A Yeast Heterogeneous Nuclear Ribonucleoprotein Complex Associated With RNA Polymerase II
Nicholas K. Conrad, Scott M. Wilson, Eric J. Steinmetz, Meera Patturajan, David A. Brow, Maurice S. Swanson, Jeffry L. Corden


Recent evidence suggests a role for the carboxyl-terminal domain (CTD) of the largest subunit of RNA polymerase II (pol II) in pre-mRNA processing. The yeast NRD1 gene encodes an essential RNA-binding protein that shares homology with mammalian CTD-binding proteins and is thought to regulate mRNA abundance by binding to a specific cis-acting element. The present work demonstrates genetic and physical interactions among Nrd1p, the pol II CTD, Nab3p, and the CTD kinase CTDK-I. Previous studies have shown that Nrd1p associates with the CTD of pol II in yeast two-hybrid assays via its CTD-interaction domain (CID). We show that nrd1 temperature-sensitive alleles are synthetically lethal with truncation of the CTD to 9 or 10 repeats. Nab3p, a yeast hnRNP, is a high-copy suppressor of some nrd1 temperature-sensitive alleles, interacts with Nrd1p in a yeast two-hybrid assay, and coimmunoprecipitates with Nrd1p. Temperature-sensitive alleles of NAB3 are suppressed by deletion of CTK1, a kinase that has been shown to phosphorylate the CTD and increase elongation efficiency in vitro. This set of genetic and physical interactions suggests a role for yeast RNA-binding proteins in transcriptional regulation.

EUKARYOTIC messenger RNA biogenesis is a complex, multistep process carried out by molecular machines consisting of hundreds of polypeptides. While many components of the transcription and splicing machines have been identified and assigned functional roles, less is known about how the synthesis of pre-mRNA is synchronized with its processing and transport. As nascent pre-mRNA emerges from the elongating RNA polymerase II (pol II), its postsynthetic fate is determined by a large number of RNA-binding proteins. Primary among these proteins are the heterogeneous nuclear ribonucleoproteins (hnRNPs) that have been shown to play multiple roles in processing and transport (Siomi and Dreyfuss 1997; Krecic and Swanson 1999). These proteins are thought to be the first to bind nascent transcripts, but how, or indeed whether, they interact with the transcription machinery as RNA-bound factors has not been investigated previously.

Recent studies have indicated that the carboxyl-terminal domain (CTD) of the largest subunit of pol II interacts with multiple pre-mRNA processing components. Direct interaction between the CTD and capping (Cho et al. 1997, 1998; McCrackenet al. 1997a; Yueet al. 1997; Hoet al. 1998; Ho and Shuman 1999; Pillutlaet al. 1998) and 3′-end processing (McCrackenet al. 1997b) components has been demonstrated. A highly phosphorylated form of pol II associates with spliceosome components (Mortillaroet al. 1996; Yuryevet al. 1996; Kimet al. 1997), and transcripts synthesized by pol II containing a deleted CTD are poorly spliced in vivo (McCrackenet al. 1997b). The CTD is necessary for recruitment of pre-mRNA splicing factors to transcription sites in vivo (Misteli and Spector 1999). In vitro, a purified pol II complex with a hyperphosphorylated CTD stimulates splicing (Hiroseet al. 1999), and 3′ processing can be stimulated by recombinant phosphorylated CTD (Hirose and Manley 1998). A recently identified set of serine-arginine (SR)-rich proteins termed SCAFs (SR-like CTD-associated factors, Yuryevet al. 1996; Corden and Patturajan 1997) can interact with the phosphorylated CTD (Patturajanet al. 1998b). These proteins are excellent candidates for coupling transcription and pre-mRNA processing.

The yeast NRD1 gene encodes a protein with homology to mammalian SCAF8 and SCAF4 (Steinmetz and Brow 1996, 1998; Steinmetz 1997; Yuryevet al. 1996). These proteins all contain an N-terminal CTD-interacting domain (CID), a single RNA recognition motif (RRM), and a region rich in alternating charged residues (Corden and Patturajan 1997). In the mammalian proteins, the latter region is comprised mainly of alternating (phospho) serine and arginine residues similar to the SR protein family (Fu 1995). The yeast Nrd1p RE/RS motif contains relatively few arg/ser dipeptides, but is rich in arg/glu dipeptides.

NRD1 was identified originally as a gene encoding an hnRNP that mediated the downregulation of a reporter gene containing an exogenous sequence element in its intron (Steinmetz and Brow 1996). This artificial cis-acting RNA sequence, designated U6R*, comprises 112 bases derived from an antisense U6 RNA gene together with unrelated 5′ flanking sequences. The C-terminal portion of Nrd1p, containing the RRM, binds specifically to this RNA element in vitro (Steinmetz and Brow 1998). In vivo, truncated reporter RNA species with 3′ ends mapping to the U6R* element further suggest that Nrd1p plays a role in transcript elongation, termination, and/or 3′-end formation (Steinmetz 1997). The evolutionarily conserved N-terminal domain of Nrd1p interacts with the pol II CTD in a two-hybrid assay (Yuryevet al. 1996). More recent studies have shown that the mammalian SCAF8-CTD interaction is dependent on CTD phosphorylation (Patturajanet al. 1998b). Interaction with the phosphorylated CTD would place Nrd1p in transcription elongation complexes where it could monitor nascent transcripts for the presence of RNA signal sequences leading to 3′-end formation. Consistent with this model, deletion of most of the CTD-interacting domain eliminates Nrd1p-dependent downregulation of U6R*-containing pre-mRNA (Steinmetz and Brow 1998).

While NRD1 is an essential gene, the CTD-interacting domain is not required for viability (Steinmetz and Brow 1998). To further investigate the essential cellular function of NRD1, we have isolated and characterized a set of temperature-sensitive nrd1 mutants. We show here that several nrd1 temperature-sensitive alleles are synthetically lethal with several CTD truncation mutations. In addition, we show that overexpression of another hnRNP, Nab3p, suppresses some nrd1 temperature-sensitive alleles. A genome-wide two-hybrid screen with Nrd1p as bait yielded Nab3p as an interacting protein, and the region of Nrd1p necessary for interaction with Nab3p was mapped to <100 amino acid residues between the CID and RE/RS domains. Finally, we show that a nab3 temperature-sensitive mutant is suppressed by a mutation in CTK1, a gene encoding the catalytic subunit of a kinase that phosphorylates the CTD. Together, these results suggest that the pol II CTD, Nrd1p, Nab3p, and the CTD kinase CTDK-I function together at the interface between transcription and pre-mRNA processing.


Yeast strains, genetic methods, and media: Saccharomyces cerevisiae strains used in this study are listed in Table 1. BY4743, a derivative of S288C produced by mating BY4741 and BY4742 (Brachmannet al. 1998), is the parent of strains used in all nrd1 domain deletion and temperature-sensitive experiments except for the CTD synthetic lethality analysis. The latter used ZAYA, which was derived from the mating of Z26 (Nonetet al. 1987) with AYA1 (Yuryev and Corden 1996). NRD1 was targeted for disruption using a fragment generated by PCR of pUG6 with oligos JC827 (5′-AAAGG AACCG GAAAG CAACA AACAT ACTAA ACATC CCATA CCAGC TGAAG CTTCG TACGC-3′) and JC828 (5′-TTTTA TGTAC TATGA GCAAA TAAAG GGTGG AGTAA AGATC GCATA GGCCA CTAGT GGATC TG-3′), which amplified a product encoding a Kanr gene flanked by 40 bp of homology to sequences upstream and downstream of NRD1 coding sequence (Guldeneret al. 1996). After transformation of the diploid strain and selection for Kanr, disruption was confirmed via Southern blot, and sporulation verified that NRD1 is essential in the background. Diploids were further transformed with a URA3-marked plasmid (pNRD-HA) encoding an epitope-tagged wild-type NRD1 gene and sporulated, and the derived Kanr/Ura+ strains (YJC582 and YJC546) were used for subsequent experiments. Derivatives of these strains were made via transformation and counterselection on 0.1% 5-fluoroorotic acid (5-FOA) (Boekeet al. 1984).

