAs is more prominently altered in nab3 mutants compared to nrd1 mutants. In particular, the level of an FLC1-FMP40 intergenic CUT with three Nab3 binding sites is elevated in nab3-11 RRM mutant cells, but unchanged in nrd1-102 RRM mutant cells [25]. The processing of 23S/20S pre-rRNA is also impacted in nab3-11 cells, but unaltered in nrd1-102 cells [46]. Finally, the level of 5'-extended pre-tRNA Arg(UCU) is greater in Nab3-depleted cells than it is in Nrd1-depleted cells [23].
The importance of the NNS complex in regulating the exosome, TRAMP, and RNA processing in yeast and conservation of the exosome and TRAMP components in humans raises the question of whether the NNS exosome cofactor is also conserved in humans. In strong support, the human Sen1 helicase, Senataxin, catalyzes termination in human cells [35]. The human SCAF8 RRM protein has also been proposed to be a Nrd1 orthologue based on sequence identity [53]. However, to date, no human Nrd1 or Nab3 orthologue has been functionally characterized. Here, we find that expression of the human RALY (hRALY) RRMcontaining protein, which shares homology with Nab3 and contains an RRM with 31% identity to the Nab3 RRM, like Nab3, can improve the growth of air1/2 TRAMP mutant cells, suggesting that hRALY can modulate a function performed by TRAMP in yeast cells. Consistent with a potential role for hRALY in processing of ncRNAs, nuclear hRALY interacts with numerous RNA binding proteins involved in ncRNA processing/stability in human cells [54].
Combined, the data presented here indicate that Nab3 and Nrd1 can work independently to recruit the exosome to ncRNA targets, providing combinatorial flexibility in RNA processing.

Chemicals and media
All chemicals were obtained from Sigma-Aldrich (St. Louis, MO), United States Biological (Swampscott, MA), or Fisher Scientific (Pittsburgh, PA) unless otherwise noted. All media were prepared by standard procedures [55].

Strains
All strains used in this study are listed in S1 Table. The air1 (ACY1090), air2Δ (ACY1091), and trf4Δ (ACY2149) strains were obtained from Research Genetics. The NRD1-TAP (ACY2293), RRP6-TAP (ACY1063), and DIS3-TAP (ACY1926) strains were obtained from Thermo Scientific (Open Biosystems). The air1-C178R air2Δ (ACY2020) strain was constructed by insertion of the C178R mutation into the AIR1 ORF in the W303 strain by the 'delitto perfetto' method [59] and deletion of the AIR2 ORF by homologous recombination with AIR2-NATMX PCR product. The trf4Δ (ACY2154) strain was constructed by deletion of the TRF4 ORF in the W303 strain by homologous recombination with TRF4-NATMX PCR product. The nab3Δ (ACY2181) strain was constructed by transformation of a URA3 NAB3 (pAC3285) plasmid into the W303 strain and deletion of the NAB3 ORF by homologous recombination with NAB3-NATMX PCR product. The air1-C178R air2Δ rrp6Δ strain (ACY2294) was constructed by deletion of the RRP6 ORF in the ACY2020 strain by homologous recombination with RRP6-UTR KANMX PCR product. The air1-C178R air2Δ nrd1Δ strain (ACY2320) was constructed by transformation of a URA3 NRD1 (pAC3285) plasmid into the ACY2020 strain and deletion of the NRD1 ORF by homologous recombination with NRD1-UTR KANMX PCR product. The air1-C178R air2Δ dis3Δ strain (ACY2119) was constructed by transformation of a URA3 DIS3 (pAC2681) plasmid into the ACY2020 strain and deletion of the DIS3 ORF by homologous recombination with DIS3-UTR KANMX PCR product. The air1Δ air2Δ (ACY2036) strain was constructed by transformation of a URA3 AIR2 (pAC1614) plasmid into the W303 strain and consecutive deletion of AIR1 and AIR2 ORFs by homologous recombination with AIR1-NATMX and AIR2-HPHMX PCR products. The air1-C178R-TAP air2Δ strain (ACY2051) was constructed by insertion of C-terminal TAP-Sphis5+ PCR product into the air1-C178R ORF in the air1-C178R air2Δ (ACY2020) strain by homologous recombination.

