The Lsm Proteins: Ring Architectures for RNA Capture

1. Introduction

It is today recognized that the vast majority of the cellular pool of RNA (nearly 98% in humans) comprises non-coding RNA (ncRNA) species (Mattick, 2001), with only a small proportion serving as direct template for protein synthesis. The diverse ncRNA forms are themselves capable of function, involved in a plethora of tasks such as protein scaffolding, cis and trans regulatory roles and catalysis (Lilley, 2005; Mattick & Makunin, 2006). Many of these functions are carried out in tight partnership with specific ancillary proteins within large ribonucleo-protein complexes (RNPs) (Eddy, 2001).

Various types of ncRNA, as well as RNPs containing tRNA, rRNA or snRNA, directly interact with mRNA at different stages of its life. Figure 1 presents an overview of the maturation of pre-mRNA and the fate of the mRNA generated. Pre-mRNA initially undergoes modification to enhance its stability: a 5’ methyl guanosine (m₇G) cap added during transcription (Wen & Shatkin, 1999) and a poly(A)-tail placed in the 3’ region by the polyadenylation machinery (Proudfoot et al., 2002; Balbo & Bohm, 2007). Following initiation of spliceosomal assembly by recruitment of core particles in the cytoplasm, non-coding introns are spliced from the pre-mRNA sequence by the mature spliceosome in the nucleus (Crick, 1979; Pozzoli et al., 2002). This multi-megadalton complex itself contains 170 protein components and various types of snRNA, rivaling the ribosome in molecular complexity (Wahl et al., 2009).

Within the spliceosome, several distinct small nuclear RNP (snRNP) core complexes each contain snRNA organized around specific ring-structured protein assemblies. For those known as U1-, U2-, U4- and U5-snRNPs, these ring scaffolds are provided by members of the Sm protein family (Luhrmann et al., 1990), recruited to their specific snRNA partners in the cytoplasm at a distinct Sm-site of bases (Urlaub et al., 2001; Peng & Gallwitz 2004). The core snRNPs are reimported into the nucleus for further processing and spliceosome assembly (Will & Luhrmann, 2001; Patel & Bellini 2008). In contrast, U6 snRNA is first modified within the nucleoli and then engages with a related protein ring, in this case containing Lsm (“Sm-like”) proteins Lsm2-Lsm8. Together with the U1-U5 particles, the U6 snRNP is translocated to Cajal bodies for formation of the U4/U6*U5 tri-snRNP (Patel & Bellini, 2008). The mature snRNPs eventually assemble on pre-mRNA for intron removal steps (Will & Luhrmann, 2001; Patel & Bellini 2008).

Figure 1.

Lifecycle of mRNA from transcription to decay.

Following excision of introns, mRNA enters the cytoplasm via the nuclear pore complex to be either translated or degraded. In eukaryotes, two pathways are utilized for mRNA decay: i) 3’-to-5’ degradation by the exosome or ii) 5’-decapping, followed by 5’-to-3’ exonuclease degradation (Garneau et al., 2007). In either event, decay is initiated by shortening of the poly(A)-tail by deadenylases (Tucker et al., 2001; Garneau et al., 2007; Nissan et al., 2010). Protein machinery required for the 5’-decapping pathway is found enriched in cytoplasmic foci known as processing or P-bodies (Sheth & Parker 2003), which appear to control the sorting and storage of mRNA. Within P-bodies, a specific assembly of Lsm proteins (Lsm1-Lsm7) and ancilliary protein factors expedites mRNA decapping and subsequent breakdown by ribonuclease (Nissan et al., 2010). While the extent to which mRNA decay is restricted to P-bodies is unclear, sequestered mRNA species are observed to leave P-bodies and may re-enter translation (Brengues et al., 2005).