Domain deletion analysis was carried out by transforming YJC582 with deletion constructs and retaining selection for both wild-type and deletion plasmids. Multiple independent transformants were grown to stationary phase in CSM-Leu-Ura. Dilution series were made (eightfold) and plated onto CSM-Leu 5-FOA plates at 15, 20, 25, 30, and 37°.

Rich (YPD) and minimal media used were prepared as described (Roseet al. 1990), except CSM dropout mix was obtained from Bio 101 (Vista, CA). Transformation of yeast was performed in lithium acetate as described (Schiestl and Gietz 1989).

Plasmids: Plasmids used in this study are summarized in Table 2. Most of the constructs generated in the present study used the pRS series of shuttle vectors (Sikorski and Hieter 1989; Christiansonet al. 1992). All NRD1 constructs in this study are derived from an ~5-kb genomic HindIII fragment containing the entire NRD1 coding sequence and 1 kb of upstream sequence in pRS316 (Steinmetz and Brow 1996). Plasmid pNrd4 was created by subcloning the HindIII fragment into the HindIII site of pRS426 in the orientation that places the upstream sequence nearest the XhoI site of the vector. To remove ~2 kb of downstream sequence and engineer a NotI site at the termination codon of NRD1, PCR products using pNrd4 as template and primers JC811 (5′-GACCC CACTG CTCAA TT GAA TTCTT-3′), which includes an endogenous EcoRI site, and JC813 (5′-TTA GCGGCCGC TTTGT TGTTG TTGC-3′), which includes the engineered NotI terminator, were mixed with amplification products of pNrd4 with JC810 (5′-CCGCGAGCTCA AAGTT TATTG ATTGT TTAAC TG-3′) and JC812 (5′-AGCGGCCGCTA AGATCT TTACT CCA-3′). The latter products include the same NotI site (12-bp overlap) and ~280 bp of 3′ sequence. Relevant restriction sites in the primers are underlined. The mixed PCR products were used as a template for PCR with primers JC811/JC810, and the resulting fragment was cut with EcoRI and SacI and ligated to EcoRI/SacI-digested pNrd4. The resulting clone, pNrd1-Not, and all other constructs made via PCR were sequenced to exclude the possibility of PCR-induced mutations.

We tagged NRD1 using the method of Tyers et al. (1992). An attempt to produce a triple-HA-tagged version of NRD1 produced instead a double-tagged version of NRD1 with a fortuitous termination codon following the internal BamHI site (pNRD1-HA). Plasmids pJC504 and pJC580 were produced by transfer of XhoI/SacI pNRD1-HA fragment to XhoI/SacI-digested pRS414 and pRS415, respectively.

View this table:

Yeast strains used in this study

A NRD1-promoter vector, pJC652, was designed to construct the nrd1-CIDΔ and nrd1-CID constructs and for use in generation of temperature-sensitive mutations. This vector is comprised of 1 kb of upstream sequence (up to and including the NRD1 initiation codon). Adjacent to the initiation codon, an artificial SmaI site was engineered, followed by the endogenous 3′ EcoRI site and the HA-tag. The vector permits easy inframe subcloning using SmaI/EcoRI-digested PCR fragments and provides a convenient restriction site, which was used for gap repair in the generation of temperature-sensitive mutations. Oligos JC835 (5′-GCCGAATTCT GCCCG GGCAT TATGG GATGT TTAGT AT-3′) and JC836 (5′-CGA TAAGCTTG AAAG CCGCT TTA-3′) were used to generate a product that spanned the natural HindIII site 1 kb upstream of NRD1 coding sequence to the initiation codon. Immediately downstream of the initiation codon, JC835 contains SmaI and EcoRI sites. The fragment was cut with EcoRI and HindIII and ligated to a pNRD1-HA EcoRI/HindIII fragment. The resulting clone was transferred to pRS415 via XhoI/SacI digestion and ligation.

Domain deletion constructs were designed to test the importance of Nrd1p motifs for viability and were used to map the positions of temperature-sensitive mutations. SmaI/EcoRI-digested pJC652 was ligated to EcoRI/SmaI-digested PCR products from JC841 (5′-GGCGAATTC TTGGG ATCCAG TGATA GTTGT-3′) and JC842 (5′-TCCCCCGGG GACGA CGATT TTCAA AA-3′) and JC1111 (5′-AAG CCCGGG CAACT ATCACT GGATCC CAA-3′) and SD2486 (5′-CAGAA ATTAT ATATA GAGGT-3′) to produce pJC583 (nrd1-CID) and pJC643 (nrd1-CIDΔ), respectively. In order to verify that the pJC652-derived N-terminal sequence alterations had no phenotype of their own, an otherwise wild-type construct was generated by ligating EcoRI-digested JC1110 (5′-AAGAATTC ACAACT ATCACT GGATCC CAA-3′) and JC1120 (5′-GAGCA AATAA AGGGT GGAGT-3′)-amplified PCR fragment into EcoRI-digested pJC583. No phenotype was observed when this was the sole NRD1 allele. Deletion of the RRM motif was achieved by PCR amplification of pNRD1-HA with oligos SD2485 (5′-GTACT TTTCTC CAAGC ACGA-3′) and JC1103 (5′-GGACTAGT ATCTGG TGGCAA TGTAG AGT-3′) and SD2486 and JC1102 (5′-GGACTAGT ACTAGG TGGGG AGTTG GTTT-3′). The two products were cut with SpeI, ligated overnight at 14°, and reamplified with JC1104 (5′-GAAA CACCT TCACC AACGAT-3′) and SD2485. Products were then cut with BstBI/EheI and ligated to BstBI/EheI-cut pJC504. Similarly, deletion of the RE/RS motif used SD2485 and JC1101 (5′-GGACTAGT ACCAT ATTCG GTAAT TGTAT-3′) and SD2486 and JC1100 (5′-GGACTAGT CCTCC GGCACC ATTTT CTCA-3′) in the initial PCR reactions, and XcmI/BstBI in the final step. Finally, pJC605 (nrd1-RSΔ) and pJC606 (nrd1-RRMΔ) were generated by XhoI/SacI transfer to pRS415.

View this table:

Plasmids used in this study

Plasmids were recovered from temperature-sensitive strains by standard laboratory procedures (Ausubelet al. 1995). Sequenced fragments of temperature-sensitive alleles nrd1-51 and nrd1-60 were subcloned to generate nrd1-101 and nrd1-102, respectively; both of these alleles include only one mutation. Plasmids pJC580 (NRD1-HA) and pJC951 (nrd1-51) were transformed into dam dcm Escherichia coli strain DM1 (Life Technologies, Gaithersburg, MD) and recovered; nonmethylated plasmids were cut with BclI/XhoI, and 7.9-kb (derived from pJC580) and 1.1-kb (derived from pJC951) products were ligated to generate pJC719. Plasmids pJC504 (NRD1-HA) and pJC955 (nrd1-60) were cut with EheI/BstBI; 7.3- and 0.5-kb fragments, respectively, were ligated. The resulting clone was subcloned to pRS415 by XhoI/SacI fragment ligation to produce pJC720.

Isolation of nrd1 temperature-sensitive alleles: Temperature-sensitive alleles of nrd1 were isolated via PCR mutagenesis and gap repair (Muhlradet al. 1992). Mutagenic PCR was carried out in a 100-μl reaction containing 5 units Taq polymerase; 10 mm Tris-HCl (pH 8.3); 50 mm KCl; 1.5 mm MgCl2; 0.2 mm each dATP, dGTP; 1 mm each dCTP, dTTP; 1 μm each JC1120 and JC1121 (5′-CC-CG CTCCTT CCCTT TACGA-3 ′), and 10 ng pJC504 template. After 10 cycles, MnCl2 was added to 100 μm, and the reaction was carried out for an additional 30 cycles with an annealing temperature of 55°. Reaction products were cotransformed in a 10:1 molar ratio with SmaI-digested pJC652 into YJC582. Transformants were selected on CSM-Leu plates, replica plated to CSM-Leu plates containing 0.1% 5-FOA, and then replica plated to CSM-Leu plates at 25° and 37°. Approximately 104 colonies were screened, yielding 50 candidates (nrd1-51 through nrd1-100). Six of these candidates (nrd1-51, nrd1-52, nrd1-53, nrd1-54, nrd1-60, and nrd1-83) were transformed with pNRD1-HA, which complemented the phenotype in all six cases. No other candidates were tested for complementation by pNRD1-HA.

