La proteins couple use of sequence-specific and non-specific binding modes to engage RNA substrates

ABSTRACT

La shuttles between the nucleus and cytoplasm where it binds nascent RNA polymerase III (pol III) transcripts and mRNAs, respectively. La protects the 3ʹ end of pol III transcribed RNA precursors, such as pre-tRNAs, through the use of a well-characterized UUU-3ʹOH binding mode. La proteins are also RNA chaperones, and La-dependent RNA chaperone activity is hypothesized to promote pre-tRNA maturation and translation at cellular and viral internal ribosome entry sites via binding sites distinct from those used for UUU-3ʹOH recognition. Since the publication of La-UUU-3ʹOH co-crystal structures, biochemical and genetic experiments have expanded our understanding of how La proteins use UUU-3ʹOH-independent binding modes to make sequence-independent contacts that can increase affinity for ligands and promote RNA remodeling. Other recent work has also expanded our understanding of how La binds mRNAs through contacts to the poly(A) tail. In this review, we focus on advances in the study of La protein-RNA complex surfaces beyond the description of the La-UUU-3ʹOH binding mode. We highlight recent advances in the functions of expected canonical nucleic acid interaction surfaces, a heightened appreciation of disordered C-terminal regions, and the nature of sequence-independent RNA determinants in La-RNA target binding. We further discuss how these RNA binding modes may have relevance to the function of the La-related proteins.

KEYWORDS:

• La protein
• La –related protein
• RNA binding protein
• precursor tRNA
• RNA processing
• protein domain
• RNA-protein interaction
• translational control
• RNA chaperone

Introduction

La (also referred to as Sjogren’s syndrome antigen B, or SS-B) was first discovered as an auto-antigen in patients suffering from Sjogren’s syndrome, neonatal lupus, and systemic lupus erythematosus [1,2]. Although it is still unclear why La is targeted as an autoantigen, several decades of research have shed light on the multifaceted functions of this RNA binding protein (RBP), which has also helped to further our understanding of RNA-RBP interactions more generally. La is essential in all metazoans investigated to date, including T. brucei, A. thaliana, D. melanogaster, and M. musculus [3–6]. However, La is dispensable in both budding and fission yeast [7,8], and these model organisms have made extensive contributions to our understanding of conserved aspects of La function through genetic and loss of function/rescue studies. The first insights into La function came from immunoprecipitation of La using antibodies from human patients, where it was discovered that La is an RNA binding protein [9,10]. Follow up studies indicated that the most abundant RNAs associated with human La were nascent RNA polymerase III (pol III) transcripts, and more specifically, these RNAs still include 3ʹ uridylates that are later processed (removed) during maturation [11,12]. Further work identified the UUU-3ʹOH motif common to all nascent Pol III transcripts as the first sequence-specific La-RNA binding motif, with a minimum uridylate tail length of three U’s for high affinity binding in human La (hLa) [13,14], and four or five U’s forming the minimum preferred binding length in Schizosaccharomyces pombe (Sla1p) [15].

These and future experiments served to identify the simplest of La functions: by binding to the UUU-3ʹOH motif, La protects the 3ʹ end of nascent RNA polymerase III transcripts from 3ʹ to 5ʹ decay by exonucleases such as Rex1p and the nuclear exosome [16,17]. Since that time, however, a number of other activities have been assigned through which La uses RNA binding modes that are at least partially distinct from UUU-3ʹOH binding. Foremost among these UUU-3ʹOH independent functions is the recurring theme that La uses alternative RNA binding surfaces to both increase affinity for targets [18–21], as well as serve as an RNA chaperone, with La RNA chaperone activity being best characterized in the promotion of folding and the efficient maturation of pre-tRNAs [17,18,22–30]. Among the UUU-3ʹOH lacking cohort of La-RNA targets, La has been shown to bind a number of cellular coding substrates and viral coding and non-coding RNAs [31–40]. La function as an RNA chaperone has been hypothesized to promote translation of mRNAs, especially in the context of higher-order RNA structures in 5ʹ UTRs whose precise fold is required for function, such as internal ribosome entry sites (IRESs) [41,42]. Other functions associated with La include the binding and protection of virally encoded RNAs to promote viral infection [38,40,43], promotion of multiple turnover by RISC [44], as well as prevention of aberrant loading of tRNA fragments into Argonaute, among others [45], but the RNA binding modes used by La in such functions are not well understood. Thus, La proteins have been associated with a number of functions in RNA processing, RNA stabilization and mRNA translation, but the RNA binding modes associated with La function beyond UUU-3ʹOH recognition are still a matter of investigation.