2. Phylogeny of Lsm protein sequences

The Lsm proteins recur as molecular chaperones for RNA during the many steps of its processing, sorting and regulation (Beggs, 2005). While Sm proteins were first found enriched in a patient with systemic lupus erythematosis (Lerner & Steitz 1979), the wider protein family has since been described across all domains of life (Beggs, 2005; Ma et al., 2005). Members include eukaryotic Lsm (Salgado-Garrido et al., 1999), Sm (Kambach et al., 1999) and SMN/Gemin proteins (Selenko et al., 2001; Ma et al., 2005), archaeal Lsm proteins (Collins et al., 2001), the bacterial protein Hfq (Schumacher et al., 2002) and a recently identified Lsm homolog of cyanophage origin (Das et al., 2009). Eukaryotic genomes can contain up to 16 Lsm and 7 Sm proteins (Albrecht & Lengauer 2004), yet 2-3 Lsm proteins are generally encoded in archaea (Collins et al., 2001; Toro et al., 2002; Mura et al., 2003) and only a single form is evident in bacteria and cyanophage (Schumacher et al., 2002; Das et al., 2009).

A characteristic feature of the Lsm proteins is their natural tendency to form ring-shaped quaternary complexes, each of a precise composition related to cellular location and RNA target (Beggs, 2005; Spiller et al., 2007). In prokaryotes and archaea, homomeric complexes of six or seven Lsm protomers appear to be functional, whilst discrete heteromeric assemblies of seven distinct Lsm proteins are found in eukaryotes. The individual Lsm proteins vary in size from 8-25 kDa (78-240 amino acids); representative sequences are depicted in Figure 2. Within each, a bipartite consensus sequence (designated Sm1 and Sm2 motifs) can be identified. These motifs arise from strands 1-3 and 4-5 of the core -sheet structure, respectively. A variable stretch of residues between these conserved segments is created by a surface-exposed interconnecting loop (Kambach et al., 1999; Collins et al., 2001).

The N- and C-terminal tail regions of each Lsm sequence are often highly charged and differ markedly between members; these are considered to provide contact points for additional protein or RNA interactions (Reijns et al., 2008; Reijns et al., 2009; Weber et al., 2010). In the case of the eukaryotic Lsm1 and Lsm4 proteins, these tail segments are notably elongated.

The most highly conserved sequence segments across the Lsm family include specific amino acid sidechains implicated in RNA-binding. These are localized to two specific loop features, as outlined in Figure 2. For archaeal and eukaryotic Lsm proteins, sequence motifs Asp-x---Asn ( = hydrophobic) and Arg-Gly-(Asp) (Kambach et al., 1999; Collins et al., 2001; Toro et al., 2001) are characteristic of loops L3 and L5, respectively. In bacterial Hfq, these RNA-binding segments occur as Asp-x--- (L3) and Tyr-Lys-His (L5) (Schumacher et al., 2002).

Figure 2.

Structure-based Lsm protein sequence alignment. Sequences displayed are for S. cerevisiae Lsm3 (yLsm3), H. sapiens Lsm3 (hLsm3), S. cerevisiae SmD2 (ySmD2), human SmD2 (hSmD2), M. thermoautotrophicum Lsm (MtLsm) and E. coli Hfq (EcHfq). Shaded residues represent areas with 80% sequence homology. Secondary structure assignment is based on the crystal structure of yLsm3 (Naidoo et al., 2008). Red bars indicate conserved residues implicated in RNA binding. # indicates additional truncated residues not displayed. Boxed insert shows organization of other Lsm multidomain proteins: AD, anticodon binding domain; MTD, methyl transferase domain; DFDF, DFDF‐x(7)‐F containing domain; FFD, Y‐x‐K‐x(3)‐FFD‐x‐(IL)‐S containing motif; TFG: [RKH]‐x(2‐5)‐E‐x(0‐2)‐[RK]‐x(3‐4)‐[DE]‐TFG containing domain. CTD, C-terminal domain2.

For this bacterial ortholog, a highly conserved Gln residue on the N-terminal -helix is also implicated in RNA-binding (Schumacher et al., 2002).

