The U6 small nuclear ribonucleoprotein is a critical component of the eukaryotic spliceosome. The first protein that binds the U6 snRNA is the La protein, an abundant phosphoprotein that binds the 3′ end of many nascent small RNAs. A complex of seven Sm-like proteins, Lsm2-Lsm8, also binds the 3′ end of U6 snRNA. A mutation within the Sm motif of Lsm8p causes Saccharomyces cerevisiae cells to require the La protein Lhp1p to stabilize nascent U6 snRNA. Here we describe functional interactions between Lhp1p, the Lsm proteins, and U6 snRNA. LSM2 and LSM4, but not other LSM genes, act as allele-specific, low-copy suppressors of mutations in Lsm8p. Overexpression of LSM2 in the lsm8 mutant strain increases the levels of both Lsm8p and U6 snRNPs. In the presence of extra U6 snRNA genes, LSM8 becomes dispensable for growth, suggesting that the only essential function of LSM8 is in U6 RNA biogenesis or function. Furthermore, deletions of LSM5, LSM6, or LSM7 cause LHP1 to become required for growth. Our experiments are consistent with a model in which Lsm2p and Lsm4p contact Lsm8p in the Lsm2-Lsm8 ring and suggest that Lhp1p acts redundantly with the entire Lsm2-Lsm8 complex to stabilize nascent U6 snRNA.
THE process of pre-mRNA splicing requires five small ribonucleoprotein particles, the U1, U2, U4, U5, and U6 snRNPs. These small RNPs associate with each other, with the mRNA, and with a large number of splicing factors to form the spliceosome (reviewed by Burgeet al. 1998). The U1, U2, U4, and U5 snRNPs each consist of an RNA molecule, seven common proteins known as Sm proteins, and additional snRNP-specific proteins. In vertebrates, binding of the Sm proteins to the snRNA occurs in the cytoplasm and is required for hypermethylation of the 5′ cap structure and reimport of the snRNPs into the nucleus (Kambachet al. 1999a). All Sm proteins share a conserved Sm motif consisting of two short submotifs, Sm1 and Sm2. The crystal structures of two Sm heterodimers have revealed that the Sm motif folds into an N-terminal α-helix, followed by a five-stranded antiparallel β sheet (Kambachet al. 1999b). From these structures, a model has been proposed in which the seven Sm proteins interact via their Sm motifs to form a heptameric ring around the snRNA (Kambachet al. 1999b).
In addition to the canonical Sm proteins, other Sm motif-containing proteins have been identified in many eukaryotes and certain archaebacteria (Cooperet al. 1995; Hermannet al. 1995; Seraphin 1995; Fromont-Racineet al. 1997; Achselet al. 1999; Mayeset al. 1999; Salgado-Garridoet al. 1999). In yeast, there are nine of these Sm-like proteins (named Lsm1-Lsm9 for Like Sm). Two distinct heptameric complexes of these Sm-like proteins have been described. One complex, consisting of the Lsm2-Lsm8 proteins, binds to the 3′ end of the U6 snRNA and is required for the stable accumulation of U6 snRNPs (Pannoneet al. 1998; Achselet al. 1999; Mayeset al. 1999; Salgado-Garridoet al. 1999; Vidalet al. 1999). A second complex, consisting of the Lsm1-Lsm7 proteins, functions in mRNA degradation, most likely at the decapping step (Boecket al. 1998; Bouveretet al. 2000; Tharunet al. 2000). The Lsm2-Lsm8 complex was recently purified from human cells and found to resemble a doughnut, similar in size and shape to the core Sm snRNPs (Achselet al. 1999). This finding, coupled with the fact that each of the Lsm2-Lsm8 proteins can be specifically aligned with one of the bona fide Sm proteins (Fromont-Racineet al. 1997; Salgado-Garridoet al. 1999), suggests that the Lsm proteins assemble into analogous heptameric ring structures. The order of the Lsm2-Lsm8 proteins within the ring is unknown.
Although the Lsm2-Lsm8 complex binds the 3′ end of U6 RNA and is required for the stable accumulation of U6 snRNPs, the precise role of this complex in U6 biogenesis and function is unknown. The first protein known to bind U6 RNA is the La protein, an abundant nuclear phosphoprotein that binds and stabilizes many newly synthesized small RNAs (Rinke and Steitz 1982, 1985; Pannoneet al. 1998; Kufelet al. 2000; Xueet al. 2000). The La protein binds the UUUOH at the 3′ end of these RNAs (Stefano 1984), which, for U6 RNA, overlaps the binding site of the Lsm2-Lsm8 complex (Achselet al. 1999; Vidalet al. 1999). Thus, binding by the Lsm2-Lsm8 complex to the 3′ end of U6 RNA may displace the La protein. Consistent with this idea, yeast cells containing a mutation in LSM8 require Lhp1p, the yeast La protein homologue, to stabilize newly synthesized U6 RNA (Pannoneet al. 1998). However, it is unclear whether mutations in other Lsm proteins cause cells to require Lhp1p.
