Copyright © 2005 by the Genetics Society of America DOI: 10.1534/genetics.104.034322

Mutations in the Saccharomyces cerevisiae LSM1 Gene That Affect mRNA Decapping and 3؅ End Protection

Sundaresan Tharun,*,1 Denise Muhlrad,† Ashis Chowdhury* and Roy Parker† *Department of Biochemistry, Uniformed Services University of the Health Sciences, Bethesda, Maryland 20814-4799 and †Department of Molecular and Cellular Biology and Howard Hughes Medical Institute, University of Arizona, Tucson, Arizona 85721 Manuscript received August 6, 2004 Accepted for publication January 20, 2005

ABSTRACT The decapping of eukaryotic mRNAs is a key step in their degradation. The heteroheptameric Lsm1p–7p complex is a general activator of decapping and also functions in protecting the 3Ј ends of deadenylated mRNAs from a 3Ј-trimming reaction. Lsm1p is the unique member of the Lsm1p–7p complex, distinguish- ing that complex from the functionally different Lsm2p–8p complex. To understand the function of Lsm1p, we constructed a series of deletion and point mutations of the LSM1 gene and examined their effects on phenotype. These studies revealed the following: (i) Mutations affecting the predicted RNA- binding and inter-subunit interaction residues of Lsm1p led to impairment of mRNA decay, suggesting that the integrity of the Lsm1p–7p complex and the ability of the Lsm1p–7p complex to interact with mRNA are important for mRNA decay function; (ii) mutations affecting the predicted RNA contact residues did not affect the localization of the Lsm1p–7p complex to the P-bodies; (iii) mRNA 3Ј-end protection could be indicative of the binding of the Lsm1p–7p complex to the mRNA prior to activation of decapping, since all the mutants defective in mRNA 3Ј end protection were also blocked in mRNA decay; and (iv) in addition to the Sm domain, the C-terminal domain of Lsm1p is also important for mRNA decay function.

ESSENGER RNA (mRNA) turnover is an impor- 1992; Decker and Parker 1993; Hsu and Stevens 1993; M tant control point in the regulation of gene ex- Muhlrad et al. 1994, 1995; Beelman et al. 1996; Dunckley pression. Numerous examples of genes regulated at the and Parker 1999; Tucker et al. 2001). In the 3Ј to 5Ј level of mRNA stability are known to date (Hentze pathway, deadenylated mRNAs are degraded in a 3Ј to 1991; Peltz et al. 1991; Abler and Green 1996; Guhan- 5Ј exonucleolytic manner by the exosome (Muhlrad iyogi and Brewer 2001; Tharun and Parker 2001b). et al. 1995; Anderson and Parker 1998). Several studies have shown that mRNA decay factors Decapping is a crucial step in the 5Ј to 3Ј decay path- and decay pathways are well conserved in eukaryotes way because it is the site of several regulatory inputs from yeast to humans (Seraphin 1992; Couttet et al. (Coller and Parker 2004). Moreover, a large number 1997; Gao et al. 2001; Wang and Kiledjian 2001; of proteins affect decapping in addition to the decap- Decker and Parker 2002; Liu et al. 2002; Lykke- ping enzyme. The translation initiation machinery and Andersen 2002; Van Dijk et al. 2002, 2003; Wang et the poly(A)-binding protein are antagonistic to decap- al. 2002; Piccirillo et al. 2003; Tseng-Rogenski et al. ping (Caponigro and Parker 1995; Coller et al. 1998; 2003). Schwartz and Parker 1999, 2000; Vilela et al. 2000; In eukaryotes, wild-type mRNAs primarily degrade via Wilusz et al. 2001; Ramirez et al. 2002; Khanna and deadenylation-dependent decay pathways that involve Kiledjian 2004). Conversely, several other factors func- Ј Ј Ј Ј 5 to 3 or 3 to 5 mode of degradation of the mRNA tion as activators of decapping. They include Pat1p, body. In yeast, deadenylation is carried out by the Ccr4p Dhh1p, Lsm1p–7p complex, Edc1p, Edc2p, and Edc3p Ј Ј complex (Tucker et al. 2001). In the 5 to 3 decay (Hatfield et al. 1996; Boeck et al. 1998; Bonnerot et pathway (also known as the “deadenylation-dependent al. 2000; Bouveret et al. 2000; Tharun et al. 2000; decapping pathway”), this is followed by decapping by Wyers et al. 2000; Coller et al. 2001; Dunckley et al. the decapping enzyme (Dcp1p–2p complex) that ex- 2001; Fischer and Weis 2002; Schwartz et al. 2003; Ј Ј poses the body of the message to 5 to 3 exonucleolytic Kshirsagar and Parker 2004). decay by the exonuclease Xrn1p (Muhlrad and Parker The Lsm1p–7p complex is particularly interesting as a decapping activator for several reasons. First, it is a highly conserved component of eukaryotic mRNA decay 1Corresponding author: Department of Biochemistry, Uniformed Ser- vices University of the Health Sciences, 4301 Jones Bridge Rd., machinery. Second, it physically interacts with several Bethesda, MD 20814-4799. E-mail: [email protected] other factors involved in the major mRNA decay path-

Genetics 170: 33–46 (May 2005) 34 S. Tharun et al. way, including the decapping activators Pat1p and Dhh1p subunit that is either Lsm1p or Lsm8p. However, it is and the 5Ј to 3Ј exonuclease Xrn1p (Bonnerot et al. not known what features of Lsm1p and Lsm8p dictate 2000; Bouveret et al. 2000; Tharun et al. 2000; Coller their respective roles. et al. 2001). Third, in vivo, it associates with a pool of The mechanism by which Lsm1p–7p complex acti- deadenylated mRNPs that is bound to the decapping en- vates decapping is not clear. It is also not known how the zyme and targeted for decay but distinct from translating decapping activation function is related to the mRNA 3Ј mRNPs (Tharun and Parker 2001a). Fourth, several end protection function. Given the structural similarity studies suggest that in addition to mRNA decapping, this of the Lsm1p–7p complex to the Lsm2p–8p and the complex is also involved in protecting mRNA 3Ј ends Sm complexes, a reasonable model would be that simi- from trimming (Boeck et al. 1998; He and Parker 2001). larity also exists at a mechanistic level in the manner in Finally, mammalian cells overexpressing Lsm1p show a which these complexes function. Importantly, given that transformed phenotype, suggesting a possible role for the Sm complex and several Sm-like protein complexes Lsm1p in growth control (Schweinfest et al. 1997; (including the Lsm2p–8p complex) have been shown Gumbs et al. 2002). to directly bind to their RNA substrates to execute their The Lsm1p–7p complex, which is conserved in all functions (Achsel et al. 1999, 2001; Raker et al. 1999; eukaryotes, is a heptameric complex made of seven pro- Toro et al. 2001; Moller et al. 2002; Schumacher et teins, Lsm1p through Lsm7p. The Lsm (“Like Sm”) al. 2002), it is likely that the Lsm1p–7p complex also proteins are homologous to the Sm proteins and these directly binds to mRNAs to protect their 3Ј ends and proteins form a family of protein complexes that in- to activate their decapping. This is consistent with the cludes complexes found in eubacteria, archaea, and earlier co-immunoprecipitation analyses, which re- eukaryotes (Achsel et al. 1999, 2001; Kambach et al. vealed that the Lsm1p–7p complex associates with dead- 1999; Collins et al. 2001; Mura et al. 2001; Toro et al. enylated mRNAs in vivo (Tharun et al. 2000; Tharun 2001; Moller et al. 2002; Schumacher et al. 2002). and Parker 2001a). If this is true, then, disrupting the Individual members of this protein family are character- interaction of the Lsm1p–7p complex with RNA should ized by the presence of a bipartite sequence motif, re- impair both mRNA 3Ј end protection and decapping ferred to as the Sm domain. The Sm domain consists activation functions. of two conserved segments of amino acids (Sm motifs Recent studies have shown that the decapping and I and II) separated by a nonconserved segment of vari- 5Ј to 3Ј exonucleolysis of mRNAs occur in discrete cyto- able length (Cooper et al. 1995; Hermann et al. 1995; plasmic structures called processing bodies (P-bodies), Seraphin 1995). where Xrn1p, the various decapping factors including These Sm and Lsm proteins also show similarity in the Lsm1p through Lsm7p proteins, and mRNA mole- their tertiary structure and the quarternary structure of cules targeted for decay are localized (Ingelfinger et the complexes they form. Typically, they exist in cells al. 2002; Van Dijk et al. 2002; Sheth and Parker 2003; in the form of ring-shaped hexa or heptameric com- Cougot et al. 2004). However, it is not known how the plexes (Achsel et al. 1999, 2001; Kambach et al. 1999; Lsm1p–7p complex and the other decay factors are Collins et al. 2001; Mura et al. 2001; Toro et al. 2001; recruited to the P-bodies. Moller et al. 2002; Schumacher et al. 2002; Zaric et To identify the critical regions of the Lsm1p protein al. 2005). For example, the seven Sm proteins conserved and to determine how they are related to the functions in all eukaryotes associate into a ring-shaped heptameric of Lsm1p, we undertook a mutagenic analysis of Lsm1p. “Sm complex” (Kambach et al. 1999) that assembles Structural and biochemical studies in the past have iden- onto the U1, U2, U4, and U5 snRNAs to form the cores tified residues in the Sm domain of Sm proteins and of the corresponding spliceosomal snRNPs (Lerner archaeal Sm-like proteins that are involved in RNA and and Steitz 1981; Luhrmann et al. 1990). On the other inter-subunit interactions (Kambach et al. 1999; Mura hand, there are eight Lsm proteins (Lsm1p through et al. 2001, 2003; Toro et al. 2001; Urlaub et al. 2001). Lsm8p) conserved in all eukaryotes and they appear to Therefore, we predicted that the homologous residues form two heptameric complexes (Salgado-Garrido et in Lsm1p could have similar roles, mutated such resi- al. 1999; Bouveret et al. 2000; Tharun et al. 2000). dues, and examined the effect of such mutations on the One complex, consisting of Lsm1p through Lsm7p pro- functions of Lsm1p. In addition, we also examined the teins, is an activator of mRNA decapping (Boeck et al. roles of other residues within the Lsm1p, including 1998; Bouveret et al. 2000; Tharun et al. 2000), while those in the conserved C-terminal domain following the the other made of Lsm2p through Lsm8p proteins func- Sm domain. tions in pre-mRNA splicing by forming the core of U6 These studies revealed the following: snRNP (Cooper et al. 1995; Pannone et al. 1998, 2001; Achsel et al. 1999; Mayes et al. 1999; Salgado-Gar- i. Mutations affecting the predicted RNA-binding sur- rido et al. 1999). Thus, although involved in very differ- face of Lsm1p lead to impairment of mRNA decay ent functions, the Lsm1p–7p and Lsm2p–8p complexes without significantly affecting the localization of the share six of their seven subunits and differ in only one Lsm1p–7p complex to the P-bodies. Thus, interac- Mutagenic Analysis of Yeast Lsm1p 35