High copy suppressor screen
To identify high copy suppressors of the temperature sensitive growth of air1-C178R air2Δ cells at 30°C, air1-C178R air2Δ cells (ACY2020) were transformed with a 2 μ URA3 yeast genomic DNA plasmid library and plated on Ura − minimal media plates. As controls, the cells were also transformed with 2 μ URA3 vector (pRS426) or AIR1 (pAC1613). The cells were grown at 25°C for one day and then shifted 30°C for 2-4 days select for suppressors. Approximately 18,000 transformants containing library plasmids were screened at 30°C. Library plasmids from transformants that showed improved growth at 30°C relative to vector alone were isolated and retransformed into air1-C178R air2Δ cells to confirm that these plasmids cause suppression of the thermosensitive growth of the cells at 30°C. The confirmed suppressor plasmids were sequenced to identify the genomic DNA inserts.

Quantitation of immunoblots and Northern blots
The band intensities/areas from all immunoblots and Northern blots were quantitated using ImageJ v1.4 software (National Institute of Health, MD; http://rsb.info.nih.gov/ij/) and relevant percentages of protein or RNA were calculated in Microsoft Excel for Mac 2011 (Microsoft Corporation). To quantitate the fold overexpression of Nab3 and Nrd1 protein in air1-C178R air2Δ cells containing NAB3 or NRD1 relative to cells containing vector alone, the Nab3/Nrd1 intensity in cells containing NAB3 or NRD1 was normalized to Pgk1 intensity and Nab3/Nrd1 intensity in cells containing vector alone. To quantitate the percentage of air1-C178R-TAP protein in air1-C178R-TAP air2Δ cells containing TRF4 or NAB3 relative to cells containing vector alone, the air1-C178R-TAP intensity in cells containing TRF4 or NAB3 was normalized to Pgk1 intensity and air1-C178R-TAP intensity in cells containing vector alone. To quantitate the percentage of bound nab3-Myc mutant protein relative to bound Nab3-Myc wild-type protein in Nrd1-TAP binding assays, the bound Nab3/nab3-Myc intensity was normalized to bound Nrd1-TAP intensity, input Nab3/nab3-Myc intensity (normalized to Pgk1 intensity), and bound Nab3-Myc wild-type protein intensity. To quantitate the percentage of input nab3-Myc mutant protein relative to input Nab3-Myc wild-type protein, input nab3-Myc intensity were normalized to Pgk1 intensity and Nab3-Myc intensity. To quantitate the percentage of GFP protein in air1-C178R air2Δ cells containing pREF-GFP reporter and NAB3, nab3 mutants, NRD1 or SEN1, relative to cells containing pREF-GFP reporter and vector alone in terminator readthrough reporter assays, the GFP intensity in cells containing NAB3, nab3 mutants, NRD1 or SEN1 was normalized to Pgk1 intensity and GFP intensity in cells containing vector alone. To quantitate the percentage of IMD2 CUT and readthrough RNA in W303, air1-C178R air2Δ, trf4Δ, and air1-C178R air2Δ rrp6Δ containing vector, AIR1, TRF4, NAB3 or NRD1 in Northern blots, IMD2 CUT and readthrough RNA intensity was normalized to ACT1 or scR1 RNA intensity and IMD2 CUT/readthrough intensity in cells containing vector alone. Quantitation refers to specific experiments shown in Figures but is representative of multiple experiments.