To better understand the functions of individual members of the Lsm2-Lsm8 complex, we examined the genetic interactions between Lhp1p, U6 RNA, and the Lsm proteins in the yeast Saccharomyces cerevisiae. We report that LSM2 and LSM4 function as allele-specific suppressors of two mutations in LSM8, consistent with the speculation (He and Parker 2000; Pannone and Wolin 2000) that Lsm2p and Lsm4p directly contact Lsm8p in the Lsm2-Lsm8 ring. Consistent with a direct interaction, overexpression of LSM2 in the lsm8-1 mutant strain increases the levels of the mutant Lsm8 protein. In the presence of high levels of U6 RNA, the normally essential LSM8 becomes dispensable for growth, indicating that the only essential function of LSM8 is in U6 RNA biogenesis and/or stability. Furthermore, deletions of LSM5, LSM6, or LSM7, but not LSM1, are synthetically lethal with deletions of LHP1. Our experiments support the idea (Achselet al. 1999) that Lhp1p acts redundantly with the assembled Lsm2-Lsm8 complex to stabilize newly synthesized U6 RNA.
MATERIALS AND METHODS
Yeast media, strains, and plasmids: Yeast media and manipulations were as described in Sherman (1991). The strains used in this study are listed in Table 1. The lsm1Δ strain (Diezet al. 2000) was a gift of P. Ahlquist (University of Wisconsin).
Synthetic lethal screen: The synthetic lethal screen was performed as described by Pannone et al. (1998) with the following modifications. CY2 cells carrying pATL, a centromeric plasmid containing LHP1, TRP1, and ADE2 (Pannoneet al. 1998), were mutagenized with ethylmethane sulfonate to 15% survival and plated on synthetic complete media containing limiting amounts of adenine (SCiade). Colonies were screened at 25° for the inability to lose pATL and form red sectors. Of 185,000 colonies screened, 68 did not form red sectors. These candidates were transformed with a second plasmid, pSLL28 (Yoo and Wolin 1997), which contains LHP1, URA3, and LYS2, and tested for the ability to lose pATL. Forty-nine candidates were able to form sectoring colonies. Of the 49 that sectored, 15 were dead on media containing 1 μg/ml 5-fluoroorotic acid, indicating they could not lose pSLL28. Backcrossing to CY4 revealed that five strains carried a single recessive mutation that caused them to require pATL. Complementation analysis with BP2 demonstrated that one strain failed to complement the lsm8-1 mutation. This strain was backcrossed three times to CY2 to generate BP26a and BP26α. The characterization of the other strains will be presented elsewhere (S. D. Kim and S. L. Wolin, unpublished results).
Subcloning LSM genes into pRS316: Each LSM gene was subcloned into pRS316 (CEN, URA3; Sikorski and Hieter 1989) as described below. Plasmids carrying the LSM3, LSM5, LSM6, and LSM7 genes fused to two IgG-binding domains of Staphylococcus aureus protein A (PrA) were gifts of B. Seraphin (Seraphin 1995; Salgado-Garridoet al. 1999). The plasmid pBS959 was digested with BamHI and HindIII and the LSM5-PrA-containing fragment was subcloned into the same sites of pRS316 to create pLSM5-PrA. The plasmid pBS1296 was digested with BamHI and SacI and the fragment containing LSM6-PrA was ligated into the BamHI-SacI sites of pRS316 to create pLSM6-PrA. The plasmid pBS957 was digested with BamHI and HindIII and the fragment containing LSM7-ProtA was subcloned into the BamHI-HindIII sites of pRS316 to create pLSM7-PrA. The plasmid pBS867 contains LSM3-PrA in vector pRS316. To create untagged versions, PCR was used to amplify the LSM genes from plasmids pLSM5-PrA, pLSM6-PrA, pLSM7-PrA, and pBS867. In each case, the oligonucleotide used to amplify the 3′ portion of each LSM gene introduced a stop codon at the end of the LSM coding sequence. The resulting DNAs were cloned into the BamHI-EcoRI sites of pRS316 to create pLSM3, pLSM5, pLSM6, and pLSM7. Plasmid pSDB23-1 (Cooperet al. 1995; a gift from J. Beggs) was digested with HindIII and the 1.4-kb fragment containing LSM4 was subcloned into the HindIII site in pRS316 to create pLSM4. Plasmid pJD09 (a gift from P. Ahlquist, University of Wisconsin) was digested with HindIII and EcoRI and the 1.5-kb LSM1 fragment was subcloned into the HindIII-EcoRI sites of pRS316 to create pLSM1. Plasmids p22U and pSNPU (Pannoneet al. 1998) contain the LSM8 and LSM2 genes, respectively, in pRS316. pLSM1, pSNPU, pBS867, pLSM4, pLSM5, pLSM6, pLSM7, p22U, pRS426-SNR6 (a gift from D. Brow, University of Wisconsin), and pRS316 were transformed individually into strains BP5 and BP8.