tions with mRNA are critical for activation of decap- For all experiments, cells were grown at 25Њ in Ϫura medium ping but not for recruitment of the Lsm1p–7p com- and collected at log phase. For examination of P-bodies (experi- ments in Figure 5 and Figure 6), cells grown to log phase plex to P-bodies. were collected, incubated in glucose-free medium for 10 min, ii. Lesions in the predicted inter-subunit interaction do- and then observed using confocal microscope. Confocal mi- mains also lead to significant impairment of mRNA croscopy was done as described (Sheth and Parker 2003; decay indicating the importance of the Lsm1p–7p Teixeira et al. 2005). complex integrity for mRNA decay. iii. All the mutants that are defective in mRNA-3Ј end RESULTS protection are also defective in mRNA decay, sug- gesting that 3Ј end protection could be indicative Mutagenesis of LSM1: Structural and biochemical stud- of the Lsm1p–7p complex-mRNA interaction that ies have been conducted on the human Sm subcom- leads to decapping activation. plexes and several archaebacterial Sm-like protein com- iv. In addition to the Sm domain, the extended C-termi- plexes in the past (Kambach et al. 1999; Mura et al. nal domain of Lsm1p is also important for the func- 2001, 2003; Toro et al. 2001; Urlaub et al. 2001). Such tion of the Lsm1p–7p complex. studies revealed the crucial amino acid residues—involved in RNA contact and inter-subunit contacts—of the Sm and Sm-like proteins that form these complexes. Using MATERIALS AND METHODS this information and the homology of these proteins to yeast Lsm1p, we predicted the residues of Lsm1p that Yeast strains, plasmids, and mutagenesis: All the mutant ver- sions of LSM1 were made following the Quikchange mutagenesis are likely to be involved in RNA contact and inter-sub- protocol (Stratagene, La Jolla, CA) using plasmid pST11 (car- unit contacts and targeted them (Figure 1, a and b) in rying the wild-type LSM1 gene) as the template. This generated our first set of mutants (lsm1-6 through lsm1-9, lsm1-13, the plasmids, pST21 through pST49 and pST18, each of which and lsm1-14). carried a different lsm1 allele (Tables 1 and 2). pST11 was Structural studies on several members of the Sm-like made by amplifying the LSM1 gene (containing the ORF along with 619 and 543 nucleotides, respectively, of the 5Ј- and 3Ј- protein family have revealed that the Sm domain of flanking regions) by polymerase chain reaction (PCR) from these proteins folds into a conserved structure (“Sm yeast genomic DNA and cloning it into pRS416 that is a CEN fold”) consisting of an N-terminal ␣-helix followed by vector with URA3 marker (Sikorski and Hieter 1989). All a five-stranded, strongly bent ␤-sheet (Kambach et al. the plasmids were sequenced to make sure that unintended 1999; Mura et al. 2001, 2003; Toro et al. 2001). Here mutations were not present in the LSM1 gene. Plasmids car- rying wild-type (pST11) and mutant versions of LSM1 were the key residues involved in contacting the RNA are transformed into the lsm1⌬ strain yRP1365 (MAT␣ trp1 leu2 located in loops 3 (between ␤-strands 2 and 3) and 5 ura3 lys2 cup1⌬::LEU2(PM) lsm1⌬::TRP1)(Tharun et al. 2000) (between ␤-strands 4 and 5), while many of the residues and the resulting transformants were used for the experi- involved in inter-subunit interactions are located in ments. yRP1365 carrying pRS416 served as the negative con- ␤-strands 2, 3, and 4. The 3D structure of the Sm domain trol. For the experiments in Figure 5, an lsm1⌬ strain express- of the yeast Lsm1p generated using the homology-based ing GFP fusions of LSM1 or lsm1-9 or lsm1-14 under LSM1’s modeling program 3D-JIGSAW (Bates and Sternberg native promoter from a CEN vector was used. Plasmid pRP1176 1999; Bates et al. 2001; Contreras-Moreira and Bates (J. Coller and R. Parker, unpublished results) is a CEN 2002) supports the idea that the homologous residues vector expressing LSM1-GFP from LSM1’s native promoter. in Lsm1p are organized similarly (Figure 1b). Mutations of lsm1-9 and lsm1-14 were introduced into the LSM1-GFP cassette of this plasmid by Quikchange mutagenesis The Lsm1p residues implicated (on the basis of homol- protocol (Stratagene) to generate plasmids pST91 and pST92, ogy) in inter-subunit contacts included both charged and respectively. pRP1176, pST91, and pST92 were separately trans- hydrophobic amino acids. R59 and R62 located in loop ⌬ ␣ formed into the Euroscarf lsm1 strain, ES11301 (MAT , his3, 2 of the modeled 3D structure were mutated in the ⌬ r leu2, ura3, lys2, lsm1 ::NEO ), to get the strains needed for the allele lsm1-6. L64 and G66 located in ␤-strand 2 and experiments shown in Figure 5. Strains used in experiments for Figure 6 were made by introducing each of the plasmids R62 were changed in the allele lsm1-7. Residues 101 to pST11, pST 29, or pST34 (coding for LSM1, lsm1-9, and lsm1- 104 (I, F, M, and I) located in ␤-strand 4 were changed 14, respectively; see Table 1) separately into the strain yRP in lsm1-13. The charged residues R59 and R62 were 2008 (genotype the same as ES11301 except that it is MATa replaced with alanines. On the other hand, residues 101 and LSM7-GFP-HIS3) or yRP 2010 (genotype the same as to 104 (I, F, M, and I) and L64 were changed to charged ES11301 except that it is MATa and LSM2-GFP-HIS3). yRP 2008 and yRP 2010 were made by crossing the LSM2-GFP and residues, since hydrophobicity was predicted to be cru- LSM7-GFP strains of Euroscarf with the Euroscarf lsm1⌬ strain cial for their function. G66 was replaced by a bulky ES11301. tryptophan (Figure 1, a and b). RNA and protein analyses and microscopy: Transcription Lsm1p residues implicated in RNA binding are lo- shut-off experiments, RNA preparation, Northern analysis, de- cated in loops 3 and 5 of the modeled 3D structure termination of MFA2pG mRNA half-life, quantitation of poly(G) fragment accumulation, preparation of yeast cell lysates, and (Figure 1b). The loop 3 residues Y74 and N76 are mu- Western analysis for Lsm1p were all performed as described tated in lsm1-9, while the loop 5 residues R105, G106, earlier (Tharun and Parker 1999, 2001a; Tharun et al. 2000). and E107 are mutated in lsm1-14. The loop 3 residue 36 S. Tharun et al.