General architecture and structural studies of La associated domains

The genuine La proteins were among the founding cohort of RNA binding proteins from which the most common RNA-binding domain, the RNA-recognition motif (RRM), was identified [46]. All genuine La proteins share a conserved N-terminal domain (NTD) in which a winged-helix fold containing domain known as the La motif (LAM) is juxtaposed with an RRM separated by a short linker in a tandem arrangement known as the La module (Figures 1 and 2 [47,48];). These two structural domains are the only ones found in La proteins from lower eukaryotes such as yeast; in S. cerevisiae a nuclear localization signal (NLS) is embedded in the RRM [49] while in S. pombe and other examined organisms the NLS is closer to the C-terminus [50]. In higher eukaryotes, a second RRM (RRM2 or RRMc) is found following RRM1 and a linker. Finally, La proteins from both yeast and higher systems are noted to have a disordered region C-terminal to the last RRM (RRM1 in yeast or RRM2 in higher systems), which also contains stretches of charged amino acids such as the short basic motif (SBM) previously characterized for human La (reviewed in [48]).

Figure 1. Regions of the La module associated with RNA interaction. Top: schematic comparing domain organization of human La and S. pombe La. NRE: nuclear retention element. SBM: short basic motif. S366: site of phosphorylation. NLS: nuclear localization signal. Bottom: High-resolution structure of the human La motif (light grey) and RNA recognition motif 1 (RRM1; dark grey) with highlighted amino acids, including alignments showing conservation of residues linked to RNA function. Pink: amino acids previously shown to make direct contacts to UUU-3ʹOH in La module-UUU-3ʹOH co-crystal structures [55,56]. Note I140 on RRM1 colored in pink as this contacts UUU-3ʹOH, but side chain and conservation of I140 not included as this contact is to peptide backbone. Tan: amino acids in loop 3 of RRM1 whose mutation results in reduced tRNA binding, defective tRNA mediated suppression and impaired RNA chaperone activity [19]. Blue: β-sheet and canonical RNA binding surface of RRM1; mutation of aromatic amino acids conserved in RNP1 and RNP2 motifs (highlighted) causes defective tRNA mediated suppression and decreased RNA chaperone activity [17,24]. Green: alpha helical extension C-terminal to canonical RRM fold; mutation of highlighted amino acids results in defective tRNA mediated suppression and decreased RNA chaperone activity [24]. Orange: amino acids in first wing of winged-helix fold of La motif; mutation results in defects in binding to A20 and binding to poly(A) in human cells. PDB for figure: 2VON [56]. Schematic generated using [51]

Figure 2. Hypothesized arrangement of structural domains during La binding to pre-tRNAs and poly(A). Top: schematic of domains in human La. Bottom left: arrangement of domains during binding to poly(A). Bottom right: arrangement of domains during binding to pre-tRNA. Protein-RNA contacts validated by co-crystal structures (UUU-3ʹOH) shown in black dashed lines [55,56]. Protein-protein or protein-RNA contacts hypothesized from mutagenesis experiments shown in green and red dashed lines, respectively. During binding to UUU-3ʹOH or pre-tRNA, interdomain contact between RRM1-α1 (green) and RRM2-α3/NRE (blue) that is present during poly(A) binding is disrupted [75]. Binding to pre-tRNA also involves contacts to α3 helix of RRM2 (possibly through the 5ʹ leader [75],) and loop 3 of RRM1 [19]. Binding to poly(A) hypothesized to involve winged-helix face of La motif and relies also on contact to RRM1, although specific amino acids of RRM1 important for this are not yet known (see ‘?’ [42];). Similarly, the region of the tRNA contacted by the β2-β3 loop of RRM1 is not known (see ‘?’ [18];) Poly(A) not drawn to scale