Anti-Lsm8p antibody synthesis: Anti-Lsm8p peptide antibodies were generated by AnaSpec. A C-terminal Lsm8p peptide (H-Cys-Asn-Lys-Ile-Glu-Asn-Glu-His-Val-Ile-Trp-Glu-Lys-Val-Tyr-Glu-Ser-Lys-Thr-Lys-OH) was conjugated to BSA and injected into rabbits. The antisera was affinity purified against the peptide following the fifth bleed.
Native gels, Northern blots, and oligonucleotides: Extracts were prepared by vortexing 5 OD600 units of cells in 4 volumes of buffer A (50 mm Tris HCl pH 7.5, 25 mm NaCl, 5 mm MgCl2) with 0.25 μm phenylmethylsulfonyl fluoride in the presence of glass beads. The extract was sedimented at 100,000 × g in a Beckman TLA100 rotor for 20 min. A total of 0.1 OD260 units of each supernatant was mixed with an equal volume of buffer A containing 8% glycerol and loaded on a 4% polyacrylamide gel (80:1 acrylamide:bis) in 25 mm Tris, 25 mm boric acid, and 1 mm EDTA that had been prerun at 250 V for 30 min at 4°. Gels were run at 300 V until the bromophenol blue dye reached the bottom. RNA was transferred to ZetaProbe GT membranes (Bio-Rad) in 0.5× TBE at 150 mA for 16 hr. Hybridization with [γ-32P]ATP-labeled oligonucleotides was done as described by Tarn et al. (1995). Oligonucleotides used were U4: 5′-AGGTATTCCAAAAATTCCCTAC-3′ and U6D2: 5′-CGAAATAAATCTCTTTGTAAAACGG-3′.
Disruption of the LSM5, LSM6, and LSM7 genes: To disrupt LSM5, the HIS3 gene from pRS313 (Sikorski and Hieter 1989) was amplified using forward and reverse primers that contained 45 and 49 nucleotides (nt) of LSM5 sequence at their 5′ ends, respectively. This PCR product was transformed into CYY0, where it replaced one allele of the LSM5 open reading frame with HIS3, leaving only 11 nt of LSM5 coding sequence at the 5′ terminus and 13 nt at the 3′ terminus. This generated strain DNY8. To disrupt LSM6, the URA3 gene from pRS316 (Sikorski and Hieter 1989) was amplified using forward and reverse primers that contained 39 and 53 nt of LSM6 sequence at their 5′ ends, respectively. This PCR product was transformed into CYY0 where it replaced one allele of LSM6 with URA3, leaving only 17 nt of LSM6 coding sequence at the 5′ terminus and 15 nt at the 3′ terminus. This generated strain DNY4. To disrupt LSM7, the TRP1 gene from pRS314 (Sikorski and Hieter 1989) was amplified using forward and reverse primers that contained 48 and 50 nt of LSM7 sequence at their 5′ ends, respectively. This PCR product was transformed into CYY0 where it replaced one allele of LSM7 with TRP1, leaving 10 nt of LSM7 coding sequence at both the 5′ and 3′ termini. This generated strain DNY9.
Allele-specific suppression of lsm8 mutations by Lsm proteins: To identify additional mutations that cause cells to require LHP1, we performed a genetic screen using a previously described strategy (Pannoneet al. 1998; Xueet al. 2000). An ade2 strain lacking LHP1 in the genome, but containing LHP1, ADE2, and TRP1 on a centromeric plasmid, was mutagenized with ethylmethane sulfonate and screened for the inability to lose the plasmid. As a red pigment accumulates in ade2 strains, cells that lose the plasmid are red, while cells that retain it are white. To determine if any mutants were allelic to previously isolated mutations (Yoo and Wolin 1997; Pannoneet al. 1998; Xueet al. 2000), we performed complementation analyses. Each mutant strain was mated to our previously isolated mutant strains, and the diploids were tested for the ability to lose the LHP1-containing plasmid. One strain, which we refer to as lsm8-2, failed to complement the lsm8-1 strain. Sequencing of LSM8 in the strain revealed a single base change that converted the tryptophan at position 119 to a stop codon. Because the requirement for LHP1 in this strain could be complemented by a low-copy plasmid containing LSM8, we conclude that the requirement for LHP1 is due to truncation of Lsm8p. The positions of the lsm8-1 and lsm8-2 mutations are shown in Figure 1. Similar to the lsm8-1 strain (Pannoneet al. 1998), the lsm8-2 strain was slightly cold sensitive for growth (data not shown).