Mapping and sequencing nrd1 temperature-sensitive mutations: In order to map the location of mutations imparting the temperature-sensitive phenotype, we performed complementation by gap repair. Linearized plasmids were transformed into temperature-sensitive strains and selection was carried out at 37°. Plasmids used were: pJC605 (SpeI cut), pJC606 (SpeI cut); pJC583 (BamHI cut); and pJC643 (SmaI cut). Complementing plasmids showed 500 to >1000 colonies per plate, while noncomplementing plasmids showed two to three orders of magnitude fewer colonies. Using this data, we deduced the approximate domain of the temperature-sensitive mutation and sequenced that region. We subcloned the sequenced areas of nrd1-51 and nrd1-60 as outlined in the Plasmids section, and we renamed the resulting temperature-sensitive alleles nrd1-101 and nrd1-102, respectively.

Synthetic lethality with CTD truncation mutations: Strains used in CTD synthetic lethality assays were made by transforming YJC546 with a pRS414-derived (Sikorski and Hieter 1989) plasmid encoding the appropriate NRD1 allele followed by counterselection on 5-FOA. Resulting strains were transformed with pRP112 (Nonetet al. 1987) and selected for Ura+ phenotype with no selection for Leu+; Ura+/Leu strains carrying NRD-HA, nrd1-51, and nrd1-83 were named YJC1111, YJC1115, and YJC1119, respectively. These strains were then transformed with empty vector (CEN/LEU2) or plasmids encoding wild-type RPB1, or rpb1 derivatives with CTD truncations to 10 or 9 repeats. Multiple independent transformants were tested for viability on 5-FOA as in the domain deletion analysis. No significant difference was seen in the results if the strains were grown overnight in CSM-Leu or CSM-Ura-Leu prior to 5-FOA selection.

High-copy suppression of nrd1 temperature-sensitive alleles: A high-copy suppressor screen was performed using a 2μ/URA3 library (average insert is ~7 kb) generously provided by C. Connelly (Connelly and Hieter 1996). In the first screen, YJC818, a nrd1-51 strain, was transformed with the library, plated on CSM-Ura, and grown for 24 hr at 25° to allow accumulation of high-copy number. Plates were subsequently transferred to 37° to select for the presence of high-copy suppressors. One plate was left at 25° to determine transformation efficiency. Of ~104 transformants screened, eight colonies grew at 37°, and the appropriate high-copy plasmid was recovered. Multiple restriction digests indicated the presence of three different plasmids, pSUP1, pSUP5, and pSUP8, which were obtained one, four, and three times, respectively. Approximately 500 bp from each end of the insert was sequenced. Similar high-copy screens were performed on nrd1-53 and nrd1-60, but no suppressors were found.

Two-hybrid analysis of Nrd1p-Nab3p interactions: The yeast two-hybrid system of James et al. (1996), consisting of the efficient selection strain PJ69-4a (harboring selectable GAL UAS-dependent HIS3 and ADE2 reporter genes) and improved Gal4 activation domain (GAD) and DNA-binding domain (GBD) fusion plasmids, was used to identify Nrd1p interaction partners. GBD-Nrd1p fusions were constructed in the URA3-marked plasmids pGBDU-C1 and pGBDU-C3. A BamHI/EcoRI fragment of NRD1 encoding residues 169–560 was excised as a BamHI/SalI fragment from pET21b-Nrd1(169–560) (Steinmetz and Brow 1998) and cloned into BamHI/SalI-digested pGBDU-C3 to generate pGBDU-Nrd1(169–560). To create pGBDU-Nrd1(1–560), a fragment encoding Nrd1p residues 1–169 was generated by PCR of a NRD1 plasmid with primers Nrd1–5′NdeI (5′-CTAAA CATCC CATATG CAGCA GGACG AC-3′) and Nrd1-Bam169 (5′-GCTTG GGATCC AGTGA TA-3′). The NdeI-digested PCR product was blunted by filling in with Klenow, then digested with BamHI and ligated into SmaI/BamHI-digested pGBDU-Nrd1(169–560). pGBDU-Nrd1(169–265) was created by deleting a BglII fragment bounded by sites in the Nrd1 coding sequence and in the polylinker of pGBDU-Nrd1(169–560). pGBDU-Nrd1(265–560) was created by subcloning a BglII/SalI fragment from pGBDU-Nrd1(169–560) into pGBDU-C1 to obtain the correct reading frame fusion. pGBDU-Nrd1(169–560RSΔ) was constructed by substituting a BamHI/KpnI fragment from the nrd1-RSΔ plasmid, pJC605, into pGBDU-Nrd1(169–560).

The set of yeast two-hybrid libraries constructed by James et al. (1996), which have yeast genomic fragments fused to the Gal4 activation domain in all three reading frames in the LEU2-marked plasmids pGAD-C1, pGAD-C2, and pGAD-C3, was used to transform strain PJ69-4a harboring pGBDU-Nrd1(1–560). A portion of each transformation reaction was plated onto −Ura −Leu medium to determine the number of transformants, and the remainder was plated on −Ura −Leu −His to select for transformants that received GAD fusions allowing activation of the HIS3 reporter gene. After growth at 30° for 1 wk under these low-stringency conditions, colonies were replica plated onto −Ura −Leu −Ade medium to select for those transformants that also allow activation of the more stringent ADE2 reporter. DNA was isolated from colonies appearing 1–5 days after replica plating, and the genomic fragment fused to GAD identified by PCR amplification followed by sequencing with primers GAD5′ (5′-AAGAT ACCCC ACCAA ACCC-3′) and GXD3′ (5′-TTGAG ATGGT GCACG ATGC-3′). Dependence on the GBD-Nrd1 bait for growth on −Ade and −His was verified after selection against pGBDU-Nrd1(1–560) on 5-FOA.

Isolation of nab3 temperature-sensitive alleles: The nab3-1, nab3-3, and nab3-4 mutant alleles were generated by mutagenic PCR using MSS44 (5′-TAATA CGACT CACTA TAGGG AGA-3′) and MSS45 (5′-CATAC GATTT AGGTG ACACT ATAG-3′) and pNAB3.8 (Wilsonet al. 1994) as described previously (Minvielle-Sebastiaet al. 1998). The mutagenized DNA and StyI-cut pNAB3.14 were cotransformed into YSW501-1C, and the cells were plated at 24° on SD-Trp. After 3 days, visible colonies were replica plated onto 5-FOA plates and incubated at either 24° or 37°. Colonies that grew at 24° but not 37° were restreaked on SD-Trp. The temperature-sensitive phenotype was confirmed by isolating nab3 plasmids and recloning of the mutant alleles into pRS314 followed by transformation into YSW501-1C. All mutant alleles were sequenced completely. Several nab3 mutant strains carried alleles with multiple missense mutations including nab3-1 (F371I, Q520H), nab3-3 (I84T, E128D, E129G, F371L, H450R), and nab3-4 (A312G, L314S, M388V). To confirm that the temperature-sensitive mutation in nab3-3 was F371L, the nab3-10 allele was constructed using mutagenic oligonucleotides and subsequently integrated into the NAB3 chromosomal locus in both L4717 and L4718 strains. Plasmid pNAB3.16 was used as a template for primary PCR reactions using MSS639 (5′-CCCAT CGATTT CTGCA CAGCA ACAAGC-3′) and MSS642 (5′-GCTTT GAGGG TTGTC AAGCT GAATG AATCCA AAGGC-3′) to generate the nab3-10 5′ fragment and MSS641 (5′-CATTC AGCTT GACAA CCCTC AAAGC GTTAG AGATGC-3′) and MSS640 (5′-CCCGG ATCCG CTGTT CTTTA AAGC GCC-3′) for the 3′ fragment (mutations are underlined). For the secondary PCR reaction, primary PCR products were used as the template and MSS639 and MSS640 as primers. The resulting DNA was cut with ClaI and BamHI and ligated to ClaI/BamHI-digested YIp5. The resulting pNAB3.60 plasmid was linearized with either NheI or HindIII and transformed into L4717 and L4718. Following passage on SD-Ura and 5-FOA plates, transformants were tested for growth at 24°, 30°, and 36°. Temperature-sensitive alleles were recovered by gap repair, and the nab3-10 gene was sequenced completely to confirm the presence of the single F371L missense mutation. The growth properties of nab3-3 and nab3-10 were identical, demonstrating that the single missense mutation was responsible for temperature-sensitive growth.