Figure 1.—Regions of LSM1 targeted in our study. (a) Position of the mutated residues in the primary sequence. Regions of primary sequence targeted in the mutant alleles, lsm1-1 through lsm1-24, are underlined. The allele number is given directly below the part of primary sequence targeted in each of these mutant alleles. Residues replacing the wild-type sequence are indicated in italics above the appropriate wild-type residues. Triangles following (alleles 25 and 26) or preceding (27–29) allele numbers point to the last of the residues deleted from the N and C termini, respectively, in those deletion alleles. Amino acid residues shown in boldface and italicized boldface letters are predicted to be involved in inter-subunit or RNA contacts, respectively. Regions of Lsm1p predicted to form the secondary structural elements (N-terminal ␣-helix, ␣1, and the five ␤-strands, ␤1 through ␤5) characteristic of the Sm and Lsm proteins are also indicated. Primary sequence stretches indicated by long and short shaded areas are Sm motifs I and II, respectively, of the Sm domain. Amino acid residue numbers are indicated on the right of each line. (b) Location of the residues implicated in inter-subunit and RNA contacts in the predicted three-dimensional structure of the Sm domain of Lsm1p. Tertiary structure of the Sm domain of Lsm1p was generated using the homology-based modeling program 3D-JIGSAW (Bates and Sternberg 1999; Bates et al. 2001; Contreras-Moreira and Bates 2002) using the known tertiary structure of the archebacterial Sm-like protein, Smap1, from Pyrobaculum aerophilum as the template. Regions in green and pink contain residues implicated in inter-subunit contacts and RNA contacts, respectively. Residues targeted in our study are indicated in boldface letters. Various secondary structural elements like the N-terminal ␣-helix (␣1), the five ␤-strands (␤1 through ␤5), and the loop regions are also indicated. (c) Alignment of yeast Lsm1p with human Lsm1p using the ClustalW program. Long and short shaded areas indicate Sm motifs I and II of yeast Lsm1p. Residues in boldface or italicized boldface letters indicate those predicted to be involved in inter-subunit and RNA contacts, respectively. Portions of the primary sequence forming the Sm domain and the N- and C-terminal extensions are indicated above the primary sequence. The bottom line in each block indicates the yeast Lsm1p residues that are either identical (*) or replaced with a similar residue (o) in the human Lsm1p. Amino acid residue numbers are indicated on the left and right (for the lower block) of the primary sequence. For both proteins, the entire primary sequence is shown.

D72 was mutated along with R69 located in ␤-strand 2 inside the Sm domain. Changes were restricted to four in lsm1-8. N76, Y74, R105, and E107 were predicted to or fewer neighboring residues in all the mutants. form the RNA-binding pocket while D72 and G106 were Finally, to address the role of the regions of Lsm1p predicted to be important in holding N76 in the proper that are outside the Sm-domain, we also made five dele- orientation. While G106 was changed to a tryptophan, tion mutants. These included lsm1-25 and lsm1-26 that other residues were replaced with alanines (Figure 1, a had, respectively, 17 and 36 amino acids deleted from and b). the N terminus and lsm1-27 through lsm1-29 that had All the clusters of charged residues in Lsm1p whose 55, 43, and 28 amino acids, respectively, deleted from function could not be readily predicted on the basis the C terminus (Figure 1a). Importantly, in these mu- of sequence homology were replaced with alanines to tants, the Sm domain was left intact so that the mutant generate the mutants lsm1-1 through lsm1-5, lsm1-10 proteins are likely to fold into the proper tertiary struc- through lsm1-12, and lsm1-15 through lsm1-24 (Figure ture that is characteristic of the Sm-like proteins. 1a). These included residues in the regions flanking or Mutants were made as described in materials and Mutagenic Analysis of Yeast Lsm1p 37

TABLE 1 Description of the changes introduced in the various alleles and their associated phenotypes: mutations in the Sm domain

LSM1 Plasmid Growth mRNA mRNA 3Ј end allele borne Amino acid residue changes at 36Њ a decay b protection b a. lsm1 alleles affected in predicted inter-subunit contact residues lsm1-6 pST26 R59, D60, and R62 to A’s Ϫϩϩϩϩ lsm1-7 c pST27 R62, L64, and G66 to A, D, Ϫϩ ϩ and W, respectively lsm1-13 pST33 I101, F102, M103, and I104 Ϫϩ ϩ to R, E, K, and D, respectively

b. lsm1 alleles affected in predicted RNA-contact residues lsm1-8 pST28 R69 and D72 to A’s Ϫϩ ϩ lsm1-9 pST29 Y74 and N76 to A’s Ϫ/ϩϩϩϩ lsm1-14 pST34 R105, G106, and E107 to A, W, Ϫϩϩϩ and A, respectively

c. lsm1 alleles affected in other Sm domain residues lsm1-5 pST25 D51, R52, and K53 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-11 pST31 E90, E91, and K93 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-12 pST32 E96, E97, D98, and R99 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-15 pST35 E114 and D116 to A’s ϩ ϩϩϩ ϩϩϩ a Growth, no growth, and growth that is significantly slower than wild-type cells are indicated by ϩ, Ϫ, and Ϫ/ϩ symbols, respectively. b Strong (comparable to lsm1⌬) and moderate block to these functions are indicated by ϩ and ϩϩ, respec- tively. ϩϩϩ indicates normal (wild type) function. c Mutant Lsm1p protein bank as very weak or undetectable (by Western analysis) in the lysates of these alleles. methods by introducing the changes in the LSM1 gene First, we observed that among the three lsm1 alleles (with its native promoter and 3Ј-UTR sequences) cloned bearing mutations in the predicted RNA-binding resi- in a CEN vector with URA3 marker. The plasmids car- dues (i.e., lsm1-8, lsm1-9, and lsm1-14), lsm1-8 and lsm1- rying the wild-type and the various mutant versions of 14 were unable to grow at 36Њ, while lsm1-9 was able LSM1 were transformed into the lsm1⌬ strain yRP1365 to grow very slowly. These results provide the first so that the plasmid-borne gene is the only source of indication that these residues are important for Lsm1p protein in the cell. The lsm1⌬ strain transformed Lsm1p function. However, the ability of lsm1-9 cells with the empty vector pRS416 and the recombinant to grow at 36Њ, albeit slowly, suggests that at least vector carrying the wild-type LSM1 gene were used as some parts of the RNA-binding surface can undergo the negative and positive controls, respectively, in all mutation without affecting viability at high tempera- experiments and referred to as lsm1⌬ and wild type, ture. respectively. The mutants were studied for temperature Second, we observed that all the three lsm1 alleles bear- sensitivity, mRNA decay, and mRNA 3Ј end protection ing mutations in the predicted inter-subunit contact as described below. residues, lsm1-6, lsm1-7, and lsm1-13 were unable to LSM1 mutations resulting in temperature sensitivity grow at 36Њ just like lsm1⌬. These results support the of growth: lsm1⌬ cells are temperature sensitive for idea that these residues are required for Lsm1p func- growth at higher temperatures (Mayes et al. 1999). Ear- tion. lier studies had suggested that the temperature sensitiv- Third, we observed that large deletions in the region of ity of the lsm1⌬ cells is likely to be related to their mRNA Lsm1p that is C-terminal to the Sm domain lead to 3Ј-end-trimming phenotype (He and Parker 2001). temperature-sensitive growth as shown by the inability Therefore, as a first assay for the function of the mutant of the alleles lsm1-27 and lsm1-28 to grow at 36Њ. The Lsm1p proteins, we studied their ability to support growth mutant lsm1-29, which bears the shortest deletion of at high temperature. To this end, growth of the mutant this C-terminal region, grew slowly at 36Њ. These obser- and the wild-type cells on Ϫura plates was examined at vations indicate that residues outside the Sm domain 25Њ and 36Њ. As expected, at 25Њ both of the control strains are also likely to be important for Lsm1p function. LSM1 and lsm1⌬ were able to grow, while at 36Њ, lsm1⌬ Fourth, we observed that mutations in or deletion of cells were unable to grow. Analysis of the growth of the the stretch of amino acids N-terminal to the Sm do- mutants revealed the following (Tables 1 and 2): main did not affect growth at any temperature, sug- 38 S. Tharun et al.

TABLE 2 Description of the changes introduced in the various alleles and their associated phenotypes: mutations outside the Sm domain

Plasmid Growth mRNA mRNA 3Ј end LSM1 allelle borne Amino acid residue changes at 36Њ a decay b protection b a. lsm1 alleles with mutations in the N-terminal extension lsm1-1 PST21 K6, D7, and R8 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-2 pST22 D14, K16, and R17 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-3 pST23 K24 and K25 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-4 pST24 E28, E30, and D32 to A’s ϩ ϩϩϩ ϩϩϩ

b. lsm1 alleles with mutations in the C-terminal extension lsm1-16 pST36 D118, K119, E120, and D121 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-17 pST37 E125, E128, and R129 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-18 pST38 K133 and E134 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-19 pST39 K139 and K141 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-20 pST40 D143, E144, K145, and R146 to A’s, respectively ϩ ϩϩϩ ϩϩϩ lsm1-21 pST41 K148, E149, and E150 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-22 pST42 H152, K153, K155, and K156 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-23 pST43 R159 and H160 to A’s ϩ ϩϩϩ ϩϩϩ lsm1-24 pST44 D165, H167, K168, and D170 to A’s ϩ ϩϩϩ ϩϩϩ

c. lsm1 alleles with N-terminal truncations lsm1-25 pST45 N terminus to R17 deleted ϩ ϩϩϩ ϩϩϩ lsm1-26 pST46 N terminus to D36 deleted ϩ ϩϩϩ ϩϩϩ

d. lsm1 alleles with C-terminal truncations/insertion lsm1-27 c pST47 D118 to C terminus (55 residues) deleted Ϫϩϩϩϩ lsm1-28 c pST48 I130 to C terminus (43 residues) deleted Ϫϩ ϩ lsm1-29 pST49 K145 to C terminus (28 residues) deleted Ϫ/ϩϩϩϩϩ lsm1-31 pST18 2ϫ FLAG epitope fusion at the C terminus Ϫϩϩϩϩ a Growth, no growth, and growth that is significantly slower than wild-type cells are indicated by ϩ, Ϫ, and Ϫ/ϩ symbols, respectively. b Strong (comparable to lsm1⌬) and moderate block to these functions are indicated by ϩ and ϩϩ, respectively. ϩϩϩ indicates normal (wild type) function. c Mutant Lsm1p protein band was very weak or undetectable (by Western analysis) in the lysates of these alleles.