Previously, we identified LSM2, which encodes another Sm-like U6 snRNP protein, as a low-copy suppressor of the lsm8-1 mutation (Pannoneet al. 1998). To determine whether other Sm-like proteins were able to eliminate the requirement for LHP1, we performed suppression analyses. Each of the LSM1-LSM8 genes was subcloned into the low-copy plasmid pRS316 and tested for the ability to eliminate the requirement for LHP1 in the lsm8 strains. For these experiments, we used lsm8 mutant strains that contained LHP1 under control of the MET3 promoter. Because the MET3 promoter is repressed by high concentrations of methionine (Cherestet al. 1987), the lsm8 strains are unable to grow on plates containing 2 mm methionine (not shown, but see Figure 2, A and B). In our initial experiments, LSM3, LSM5, LSM6, and LSM7 were tested as fusion proteins, as we used plasmids in which the C terminus of each coding sequence was fused to two IgG-binding domains of the S. aureus protein A (PrA; Salgado-Garridoet al. 1999). For the lsm8-1 mutant strain, introduction of either LSM2, LSM3-PrA, or LSM8 on the low-copy plasmid eliminated the requirement for LHP1 (Figure 2A), as these strains were able to grow on high methionine-containing media. Overexpression of LSM1, LSM4, LSM5-PrA, LSM6-PrA, LSM7-PrA, or the parent vector failed to eliminate the requirement for LHP1, although the background growth was reproducibly higher in the presence of extra copies of LSM4 (Figure 2A).
To determine whether the protein A tags affected the ability of LSM3 and LSM5-7 to suppress the lsm8 mutations, each of these genes was cloned without the tag into pRS316 and tested in the mutant strains. This revealed that, while the LSM3-PrA fusion suppressed the requirement for LHP1, wild-type LSM3 did not (Figure 2A, fourth panel). Similar to the result obtained with the PrA fusions, the wild-type LSM5-LSM7 failed to suppress the requirement for LHP1 in the mutant strain (data not shown).
Interestingly, the second lsm8 mutant allele, lsm8-2, exhibited a different pattern of suppression by LSM genes. Neither LSM2, LSM3, or LSM3-PrA, when present on the low-copy plasmid, allowed the mutant cells to grow on the 2-mm methionine medium (Figure 2B, also data not shown). In contrast, expression of LSM4 allowed some growth, although less than when LSM8 was expressed in the mutant cells (Figure 2B). Thus, while both LSM2 and LSM3-PrA are able to suppress the requirement for LHP1 in the lsm8-1 mutant cells, only LSM4 partially suppresses the LHP1 requirement in the lsm8-2 cells. However, as previously reported for the lsm8-1 allele (Pannoneet al. 1998), expression of U6 snRNA on the high-copy plasmid pRS426 eliminates the LHP1 requirement of the lsm8-2 strain (Figure 2B, pSNR6 sector). Thus, despite the differences in suppression of the two alleles by LSM genes, at least part of the requirement for LHP1 in both strains is likely due to defects in U6 snRNP biogenesis or stability.
Extra copies of LSM2 and LSM3-PrA increase pre-mRNA splicing efficiency in the lsm8-1 mutant strain: To explore the mechanism by which extra copies of LSM and U6 genes eliminated the requirement for Lhp1p, we created lsm8 strains in which LHP1 was replaced by extra copies of these genes. We started with lsm8 ade2 strains in which the only copy of LHP1 was supplied on an ADE2-containing plasmid. These strains are white as long as the ADE2 plasmid is present. Following transformation with plasmids containing LSM or U6 RNA genes, the strains were screened for the ability to lose the LHP1 and ADE2-containing plasmid and form colonies with red sectors. In the presence of either LSM2 or LSM3-PrA on a centromeric plasmid, or SNR6 on a high-copy plasmid, the lsm8-1 strains were able to lose the LHP1-containing plasmid. However, when LSM4 was expressed in the lsm8-2 strain, the strain was unable to lose the LHP1 plasmid and remained white, suggesting that at least a low level of LHP1 expression was required for efficient growth. Thus, while expression of LSM4 allows lsm8-2 strains to grow under conditions where LHP1 transcription is repressed (Figure 2B), it may not completely eliminate the requirement for LHP1. Even though the MET3 promoter is tightly regulated by methionine (Cherestet al. 1987), a small amount of Lhp1p may still be synthesized under repressive conditions.