Total RNA isolation
To prepare S. cerevisiae total RNA from cell pellets of 10 ml cultures grown to OD 600 = 0.5-0.7, glass beads (2-3 x 100 μ l) were added to each cell pellet in 2 ml screw-cap tube, 1 ml TRIzol (Invitrogen) was added and cell sample was vigorously disrupted in Mini Bead Beater 16 Cell Disrupter (Biospec) for 2 min at 25°C. For each sample, 100 μ l of 1-bromo-3-chloropropane (BCP) was added, the sample was vortexed for 15 sec, and incubated at 25°C for 2 min. Each sample was then centrifuged at 16,300 x g for 8 min at 4°C and the upper layer was transferred to a fresh microfuge tube. RNA was precipitated with 500 μ l isopropanol and the sample was vortexed for 10 sec to mix. Total RNA was pelleted by centrifugation at 16,300 x g for 8 min at 4°C. Supernatant was decanted and 1 ml of 75% ethanol was added to wash RNA pellet. Sample was centrifuged at 16,300 x g for 5 min at 4°C. Supernatant was decanted, remaining ethanol was removed, and the RNA pellet was air dried for 15 min. Total RNA was resuspended in 50 μ l diethylpyrocarbonate (DEPC (Sigma))-treated water and stored at-80°C.

Recombinant protein expression and purification
Recombinant GST fusion and His-tagged proteins were expressed in bacteria and purified. GST (pGEX-TEV), GST-Rrp47 (pAC3311), GST-Nab3 (pAC3312), and His-Rrp6 (pAC3313) were expressed in E. coli DE3 cells and purified by batch purification. Overnight cultures were used to inoculate 100 ml LB media supplemented with 100 μ g/ml ampicillin or 50 μ g/ml kanamycin. Cultures were grown at 37°C to an OD 600 of 0.6-0.8, induced with 200 μ M IPTG and grown at 25°C overnight. For batch purification of GST fusion proteins, cells were collected and lysed in 10 ml phosphate buffered saline (PBS) supplemented with protease inhibitor mixture (1 mM phenylmethylsulfonyl fluoride, 3 ng/ml pepstatin A, leupeptin, aprotinin, and chymostatin) and 2 mM dithiothreitol (DTT) by incubation with 10 mg lysozyme for 30 min on ice and sonication. Lysates were cleared by centrifugation at 12,000 x g for 10 min and incubated with glutathione Sepharose 4B (GE Healthcare) for 2 hr at 4°C with mixing. The beads were then washed once with 10 ml PBS supplemented with 0.5% Triton-X-100 and 2 mM DTT and twice with 10 ml PBS supplemented with 2 mM DTT. GST proteins were eluted from beads with 1 ml 50 mM Tris-HCl, pH 8 containing 10 mM reduced glutathione and dialyzed into PBS supplemented with 2 mM DTT. For batch purification of His-tagged Rrp6 protein, cells were collected and lysed in 10 ml lysis buffer (50 mM NaH 2 PO 4 , pH 7.4, 300 mM NaCl, 10 mM imidazole) supplemented with protease inhibitor mixture and 2 mM DTT by incubation with 10 mg lysozyme and sonication. Lysates were cleared by centrifugation at 12,000 x g for 10 min and incubated with nickel-NTA agarose (Qiagen) in lysis buffer for 2 hr at 4°C with mixing. The beads were then washed twice with 10 ml wash buffer (50 mM NaH 2 PO 4 , pH 7.4, 300 mM NaCl, 20 mM imidazole). His-tagged Rrp6 protein was eluted from agarose with 1 ml elution buffer (50 mM NaH 2 PO 4 , pH 7.4, 300 mM NaCl, 250 mM imidazole) and dialyzed into PBS supplemented with 2 mM DTT.