gesting that this region is not critical for Lsm1p func- a chromosomal location in our strains (Hatfield et al. tion. Similarly, several other mutations affecting small 1996). The MFA2pG mRNA decays by deadenylation- clusters of charged residues distributed throughout dependent decapping followed by 5Ј to 3Ј exonucleo- the protein were all tolerated, suggesting that none lytic digestion by Xrn1p (Muhlrad et al. 1994). The of these regions by itself is of major importance for MFA2pG mRNA also contains a poly(G) insertion in its the function of Lsm1p. 3Ј-UTR that forms a strong secondary structure in vivo Finally, although all the strains were able to grow in and blocks Xrn1p in cis. This results in the stable accu- Ϫura plates at 25Њ, differences in growth rates could mulation of the degradation intermediate called poly(G) be observed. As expected, lsm1⌬ cells grew much fragment in vivo. As a result, in strains where decapping more slowly than the wild-type cells. Further, a subset is defective, the accumulation of the poly(G) fragment of the mutants unable to grow at 36Њ showed slow growth is decreased. Thus, the relative levels of the poly(G) at 25Њ, similar to lsm1⌬. They were lsm1-7, lsm1-8, lsm1- fragment and the full-length mRNA at steady state in a 13, and lsm1-27 (not shown). These strains and lsm1-9 strain can be used as a first approximation of the effi- and lsm1-14 were also slow growing in liquid cul- ciency of decapping of this reporter mRNA in that strain tures at 25Њ. (Hatfield et al. 1996; Cao and Parker 2001). There- fore, we first identified the lsm1 mutants defective in Effect of LSM1 mutations on mRNA decay: A key set mRNA decay on the basis of their inability to accumulate of experiments examined the effect of the lsm1 muta- the poly(G) fragment at sufficient levels. This was fol- tions on mRNA decay. For this experiment, we utilized lowed by a direct measurement of the MFA2pG mRNA the MFA2pG reporter mRNA system. This mRNA is ex- half-life in such mutants to fully confirm the mRNA pressed under the control of the GAL promoter from decay phenotype. Mutagenic Analysis of Yeast Lsm1p 39

Figure 2.—Relative lev- els of poly(G) fragment and full-length MFA2pG mRNA in lsm1 mutants. Cells were grown in Ϫura medium with galactose as the sole carbon source to log phase when they were collected and RNA ex- tracted. RNA was separated on polyacrylamide gel and subjected to Northern anal- ysis using oligo probe for MFA2pG mRNA. Bands were visualized and quantitated us- ing the PhosphorImager. lsm1 alleles carried by the strains from which RNA was made are indicated above the lanes. Predicted function of the residues targeted in the important mutants and the nature of the deletions in the deletion mutants are also indicated above the allele names. All strains are deleted for the chromosomal copy of LSM1 and carry the indi- cated alleles on a CEN vector as mentioned in materials and methods. The lsm1⌬ strain carries empty vector. LSM1ϩdT indicates RNA from wild-type cells subjected to oligo(dT)-mediated RNaseH reaction before running on the gel. Positions of normal and trimmed full-length MFA2pG mRNA and poly(G) fragment are indicated on the left. % poly(G) fragment and % trimmed poly(G) fragment are given below the lanes. % poly(G) fragment is calculated by taking the total signal [i.e., total full length plus total poly(G) fragment] as 100%. % trimmed poly(G) fragment is calculated by taking the total poly(G) fragment (i.e., trimmed fragment plus normal fragment) as 100%.

As shown in Figure 2, fragment levels relative to the ing a clear defect in mRNA decapping (Figure 3). full length (quantitated using the PhosphorImager) While lsm1-8 showed a strong phenotype (comparable -of total signal) to lsm1⌬), lsm1-9 and lsm1-14 showed a moderate phe %15ف) were very low in lsm1⌬ cells -of total signal). notype. These results argue that RNA binding is re %40ف) compared to the wild-type cells It should be noted that because both the MFA2pG full- quired for Lsm1p’s role in mRNA decapping. length mRNA and the poly(G) fragment accumulate 3Ј- Second, we observed that mutations in the predicted trimmed species in addition to the normal species (see inter-subunit contact residues, lsm1-6, lsm1-7, and below) in lsm mutants (He and Parker 2001), the inten- lsm1-13 all showed defects in decapping as assessed sities of normal and trimmed species were added up to by both the levels of mRNA decay fragment and the determine the levels of the full-length mRNA and the accumulation of deadenylated full-length mRNA (Fig- fragment. ure 2). Again, measurement of the MFA2pG mRNA Figure 2 also reveals the strong accumulation in the decay rate in these mutants revealed a prolonged half- ⌬ lsm1 strain of the deadenylated full-length MFA2pG life, indicating a clear defect in mRNA decapping mRNA. This is characteristic of strains defective in de- (Figure 3). The lsm1-7 and lsm1-13 alleles were almost capping (Beelman et al. 1996; Dunckley and Parker as strong as a null allele, whereas the lsm1-6 allele 1999; Tharun et al. 2000). showed a moderate phenotype. Thus, these results Examination of the various mutants revealed the fol- support the idea that maintenance of inter-subunit lowing: contacts is of critical importance for Lsm1p’s func- First, mutants with lesions in the predicted RNA-binding tion. surface, lsm1-8, lsm1-9, and lsm1-14 all showed defects Third, we observed that deletions within the C-terminal in mRNA decay. Each mutant showed a decreased extension (lsm1-27, lsm1-28, and lsm1-29) resulted in accumulation of the mRNA decay intermediate rela- a clear inhibition of mRNA decapping as assessed by tive to the full-length mRNA. In addition, all three both the poly(G) fragment levels and the accumula- alleles accumulated deadenylated full-length mRNAs, tion of deadenylated mRNA (Figure 2). In fact, even consistent with a defect in decapping (Figure 2). More- fusing a FLAG-peptide at the C terminus of otherwise over, measurement of MFA2pG mRNA decay rate in wild-type Lsm1p (lsm1-31; Table 2) led to a partial these mutants showed a prolonged half-life, indicat- defect in mRNA decapping (Figure 2). These observa- 40 S. Tharun et al.

ever, the alleles lsm1-16 through lsm1-24, which carry lesions in different clusters of charged amino acids residing in the C-terminal domain, failed to show any significant growth or mRNA decay defects, indicating that smaller perturbations in this domain can be toler- ated (Table 2). Fourth, we observed that mutations (lsm1-1 through lsm1-4) or deletions (lsm1-25 and lsm1-26) in the Lsm1p primary sequence that is N-terminal to the Sm domain did not lead to any significant decapping defect as assessed from the accumulation of the poly(G) fragment or deadenylated full-length mRNA (Figure 2 and Table 2). This observation was consistent with these lesions not affecting growth at high temperature (Table 2). Similarly, mutations in the nonconserved region that is located between the Sm motifs I and II (lsm1-11 and lsm1-12) also did not lead to any sig- nificant phenotype (Table 1). This suggests that the residues affected in these mutants do not play a major role in Lsm1p function. -Effect of LSM1 mutations on mRNA-3؅ end protec tion: We also examined the effect of the lsm1 alleles on the protection of the mRNA 3Ј end from a trimming reaction. Specifically, earlier studies (Boeck et al. 1998; He and Parker 2001) showed that in lsm1 through lsm7 mutants several mRNAs accumulate species that are to 20 nucleotides 10ف truncated at their 3Ј ends by (“trimmed species”) in addition to the normal full- length species in vivo. This indicates that the Lsm1p–7p complex is involved in protecting mRNA 3Ј ends in vivo. In the case of MFA2pG and PGK1pG mRNAs, poly(G) fragments generated from these mRNAs also accumu- late as normal and trimmed species in the lsm mutants (He and Parker 2001). While the trimmed and normal species of full-length mRNA run very close together in the gel, the trimmed poly(G) fragment and normal poly(G) fragment are more easily separable. Therefore, Figure 3.—MFA2pG mRNA half-life in lsm1 mutants. Cells were grown as described for Figure 2. At log phase, they were determining the ratio of the two species of the poly(G) collected and resuspended in fresh Ϫura medium containing fragment is a convenient way to assess the degree of glucose but no galactose. Aliquots of culture were removed mRNA-3Ј end protection occurring in a given strain. at specified time intervals (indicated above the lanes) follow- This was quantitated using the PhosphorImager from ing shift to glucose medium and cells were collected from Northern blots containing steady state RNA samples. As them. RNA was extracted from these cells and subjected to Northern analysis after separation on polyacrylamide gels as seen in Figure 2, wild-type cells had very little trimmed described for Figure 2. Bands were visualized and the intensity fragment (Յ10% of total fragment). However, lsm1⌬ to 60% 55ف) of full-length MFA2pG mRNA bands quantitated using the cells had high levels of trimmed fragment PhosphorImager. Blots were stripped and reprobed for 7S of the total fragment levels). RNA. 7S RNA bands were also quantitated in each lane with An important result was that all of the lesions that af- the PhosphorImager and those values were used for normaliz- ing the MFA2pG full-length mRNA levels. Calculated half-life fected mRNA decapping also affected mRNA trimming. values are indicated on the right. lsm1 alleles carried by the Specifically, all the mutants affecting the residues pre- strains used for the experiments are indicated on the left. dicted to be involved in RNA and inter-subunit contacts showed significant accumulations of trimmed fragment, with lsm1-7, lsm1-8, lsm1-9, lsm1-13, and lsm1-14 showing tions were further supported by the longer half-life strong defects. The mutant lsm1-6 showed a moderate of MFA2pG mRNA in these mutants (Figure 3). These accumulation of the trimmed fragment that is consistent results indicate that the C terminus has an important with the moderate decapping defect seen in this mutant. function in Lsm1p’s role in mRNA decapping. How- Further, the C-terminal deletion mutants lsm1-27, lsm1- Mutagenic Analysis of Yeast Lsm1p 41

Figure 4.—Comparison of the Lsm1p protein levels in wild-type and lsm1 mutant cells. Cells were grown in Ϫura liquid medium at 25Њ and collected at log phase. Lysates were made from the various strains (indicated on top of the lanes) and subjected to Western analysis as described in materials and methods to visualize the Lsm1p protein. Species of higher mobility seen in several lanes (marked by arrowheads) are presumably degraded fragments of Lsm1p. The mutant protein produced in lsm1-13 shows an aberrant lower mobility than wild-type Lsm1p in spite of being equal to the wild-type protein in length.