The nab3-9 strain was isolated during a synthetic lethal screen. Briefly, Y388 and YSW550-1D were mated, the resulting diploids were dissected, and haploid ade2 ade3 nab3-3 progeny were identified following transformation with pNAB3.48. Dissected haploids were scored for deletion of NAB3 (Leu+), presence of nab3-3 (Trp+), and positive sectoring (Sec+). One of these haploids, YSW900, was plated at a density of 2000 cells/plate and immediately irradiated with UV light in a Stratalinker (Stratagene, La Jolla, CA) to ~10% survival. Following mutagenesis, cells were incubated for 3–5 days at 30° and scored for their ability to sector. The majority of the cells (>95%) remained Sec+. Nonsectoring (Sec) colonies were restreaked twice on YPD and tested for temperature-dependent sectoring. One of these strains, YSW901, was unable to sector at 30° but sectored at 24°. This temperature-dependent sectoring was complemented by either a NAB3 or nab3-3 plasmid-borne allele, suggesting that YSW901 carried a new nab3 temperature-sensitive allele. The mutant allele (nab3-9) was recovered by plasmid rescue and was found to contain a new missense mutation in the RRM, P374L, in addition to the five nab3-3 missense mutations. The nab3-11 strain, which contains only two missense mutations (F371L and P374L), was constructed essentially as described for nab3-10, except that the primary PCR reaction used MSS639 and MSS644 (5′-GCTTT GAGTG TTGTC AAGCT GAATG AATCC AAAGG C-3′) for the 5′ fragment and MSS643 (5′-CATTC AGCTT GACAA CACTC AAAGC GTTAG AGATGC-3′) and MSS640 for the 3′ fragment. The growth characteristics of nab3-9 and nab3-11 were identical at 24°, 30°, and 36°.

Extragenic suppression of nab3-9: Spontaneous extragenic suppressors were isolated using the nab3-9 strain, which is growth defective at 30°. Briefly, 5 × 107 YSW901 cells were plated at 30° for 3–4 days. Over 800 Ts+ colonies were selected, restreaked, and then tested for growth at 24° and 14°. Of the 30 Cs colonies recovered, five showed linkage of the Ts+/Cs phenotypes when backcrossed to YSW901. Complementation analysis was performed, and representatives from each group were cloned by complementation of the Cs phenotype using a YCp50 yeast genomic library.

Immunoprecipitation analysis: Monoclonal antibody 12CA5 was coupled to protein A/G beads (Pierce, Rockford, IL). A total of 60 μl of beads was bound per 10 ml of 12CA5 culture supernatant and crosslinked to dimethylpimelimidate essentially as described (Harlow and Lane 1988). Coupled beads (40 μl) were incubated with 1.25 mg of yeast whole cell extract in 1 mg/ml BSA, 50 mm Tris-HCl (pH 7.6), 0.05% NP-40, 2 mm MgCl2, 100 mm NaCl with protease (1 mm phenylmethylsulfonyl fluoride, 2 mm pepstatin, 0.6 mm leupeptin, 2 mm benzamidine HCl) and phosphatase (0.2 mm Na3VO4, 50 mm KF, 15 mm β-glycerolphosphate) inhibitors in a reaction volume of 2 ml for 2 hr at 4° in the presence or absence of 100 ng/μl RNase A. Beads were then washed batch-wise five times in 500 μl binding buffer (without BSA). Elution was carried out by boiling for 3 min in 60 μl 2× Laemmli buffer without β-mercaptoethanol.

Preparation of extracts and Western blot analysis: Yeast total protein extracts were prepared essentially as described (Patturajanet al. 1998a). Cell pellets were resuspended in an approximately equal volume of buffer A (200 mm Tris-HCl, pH 8.0, 320 mm ammonium sulfate, 5 mm MgCl2, 10 mm EGTA, 20 mm EDTA, 1 mm dithiothreitol, 20% glycerol) with protease inhibitors and phosphatase inhibitors. Acid-washed 425-660 micron beads (Sigma, St. Louis) were added up to the meniscus and vortexed 10 times for 20 sec each with intervals of 30 sec on ice. The lysate was cleared by centrifugation at 10,000 × g for 10 min. Western blot analysis was performed with tissue culture supernatant of mAb 12CA5 (Wilsonet al. 1984) diluted 1:10 in 1× PBS with 0.05% Tween-20, 5.0% dried milk.

Dephosphorylation of yeast protein extracts: Yeast total protein extract (~100 μg) was diluted 10-fold and dialyzed twice for 1.5 hr each against 5 × 103 volumes of λ phosphatase buffer (50 mm Tris-HCl, pH 7.5, 0.1 mm EDTA, 5 mm DTT, 0.01% Brij 35, 10% glycerol). Each extract was divided in half, brought to 2 mm MnCl2, and incubated in the presence or absence of 400 units of λ phosphatase (New England Biolabs, Beverly, MA) for 1 hr at 30°.


Identification of essential Nrd1p domains: Sequence analysis reveals three conserved sequence domains in Nrd1p (Steinmetz and Brow 1996). The N terminus of the protein contains similarity to the CID of pol II CTD-binding proteins called SCAFs (Patturajanet al. 1998b), and has been shown to bind to the CTD by yeast two-hybrid analysis (Yuryevet al. 1996). Also evident are an RRM and a 21-amino acid RE/RS domain, rich in alternating arginine and glutamate or serine dipeptides. To determine which of these domains is necessary for the essential function of Nrd1p, we deleted each one and tested for viability by plasmid shuffle. Strain YJC582 (nrd1Δ, NRD1-HA on a 2μ/URA3 plasmid) was transformed with domain deletion constructs and counterselected on 5-FOA at multiple temperatures (15°, 20°, 25°, 30°, and 37°). Expression of the truncated proteins was verified by Western blot (data not shown).

The results of this analysis are summarized in Figure 1. Deletion of the RRM does not support viability, a result consistent with an essential role for RNA binding (Steinmetz and Brow 1998) by Nrd1p in vivo. The only other lethal deletion removes all sequences C-terminal to the CID, including the RRM.

As previously observed, deletion of the CID is not lethal (Steinmetz and Brow 1998). However, we observed that deletion of this domain confers partial cold- and temperature-sensitive phenotypes as well as an increase in doubling time. Colonies grown on rich or minimal plates are heterogeneous in size, and the proportion of cells that are viable upon counterselection of the wild-type allele is much lower than that of other viable alleles. The nrd1Δ39-169 allele (Steinmetz and Brow 1998) did not show these phenotypes, suggesting that the amino terminal 38 residues may provide partial CID function. The importance of the first 38 amino acid residues is supported by the isolation of a temperature-sensitive mutation in this region (see below). Since nrd1Δ39-169 is not HA-tagged, it remains possible that the HA-tag present on the nrd1-CIDΔ allele compromises Nrd1p function. However, we feel that this is unlikely because the HA-tagged wild-type allele shows no abnormal growth phenotype.

Figure 1.