Overall, the bacterial protein Hfq shows little sequence conservation with its archaeal and eukaryotic orthologs, yet the archaeal and eukaryotic Lsm proteins share some limited sequence similarity (20 %). The following Lsm-Sm protein paralogs are identifiable: Lsm1-SmB, Lsm2-SmD1, Lsm3-SmD2, Lsm4-SmD3, Lsm5-SmE, Lsm6-SmF, Lsm7-SmG, Lsm8-SmB (Fromont-Racine et al., 2000). These specific sequence relationships suggest the eukaryotic Lsm proteins to have evolved from a common archaeal ancestor in two waves (Khusial et al., 2005; Veretnik et al., 2009). A first gene duplication event likely created eight distinct Lsm proteins, from which later evolved the Sm protein group. The diversity of biological activities of Lsm proteins compared to their more specialized Sm counterparts supports this two-step evolution model (Beggs, 2005; Khusial et al., 2005). The presence of up to three Lsm proteins in archaea, as well as an Hfq-like protein in archaeal M. jannaschii, further supports a common ancestor of eukaryotic and archaeal Lsm proteins (Fischer et al., 2011).

A few multidomain proteins incorporating Lsm components have been observed (summarized, Figure 2). Lsm12 includes t-RNA and methyltransferase domains (Albrecht & Lengauer, 2004), and Lsm13, Lsm14 and Lsm15 all contain a central DFDF-x(7)-F domain (Albrecht & Lengauer, 2004; Anantharaman & Aravind, 2004). Lsm16 features a remarkably disrupted Lsm variant (lacking both the N-terminal -helix and a complete 4 strand) in addition to FDF and YjeF-N domains (Albrecht & Lengauer, 2004; Tritschler et al., 2007). This protein is suggested to be dimeric in solution (Ling et al., 2008). The archaeal protein Pa-Sm3 contains an Lsm-like domain in addition to a C-terminal domain of unknown function adopting an /-fold (Mura et al., 2003).

3. Structures of Lsm protein ring complexes

Crystal structures of Lsm and Sm proteins from diverse sources today provide many high-resolution views of the ring morphology of their assemblies. As shown in Figure 3, Lsm rings have been observed to range 58-75 Å in diameter and to contain a central pore of 6-15 Å. Some crystal structures solved to date (Table 1) have been obtained in the presence of specific RNA partners. The recent solving of the human U1-snRNP structures containing the Sm assembly bound together with U1 snRNA and proteins U1-70K and U1-A have been significant and exciting advances (Pomeranz Krummel et al., 2009; Weber et al., 2010). These provide the first molecular detail of L/Sm rings bound to the highly intertwined protein-RNA network within RNP complexes.

Within the various Lsm ring assemblies, each protomer occurs as a highly bent five-stranded antiparallel -sheet overlaid in most cases by an N-terminal -helix (Figure 4A). The pronounced twist of the -sheet aligns strand 5 against 1, so forming an SH3-type barrel loosely related to the OB-fold (Kambach et al., 1999; Collins et al. 2001). Strands 4 and 5 each present on opposite ends of the module, so providing interaction sites for adjacent Lsm subunits via 4-5’ pairing (Figure 4). Stacking of five to eight protomers in such a manner ultimately results in the formation of the toroid assembly characteristic of all Lsm assemblies (Figure 4).

Within this ring organisation, the N-terminal amphipathic -helices of each Lsm component are gathered across one face of the toroid, from which also project the unstructured N- and

Figure 3.