Chromatin immunoprecipitation
To analyze RNA Pol II association with IMD2 or snR13 gene in air1/2 cells, chromatin was prepared from yeast cells and immunoprecipitated with an RNA Pol II antibody according to the protocols of Keogh and Buratowski [60] and Chen et al. [61]. The relative Pol II occupancy across IMD2 or snR13-TRS31 locus was determined by quantitative PCR on chromatin immunoprecipitation and input samples using five IMD2 or snR13 primer pairs (S2 Table) and calculation of percentage input. Triplicate 100 ml cultures of air1-C178R air2Δ (ACY2020) cells containing vector (pRS426), AIR1 (pAC1613) or NAB3 (pAC2880) in Ura − media were grown to OD 600 = 0.5 at 25°C. air1-C178R air2Δ cultures were shifted to 30°C for 1 hr 30 min and grown to OD 600 = 0.8. Cells were cross-linked with 2.7 ml 37% formaldehyde (1% final) at 25°C for 20 min, incubated with 10 ml glycine stop solution (3 M glycine, 20 mM Tris base) for 5 min at 25°C, collected by centrifugation at 1500 x g for 4 min, washed twice with 50 ml TBS (20 mM Tris-HCl, pH 7.5, 150 mM NaCl), washed once with 10 ml FA lysis buffer (50 mM HEPES: KOH, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate) supplemented with 0.1% SDS, and collected by centrifugation. Cross-linked cell pellets were lysed in 300 μ l 300 mM FA Lysis buffer (FA buffer containing 300 mM NaCl) by bead beating. Chromatin extract was sonicated using a Bioruptor sonicator (set to high) for 7 min in 30 sec pulses. Sonicated chromatin extract was centrifuged at 4,000 x g for 15 min at 4°C and 500 μg chromatin extract was utilized per immunoprecipitation with an anti-Pol II monoclonal antibody (Clone 4H8; Active Motif). Immune complexes were captured using Protein A-conjugated agarose beads (Santa Cruz Biotechnology) and washed for 5 min each in 300 mM FA lysis buffer, 500 mM FA buffer (FA buffer containing 500 mM NaCl), LiCl buffer (10 mM Tris-HCl, pH 8.0, 250 mM LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA), and TE. Samples were digested with RNase A and eluted from the Protein A-agarose beads in elution buffer (0.1 M sodium bicarbonate, 1% SDS) by two 200 μ l room temperature incubations with rotation. IP samples and input samples (10% total chromatin extract) were incubated at 65°C for at least 4 hr to reverse cross-links. DNA was purified using miniprep columns (Qiagen), IP DNA was eluted in 50 μ l distilled water, and input DNA in 100 μ l distilled water. Quantitative PCR (qPCR) on technical duplicates of IP or input DNA with IMD2 or snR13 primers (S2 Table) was performed in 20 μ l reactions containing QuantiTect SYBR Green PCR master mix (Qiagen), 0.5 mM primers and 2 μ l of IP or input DNA. Each qPCR experiment was performed on a StepOnePlus Real-Time PCR machine (Applied Biosystems) using T anneal = 55°C and 44 cycles. Relative Pol II occupancy was calculated as a percentage of input using the equation: ΔCt = 2^-(IP Ct -Input Ct ). The mean relative occupancy values from three independent experiments were calculated and the mean occupancy values at Primer Pair 1-5 locations were normalized to the value obtained with Primer Pair 1 at the IMD2 CUT or snR13 gene in each strain. The normalized mean Pol II occupancy (relative Pol II occupancy) is displayed with error bars that represent the standard error of mean.

Localization of GFP-tagged proteins
To localize Nab3-GFP and nab3-1-448 proteins, air1-C178R air2Δ cells (ACY2020) expressing GFP-tagged Nab3 protein (pAC3281) or nab3-1-448 (pAC3282) were grown in Leu − media with 2% glucose overnight at 25°C, transferred to Ura − Leu − media with 2% glucose and grown to log phase at 25°C. GFP fusion proteins were visualized by direct fluorescence microscopy using an Olympus BX60 direct fluorescence microscope equipped with a photometric Quantix digital camera from Roper Scientific (Tucson, AZ) and filters from Chroma Technology (Brattleboro, VT). All images were captured using IP Lab Spectrum software.