Schematic representation of results from Nrd1p domain-deletion analysis. Plasmid shuffle assay was used to examine the viability of yeast strains carrying the indicated domain deletion. Domains include the N-terminal CID, a motif rich in arginine/glutamate and arginine/serine dipeptides (RE/RS), an RRM, a C-terminal region rich in proline and glutamine (P/Q), and an HA-tag.

The RE/RS domain is expendable for normal growth. However, while indirect immunofluorescence shows wild-type Nrd1p is nuclear at steady state, deletion of the RE/RS motif leads to increased Nrd1p levels in the cytoplasm (data not shown). Interestingly, similar “leaky” nuclear localization has been observed upon deletion of the RS domains of mammalian SF2/ASF, SRp40, SC35, and SRp20 (Cacereset al. 1998).

Temperature-sensitive alleles of nrd1: Previously, a temperature-sensitive missense mutation was isolated in the Nrd1p RRM, V368G (Steinmetz and Brow 1996). To allow further characterization of functional domains in Nrd1p, additional temperature-sensitive alleles of nrd1 were isolated using PCR mutagenesis and gap repair (Muhlradet al. 1992). Approximately 104 transformants were screened, and 50 candidate strains were isolated based on their inability to grow at 37°. Of these strains, 6 were tested for complementation by pNRD1-HA, and all 6 strains were shown to carry recessive temperature-sensitive nrd1 alleles (nrd1-51, nrd1-52, nrd1-53, nrd1-54, nrd1-60, and nrd1-83, data not shown). For 4 of these alleles, nrd1-51, nrd1-54, nrd1-60, and nrd1-83, the regions with mutations responsible for the temperature-sensitive phenotype were deduced by gap repair with linearized domain deletion constructs as described in materials and methods. Alleles nrd1-51 and nrd1-54 were complemented by the linearized nrd1-CID construct, as well as nrd1-RSΔ and nrd1-RRMΔ, but not by the linearized nrd1-CIDΔ construct, suggesting that these alleles contain mutations in the CID. Similarly, temperature sensitivity of nrd1-60 and nrd1-83 was complemented by every construct tested except for SpeI-digested nrd1-RRMΔ construct, suggesting that RRM mutations confer temperature sensitivity in these alleles. To verify the presence of mutations at the indicated domains, the appropriate domain was sequenced (Figure 2A).

To establish whether the conditional phenotypes of nrd1-51 and nrd1-60 are due to the identified mutations, the sequenced portions of these alleles were subcloned into an otherwise normal NRD1-HA plasmid. The resulting alleles were named nrd1-101 (S16P) and nrd1-102 (V379G), respectively. These particular mutant alleles were chosen because serine 16 is conserved between Nrd1p and SCAF8, and valine 379 is in the predicted ribonucleoprotein (RNP1) motif of the RRM. While there was no phenotypic difference observed between nrd1-60 and nrd1-102, nrd1-101 was more leaky at 37° than nrd1-51 and showed even less temperature sensitivity when plated on galactose plates (Figures 2B and 4B, and data not shown). The presence of multiple mutations in nrd1-51 is indicated by these observations.

Figure 2.

Temperature-sensitive alleles of nrd1 used in this study. (A) Schematic representation of nrd1 temperature-sensitive alleles. Regions outlined with a question mark were not sequenced. Mutations in the CID of nrd1-51 and in the RRM of nrd1-60 were subcloned to otherwise wild-type NRD1-HA to generate nrd1-101 and nrd1-102, respectively. (B) Wild-type and temperature-sensitive strains were grown to saturation, diluted approximately twofold to equal OD600, diluted serially (eightfold), and grown for 48 hr on YPD at indicated temperatures. Growth of strain harboring the nrd1-54 is similar to that of nrd1-101 (data not shown).

Temperature-sensitive alleles of nrd1 are synthetic lethal with truncations of the CTD: Because Nrd1p is similar to the mammalian SCAF8 family of CTD-binding proteins and has a similar organization of domains, we examined potential genetic interactions between NRD1 and the pol II CTD. Strains YJC1111 (NRD1-HA), YJC1115 (nrd1-51), and YJC1119 (nrd1-83) are congenic and have chromosomal disruptions of the RPB1 and NRD1 loci that are covered with plasmids carrying wild-type RPB1 (CEN/URA3) and the indicated nrd1 (CEN/TRP1) allele. Each of these strains was transformed with vector alone or derivatives of RPB1 with 26 (wild-type), 10, or 9 repeats of the CTD heptad consensus sequence on a LEU2-marked CEN plasmid (West and Corden 1995). Transformants were serially diluted and plated onto CSM-Leu 5-FOA plates to counterselect the wild-type RPB1 allele. Most combinations of CTD truncation with nrd1 temperature-sensitive alleles do not support viability (Figure 3), demonstrating a synthetically lethal phenotype. A small percentage of cells survive in the nrd1-51/CTD10 strains. These may be revertants caused by CTD expansion as observed previously (West and Corden 1995). Alternatively, some combination of copy numbers of the plasmid-borne CTD10 and nrd1-51 alleles may suppress the synthetically lethal phenotype allowing growth of these cells.

Figure 3.

Synthetic lethality between nrd1 temperature-sensitive mutations and CTD truncations. Strains YJC1111 (NRD1-HA, top), YJC1115 (nrd1-51, bottom right), and YJC1119 (nrd1-83, bottom left) were transformed with vector alone or versions of RPB1 with 26 (WT CTD), 10 (CTD10), or 9 (CTD9) repeats of the CTD heptad consensus sequence on a LEU2-marked plasmid. Transformants were grown to saturation in CSM-Leu-Ura media, diluted serially (eightfold), and plated onto 5-FOA plates at the indicated temperatures. Strain YJC1119 was plated at 25° because nrd1-83 strains grow poorly at 30°.

NAB3 is an allele-specific high-copy suppressor of nrd1-51: A high-copy suppressor screen of nrd1-51 was performed using a 2μ/URA3 genomic library with an average insert size of 7 kb (Connelly and Hieter 1996). Eight suppressor strains were isolated and high-copy plasmids were recovered. One strain contained a NRD1 plasmid, and the remaining seven had one of two different plasmids (pSUP5 or pSUP8). Partial sequences indicated that both of these plasmids carry a copy of NAB3, a yeast hnRNP gene (Wilsonet al. 1994). To verify that Nab3p overexpression suppressed the nrd1-51 allele, YJC818 was transformed with vector alone or pNAB3.18, which has the NAB3 gene under the control of the GAL1 promoter. As expected, pNAB3.18 suppressed the nrd1-51 temperature-sensitive phenotype on galactose plates, but not on glucose plates (Figure 4A).

Since nrd1-51 most likely contains mutation(s) other than S16P, we wanted to verify that Nab3p overexpression suppresses the S16P mutation. However, we could not achieve this using pNAB3.18 because the nrd1-101 allele is leaky on galactose media. Therefore, we used pSUP8 to examine suppression of nrd1-101. Yeast strains carrying the wild-type, nrd1-51, or nrd1-101 alleles were transformed with either empty vector or pSUP8. The growth of the pSUP8-transformed nrd1-101 strain was indistinguishable from wild type at 37° (Figure 4B). Interestingly, pSUP8 also suppressed the temperature-sensitive phenotype of nrd1-54, which, like nrd1-101, has a mutation that alters a serine residue (S116R) in the CID and is leaky on galactose (data not shown). Alleles with mutations mapping outside the CID were not suppressed by Nab3p overexpression (nrd1-60, nrd1-83, and nrd1-102; data not shown). The allele specificity of Nab3p overexpression is consistent with a model in which the CID of Nrd1p has a function that is distinct from the rest of the protein and is linked to the yeast hnRNP, Nab3p. Further high-copy screens were performed on nrd1-53 and nrd1-60, but no high-copy suppressors were found (NRD1 was recovered 10 and 4 times, respectively).