Selected crystal structures solved for Lsm assemblies. A) Pentamer, cyanophage ECX21941 (PDB 3BY7) 60 Å ring, 9 Å pore. B) Hexamer, of C. parvum Lsm5 (PDB 3PGG) 60 Å ring, 10 Å pore. C) Hexamer, S. aureus Hfq (PDB 1KQ1) 65 Å ring, 11 Å pore. D) Hexamer, A. fulgidus Sm2 (PDB 1LJO) 58 Å ring, 6 Å pore. E) Hexamer, A. fulgidus Sm2 (PDB 1LJO) 58 Å ring, 6 Å pore. F) Heptamer, A. fulgidus Sm1 (PDB 1I4K) 65 Å ring, 13 Å pore. G) Heptamer, M. thermoautotrophicum Lsm (PDB 1I81) 65 Å ring, 10-15 Å pore. F) Heptamer, S. cerevisiae Sm-F (PDB 1N9R) 65 Å ring, 10-15 Å pore. H) Octamer, S. cerevisiae Lsm3 (PDB 3BW1) 75 Å ring, 15 Å pore.

	Protein^1	PDB ID	Resolution (Å)	Organism	Reference
Hexameric	HsSmD3B	1D3B	2.00	H. sapiens	Kambach et al., 1999
SaHfq	1KQ1	1.55	S. aureus	Schumacher et al., 2002
SaHfq*	1KQ2	2.71	S. aureus	Schumacher et al., 2002
AfSm2	1LJO	1.95	A. fulgidus	Toro et al., 2002
EcHfq	1HK9	2.15	E. coli	Sauter et al., 2003
PaHfq	1U1S	1.60	P. aeruginosa	Nikulin et al., 2003
PaHfq	1U1T	1.90	P. aeruginosa	Nikulin et al., 2003
MjSm	2QTX	2.50	M. jannaschii	Nielsen et al., 2007
CpLsm5	3PGG	2.14	C. parvum	Vedadi et al., 2007
AHfq	3HFN	2.31	Anabena sp.	Boggild et al., 2009
EcHfq*	3GIB	2.40	E. coli	Link et al., 2009
SHfq	3HFO	1.30	Synchocystis sp.	Boggild et al., 2009
PaH57THfq	3INZ	1.70	P. aeruginosa	Moskaleva et al., 2010
PaH57AHfq	3M4G	2.05	P. aeruginosa	Moskaleva et al., 2010
BsHfq	3HSB	2.20	B. subtilis	Someya et al., 2010^3
Heptameric	MtLsm	1I81, 1MGQ	2.00, 1.70	M. thermoautotrophicum	Collins et al., 2001
PaeSm1	1I8F	1.75	P. aerophilum	Mura et al., 2001
AfSm1	1I4K	2.50	A. fulgidus	Toro et al., 2001
AfSm1*	1I5L	2.75	A. fulgidus	Toro et al., 2001
MtLsm	1JBM	1.85	M. thermoautotrophicum	Mura et al., 2003b
PaeSm1	1JRI	1.75	P. aerophilum	Mura et al., 2003b
PaeSm1	1LNX	2.05	P. aerophilum	Mura et al., 2003b
PabSm1	1H64	1.90	P. abysii	Thore et al., 2003
PabSm1*	1M8V	2.60	P. abysii	Thore et al., 2003
PaeSm3	1M5Q	2.00	P. aerophilum	Mura et al., 2003a
PaeSm1	1LOJ	1.90	M. thermoautotrophicum	Mura et al., 2003b
ScSmF	1N9R	2.80	S. cerevisiae	Collins et al., 2003
ScSmF	1N9S	3.50	S. cerevisiae	Collins et al., 2003
SsSm1	1TH7	1.68	S. solfataricus	Kilic et al., 2005
U1-snRNP*	3CW1	5.49	H. sapiens	Pomeranz Krummel; et al., 2009
U1-snRNP*	3PGW	4.40	H. sapiens	Weber et al., 2010
Other	CphLsm	3BY7	2.60	Cyanophage	Das et al., 2009
ScLsm3	3BW1	2.50	S. cerevisiae	Naidoo et al., 2008
PfuQ8TZN2^2	1YCY	2.80	P. furiosus	Huang et al., 2004^3

Table 1.