Supporting Information S1 Fig. NAB3 suppresses the slow growth of trf4 Δ mutant cells and RNA binding protein genes do not suppress the thermosensitive growth of air1/2 cells. Related to Figs. 1-2. (A) NAB3 and nab3-Δ NBD mutant suppress trf4Δ slow growth at 25°C, relative to cells containing vector alone, but nab3-11, nab3-R331A, nab3-F333A, and nab3-S399A RRM mutants, NRD1, and SEN1 do not suppress trf4Δ slow growth at 25°C. The trf4Δ cells containing vector, NAB3, nab3 RRM mutants, nab3-ΔNBD mutant, NRD1 or SEN1 2μ URA3 plasmid were grown to saturation, serially diluted and spotted on plates, and grown at indicated temperatures. (B) RNA binding protein genes, NPL3, HRP1, PUB1, and NAB2, do not suppress the thermosensitive growth of air1-C178R air2Δ cells at 30°C. The air1-C178R air2Δ cells containing vector, AIR1, TRF4, GCD14, NAB3, NPL3, HRP1, PUB1 or NAB2 2 μ URA3 plasmid were grown to saturation, serially diluted and spotted on plates, and grown at indicated temperatures. See Materials and Methods for details. (TIF) S2 Fig. Exogenous Nrd1 and Nab3 are overexpressed relative to endogenous Nrd1 and Nab3 in air1/2 cells. Related to Fig. 1. Lysates of air1-C178R air2Δ cells containing vector, AIR1, TRF4, NAB3 or NRD1 2 μ URA3 plasmid at 30°C were analyzed by immunoblotting to detect Nab3 and Nrd1 and 3-phosphoglycerate kinase (Pgk1) as a loading control. Fold overexpression of Nab3 and Nrd1 relative to Pgk1 loading control and cells containing vector alone (Fold Nab3/Nrd1 Rel Vector) is shown below lanes and was calculated as described in Materials and Methods. Nonadjacent lanes in the same immunoblot are separated by white space. The nab3-R331A and nab3-S399A RRM mutant proteins show binding to Nrd1 similar to wild-type Nab3, but the nab3-ΔNBD mutant protein shows greatly reduced binding to Nrd1. TAP-tagged Nrd1 was precipitated from lysates of NRD1-TAP cells expressing Myctagged Nab3, nab3-R331A, nab3-S399A, or nab3-ΔNBD and bound (B), unbound (U), and input fractions were analyzed by immunoblotting to detect Nab3-Myc proteins, Nrd1-TAP proteins and 3-phosphoglycerate kinase (Pgk1) as a loading control. The percentage of bound Nab3 relative to input protein and bound wild-type Nab3 (% Bound) is shown below the bound lanes. The percentage of input Nab3 protein relative to input wild-type Nab3 protein (% Input) is shown below the input lanes. The percentages of protein were calculated as described in Materials and Methods. Quantitation refers to specific experiment shown but is representative of multiple experiments. Immunoblots in Figs. 2B and 3D derive from this original immunoblot. (B) Wild-type cells containing vector, NAB3, nab3-11, nab3-R331A, nab3-F333A, nab3-S399A, nab3-S400A RRM mutants or nab3-ΔNBD mutant 2 μ URA3 plasmid were grown to saturation, serially diluted and spotted on plates, and grown at indicated temperatures. See Materials and Methods for details. Cells spotted in upper and lower panels are on different plates. (TIF) S4 Fig. nab3-ΔNBD mutant decreases levels of native IMD2 CUT and readthrough product similar to NAB3 in air1/2 cells, but most nab3 RRM mutants decrease the level of the IMD2 CUT and readthrough product to a lesser extent than NAB3, and NAB3 does not significantly affect the termination of native snR13 snoRNA gene in air1/2 cells. Related to Fig. 4. (A) Northern blot of total RNA from air1-C178R air2Δ cells expressing vector, NAB3, nab3-ΔNBD mutant or nab3 RRM mutants, nab3-11, nab3-R331A, nab3-F333A, and nab3-S399A, grown at 30°C was probed with an IMD2 CUT-specific probe. Ethidium bromide-stained 25S rRNA on the Northern blot is shown as a loading control. The IMD2 CUT (IMD2 Short/CUT) and IMD2 CUT readthrough product (IMD2 Long/RT) are labeled. A longer exposure of IMD2 CUT readthrough product is shown above. (B) NAB3 does not significantly affect Pol II occupancy downstream of snR13 gene at Primer Pair 2-5 positions in air1-C178R air2Δ cells relative to air1/2 cells containing vector alone (p-value 0.3), suggesting that Nab3 overexpression does not significantly