Since the lsm8-1 mutation results in a pre-mRNA splicing defect (Pannoneet al. 1998), we determined whether extra copies of LSM2 and LSM3-PrA increased the efficiency of pre-mRNA splicing in the mutant strains. Total RNA was extracted from each strain and subjected to Northern analysis to detect the U3 small nucleolar RNA, since this RNA contains an mRNA-type intron (Myslinskiet al. 1990). As previously described (Pannoneet al. 1998), cells containing the lsm8-1 mutant allele accumulate unspliced pre-U3 RNA (Figure 3A, lane 3). In the presence of extra copies of LSM2 or LSM3-PrA, the splicing defect was less evident (lanes 4 and 5), consistent with the finding that these genes function as low-copy suppressors of the lsm8-1 mutation. However, while extra copies of SNR6 also allow lsm8-1 cells to grow in the absence of LHP1, only a slight decrease in pre-U3 RNA was detected in this strain (lane 7). Thus, while extra copies of LSM2, LSM3-PrA, and SNR6 all allow lsm8-1 cells to grow in the absence of LHP1, LSM2 and LSM3-PrA are more efficient at suppressing the pre-mRNA splicing defect than is SNR6.
The level of the lsm8-1 protein increases in the presence of extra copies of LSM2: To understand how extra copies of LSM2, LSM3-PrA, and SNR6 suppress the requirement for Lhp1p in lsm8-1 cells, we examined the levels of Lsm8p in the various strains. To this end, we prepared antibodies against the C-terminal 19 amino acids of Lsm8p and performed protein immunoblots (Figure 3B). The antibodies detected a polypeptide of 14 kD in wild-type extracts (lane 1), consistent with the predicted molecular weight of 14.5 kD. This band was greatly reduced in extracts from lsm8-1 mutant cells (lanes 2 and 3). To confirm the identification of the 14-kD protein as Lsm8p, we examined extracts from a strain in which three copies of the human c-myc epitope were fused to the C terminus of Lsm8p (Pannoneet al. 1998). As expected, the 14-kD band was replaced by a band of ∼20 kD (lane 4). Because the mutation in lsm8-2 cells truncates Lsm8p, we were unable to detect Lsm8p in these cells with the antibody against the C terminus (data not shown).
We next determined the levels of Lsm8p in the lsm8-1 mutant strains. In the presence of the LSM2-containing plasmid, the levels of the mutant Lsm8p increased, although not to wild-type levels (Figure 3C, compare lanes 3 and 4). Expression of LSM3-PrA on the centromeric plasmid did not result in a significant increase in the levels of Lsm8p (lane 5), while LSM8 restored Lsm8p to wild-type levels (lane 6), as expected. Curiously, in the presence of extra copies of SNR6, the mutant Lsm8p was further reduced (lane 7). Thus, expression of LSM2, but not LSM3-PrA or SNR6, in the lsm8-1 mutant strain results in an increase in the steady-state level of the mutant protein. Although we have not established the molecular mechanism by which this occurs, we note that Lsm8p and Lsm2p interact strongly in two-hybrid analyses (Fromont-Racine et al. 1997, 2000; Mayeset al. 1999; Uetzet al. 2000) and that these proteins have been proposed to be adjacent in the Lsm2-Lsm8 complex (He and Parker 2000; Pannone and Wolin 2000). Thus, binding of Lsm2p to Lsm8p may stabilize the mutant Lsm8 protein.
LSM8 is not essential in the presence of multiple U6 RNA genes: The observation that the levels of the mutant Lsm8p were drastically reduced in the presence of extra copies of SNR6 was surprising, since LSM8 is essential for yeast viability (Fromont-Racineet al. 1997; Pannoneet al. 1998). We thus determined whether LSM8 becomes dispensable for growth in the presence of multiple copies of SNR6. We transformed the high-copy plasmid containing SNR6, or the empty vector alone, into a diploid strain in which one allele of LSM8 was replaced with HIS3 (Pannoneet al. 1998). In the presence of the empty vector, sporulation of the diploid and tetrad dissection yielded only two viable segregants per tetrad (Figure 4A), all of which required histidine for growth (data not shown). However, when SNR6 was present on the high-copy plasmid, tetrad dissection yielded two, three, or four viable segregants per tetrad (Figure 4B). A total of 26 tetrads were dissected. In all cases, two segregants per tetrad were auxotrophic for histidine, indicating that they contained the wild-type LSM8 gene. In addition, 12 tetrads contained one or two additional viable progeny, all of which were His+. The His+ segregants were variable in size (arrowheads, Figure 4B), but tended to be smaller than the segregants containing LSM8. One explanation for the size variation may be that the lsm8::HIS3 progeny contain different numbers of the SNR6 plasmid, since the copy number of these plasmids ranges from 10 to 40 per cell (Sherman 1997). Consistent with this, in all tetrads that contained only two viable segregants, the cells required both histidine and uracil for growth, revealing that they also lacked the SNR6-containing plasmid. Thus, those diploid cells that gave rise to only two his- ura- progeny may have lost the SNR6 plasmid during sporulation. Nonetheless, the fact that a large fraction of the tetrads yielded viable His+ progeny reveals that LSM8 is dispensable for growth in the presence of multiple copies of SNR6. This result strongly suggests that the only essential function of Lsm8p is in U6 snRNP biogenesis or stability.