Table 1. Crystal structures solved for Lsm assemblies (to 2010)¹Proteins are named by the first letters of the species, followed by the type of protein. Asterisked entries indicate structures solved in the presence of RNA. ²Hypothetical protein adopting an Lsm fold.³Structure deposited without supporting publication.

C-terminal extensions. The opposite face of the ring, named the distal face, is predominantly composed of residues of the variable loop L4 segments. All the Lsm ring structures (across eukarya, archaea and bacteria) reveal clusters of positive residues lining the internal pore, as well as pronounced positive elements on the distal face (Toro et al., 2001; Brennan & Link, 2007; Naidoo et al., 2008).

The body of structural data adds to biochemical understanding concerning L/Sm-RNA interactions, and distinct RNA sites within the protein oligomer. These include i) a binding site within the lumen of the ring, ii) an external contact site on the helix face and iii) residues located on the distal face of the complex (Figure 4). The first of these sites engages residues from loops L3 and L5, contributed from all Lsm components to create a nucleotide-binding pocket running around the inner rim (Weber et al., 2010). The specific architecture and repeated circular location of these specific, highly conserved, sidechains enables one nucleotide base to be bound per L/Sm protomer. Crystal structures of archaeal and bacterial Lsm complexed with RNA clearly show the oligonucleotides to be threaded around this rim of the toroid (Toro et al., 2001; Schumacher et al., 2002). Each binding “slot” allows specific base stacking to a hydrophobic sidechain of loop L3, as well as contact with the signature Arg residues of loop L5 and H-bonding with Asn residues (strand 4). Further electrostatic contacts (involving conserved Asp (strand 2), Arg (loop L5) and Gly (loop L5) residues) enhance the stability of the Lsm-RNA complex (Toro et al., 2001). Figure 5 displays these relevant binding interactions for U₅ within the lumen site of archaeal AfSm1.

An external contact site for RNA at the helix face of the Lsm toroid (site ii) is suggested by the crystal structure of PaSm1 bound with U₇ oligonucleotide (Thore et al. 2003). In this case, each of two sandwiched Lsm rings engage two nucleotides at the N-terminal -helix (Arg, His) and strands 2 (Tyr) via base stacking and H-bonding.

A third distinct RNA-binding site (iii) is likely to be unique to the bacterial Hfq assembly, and its tripartite form has been detailed in the crystal structure of Hfq bound to poly(A) RNA (Link et al. 2009). The protein Hfq engages poly(A) sequences on its distal face via specific residues exposed from strands 2 and 4. There is, however, no evidence for poly(A) binding by eukaryotic Lsm proteins. In the structure of the Hfq/RNA complex, RNA contacts include electrostatic interactions from Lys (strand 2) and Gln (strand 4) sidechains, as well as stacking of bases between Tyr, Leu (strand 2) and Leu and Ile (strand 2’) of adjacent subunits. It is in this region of the toroid that sequence variability of the loop L4 across the Lsm family results in non-conservation of distal face chemistry, so explaining the unique binding properties of Hfq.

Within the crystal structures of the human U1-snRNP complex, multiple RNA interactions made by the ring of Sm proteins include binding sites i) and ii) outlined above (Weber et al., 2010). However, the U1-snRNP structure also clearly demonstrates the role of the Sm sequence extensions and loop regions as additional interaction sites, particularly the C-terminal extensions of SmD3 and SmB. In the lumen of the toroid (i.e. site i), snRNA threads

to stack single nucleotides of the Sm site against the key loop L3 and L5 residues, noteably the aromatic sidechains. From the helix face of the ring are projected residues of the N-terminal -helix and loop L3 of SmD2, forming an external contact site (reminiscent of site ii) that guides the snRNA into the ring pore. Residues from the loop L2 regions of SmD1 and SmD2 appear to guide RNA out from the Sm ring. Protruding beyond the distal face, residues of the elongated L4 loops of SmD2 and SmB provide another important interaction point to clamp and secure a stem-loop of the snRNA.