Two-hybrid interactions between Nrd1p and Nab3p: A yeast two-hybrid screen was conducted to identify proteins that interact with Nrd1p in vivo, using the improved reporter strain and libraries of James et al. (1996). A “bait” construct encoding the first 560 residues of Nrd1p fused to the GBD was used to select Nrd1p interaction partners from a library of yeast genomic fragments expressed as fusions to the GAD. From ~106 transformants subjected to this selection, 13 that showed Nrd1p bait-dependent growth on −his and −ade were analyzed. Two transformants exhibiting robust growth on −his and −ade media were found to contain in-frame GAD fusions to residues 99-766 of Nab3p or to the entirety of Nab3p C-terminal to residue 35, respectively. None of the other 11 positive clones contains an in-frame fusion.

A series of bait fusions expressing different fragments of Nrd1p was used to show that the region between residues 169 and 265 containing the RE/RS domain (residues 245–265) is necessary and sufficient for productive two-hybrid interactions with the GAD-Nab3 fusion (Figure 5). The RE/RS sequence itself, however, is not required for interaction with Nab3p, since the deletion of this sequence from the GBD-Nrd1p(1–560) fusion still allows for productive two-hybrid interactions with GAD-Nab3p. These results define a new functional domain of Nrd1p: a Nab3p-interaction domain located between residues 169 and 244 and situated between the CID and RE/RS domains.

Nrd1p and Nab3p co-immunoprecipitate: To further characterize the in vivo Nrd1p and Nab3p interaction, we performed coimmunoprecipitation experiments. Extracts were made from strains YJC610 (NRD1-HA) and YJC976 (NRD1) and were immunoprecipitated with the anti-HA mAb 12CA5 in the presence or absence of RNase. Proteins in the immunocomplex were separated by gel electrophoresis, were Western blotted, and were probed for the presence of Nab3p with the mAb 2F12 (Figure 6). A small fraction of the Nab3p is immunoprecipitated specifically and in an RNA-independent fashion with Nrd1-HAp.

Figure 4.

Nab3p overexpression suppresses nrd1-51 and nrd1-101. (A) Strains YJC582 (Nrd1-HA, 2μ/URA3) and YJC818 (nrd1-51) transformed with vector alone or pNAB3.18, were grown to saturation, diluted 10-fold, and streaked onto CSM-Ura plates with either glucose or galactose. (B) Strains YJC610 (NRD1-HA), YJC818 (nrd1-51), or YJC1103 (nrd1-101) were transformed with vector alone or pSUP8, a 2μ URA3-marked plasmid with a NAB3 genomic fragment. Transformants were grown to saturation in CSM-Ura media, diluted serially (8-fold), and plated onto CSM-Ura. Plates were incubated for 48 hr at 25 or 37° as indicated.

Characterization of nab3 mutants: To further examine the role of Nab3p in vivo, we generated several temperature-sensitive nab3 strains. The resulting nab3-1, nab3-3, and nab3-4 alleles were sequenced completely and were shown to contain multiple missense mutations (Figure 7A). While the growth properties of these nab3 strains were similar to a wild-type strain between 14° and 30°, nab3-1 was growth restricted at 36°, while nab3-3 and nab3-4 were inhibited at 37°. Comparison of the missense mutations in these nab3 strains suggested that RRM mutations conferred temperature sensitivity. This was confirmed in the case of nab3-3 since the growth characteristics of the nab3-10 strain, which has a single RRM mutation (Figure 7A), and nab3-3 are similar at all temperatures tested. The nab3-9 mutant, which was isolated during a nab3-3 synthetic lethal screen (see materials and methods), contains an additional RRM mutation (P374L) and was inhibited for growth at 36° (Figure 7). The two RRM mutations (F371L and P374L) are also sufficient to confer temperature sensitivity since the growth properties of nab3-11, which contains only the two RRM mutations, and nab3-9 are similar.

Extragenic suppression of nab3-9 by CTK1 kinase domain mutation: To elucidate the essential function of Nab3p, we isolated extragenic suppressors of nab3-9. Over 800 colonies that grew at 30° in 3–4 days were selected, restreaked, and tested for cold sensitivity at 14°. Of the 30 cold-sensitive strains that were identified by this analysis, only 5 were able to produce viable spores for cosegregation analysis, falling into two suppressor of nab3-3 (snb3) complementation groups. A representative from each group was used for complementation of the cold-sensitive phenotype using a YCp50 genomic library, and the complementing plasmids were recovered and sequenced. The snb32-1 strain was complemented by an 8-kb chromosome XI fragment which encodes MRP8, SDH3, TGL1, and CTK1. Subclones containing individual genes were tested for snb32-1 complementation. Only CTK1, encoding the RNA pol II carboxyl-terminal repeat domain kinase (Lee and Greenleaf 1991), was able to restore growth at 14°. To confirm that CTK1 was allelic to snb32-1, CTK1 was integrated at the ctk1 locus in the snb32-1 strain, and the resulting transformant was mated to a nab3-9 strain. Sporulation and dissection of the resulting diploid did not yield cold-sensitive progeny. In addition, the ctk1 mutant allele was recovered by gap repair and sequenced. A single missense mutation was discovered within the conserved kinase domain (K211E), suggesting that loss of Ctk1p kinase activity was responsible for nab3-9 suppression. Consistent with this possibility, disruption of the CTK1 gene (ctk1Δ) in the nab3-9 strain YSW911 suppressed the temperature-sensitive defect (Figure 7B). Similar analysis showed that the remaining complementation group (snb31) was allelic to the adenylate cyclase gene CYR1.

Figure 5.

Two-hybrid interactions between Nrd1p and Nab3p. (A) Schematic illustration of Nrd1p fragments fused to GBD and tested for interactions with Nab3p fused to GAD. Growth on −His and −Ade plates is indicated by +, while − indicates no growth on either −his or −ade. (B) Plate showing growth of PJ69-4a expressing the indicated combinations of GBD-Nrd1p and GAD-Nab3 fusions on −ade medium. The GAD-Nab3p fusion in each case includes residues 99–766 of Nab3p.

Since 25 of the 30 cold-sensitive strains did not produce viable spores for segregation analysis, these suppressor strains were mated to YSW901, and the diploids were plated at 30° to determine if suppression was dominant or recessive. This analysis indicated that 14 of these strains were recessive for suppression, and these cells were transformed with the YCp50 genomic library to identify genes that complement the cold-sensitive phenotype. Of the 14 recessive suppressor strains, 7 showed plasmid linkage for growth at 14° and loss of growth at 30°. Further analysis of these plasmids demonstrated that the gene responsible for suppression was CTK1. Therefore, out of 12 suppressor alleles identified, 8 encoded ctk1 mutant alleles. Subsequent analysis of NAB3, nab3-9, or ctk1Δ/nab3-9 cells failed to detect significant differences in the expression levels or phosphorylation status of Nab3p (data not shown).

Figure 6.

Nab3p coimmunoprecipitates with Nrd1p. Immunoprecipitation with mAb 12CA5 (anti-HA) was performed using extracts generated from log phase (OD600 = 1.0) cells expressing Nrd1p with or without the HA-tag, as indicated. The entire pellet and 5% of the load was run on a 7.5% SDS-PAGE and Western blotted using mAb 2F12 (anti-Nab3) as a probe.

The phosphorylation state of nrd1-51 is changed by Nab3p overexpression: Nrd1p migrates as multiple bands in Western blot analysis (Steinmetz and Brow 1998; and Figure 8). To test whether the different bands represent phosphoisomers of Nrd1p, we treated whole cell extracts with λ phosphatase and analyzed the resulting treated (or mock-treated) extracts by Western blotting with the HA-tag specific mAb 12CA5. Phosphatase treatment of the extracts revealed the presence of a third, faster migrating band on SDS-PAGE (Figure 8A, lanes 1 and 2), suggesting that, in vivo, Nrd1p exists in two different phosphoisoforms.