U6 snRNP levels in lsm8-1 cells are increased by extra copies of LSM2: To further examine the mechanism by which extra copies of LSM2, LSM3-PrA, and SNR6 suppress the requirement for Lhp1p in lsm8-1 cells, we examined the various U6 RNA-containing particles. We fractionated whole-cell extracts from wild-type and lsm8-1 strains using native gel electrophoresis and detected the U6 and U4 RNA-containing particles using Northern hybridization (Figure 5). Four distinct U6 RNA-containing complexes are present in wild-type extracts: the U4/U6.U5 tri-snRNP, the U4/U6 snRNP, the free U6 snRNP, and the Lhp1p/U6 RNA complex (Figure 5A, lane 1). As previously described (Pannoneet al. 1998), the free U6 snRNP is undetectable in lsm8-1 cells carrying chromosomal LHP1, and both the U4/U6 complex and U4/U6.U5 tri-snRNP are drastically reduced (lane 3). In addition, free U4 snRNPs accumulate in the mutant strain, consistent with a defect in U6 snRNP assembly (Figure 5B, lane 3; also Pannoneet al. 1998). When the sole copy of LHP1 is supplied on a centromeric plasmid (which raises the levels of Lhp1p two- to threefold), the levels of the U4/U6.U5 tri-snRNP increase, consistent with the presence of more functional U6 snRNPs in this strain (Figure 5A, lane 4; also Pannoneet al. 1998). Interestingly, expression of extra copies of LSM2 (lane 5), but not LSM3-PrA (lane 6), in the lsm8-1 strain resulted in a large increase in the levels of U6-containing snRNPs and a decrease in the levels of free U4 snRNPs. In the presence of the LSM2 plasmid, both the tri-snRNP and U4/U6 snRNP levels increased to ∼90% of wild-type levels and mature U6 snRNPs became detectable (Figure 5A, lane 5). However, while the LSM3-PrA plasmid also eliminated the requirement of lsm8-1 cells for LHP1, the only detectable change from the lsm8-1 LHP1 extracts was a small increase in the levels of the U4/U6 snRNPs (Figure 5A; compare lanes 3 and 6), with a concomitant decrease in the free U4 snRNPs (Figure 5B, lane 6). Thus, while both LSM2 and LSM3-PrA suppress the requirement for LHP1, only LSM2 restores the levels of U6-containing snRNPs in lsm8-1 cells to near wild-type levels.
Expression of the high-copy plasmid containing SNR6 in the lsm8-1 strain also did not significantly increase the level of the U4/U6.U5 tri-snRNP (Figure 5, A and B, lane 8). Instead, a heterogeneous smear of U6 RNA-containing particles migrated both with and ahead of the U4/U6 snRNP (lane 8). The smallest detectable band (asterisk, lane 8) comigrated with free U6 snRNA (data not shown). However, probing the blot to detect U4 RNA revealed that U4/U6 snRNP levels were restored to near wild-type levels (Figure 5B, lane 8). Thus, in the presence of additional copies of SNR6, U6 RNA assembles with U4 RNA to form the U4/U6 snRNP. However, assembly with the U5 snRNP remains impaired in the mutant strain.