Figure 4.

Lsm fold and quaternary structure. Ribbon diagrams of MtLsm (A, B; PDB 1I81) are displayed. A) Dimer interface of MtLsm. Chain A is represented in green, chain B in blue. Residues involved in hydrophobic packing at the dimer interface (Chain A: Ile27, Val77, Tyr78 of chain A; Chain B: Leu 30, Phe36, Leu66, Val69, Ile71) are shown in stick representation. B-C) Top and side view of heptameric MtLsm. D) Homo-heptameric MtLsm (PDB 1I81). E) Homo-octameric yeast Lsm3 (PDB 3BW1). Space-filled models highlight in red conserved residues implicated in RNA binding: Asp in 2, Asn in L3, Arg and Gly in L5.

Figure 5.

Three general sites for RNA binding within specific examples of Lsm complexes. Site i) AfSm1 (PDB 1I5L) bound to U₅ RNA viewed from helix face. Site ii) Two PaSm1 (PDB 1M8V) heptamers are bridged by uridine heptamer. Site iii) EcHfq (PDB 3GIB) bound to poly(A) viewed from distal face. U1-snRNP) Figure includes side view of the Sm-core of the human U1 snRNP structure (PDB 3PGW).

The majority of crystal structures of Lsm obtained to date portray the hexa- and heptameric protein assemblies that correspond to fully functional homomeric or heteromeric protein groupings. It is, for instance, assumed that complexes of SmD1-SmD2, SmD3-SmB and SmE-SmF-SmG can exist independently in the cytoplasm, yet rearrange into mixed heptamers in the presence of RNA during snRNP formation (Peng & Gallwitz, 2004). However, a few crystal structures suggest that other compositions, e.g. pentamers and octamers, may be stable for eukaryotic Lsm (Naidoo et al., 2008; Das et al., 2009). While it is currently not clear if these organizations are peculiar to recombinant preparations of the Lsm family, they suggest possibilities for a variety of multimeric assemblies in vivo. Our own interaction studies indicate that Lsm assemblies may be relatively dynamic in solution, providing capacity to engage in alternative protein partnerships and stable groupings (Sobti et al., 2010).

4. Functional roles for Lsm proteins

Sm and Lsm proteins are known to interact with a diversity of RNA partner species. Specific RNA sequences recognized by various Lsm complexes include the Sm-site (A₂U₅GA) (Raker et al., 1999) and U-rich stretches at the 3’ end of oligoadenylated mRNA (Chowdhury et al., 2007) and RNA polymerase III transcripts, including snRNA (Achsel et al., 1999). Other binding partners include snoRNA (Kufel et al., 2003a), P RNA (Kufel et al., 2002), tRNA (Kufel et al., 2002) and rRNA (Kufel et al., 2003b). Depletion of Lsm proteins 2-5 and 8 in yeast results in defects in post-transcriptional processing of tRNA, P RNA, rRNA, snoRNA and snRNA precursors (Kufel et al., 2002; Kufel et al., 2003b; Kufel et al., 2003a). Yet only minor (or no) effects are observed on depletion of Lsm6 and Lsm7. A summary of some specific Lsm-ncRNA interactions is presented in Table 2.