The consistent presence of both phosphoisoforms of Nrd1p from cells harvested during log phase implies a potential role for phosphorylation in the regulation of Nrd1p. Strengthening that idea, we noted that upon overexpression of Nab3p in the nrd1-51 strain, YJC818, only the top band is observed (Figure 8B). Similar results are seen with wild-type Nrd1-HAp (data not shown). Since treatment with λ phosphatase results in a band that migrates indistinguishably from the phosphatase-treated wild-type or vector-transformed YJC818 (Figure 8A, lanes 3 through 6), the change in migration is likely due to a change in phosphorylation. Deletion of CTK1 does not change the migration pattern of Nrd1p on SDS-PAGE (data not shown).

Figure 7.

nab3 temperature-sensitive alleles. (A) Schematic representation of nab3 temperature-sensitive alleles. Nab3p structural motifs include the amino terminal region rich in aspartic and glutamic acid residues (D/E), the RRM, and the carboxyl-terminal region rich in proline and glutamine residues (P/Q). All nab3 mutant alleles were completely sequenced, and the missense mutations in each allele are indicated. (B) Spot plate growth analysis of nab3, and nab3-9 suppressor, strains at 24°, 36°, and 37°.


The steady-state level of a given mRNA is determined by the relative rates of synthesis and degradation. Synthesis is a multistep process that involves transcription and processing, followed by export of mature RNA from the nucleus. Evidence has been mounting to support a model in which these steps are coupled in vivo. hnRNPs are likely to play roles in many of these processes (Siomi and Dreyfuss 1997; Krecic and Swanson 1999). In yeast, for example, Nab4p/Hrp1p has been shown to be involved in 3′ processing both in vitro and in vivo (Kessleret al. 1997; Minvielle-Sebastiaet al. 1998). Npl3p/Nab1p is genetically linked to 3′ processing factors Rna15p and Nab4p/Hrp1p and is involved in RNA export from the nucleus (Henryet al. 1996; Leeet al. 1996). The functions of other yeast hnRNPs remain to be elucidated.

Figure 8.

The phosphorylation state of Nrd1p is altered in response to Nab3p overexpression. (A) Total protein extracts were prepared from YJC610 grown in YPD and YJC818 (transformed with either vector alone or pNAB3.18) grown in CSM-Ura galactose media. Extracts were treated with λ phosphatase (+) or mock-treated (−) as indicated and run on a 10% SDS-PAGE. (B) YJC818-derived extracts described in A are shown on a 7.5% SDS-PAGE. Two independent transformants for pNAB3.18-transformed strains are shown. Extracts shown in lanes 1 and 2 were used for λ phosphatase reactions shown in lanes 3–6 of A. Western analysis was performed using mAb 12CA5.

Yeast NRD1 was first cloned based on its ability to downregulate the level of pre-mRNA transcribed from an artificial, intron-containing gene (Steinmetz and Brow 1996). While the data are inconclusive as to whether the decrease in steady-state pre-mRNA is due to a decrease in RNA synthesis or an increase in degradation, the observation of 3′ truncated products suggests that Nrd1p could be involved in premature transcriptional termination, inefficient transcriptional elongation, or degradation via a 3′ to 5′ exonuclease activity (Steinmetz 1997). The genetic interactions presented in this article support a model in which Nrd1p acts at the interface of transcription and processing without ruling out a potential role in degradation. A summary of known interactions, genetic and biochemical, is outlined in Figure 9A.

Figure 9.

(A) Schematic overview of the interactions described in this article and elsewhere. (B) A model for Nrd1p-Nab3p function. See text for details.

NRD1-CTD interaction: The similarity of Nrd1p to mammalian SCAF8 suggested initially that Nrd1p could interact with the CTD. This hypothesis has been supported by additional two-hybrid data. First, the Nrd1p CID was shown to interact with the mouse CTD (Yuryevet al. 1996). Second, in a genome-wide two-hybrid screen similar to the one described here, a yeast Nrd1p-interacting clone was isolated that encodes the Gal4 activation domain fused to the C terminus of Rpb1p. The clone includes the entire CTD as well as ~120 amino acids N-terminal to the CTD (N. K. Conrad and J. L. Corden, unpublished observations).

We have also used a genetic approach to demonstrate an interaction between Nrd1 and the CTD. Previous studies showed that many of the yeast CTD heptapeptide repeats can be deleted without phenotypic consequences (Nonetet al. 1987; West and Corden 1995). Reducing the CTD length to 8–11 repeats yields cold- and temperature-sensitive growth phenotypes (Nonetet al. 1987; West and Corden 1995) and results in altered expression of a subset of genes (Scafeet al. 1990; Meiselset al. 1995). In the current study, we show that truncation of the CTD to 9 or 10 repeats is synthetically lethal with two different temperature-sensitive alleles of nrd1. This result is consistent with previous two-hybrid and direct binding experiments with SCAF8. Together they indicate that Nrd1p and pol II function in the same pathway in vivo, most likely through direct interaction.

NRD1-NAB3 interaction: We have also demonstrated genetic and physical interactions between RNA-binding proteins Nrd1p and Nab3p. Similarities in sequence organization suggest that Nrd1p and Nab3p may play related roles in mRNA biogenesis. Both genes are essential and encode RNA-binding proteins with a single RRM and a carboxyl-terminal proline/glutamine-rich region (Wilsonet al. 1994; Steinmetz and Brow 1996). Nrd1p and Nab3p differ in that Nrd1p has RE/RS and CTD-interacting domains N-terminal to the RRM, while the corresponding region of Nab3p is highly acidic. Nrd1p is most closely related to mammalian SCAF8 (Corden and Patturajan 1997), while Nab3p is most closely related to the hnRNP C family of mammalian RNA-binding proteins (Wilsonet al. 1994).

Three lines of evidence suggest a direct interaction between Nrd1p and Nab3p in vivo. First, NAB3 is an allele-specific high-copy suppressor of the nrd1 CID mutations. Second, Nrd1p and Nab3p interact in a yeast two-hybrid assay. Third, Nrd1p and Nab3p coimmunoprecipitate. The interaction is not mediated by RNA since the two-hybrid interaction is independent of the Nrd1p RRM and extensive RNase treatment did not affect coimmunoprecipitation. The observation that only a fraction of the Nab3p coimmunoprecipitated with Nrd1p may indicate that the interaction is transient or is important for only a small subset of transcripts that contain both Nrd1p- and Nab3p-binding cis-elements. Of course, experimental conditions may also affect the efficiency of coimmunoprecipitation. For example, the proteins may dissociate upon lysis or antibody binding may destabilize the interaction. Unfortunately, the coimmunoprecipitation of Nrd1p with anti-Nab3p mAb is technically not feasible because only a small amount of Nab3p immunoprecipitated with mAb 2F12.

Both Nrd1p and Nab3p are hnRNPs that have been shown to play a role in mRNA accumulation. Nrd1p is thought to act early in transcription to mediate the downregulation of transcripts containing negative cis-elements (Steinmetz and Brow 1996, 1998; Steinmetz 1997). Nab3p overexpression has been shown to reduce the steady-state level of CLN3 mRNA (Sugimotoet al. 1995). In addition, Nab3p has been implicated in the processing of several intron-containing pre-mRNAs including CYH2 and ACT1 (Wilsonet al. 1994). However, neither nrd1 nor nab3 temperature-sensitive strains show an increase in the ratio of unspliced to spliced ACT1 or CYH2 RNAs upon shift to the nonpermissive temperature (data not shown). Both Nrd1p and Nab3p could act at the level of transcription elongation and/or termination. Alternatively, these proteins may be involved in turnover of nuclear pre-mRNA.

The synthetic lethality of nrd1 temperature-sensitive alleles with CTD-truncation mutations indicates that interaction of Nrd1p with the CTD serves an important function, yet the Nrd1p CID is dispensible for viability. The CID may provide only one of multiple redundant mechanisms to ensure interactions between Nrd1p and pol II. The allele-specific high-copy suppression of temperature-sensitive nrd1 CID mutant alleles by NAB3 suggests that Nab3p may provide another mechanism to facilitate Nrd1p-CTD interactions. If Nrd1p interacts with Nab3p and the CTD simultaneously, Nab3p may be in close proximity to the CTD and stabilize the Nrd1p-CTD interactions. This would also provide a framework within which to understand the genetic interactions between CTK1 and NAB3 reported here.