28, and lsm1-29 and the C-terminal FLAG-tagged allele dicted RNA-binding surfaces of Lsm1p affect the re- lsm1-31 also showed higher levels of trimmed fragment. cruitment of either Lsm1p or the other Lsm proteins These results demonstrate a strong correlation between to the P-bodies. the impairment of mRNA decapping and the inability First, we examined the intracellular localization of to protect mRNA 3Ј end from trimming, suggesting that the GFP-tagged forms of the mutant proteins Lsm1-9p these two functions are related (see discussion). and Lsm1-14p (which bear lesions in the predicted Expression of mutant Lsm1p proteins: It is possible RNA-binding surface of Lsm1p) or wild-type Lsm1p. that some of the mutants described above show a strong We performed the studies with cells subjected to stress phenotype due to their inability to accumulate the mu- (incubation in glucose-free medium for 10 min) since tant Lsm1p adequately. To test this possibility, we com- P-bodies are larger and more easily visualized in stressed pared Lsm1p expression in the lsm1 mutants that are cells than in cells that are not subjected to stress (Teixe- defective in mRNA decay and wild-type cells, by Western ira et al. 2005). We observed that both Lsm1-9p and analysis of the lysates prepared from the corresponding Lsm1-14p localized to P-bodies just like the wild-type strains using anti-Lsm1p antiserum. Cells grown in Ϫura Lsm1p protein (Figure 5). This indicates that the RNA- liquid medium at 25Њ and collected at log phase were binding surface of Lsm1p is not absolutely required for used for making the lysates. As seen in Figure 4, the the recruitment of Lsm1p to the P-bodies. decay-defective mutants, lsm1-6, lsm1-8, lsm1-9, lsm1-13, Next, we examined how defects in Lsm1p affect the and lsm1-14 expressed the mutant protein at levels com- localization of other components of the Lsm1p–7p com- parable to wild-type protein. On the other hand, the plex. We examined the localization of Lsm2p-GFP or results showed that the Lsm1p band was very weak or Lsm7p-GFP in LSM1, lsm1-9, lsm1-14, and lsm1⌬ cells undetectable in the lysates of mutants lsm1-7, lsm1-27, that were subjected to stress. We observed that neither and lsm1-28. At present, it is not clear whether this is Lsm7p-GFP nor Lsm2p-GFP was localized to P-bodies due to low accumulation of the mutant Lsm1p in these in lsm1⌬ cells (Figure 6). However, both Lsm2p-GFP cells or due to loss of reactivity of the mutant protein and Lsm7p-GFP localized to P-bodies in the lsm1-9 and with the antibodies as a result of the mutations/dele- lsm1-14 mutants (Figure 6). Therefore, these results sug- tions (especially in the case of lsm1-27 and lsm1-28 that gest that even when the direct contacts between Lsm1p have large deletions). and RNA are impaired, Lsm1p could be recruited to Effect of mutations affecting the predicted RNA-con- the P-bodies by virtue of its interactions with the other tacting surfaces of Lsm1p on the localization of the subunits of the Lsm1p–7p complex. Lsm proteins to P-bodies: Recent studies have suggested that decapping and the 5Ј to 3Ј degradation of mRNA occur in discrete cytoplasmic structures called P-bodies, DISCUSSION where the Lsm1p through Lsm7p proteins and many The mechanism by which the Lsm1p–7p complex other decay factors including the decapping enzyme activates mRNA decapping is not known. Since Lsm1p are concentrated in both yeast and mammalian cells is the distinguishing subunit of this complex, we have (Ingelfinger et al. 2002; Van Dijk et al. 2002; Sheth analyzed the critical functional features of this protein. and Parker 2003; Cougot et al. 2004). It is not known Our results support the idea that the functioning of how the Lsm1p through Lsm7p proteins are recruited the Lsm1p–7p complex is similar to the other Sm-like to the P-bodies, although studies in mammalian cells protein complexes at a mechanistic level. When the have suggested that these proteins need to be a part of Lsm1p residues that are likely to function in inter-sub- the Lsm1p–7p complex to be efficiently recruited to unit contacts and RNA binding were predicted on the the P-bodies (Ingelfinger et al. 2002). One possibility basis of the homology of Lsm1p to other Sm-like pro- is that the interaction of the Lsm1p–7p complex with teins (for whom such residues were identified experi- RNA is important for the recruitment of the Lsm1p–7p mentally) and mutagenized, it resulted in mRNA decay complex to the P-bodies. To test this possibility, we per- defects. This indicates that the organization of such formed experiments to determine if lesions in the pre- crucial residues is similar between Lsm1p and other 42 S. Tharun et al.

the region C-terminal to the Sm domain led to signifi- cant mRNA decay defects (Table 2). Thus, the Lsm1p protein appears to have two critical regions for function. Several observations indicate that the stretch of amino acids N-terminal to the Sm domain of Lsm1p (residues 1 to 26) is not required for Lsm1p function. First, dele- tions or point mutations in this region do not affect growth at high temperature (Table 2). Second, they also do not affect mRNA decapping as assessed by the ability to accumulate poly(G) fragment or deadenylated mRNA (Table 2 and Figure 2). Third, this region is almost completely absent in the human homolog of Lsm1p, although the region C-terminal to the Sm do- main is reasonably well conserved in the human protein (Figure 1c). In fact, even in fungal species that are very closely related to Saccharomyces cerevisiae, the N-terminal extension is not highly conserved. Our results argue that the C-terminal extension of Lsm1p is required for its function. Lsm1p has a 61- amino-acids-long region C-terminal to the Sm domain that is absent in other members of the Sm-like protein family. The following observations indicate that this re- gion is important. First, deletions in this region (lsm1-27, lsm1-28, and lsm1-29) inhibited mRNA decay (Figures 2 and 3). Although the Western analyses indicate that the mutant (truncated) Lsm1p accumulation may be severely reduced in lsm1-27 and lsm1-28, they reveal nor- mal accumulation of Lsm1-29p, indicating that at least the phenotype of lsm1-29 is attributable to the deletion (Figure 4). Second, the allele lsm1-31, which is a fusion of the FLAG peptide at the C terminus of an otherwise wild-type Lsm1p was also defective in mRNA decay and makes normal levels of the fusion protein (Figures 2, 3, and 4). Importantly, fusion of the FLAG epitope at the N terminus of Lsm1p has no effect on mRNA decay or mRNA 3Ј end protection (not shown). Finally, all the C-terminal deletion mutants were also defective in mRNA 3Ј end protection (Figure 2). The functional importance of Lsm1p’s C-terminal extension is also sup- ported by the observation that it is reasonably well con- Figure 5.—Determination of the effect of mutations affect- served in human Lsm1p (Figure 1c). Further, genetic ing the predicted RNA-binding surface of Lsm1p on the local- ization of Lsm1p to P-bodies. Yeast cells expressing GFP fu- interaction studies of Boeck et al. (1998) have shown sions of LSM1 or lsm1-9 or lsm1-14 (indicated on the left) were that an lsm1 allele (spb8-1) with intact Sm domain but observed by confocal microscopy as described in materials affected in the C-terminal domain due to a frameshift and methods. mutation is able to suppress the lethality of pab1⌬ like several other decapping mutants, including lsm1⌬ (Caponigro and Parker 1995; Boeck et al. 1998; Sm-like proteins. Further, this is also consistent with the Wyers et al. 2000). Surprisingly, our studies also indi- idea that just like the other Sm-like protein complexes, cate that smaller perturbations in the C-terminal exten- the Lsm1p–7p complex also needs to directly bind to sion of Lsm1p can be tolerated, since the alleles lsm1- the RNA substrate to act on it (i.e., activate decapping). 15 through lsm1-24 (each of which carry changes in a Our studies indicate that most of the residues that small set of neighboring charged residues in the region are crucial for Lsm1p function reside in the conserved C-terminal to the Sm domain) do not show any defect regions, Sm motifs I and II, of the Lsm1p protein. Muta- in mRNA decay. This could be because multiple redun- tions affecting any small set of neighboring residues dant sites in this region mediate functionally important located outside these regions did not lead to any signifi- interactions, perhaps as part of a larger protein–protein cant mRNA decay defects, although large deletions in interaction surface. Mutagenic Analysis of Yeast Lsm1p 43

Figure 6.—Determination of the effect of mutations affecting the predicted RNA-binding surface of Lsm1p on the localization of Lsm2p and Lsm7p to P-bodies. Wild-type (LSM1) or mutant (lsm1⌬, lsm1-9,orlsm1-14) cells (indicated on top) expressing GFP tagged versions of either Lsm2p (top) or Lsm7p (bottom) were observed by confocal microscopy as described in materials and methods.