The Lsm2-Lsm8 complex plays a key role in U6 snRNA maturation, so impacting on the formation of spliceosomal snRNPs (Karaduman et al., 2006). U6 snRNA is the most conserved of all snRNA species and key to the catalytic activity of the spliceosome (Brow, 2002). Newly transcribed U6 pre-snRNA is targeted to the nucleoli following binding of the La protein (Lhp1 in yeast) at its U-rich 3’ region (Wolin & Cedervall, 2002). Following cyclic phosphorylation, La (or Lhp1) is displaced from the U6 snRNA by the Lsm2-Lsm8 assembly (Achsel et al., 1999; Licht et al., 2008), which induces conformational changes that stimulate binding of a recycling factor (p110 or Prp24) (Rader & Guthrie, 2002; Ryan et al., 2002; Karaduman et al., 2006). These conformational changes have been suggested to assist in the formation and recycling of the U4/U6 di-snRNP by exposing single stranded nucleotides for base pairing (Beggs, 2005; Karaduman et al., 2006; Karaduman et al., 2008). The Lsm2-Lsm8 complex is also implicated in decapping steps of mRNA in the nucleus. This was suggested by the finding that Lsm6 and Lsm8 were required for nuclear mRNA decay (Kufel et al., 2004).

A specific role for Lsm1-Lsm7 concerns activation of mRNA decay in P-bodies; depletion of individual yeast Lsm proteins results in the accumulation of capped, oligoadenylated mRNA transcripts (Boeck et al., 1998; Bonnerot et al., 2000; Bouveret et al., 2000; Tharun et al., 2000). This specific Lsm complex is recruited alongside other decay factors to U-rich tracts by the protein Pat1, after its displacement of cap-binding translation factors (Parker & Sheth, 2007). It is likely that Pat1 and Lsm1-Lsm7 are then involved in subsequent activation of the Dcp1-Dcp2 enzyme (Nissan et al., 2010). A variety of studies have demonstrated the interaction of Lsm1-Lsm7 with decapping factors and exoribonuclease Xrn1 (Bonnerot et al., 2000; Bouveret et al., 2000; Tharun et al., 2000; Coller et al., 2001).

RNAspecies	Lsm function	Selected experimental evidence	References
snRNA	assembly,; processing; and nuclear; localization	Lsm2-8 binds 3’ end of U6 snRNA	Achsel et al., 1999
Lsm2-8 initiates structural; rearrangements of U6 snRNA	Karaduman et al., 2006; 2008
Depletion of Lsm2-Lsm8 results in splicing defects	Mayes et al., 1999
Splicing activity recovered through recombinant Lsm proteins	Verdone et al., 2004
Lsm2-8 localizes U6 snRNA to the nucleus	Spiller et al., 2007
tRNA	splicing,; 3’ and 5’ end-processing	Accumulation of unprocessed pre-tRNA and reduced La/Lhp1 binding upon Lsm2-Lsm5 and Lsm8 depletion	Kufel et al., 2002
Direct interaction of Lsm3 with tRNA and its splicing factors	Fromont-Racine et al., 1997
P RNA	chaperone	Depletion of Lsm2-Lsm5 and Lsm8 reduces pre-PRNA levels	Mayes et al., 1999
Reduced La/Lhp1 binding upon Lsm2-Lsm5 and Lsm8 depletion	Kufel et al., 2002
Lsm2-Lsm7 proteins coprecipitate with pre-PRNA	Salgado-Garrido et al., 1999
rRNA	3’ and 5’ end-processing	Depletion of Lsm2-Lsm5 and Lsm8 delays pre-rRNA processing and increases rRNA decay rate	Kufel et al., 2003b
Pre-rRNA coprecipitates with Lsm3 but not Lsm1	Kufel et al., 2003b
Deletion of Lsm6 and Lsm7 genes impairs 20S pre-rRNA processing	Li et al., 2009
snoRNA	3’ end-processing	Lsm2-Lsm5 and Lsm8 depletion results in U3-snoRNA degradation and loss of its 3’ extended precursor	Kufel et al., 2003a
Reduced La/Lhp1 binding upon Lsm3 or Lsm5 depletion	Kufel et al., 2003a
Lsm2-Lsm7 but not Lsm1 or Lsm8 coprecipitate with snR5 snoRNA	Fernandez et al., 2004
Lsm2-4 and 6-8 but not Lsm5 coprecipitate with U8 snoRNA	Tomasevic & Peculis, 2002

Table 2.