Interactions between CTK1, NRD1, and NAB3: Protein phosphorylation is a common thread among the interactions we describe in this article. The pol II CTD, Nrd1p, and Nab3p are all phosphorylated in vivo (Dahmus 1996; Figure 8, and data not shown). The identification of a CTD kinase as a suppressor of nab3-9 provides an additional link from NAB3 to the CTD (Figure 9A). Precisely how phosphorylation may regulate the function of the CTD, Nrd1p, and Nab3p is not known, but several clues are evident in the results presented here.

CTDK-I is a protein kinase identified initially by its ability to phosphorylate the CTD of pol II in vitro (Lee and Greenleaf 1989). This CTD kinase is comprised of three subunits, a cdc2-related catalytic subunit (Ctk1p), a cyclin-related subunit (Ctk2p), and a third subunit of unknown function (Ctk3p) (Lee and Greenleaf 1991; Sterneret al. 1995). Deletion of any of the CTDK-I subunit genes is not lethal, but the null mutants display slow growth and cold-sensitive phenotypes (Sterneret al. 1995). CTDK-I is not a part of the pol II holoenzyme (Myer and Young 1998), but rather is thought to phosphorylate the newly initiated pol II. Furthermore, CTDK-I stimulates transcript elongation in Hela nuclear extracts (Lee and Greenleaf 1997).

The genetic evidence presented here indicates that CTDK-I may play a role in negatively regulating the functions of Nrd1p and Nab3p. Deletion of CTK1 suppresses the temperature-sensitive phenotype of nab3-9. Furthermore, overexpression of Nab3p severely inhibits the growth of a ctk1Δ strain, and nrd1-102 is a weak suppressor of the cold-sensitive phenotype of ctk1Δ (data not shown). All of these results are consistent with opposing effects of Ctk1p and Nrd1p-Nab3p. Whether the interactions between Nrd1p-Nab3p and Ctk1p are direct or indirect cannot be determined from our experiments. It is possible, but not likely, that these proteins are regulated directly by CTDK-I phosphorylation. Deletion of CTK1 did not significantly change the level of radioactive phosphate labeling of Nab3p in vivo (data not shown). Similarly, the ratio of phosphoisomers of Nrd1p is not changed in a ctk1Δ strain as judged by mobility in SDS-PAGE (data not shown). However, both of these assays would likely fail to detect a change in a single phosphorylation site of a multiply phosphorylated protein. Ctk1p is more likely to regulate the function of Nrd1p and Nab3p in an indirect manner. For example, changes in the phosphorylation pattern of the CTD may alter the interaction of Nrd1p and Nab3p with the transcription apparatus.

A possible mechanism for the control of RNA synthesis by the Nrd1p-Nab3p complex: Both NAB3 and NRD1 were isolated in independent genetic screens as suppressors of (pre-)mRNA accumulation phenotypes (Sugimotoet al. 1995; Steinmetz and Brow 1996). Physical and genetic interactions among Nrd1p, Nab3p, and RNA pol II described here suggest a potential mechanism by which Nrd1p and Nab3p function together at the interface between transcription and pre-mRNA processing. In the generalized model proposed here, Nrd1p and Nab3p, perhaps in association with other hnRNPs or RNA-processing factors, form a complex that binds to specific cis-acting sites in nascent transcripts. The Nrd1p-Nab3p complex then interacts with pol II, leading to pausing or termination. Alternatively, the Nrd1p-Nab3p complex may be bound to the elongating polymerase early in the transcription cycle so that it can monitor the nascent transcript. The interaction of Nrd1p and Nab3p in a complex may allow a greater range of specificity and affinity for RNA targets than could be achieved by either protein alone, enabling a form of combinatorial control at the level of transcript elongation. Genetic evidence indicates that CTDK-I functions to oppose the activity of the Nrd1p-Nab3p complex. CTDK-I may alleviate blocks to transcript elongation by directly dissociating the Nrd1p-Nab3p RNA complex, or by disrupting the pol II-Nrd1p-Nab3p complex by changing the CTD phosphorylation pattern. This model is summarized in Figure 9B.

Previous studies showing that truncated products of the nrd1-1 and nrd1-2 nonsense mutant alleles can interfere with pre-mRNA downregulation by full-length Nrd1p (Steinmetz and Brow 1998) are consistent with this model. It was proposed that these truncated products, predicted to include amino acid residues 1–241 and 1–285, respectively, compete with full-length Nrd1p for binding to the CTD. It is now evident that these truncated products also contain the Nab3p interaction domain, and thus may also compete with intact Nrd1p for Nab3p interactions. A role for Nab3p in Nrd1p-dependent downregulation of pre-mRNA containing the U6R* RNA element could account for the observation that additional sequences outside of the 43-nt segment of U6R* shown to bind to Nrd1p in vitro are required for downregulation in vivo. These additional sequences may contain elements that interact productively with Nab3p, and the combined action of Nrd1p and Nab3p bound to RNA ligands may be required for downregulation.

Recently, mammalian complexes have been described that negatively affect transcript elongation in vitro. These negative factors include the negative elongation factor NELF (Yamaguchiet al. 1999) and the DRB sensitivity-inducing factor (DSIF; Hartzoget al. 1998; Wada et al. 1998a,b). The only known subunit of NELF is RD, a 43.2-kD protein, which, like Nrd1p, contains a single RRM and a motif rich in alternating positive/negative dipeptides (RD). DSIF is comprised of homologs of yeast Spt4p and Spt5p that interact with each other and pol II (Hartzoget al. 1998; Wadaet al. 1998a). Interestingly, the affects of DSIF and NELF can be counteracted by P-TEFb, a kinase/cyclin pair (Cdk9/cyclin T) that is closely related to CTDK-I (Zhuet al. 1997; Penget al. 1998b; Weiet al. 1998). CTDK-I and P-TEFb share amino acid homology in their kinase subunits, and both phosphorylate the CTD and stimulate elongation in vitro (Sterneret al. 1995; Lee and Greenleaf 1997; Zhuet al. 1997; Peng et al. 1998a,b; Weiet al. 1998). While the exact mechanism of NELF/DSIF and P-TEFb remains to be elucidated, the model we are proposing for Nrd1p-Nab3p is similar to that described for DSIF/NELF (Penget al. 1998a; Wadaet al. 1998b; Yamaguchiet al. 1999).

While it is unclear whether or not NELF is gene specific, in our model, Nrd1p-Nab3p acts as a gene-specific regulator of transcription. The RNA-binding specificity of Nrd1p has been demonstrated in vitro (Steinmetz and Brow 1998). The binding specificity of Nab3p is not known, but nab3 mutants have been shown to have gene-specific expression defects (Wilsonet al. 1994; Sugimotoet al. 1995). How many genes are regulated by Nrd1p-Nab3p is also not known, but neither Nrd1p nor Nab3p is abundant, suggesting a small number of essential targets.

To date, the functions attributed to yeast hnRNPs have included involvement in 3′ processing, mRNA transport, and regulation of the degradation of mRNA. Here we have demonstrated genetic and physical interactions between nuclear pre-mRNA-binding proteins and the CTD of pol II. The present work represents a starting point to examine a potential role for yeast hnRNPs in the regulation of the synthesis of pol II transcripts.


We thank Debbie Kuehn (Johns Hopkins) for assistance in the early stages of this work, Priscilla Non (University of Florida) for several yeast strains and expert technical assistance, and Phil James (University of Wisconsin) for supplying the strain, plasmids, and libraries for the yeast two-hybrid experiment. This work was supported by grants from the National Science Foundation (J.L.C., M.C.B. 9723965) and National Institutes of Health (M.S.S., GM46272; D.A.B., GM44665).


  • Communicating editor: F. Winston

  • Received July 23, 1999.
  • Accepted October 2, 1999.


View Abstract