LHP1 is essential in strains lacking LSM5, LSM6, or LSM7: Since our synthetic lethal screens identified two lsm8 alleles that cause yeast to require LHP1, but did not reveal mutations in other LSM genes, we asked whether mutations in other components of the Lsm2-Lsm8 complex would cause a requirement for Lhp1p. LSM2, LSM3, and LSM4 are all essential for viability (Cooperet al. 1995; Mayeset al. 1999; Salgado-Garridoet al. 1999), and nonlethal mutations in these genes have not been described. However, LSM6 and LSM7 are both nonessential genes (Mayeset al. 1999; Salgado-Garridoet al. 1999), and LSM5 has been reported to be both essential (Mayeset al. 1999) and nonessential (Salgado-Garridoet al. 1999). To determine whether LHP1 becomes essential in cells lacking one of these genes, we disrupted the genes encoding LSM5, LSM6, and LSM7 in a LHP1/lhp1::LEU2 diploid strain. When our diploid strain (lsm5::HIS3/LSM5, lhp1::LEU2/LHP1) was sporulated at 25°, followed by incubation of the dissected spores at 30°, we obtained two viable segregants per tetrad (Figure 6A, left), consistent with the report that LSM5 is essential (Mayeset al. 1999). However, when the dissected spores were incubated at 25°, we obtained tetrads containing either two, three, or four viable progeny (Figure 6A, right). In all cases, two segregants were large and lacked the LSM5 disruption marker. The remaining progeny were all small (Figure 6A) and His+, indicating that they contained the disrupted LSM5 gene. Thus, at 25°, segregants lacking LSM5 are viable in our strain background. Examination of the leucine requirement revealed that all His+ progeny required leucine for growth, indicating that they contained the wild-type LHP1 allele. Furthermore, all double mutants containing both the lhp1::LEU2 and lsm5::HIS3 alleles were dead (as deduced from the genotypes of the live segregants in each tetrad). Thus, LHP1 is required for viability in strains lacking LSM5. Similarly, sporulation of the lsm6::URA3/LSM6 lhp1::LEU2/ LHP1 (data not shown) and lsm7::TRP1/LSM7 lhp1:: LEU2/LHP1 (Figure 6B) diploids and tetrad dissection revealed that LHP1 was also required for viability in strains lacking either LSM6 or LSM7.
Another nonessential LSM protein, Lsm1p, associates with Lsm2-Lsm7 to form a complex that participates in mRNA degradation (Boecket al. 1998; Bouveretet al. 2000; Tharunet al. 2000). To determine whether LHP1 is required in strains lacking functional Lsm1p, we mated our lhp1::LEU2 strain to a temperature-sensitive strain carrying a large disruption within LSM1 (Diezet al. 2000). Sporulation of the diploid and dissection of the resulting tetrads revealed that lsm1 mutant progeny lacking LHP1 were able to grow at 30° (data not shown). Thus, LHP1 only exhibits genetic interactions with the LSM genes that encode components of the U6 snRNP.
In yeast, newly synthesized U6 RNA is bound and stabilized by Lhp1p, the yeast La protein. Since both Lhp1p and the Lsm2-Lsm8 complex bind the 3′ end of U6 RNA (Achselet al. 1999; Vidalet al. 1999), Lhp1p and the Lsm2-Lsm8 complex may bind consecutively to U6 RNA during U6 snRNP assembly (Pannoneet al. 1998; Achselet al. 1999). However, while Lhp1p and other La proteins preferentially bind RNAs terminating with UUUOH (Stefano 1984; Ternset al. 1992; Yoo and Wolin 1994), mature U6 RNA ends with either a 2′,3′-cyclic phosphate (in mammals) or a 3′ monophosphate (in yeast; Lund and Dahlberg 1992). Thus, while either Lhp1p or the Lsm2-Lsm8 complex can bind the nascent transcript, mature U6 RNA will only be bound by the Lsm2-Lsm8 complex. Consistent with the idea that either Lhp1p or the Lsm2-Lsm8 complex can bind nascent U6 RNA, Lhp1p is not required for U6 RNA biogenesis in otherwise wild-type yeast. However, Lhp1p becomes required to stabilize newly synthesized U6 RNA when cells contain a mutation in Lsm8p (Pannoneet al. 1998). Our experiments reveal that mutations in at least four LSM genes (LSM5, LSM6, LSM7, and LSM8) cause cells to require Lhp1p. Thus, the requirement for Lhp1p may be caused by any mutation that slows or impairs binding of the Lsm2-Lsm8 complex to U6 RNA. In this situation, Lhp1p may be required to stabilize newly synthesized U6 RNA, thus allowing assembly of the U6 snRNP.
Our result that LSM8 becomes dispensable for growth in the presence of excess U6 RNA genes strongly suggests that the only essential function of the Lsm2-Lsm8 complex is in U6 snRNA metabolism. Consistent with these results, it was previously demonstrated that U6 snRNA overproduction suppressed the growth defect caused when a strain harboring a GAL1-regulated copy of LSM8 was grown on glucose-containing medium (Mayeset al. 1999). However, in these experiments, it could not be ruled out that a small amount of Lsm8p (due to incomplete depletion and/or leakiness of the GAL1 promoter) was required for viability. The fact that LSM8 can be deleted in the presence of extra U6 snRNA genes establishes that the requirement for Lsm8p in U6 biogenesis and function can be bypassed as long as sufficient U6 RNA is present. Furthermore, the fact that Lsm8p becomes undetectable in the presence of excess U6 snRNA (Figure 3) suggests that cells may possess mechanisms for downregulating LSM8 expression in response to increased U6 RNA levels.