Table 2. Lsm binding interactions with ncRNA

In contrast to its enhancement of mRNA decay, however, the Lsm1-Lsm7 complex can also protect mRNA against 3’ end trimming (He & Parker, 2001). This may involve steric hindrance of nuclease attack at mRNA locations on which Lsm1-Lsm7 and Pat1 proteins are bound.

5. Specific functions of bacterial Hfq

Bacterial Hfq is observed to interact with bacterial sRNA and so promote the formation of sRNA-mRNA complexes (Wassarman et al., 2001; Gottesman & Storz, 2010). Bacterial sRNAs are small non-coding RNA species (50-500 nucleotides), which regulate gene expression via base pairing with mRNA transcripts in a similar mechanism to eukaryotic siRNA or miRNA (Storz et al., 2004; Majdalani et al., 2005; Livny & Waldor, 2007; Gottesman & Storz, 2010). Hfq controls gene expression either by rearranging the RNA secondary structure, or by increasing the concentration of RNA locally to promote RNA-RNA interactions (Moll et al., 2003; Lease & Woodson, 2004; Afonyushkin et al., 2005). A similar mode of binding to sRNA was recently observed for the archaeal Lsm from Haloferax volcanii (Fischer et al., 2011).

As for the eukaryotic Lsm proteins, Hfq is required for deadenylation-dependent mRNA decay. An RNase E-Hfq-sRNA complex is thought to function in translational repression and subsequent mRNA destabilization and degradation (Morita et al., 2005; Morita et al., 2006). Additional functions of Hfq include ATPase activity (Sukhodolets & Garges, 2003), cellular stress response and modulation of virulence in some bacterial strains (Tsui et al., 1994; Fantappie et al., 2009; Liu et al., 2010). Interestingly, the virulence of the multi-drug resistant human pathogen S. aureus was decreased in Hfq-deletion strains. (Liu et al., 2010).

6. Lsm proteins in human disease and viral replication

Aberrations in functions of Lsm proteins have been associated with a number of human diseases. Sm proteins are known to be targeted by auto-antibodies in systemic lupus erythematosis (Lerner & Steitz, 1979). In fact, the proteins were first identified in nuclear extracts of a patient suffering from this disease. A mutation of the SMN gene resulting in diminished assembly of snRNPs is the cause of spinal muscular atrophy (Lefebvre et al., 1995; Wan et al., 2005). Three Lsm proteins (Lsm1, Lsm3 and Lsm7) have now been directly connected to different cancer types. Lsm1 (also named cancer associated Sm-like protein, CaSm) was upregulated in pancreatic, prostate and breast cancer, as well as in several cancer-derived cell lines (Schweinfest et al., 1997; Fraser et al., 2005; Streicher et al., 2007). Remarkably, overexpression of antisense Lsm1 has been demonstrated to promote tumor reduction (Kelley et al., 2000; Kelley et al., 2001; Yan et al., 2006). Elevated levels of Lsm7 have been identified in malignant thyroid tumors, and a reduction in Lsm7 expression was observed in breast cancers (Conte et al., 2002; Rosen et al., 2005). The copy number and expression for the Lsm3 gene was found to be elevated in cervical cancer (Lyng et al., 2006).

Observations concerning Lsm proteins in viral replication underlines some interesting functional diversity. Bacterial Hfq was initially described as a host factor required for phage Qß replication (Franze de Fernandez et al., 1968). A role for Lsm1 as an effector of HIV replication has been reported (Chable-Bessia et al., 2009). It has also been suggested more recently that positive-strand RNA viruses may directly bind to the host Lsm1-7 protein complex via tRNA-like structures and A-rich stretches, so diverting normal mRNA regulation (Galao et al., 2010). The requirement of host Lsm proteins for the replication of this class of virus has additionally been demonstrated in plant brome mosaic virus (Diez et al., 2000; Noueiry et al., 2003; Mas et al., 2006) and human hepatitis C virus (Scheller et al., 2009).