Our results indicate that the residues of Lsm1p that the fact that neither the primary sequence nor the are predicted (on the basis of homology) to be involved length of this region is conserved among the members in inter-subunit contacts are indeed crucial for Lsm1p of the Sm-like protein family. Given this, the primary function. This is shown by the significant mRNA decay function of this loop may be to provide a structural defect exhibited by the mutants in which these residues turn between the third and fourth ␤-sheets in the core are changed (lsm1-6, lsm1-7, and lsm1-13; see Figures 2 structure. and 3). Although Western analyses revealed that the The Sm and Sm-like protein complexes are known accumulation of the mutant Lsm1p may be severely to bind directly to their substrate RNA molecules as impaired in lsm1-7, they showed normal levels of accu- shown by several studies (Achsel et al. 1999, 2001; mulation in the case of Lsm1–6p and Lsm1–13p. Since Raker et al. 1999; Toro et al. 2001; Moller et al. 2002; the residues mutated in these alleles are homologous Schumacher et al. 2002). Our studies suggest that in to the experimentally determined inter-subunit contact an analogous manner, the Lsm1p–7p complex could residues in other Sm-like proteins, these observations also bind directly to mRNA. This is supported by the are consistent with the idea that homologous regions observation that all the three mutants affected in the serve similar functions in the different Sm-like proteins. residues predicted to play a key role in RNA binding Importantly, these results suggest that Lsm1p needs to (lsm1-8, lsm1-9, and lsm1-14) are impaired in mRNA be a part of the Lsm1p–7p complex to execute its func- decay. This indicates that the ability to directly interact tion efficiently. Recent studies have identified cyto- with mRNA is important for the mRNA decapping func- plasmic structures, referred to as P-bodies, which are the tion of the Lsm1p–7p complex. Further, these results sites of decapping (Sheth and Parker 2003; Cougot et also suggest that the Lsm1p–7p complex is similar to the al. 2004). Mutations in human Lsm4p that are predicted Sm complex and the other Sm-like protein complexes in to disrupt its interactions with other Lsm proteins in- the organization of its RNA contacting surfaces and the hibit its entry into P-bodies (Ingelfinger et al. 2002). manner in which it contacts RNA. Therefore, it is possible that in lsm1-6 and lsm1-13, the Studies in mammalian cells have shown that human defect in decapping is due to the inefficient targeting Lsm4p cannot be recruited to P-bodies if its interactions of Lsm1p to P-bodies. with the other subunits of the Lsm1p–7p complex are Despite the clear importance of the Sm domain for disrupted by mutations (Ingelfinger et al. 2002). Our Lsm1p function, mutations in the region that separates results in Figure 6 show that Lsm2p-GFP and Lsm7p- the Sm motifs I and II failed to have any effect on mRNA GFP are not concentrated in P-bodies in lsm1⌬ cells, decay. This region corresponds to loop 4 in the modeled indicating that the recruitment of these proteins to tertiary structure of Lsm1p (Figure 1b). Both lsm1-11 P-bodies is dependent on Lsm1p. Thus, together these and lsm1-12, which bear lesions in this region, were able results support the idea that the recruitment of any of to grow at 36Њ and accumulate wild-type levels of poly(G) the subunits of the Lsm1p–7p is dependent on the abil- fragment (Table 1). This observation is consistent with ity of that subunit to be incorporated into the Lsm1p–7p 44 S. Tharun et al. complex. Therefore, it is likely that the Lsm1p–7p com- age of yeast mRNAs contain a short stretch of U plex is recruited as a unit to the P-bodies. On the other residues at their 3Ј end (Graber et al. 1999). Given hand, our results also indicate that mutations disrupting the fact that the Lsm1p–7p complex interacts with the predicted RNA-binding surfaces of Lsm1p do not and affects the decay of multiple mRNAs (Tharun affect the recruitment of the mutant Lsm1p or the other et al. 2000; Tharun and Parker 2001a), this sup- subunits of the Lsm1p–7p complex to the P-bodies (Fig- ports the idea that the Lsm1p–7p complex binds ures 5 and 6). This supports the idea that the mutations to the 3Ј ends of deadenylated mRNAs in vivo. in lsm1-9 and lsm1-14 affect the inter-subunit interac- tions and, hence, the assembly of the Lsm1p–7p com- On the basis of these observations, we propose that plex only minimally if at all. Since both lsm1-9 and lsm1- following deadenylation, the Lsm1p–7p complex di- Ј 14 mutants are defective in mRNA decay and 3Ј end rectly interacts with the 3 end of the deadenylated protection, these results further suggest that the Lsm1p mRNAs, which results in the protection of the mRNA Ј can be recruited to the P-bodies by virtue of its interac- 3 end. Further, this interaction also triggers mRNP tions with the other subunits of the Lsm1p–7p complex, rearrangement events that ultimately result in the activa- Ј even when the direct interactions between Lsm1p and tion of decapping. Importantly, mRNA 3 end protec- RNA are disrupted. tion by itself could be a crucial cellular function of the Our results document a strong correlation between Lsm1p–7p complex, since it could potentially regulate Ј Ј the ability of Lsm1p to promote decapping and its ability the rate of 3 to 5 decay of the mRNA. Such a control to protect the 3Ј end from trimming. The key observa- could be very important in the case of mRNAs whose Ј Ј tion is that all the mutants that are defective in mRNA major mode of decay is the 3 to 5 decay pathway. 3Ј protection are also defective in mRNA decay (Tables Consistent with this, the studies of He and Parker ⌬ 1 and 2 and Figures 2 and 3). This is consistent with (2001)suggest that the temperature sensitivity of lsm1 the model that an initial step in the activation of decap- cells is due to the increased susceptibility to (exosome Ј Ј ping is the interaction of the Lsm1p–7p complex with and ski proteins mediated) 3 to 5 degradation of some the 3Ј end of the mRNA substrate, and the 3Ј end protec- mRNAs that are essential at high temperature. Thus, tion is a consequence of such an interaction. for such mRNAs, the Lsm1p–7p complex could have An unresolved issue is the binding specificity of the a protective/stabilizing (rather than decay-promoting) Lsm1p–7p complex and where on the mRNA this com- function. Although recent studies have suggested that plex may interact. Several observations support the hy- mRNA decay occurs in P-bodies (Ingelfinger et al. pothesis that the Lsm1p–7p complex directly interacts 2002; Van Dijk et al. 2002; Sheth and Parker 2003; with 3Ј ends of deadenylated mRNAs in vivo: Cougot et al. 2004), it is not known what fraction of cellular mRNAs degrade outside the P-bodies. Interest- i. Co-immunoprecipitation studies indicate that Lsm1p ingly, the ski and exosomal proteins (which mediate and Lsm5p specifically associate with deadenylated the 3Ј to 5Ј mRNA decay) are not concentrated in mRNAs in vivo (Tharun and Parker 2001a). P-bodies but are uniformly distributed (Allmang et al. ii. The poly(G) decay intermediate of MFA2pG mRNA, 1999; Zanchin and Goldfarb 1999; van Hoof et al. which lacks Ͼ50% of the 5Ј portion of the mRNA 2000; Sheth and Parker 2003) in the cytoplasm, sug- (including the whole of the ORF and a part of the gesting that 3Ј to 5Ј decay is likely to occur outside 3Ј-UTR), also strongly co-immunoprecipitates with the P-bodies. Thus, the major outcome of Lsm1p–7p Lsm1p. This suggests that the 3Ј-UTR of this mRNA binding could be different for mRNAs that degrade contains a possible Lsm1p–7p complex-binding site. outside (stabilization) and inside (activation of decap- iii. The Lsm2p–8p and the Lsm2p–7p complexes that ping and decay) the P-bodies. are closely related to the Lsm1p–7p complex bind This work was supported by funds from the National Institutes of to the 3Ј ends of their substrates, the U6 snRNA Health (GM45443) and the Uniformed Services University of the and snR5, respectively (Achsel et al. 1999; Fernan- Health Sciences (C071GY). dez et al. 2004). iv. The fact that several deadenylated mRNAs accumu- Ј Ј late as 3 trimmed species (truncated at their 3 LITERATURE CITED to 20 nucleotides) in cells lacking any 10ف ends by Abler, M. L., and P. J. Green, 1996 Control of mRNA stability in of the Lsm1p through Lsm7p proteins is consistent higher plants. Plant Mol. Biol. 32: 63–78. with a lack of protection to the 3Ј ends of deadeny- Achsel, T., H. Brahms, B. Kastner, A. Bachi, M. Wilm et al., 1999 lated mRNAs in these cells. A doughnut-shaped heteromer of human Sm-like proteins binds to the 3Ј-end of U6 snRNA, thereby facilitating U4/U6 duplex v. The Sm-like protein complexes in general (includ- formation in vitro. EMBO J. 18: 5789–5802. ing the Lsm2p–8p complex) are known to bind Achsel, T., H. Stark and R. Luhrmann, 2001 The Sm domain is preferentially to U-rich regions of RNA (Achsel et an ancient RNA-binding motif with oligo(U) specificity. Proc. Natl. Acad. Sci. USA 98: 3685–3689. al. 1999, 2001; Raker et al. 1999; Toro et al. 2001; Allmang, C., E. Petfalski, A. Podtelejnikov, M. Mann, D. Toller- Schumacher et al. 2002) and a significant percent- vey et al., 1999 The yeast exosome and human PM-Scl are Mutagenic Analysis of Yeast Lsm1p 45