Although the human Lsm2-Lsm8 complex forms a ring of similar size and shape to the core Sm snRNPs, the order of the individual subunits around the rings is unknown. From the crystal structures of two Sm protein heterodimers, together with data from biochemical fractionation and two hybrid experiments, a model has been proposed in which specific Sm proteins interact with one another through their Sm motifs to form a heptameric ring (Kambachet al. 1999b). Because each of the Lsm2-Lsm8 proteins can be aligned with one of the Sm proteins (Fromont-Racineet al. 1997; Salgado-Garridoet al. 1999), Lsm proteins may interact in an analogous fashion (He and Parker 2000; Pannone and Wolin 2000; diagrammed in Figure 7). In this model, Lsm8p contacts Lsm2p and Lsm4p. We have demonstrated that both Lsm8p and U6 snRNP levels increase when Lsm2p is overexpressed in cells carrying a mutation in the Sm motif of Lsm8p. These findings are consistent with the specific interaction of these proteins within the Lsm2-Lsm8 complex. Our result that Lsm4p, when overexpressed in yeast carrying a truncation of Lsm8p, partially suppresses the requirement for Lhp1p is consistent with the idea that Lsm4p and Lsm8p directly interact within the complex. Furthermore, the fact that LSM2 does not suppress the requirement for LHP1 in the lsm8-2 truncation mutant suggests that Lsm2p may contact the C terminus of Lsm8p (in addition to the Sm motif).
Curiously, while LSM3-PrA is also a low-copy suppressor of the lsm8-1 mutation, Lsm3-PrAp overexpression did not increase Lsm8p and only moderately increased U4/U6 snRNP levels. Since LSM3-PrA overexpression can substitute for LHP1 in the lsm8-1 strain, one possibility is that the excess Lsm3-PrA protein functions similar to Lhp1p in binding and stabilizing newly synthesized U6 RNA. However, since we did not detect a Lsm3-PrA/U6 complex when the U6-containing particles were fractionated on native gels (Figure 5), this complex would have to be considerably less stable than the Lhp1p/U6 RNA complex. Since LSM3-PrA, but not wild-type LSM3, suppresses the requirement for LHP1, another possibility is that the extra C-terminal sequences in the Lsm3-PrA fusion interact with the Lsm8p tail to stabilize or enhance the function of the mutant Lsm2-Lsm8 complex. This enhancement of Lsm2-Lsm8 function could occur at a later stage of the U6 snRNP cycle, such as U4/U6 snRNP assembly or recycling of the U6 snRNP following splicing. Consistent with an interaction with the Lsm8p C terminus, excess Lsm3-PrA did not suppress the requirement for Lhp1p in the lsm8-2 truncation mutant.
Interestingly, the genetic interactions that we have identified are far more restricted than those identified using two-hybrid analyses. In two-hybrid screens, Lsm8p interacts with each of the other six members of the Lsm2-Lsm7 complex (Fromont-Racine et al. 1997, 2000; Mayeset al. 1999; Uetzet al. 2000). We note that, in two-hybrid analyses, the two proteins being tested are expressed as fusion proteins. Thus, it is possible that if the fusion proteins are incorporated into the Lsm2-Lsm8 ring, the additional sequences appended to one or both proteins could result in interactions that do not occur between the wild-type proteins. This scenario would be compatible with the observation that Lsm3p only suppresses the requirement for LHP1 when extra sequences are appended to the C terminus.
Because a small fraction of yeast pre-RNase P RNA is bound by six Sm-like proteins (Lsm2-Lsm7; Salgado-Garridoet al. 1999; B. K. Pannone and S. L. Wolin, unpublished data) and small changes in the levels of certain RNA polymerase III RNAs have been observed at late times after depletion of Lsm proteins (Mayeset al. 1999), it has been suggested that Sm-like proteins may function in the biogenesis of other small RNAs. Our result that LSM8 becomes dispensable when SNR6 is overexpressed reveals that these other possible functions are not essential roles of Lsm8p. In this regard, we note that Lsm8p, unlike Lsm2p-Lsm7p, is not detected bound to pre-RNase P RNA (Salgado-Garridoet al. 1999; B. K. Pannone and S. L. Wolin, unpublished data). Thus, there may be yet another complex of Lsm proteins, distinct from the Lsm2-Lsm8 and Lsm1-Lsm7 complexes, that functions in these other processes.
We thank Doug Rubinson for help with the synthetic lethal screen. We are grateful to Bertrand Seraphin, Jean Beggs, Paul Ahlquist, and David Brow for gifts of plasmids and yeast strains. We also thank Elisabetta Ullu for comments on the manuscript. This work was supported by grant R01-GM48410 from the National Institutes of Health. S.L.W. is an Associate Investigator of the Howard Hughes Medical Institute.
Communicating editor: S. Sandmeyer
- Received November 29, 2000.
- Accepted February 13, 2001.
- Copyright © 2001 by the Genetics Society of America