affect snR13 termination in air1/2 cells. Anti-Pol II ChIP was performed on air1-C178R air2Δ cells containing vector or NAB3 and relative Pol II occupancy was measured within and downstream of snR13 gene by qPCR with snR13 Primer Pair 1-5 as described in Material and Methods. Mean RNA Pol II occupancy values from three independent experiments normalized to Primer Pair 1 within snR13 gene are shown with error bars that represent standard error of the mean. Statistical significance of differences in mean Pol II occupancy values was determined using unpaired t test. Schematic of snR13 gene and downstream TRS31 gene is shown with positions of snR13 qPCR Primer Pairs 1-5 above and base pair distances between primer pairs below. (TIF) S5 Fig. NAB3 does not suppress air1/2 rrp6Δ cells expressing the catalytically inactive rrp6 mutant, rrp6-D238A, and Rrp6 binds to nab3-Δ1-248 NBD mutant in a similar manner to wild-type Nab3. Related to Fig. 5. (A) NAB3 does not suppress the thermosensitive growth of air1/2 rrp6Δ cells expressing the catalytically inactive rrp6 Mutant, rrp6-D238A, at 30°C. air1-C178R air2Δ rrp6Δ cells containing RRP6 or rrp6-D238A and vector, AIR1, AIR2, TRF4, NAB3, NRD1 or SEN1 2 μ URA3 plasmid were grown to saturation, serially diluted and spotted on plates, and grown at indicated temperatures. See Supplemental Experimental Procedures for details. Cells spotted in upper and lower panels are on different plates. (B) Rrp6 binds to nab3-Δ1-248 NBD mutant in a similar manner to wild-type Nab3. TAP-tagged Nab3 or nab3-Δ1-248 mutant protein was from lysates of wild-type cells expressing NAB3-TAP or nab3-Δ1-248-TAP and Myc-tagged Rrp6 and bound (B), unbound (U), and input fractions were analyzed by immunoblotting with an anti-Myc antibody to detect Rrp6-Myc proteins and an anti-Pgk1 antibody to detect 3-phosphoglycerate kinase (Pgk1) as a loading control. See Materials and Methods for details. Nonadjacent lanes in the same immunoblot are separated by white space. Different immunoblots are separated by black boxes. (TIF) S6 Fig. Alignment of Nab3 and hRALY full-length protein sequences. hRALY RNA recognition motif (RRM) has 31% identity with the Nab3 RRM and hRALY C-terminal domain has 11% identity with the Nab3 C-terminal domain. Single RRM domains are boxed and RRM consensus motifs RNP1 and RNP2 are marked with lines above. Identical residues are shaded in black and similar residues are shaded in gray. Residue numbers are shown on left. Nab3 and hRALY Isoform 1 (GenBank accession number Q9UKM9) protein sequences were aligned with ClustalW2 sequence alignment tool and shaded using BoxShade software. (TIF) S7 Fig. The nab3-1-448 mutant containing the Nab3 RRM but lacking the C-terminal domain does not suppress air1/2 thermosensitive growth, even though this mutant is expressed and localized to the nucleus. Related to Fig. 6 (A) nab3-1-448 mutant that retains RRM but lacks C-terminal domain of Nab3 does not suppress the thermosensitive growth of air1/2 cells. air1-C178R air2Δ cells containing vector, NAB3, nab3-11, or nab3-1-448 2μ URA3 plasmid were grown to saturation, serially diluted and spotted on plates, and grown at indicated temperatures. (B) nab3-1-448 mutant protein is expressed in air1/2 cells. Lysates of air1-C178R air2Δ cells containing vector or expressing nab3-1-448 mutant at 30°C were analyzed by immunoblotting with anti-Nab3 antibody to detect Nab3 protein and anti-Nrd1 antibody to detect Nrd1 as a loading control. Endogenous Nab3 and overexpressed nab3-1-448 are detected by anti-Nab3 antibody. (C) nab3-1-448 mutant protein localizes to the nucleus like Nab3 in air1/2 cells. air1-C178R air2 cells expressing C-terminally GFP-tagged Nab3 or nab3-1-448 mutant were visualized by direct fluorescence microscopy. DAPI stain shows the position of the nucleus. Differential interference contrast (DIC) images visualize the cells. See Materials and Methods for details. (TIF) S1 Table. Yeast strains and plasmids used in this study. (DOCX) S2 Table. DNA oligonucleotides used in this study. (DOCX)