related complexes of 3Ј → 5Ј exonucleases. Genes Dev. 13: 2148– Gao, M., C. J. Wilusz, S. W. Peltz and J. Wilusz, 2001 A novel 2158. mRNA-decapping activity in HeLa cytoplasmic extracts is regu- Anderson, J. S. J., and R. P. Parker, 1998 The 3Ј to 5Ј degradation lated by AU-rich elements. EMBO J. 20: 1134–1143. of yeast mRNAs is a general mechanism for mRNA turnover that Graber, J. H., C. R. Cantor, S. C. Mohr and T. F. Smith, 1999 In requires the SKI2 DEVH box protein and 3Ј to 5Ј exonucleases silico detection of control signals: mRNA 3Ј-end -processing se- of the exosome complex. EMBO J. 17: 1497–1506. quences in diverse species. Proc. Natl. Acad. Sci. USA 96: 14055– Bates, P. A., and M. J. Sternberg, 1999 Model building by compari- 14060. son at CASP3: using expert knowledge and computer automation. Guhaniyogi, J., and G. Brewer, 2001 Regulation of mRNA stability Proteins (Suppl 3): 47–54. in mammalian cells. Gene 265: 11–23. Bates, P. A., L. A. Kelley, R. M. MacCallum and M. J. Sternberg, Gumbs, A. A., C. Bassi, P. S. Moore, M. Falconi, I. Frigerio et al., 2001 Enhancement of protein modeling by human intervention 2002 Overexpression of the Sm-like proto-oncogene in primary in applying the automatic programs 3D-JIGSAW and 3D-PSSM. and metastatic pancreatic endocrine tumors. JOP 3: 109–115. Proteins (Suppl 5): 39–46. Hatfield, L., C. A. Beelman, A. Stevens and R. Parker, 1996 Muta- Beelman, C. A., A. Stevens, G. Caponigro, T. E. LaGrandeur, L. tions in trans-acting factors affecting mRNA decapping in Saccha- Hatfield et al., 1996 An essential component of the decapping romyces cerevisiae. Mol. Cell. Biol. 16: 5830–5838. enzyme required for normal rates of mRNA turnover. Nature He, W., and R. Parker, 2001 The yeast cytoplasmic LsmI/Pat1p 382: 642–646. complex protects mRNA 3Ј termini from partial degradation. Boeck, R., B. Lapeyre, C. E. Brown and A. B. Sachs, 1998 Capped Genetics 158: 1445–1455. mRNA degradation intermediates accumulate in the yeast spb8–2 Hentze, M. W., 1991 Determinants and regulation of cytoplasmic mutant. Mol. Cell. Biol. 18: 5062–5072. mRNA stability in eukaryotic cells. Biochim. Biophys. Acta 1090: Bonnerot, C., R. Boeck and B. Lapeyre, 2000 The two proteins 281–292. Pat1p (Mrt1p) and Spb8p interact in vivo, are required for mRNA Hermann, H., P. Fabrizio, V. A. Raker, K. Foulaki, H. Hornig et decay, and are functionally linked to Pab1p. Mol. Cell. Biol. 20: al., 1995 snRNP Sm proteins share two evolutionarily conserved 5939–5946. sequence motifs which are involved in Sm protein-protein interac- Bouveret, E., G. Rigaut, A. Shevchenko, M. Wilm and B. Seraphin, tions. EMBO J. 14: 2076–2088. 2000 A Sm-like protein complex that participates in mRNA Hsu, C. L., and A. Stevens, 1993 Yeast cells lacking 5Ј→3Ј exoribo- degradation. EMBO J. 19: 1661–1671. nuclease 1 contain mRNA species that are poly(A) deficient and Cao, D., and R. Parker, 2001 Computational modeling of eukaryo- partially lack the 5Ј cap structure. Mol. Cell. Biol. 13: 4826–4835. tic mRNA turnover. RNA 7: 1192–1212. Ingelfinger, D., D. J. Arndt-Jovin, R. Luhrmann and T. Achsel, Caponigro, G., and R. Parker, 1995 Multiple functions for the 2002 The human LSm1–7 proteins colocalize with the mRNA- poly(A)-binding protein in mRNA decapping and deadenylation degrading enzymes Dcp1/2 and Xrnl in distinct cytoplasmic foci. in yeast. Genes Dev. 9: 2421–2432. RNA 8: 1489–1501. Coller, J., and R. Parker, 2004 Eukaryotic mRNA decapping. Kambach, C., S. Walke, R. Young, J. M. Avis, E. de la Fortelle et Annu. Rev. Biochem. 73: 861–890. al., 1999 Crystal structures of two Sm protein complexes and Coller, J. M., N. K. Gray and M. P. Wickens, 1998 mRNA stabiliza- their implications for the assembly of the spliceosomal snRNPs. tion by poly(A)-binding protein is independent of poly(A) and Cell 96: 375–387. requires translation. Genes Dev. 12: 3226–3235. Khanna, R., and M. Kiledjian, 2004 Poly(A)-binding-protein-medi- Coller, J. M., M. Tucker, U. Sheth, M. A. Valencia-Sanchez and ated regulation of hDcp2 decapping in vitro. EMBO J. 23: 1968– R. Parker, 2001 The DEAD box helicase, Dhh1p, functions in 1976. mRNA decapping and interacts with both the decapping and Kshirsagar, M., and R. Parker, 2004 Identification of Edc3p as an deadenylase complexes. RNA 7: 1717–1727. enhancer of mRNA decapping in Saccharomyces cerevisiae. Genetics Collins, B. M., S. J. Harrop, G. D. Kornfeld, I. W. Dawes, P. M. 166: 729–739. Curmi et al., 2001 Crystal structure of a heptameric Sm-like Lerner, M. R., and J. A. Steitz, 1981 Snurps and scyrps. Cell 25: protein complex from archaea: implications for the structure 298–300. and evolution of snRNPs. J. Mol. Biol. 309: 915–923. Liu, H., N. D. Rodgers, X. Jiao and M. Kiledjian, 2002 The scaven- Contreras-Moreira, B., and P. A. Bates, 2002 Domain fishing: a ger mRNA decapping enzyme DcpS is a member of the HIT first step in protein comparative modelling. Bioinformatics 18: family of pyrophosphatases. EMBO J. 21: 4699–4708. 1141–1142. Luhrmann, R., B. Kastner and M. Bach, 1990 Structure of spliceo- Cooper, M., L. H. Johnston and J. D. Beggs, 1995 Identification somal snRNPs and their role in pre-mRNA splicing. Biochim. and characterization of Uss1p (Sdb23p): a novel U6 snRNA- Biophys. Acta 1087: 265–292. associated protein with significant similarity to core proteins of Lykke-Andersen, J., 2002 Identification of a human decapping com- small nuclear ribonucleoproteins. EMBO J. 14: 2066–2075. plex associated with hUpf proteins in nonsense-mediated decay. Cougot, N., S. Babajko and B. Seraphin, 2004 Cytoplasmic foci Mol. Cell. Biol. 22: 8114–8121. are sites of mRNA decay in human cells. J. Cell Biol. 165: 31–40. Mayes, A. E., L. Verdone, P. Legrain and J. D. Beggs, 1999 Charac- Couttet, P., M. Fromont-Racine, D. Steel, R. Pictet and T. terization of Sm-like proteins in yeast and their association with Grange, 1997 Messenger RNA deadenylylation precedes de- U6 snRNA. EMBO J. 18: 4321–4331. capping in mammalian cells. Proc. Natl. Acad. Sci. USA 94: 5628– Moller, T., T. Franch, P. Hojrup, D. R. Keene, H. P. Bachinger 5633. et al., 2002 Hfq: a bacterial Sm-like protein that mediates RNA- Decker, C. J., and R. Parker, 1993 A turnover pathway for both RNA interaction. Mol. Cell 9: 23–30. stable and unstable mRNAs in yeast: evidence for a requirement Muhlrad, D., and R. Parker, 1992 Mutations affecting stability for deadenylation. Genes Dev. 7: 1632–1643. and deadenylation of the yeast MFA2 transcript. Genes Dev. 6: Decker, C. J., and R. Parker, 2002 mRNA decay enzymes: decappers 2100–2111. conserved between yeast and mammals. Proc. Natl. Acad. Sci. Muhlrad, D., C. J. Decker and R. Parker, 1994 Deadenylation of USA 99: 12512–12514. the unstable mRNA encoded by the yeast MFA2 gene leads to Dunckley, T., and R. Parker, 1999 The DCP2 protein is required decapping followed by 5Ј→3Ј digestion of the transcript. Genes for mRNA decapping in Saccharomyces cerevisiae and contains a Dev. 8: 855–866. functional MutT motif. EMBO J. 18: 5411–5422. Muhlrad, D., C. J. Decker and R. Parker, 1995 Turnover mecha- Dunckley, T., M. Tucker and R. Parker, 2001 Two related pro- nisms of the stable yeast PGK1 mRNA. Mol. Cell. Biol. 15: 2145– teins, Edc1p and Edc2p, stimulate mRNA decapping in Saccharo- 2156. myces cerevisiae. Genetics 157: 27–37. Mura, C., D. Cascio, M. R. Sawaya and D. S. Eisenberg, 2001 The Fernandez, C. F., B. K. Pannone, X. Chen, G. Fuchs and S. L. crystal structure of a heptameric archaeal Sm protein: Implica- Wolin, 2004 An Lsm2-Lsm7 Complex in Saccharomyces cerevisiae tions for the eukaryotic snRNP core. Proc. Natl. Acad. Sci. USA Associates with the Small Nucleolar RNA snR5. Mol. Biol. Cell 98: 5532–5537. 15: 2842–2852. Mura, C., A. Kozhukhovsky, M. Gingery, M. Phillips and D. Eisen- Fischer, N., and K. Weis, 2002 The DEAD box protein Dhh1 stimu- berg, 2003 The oligomerization and ligand-binding properties lates the decapping enzyme Dcp1. EMBO J. 21: 2788–2797. of Sm-like archaeal proteins (SmAPs). Protein Sci. 12: 832–847. 46 S. Tharun et al.

Pannone, B. K., D. Xue and S. L. Wolin, 1998 A role for the yeast ping: displacement of translation factors and association of the La protein in U6 snRNP assembly: evidence that the La protein Lsm1p-7p complex on deadenylated yeast mRNAs. Mol. Cell 8: is a molecular chaperone for RNA polymerase III transcripts. 1075–1083. EMBO J. 17: 7442–7453. Tharun, S., and R. Parker, 2001b Turnover of mRNA in eukaryotic Pannone, B. K., S. D. Kim, D. A. Noe and S. L. Wolin, 2001 Multiple cells, pp. 245–257 in RNA, edited by D. Soll,S.Nishimura and functional interactions between components of the Lsm2-Lsm8 P. Moore. Pergamon, New York/Oxford. complex, U6 snRNA, and the yeast La protein. Genetics 158: Tharun, S., W. He, A. E. Mayes, P. Lennertz, J. D. Beggs et al., 187–196. 2000 Yeast Sm-like proteins function in mRNA decapping and Peltz, S. W., G. Brewer, P. Bernstein, P. A. Hart and J. Ross, decay. Nature 404: 515–518. 1991 Regulation of mRNA turnover in eukaryotic cells. Crit. Toro, I., S. Thore, C. Mayer, J. Basquin, B. Seraphin et al., 2001 Rev. Eukaryot. Gene Expr. 1: 99–126. RNA binding in an Sm core domain: X-ray structure and func- Piccirillo, C., R. Khanna and M. Kiledjian, 2003 Functional char- tional analysis of an archaeal Sm protein complex. EMBO J. 20: acterization of the mammalian mRNA decapping enzyme hDcp2. 2293–2303. RNA 9: 1138–1147. Tseng-Rogenski, S. S., J. L. Chong, C. B. Thomas, S. Enomoto, Raker, V. A., K. Hartmuth, B. Kastner and R. Luhrmann, 1999 J. Berman et al., 2003 Functional conservation of Dhh1p, a Spliceosomal U snRNP core assembly: Sm proteins assemble onto cytoplasmic DExD/H-box protein present in large complexes. an Sm site RNA nonanucleotide in a specific and thermodynami- Nucleic Acids Res. 31: 4995–5002. cally stable manner. Mol. Cell. Biol. 19: 6554–6565. Tucker, M., M. A. Valencia-Sanchez, R. R. Staples, J. Chen, C. L. Ramirez, C. V., C. Vilela, K. Berthelot and J. E. McCarthy, 2002 Denis et al., 2001 The transcription factor associated Ccr4 and Modulation of eukaryotic mRNA stability via the cap-binding Caf1 proteins are components of the major cytoplasmic mRNA translation complex eIF4F. J. Mol. Biol. 318: 951–962. deadenylase in Saccharomyces cerevisiae. Cell 104: 377–386. Salgado-Garrido, J., E. Bragado-Nilsson, S. Kandels-Lewis and Urlaub, H., V. A. Raker, S. Kostka and R. Luhrmann, 2001 Sm B. Seraphin, 1999 Sm and Sm-like proteins assemble in two protein-Sm site RNA interactions within the inner ring of the related complexes of deep evolutionary origin. EMBO J. 18: 3451– spliceosomal snRNP core structure. EMBO J. 20: 187–196. 3462. Van Dijk, E., N. Cougot, S. Meyer, S. Babajko, E. Wahle et al., Schumacher, M. A., R. F. Pearson, T. Moller, P. Valentin-Hansen 2002 Human Dcp2: a catalytically active mRNA decapping en- and R. G. Brennan, 2002 Structures of the pleiotropic transla- zyme located in specific cytoplasmic structures. EMBO J. 21: tional regulator Hfq and an Hfq-RNA complex: a bacterial Sm- 6915–6924. like protein. EMBO J. 21: 3546–3556. Van Dijk, E., H. Le Hir and B. Seraphin, 2003 DcpS can act in Schwartz, D. C., and R. Parker, 1999 Mutations in translation the 5Ј-3Ј mRNA decay pathway in addition to the 3Ј-5Ј pathway. initiation factors lead to increased rates of deadenylation and Proc. Natl. Acad. Sci. USA 100: 12081–12086. decapping of mRNAs in Saccharomyces cerevisiae. Mol. Cell. Biol. van Hoof, A., R. R. Staples, R. E. Baker and R. Parker, 2000 Func- 19: 5247–5256. tion of the ski4p (Csl4p) and Ski7p proteins in 3Ј-to-5Ј degrada- Schwartz, D. C., and R. Parker, 2000 mRNA decapping in yeast tion of mRNA. Mol. Cell. Biol. 20: 8230–8243. requires dissociation of the cap binding protein, eukaryotic trans- Vilela, C., C. Velasco, M. Ptushkina and J. E. McCarthy, 2000 lation initiation factor 4E. Mol. Cell. Biol. 20: 7933–7942. The eukaryotic mRNA decapping protein Dcp1 interacts physi- Schwartz, D., C. J. Decker and R. Parker, 2003 The enhancer of cally and functionally with the eIF4F translation initiation com- decapping proteins, Edc1p and Edc2p, bind RNA and stimulate plex. EMBO J. 19: 4372–4382. the activity of the decapping enzyme. RNA 9: 239–251. Wang, Z., and M. Kiledjian, 2001 Functional link between the Schweinfest, C. W., M. W. Graber, J. M. Chapman, T. S. Papas, P. L. mammalian exosome and mRNA decapping. Cell 107: 751–762. Baron et al., 1997 CaSm: an Sm-like protein that contributes to Wang, Z., X. Jiao, A. Carr-Schmid and M. Kiledjian, 2002 The the transformed state in cancer cells. Cancer Res. 57: 2961–2965. hDcp2 protein is a mammalian mRNA decapping enzyme. Proc. Seraphin, B., 1992 The HIT protein family: a new family of proteins Natl. Acad. Sci. USA 99: 12663–12668. present in prokaryotes, yeast and mammals. DNA Seq. 3: 177–179. Wilusz, C. J., M. Gao, C. L. Jones, J. Wilusz and S. W. Peltz, 2001 Poly(A)-binding proteins regulate both mRNA deadenylation Seraphin, B., 1995 Sm and Sm-like proteins belong to a large family: and decapping in yeast cytoplasmic extracts. RNA 7: 1416–1424. identification of proteins of the U6 as well as the U1, U2, U4 Wyers, F., M. Minet, M. E. Dufour, L. T. Vo and F. Lacroute, and U5 snRNPs. EMBO J. 14: 2089–2098. 2000 Deletion of the PAT1 gene affects translation initiation Sheth, U., and R. Parker, 2003 Decapping and decay of messenger and suppresses a PAB1 gene deletion in yeast. Mol. Cell. Biol. RNA occur in cytoplasmic processing bodies. Science 300: 805– 20: 3538–3549. 808. Zanchin, N. I., and D. S. Goldfarb, 1999 Nip7p interacts with Sikorski, R. S., and P. Hieter, 1989 A system of shuttle vectors and Nop8p, an essential nucleolar protein required for 60S ribosome yeast host strains designed for efficient manipulation of DNA in biogenesis, and the exosome subunit Rrp43p. Mol. Cell. Biol. 19: Saccharomyces cerevisiae. Genetics 122: 19–27. 1518–1525. Teixeira, D., U. Sheth, M. A. Valencia-Sanchez, M. Brengues and Zaric, B., M. Chami, H. Remigy, A. Engel, K. Ballmer-Hofer et R. Parker, 2005 Processing bodies require RNA for assembly al., 2005 Reconstitution of two recombinant Lsm protein com- and contain non-translating mRNAs. RNA 11: 371–382. plexes reveals aspects of their architecture, assembly and func- Tharun, S., and R. Parker, 1999 Analysis of mutations in the yeast tion. J. Biol. Chem. (in press). mRNA decapping enzyme. Genetics 151: 1273–1285. Tharun, S., and R. Parker, 2001a Targeting an mRNA for decap- Communicating editor: S. Sandmeyer