Thanks to several elegant experiments using high-resolution NMR and X-ray crystallography, the structures of the La motif, RRM1 and RRM2 in the absence of RNA ligand are well characterized [52–54]. In addition to these, co-crystals of the human La module complexed with an RNA ligand ending in UUU-3ʹOH have led to a detailed understanding of the UUU-3ʹOH dependent RNA binding mode [55,56]. These advances in our understanding of La structure have been reviewed previously [57–61], thus the primary purpose of this review is to discuss biochemical and genetic experiments published after these structural studies that have helped better elucidate the function and mechanisms of UUU-3ʹOH independent binding modes. Still, in order to best contextualize this more recent work, it is useful to review some key insights that were noted from the structures but whose functional significance were less clear at the time they were published. The first has to do with the binding surfaces used during UUU-3ʹOH recognition: one of the most intriguing insights gleaned from the La module/UUU-3ʹOH co-crystals was the absence of direct contacts between the RNA ligand and the expected nucleic acid interaction surfaces of the La module, namely, the winged-helix face of the La motif and the β-sheet surface of RRM1 (Figure 1; 53, 54, 57). By the time these structures were published it had already been hypothesized that La had functions beyond UUU-3ʹOH binding and associated protection from exonucleolytic degradation, and work since then has validated that some of these surfaces indeed function in such roles. A second notable observation with relevance to the more recent work was the discovery of α-helices C-terminal to the canonical RRM fold (i.e. β1α1β2β3α2β4α3) in both RRM1 and RRM2 of hLa ([52,54]; Figure 2); in the case of RRM2, this additional α-helix is sufficiently large so as to completely obscure the canonical RRM RNA binding surface, in an arrangement that is shared with the La-related proteins p65 and hLARP7 [62,63]. As will be detailed in the following sections, the unexpected arrangements of canonical RNA-binding surfaces of the La module and the unique architecture of the RRMs have proven to make important contributions to La function for client RNA targets.

RNA chaperone function and UUU-3ʹOH independent binding via RRM1

It was still early in the study of La when the first evidence of binding to targets lacking UUU-3ʹOH was described [31]. Increasingly, the use of binding modes independent of UUU-3ʹOH have become associated with La function as an RNA chaperone, with evidence for such activity coming from several directions. La has been shown to promote the translation of UUU-3ʹOH lacking but IRES containing viral and cellular mRNAs (see above). Next, deletion of La in S. cerevisiae had been shown to be synthetically-lethal with a) deletion of some tRNA modification enzymes [64–66] and b) mutations to tRNAs hypothesized to result in their misfolding [22,28]. S. pombe La (Sla1p) had also been demonstrated to rescue defective suppressor tRNAs carrying mutations hypothesized to cause misfolding [67]. Along these same lines, human La has been demonstrated to harbor strand annealing and strand displacement activities in in vitro assays for RNA chaperone function [17,23,24]. These results suggested roles for La proteins beyond simple 3ʹend protection of RNA targets, leading to the hypothesis that these other RNA binding surfaces could play important roles in these UUU-3ʹOH independent functions [55–57].

The first experiments that attempted to map RNA chaperone related functions independent of UUU-3ʹOH binding assessed the capacity of various La mutants to promote the maturation of sup3-e (tRNA-Ser^UGA) based suppressor tRNA variants in S. pombe [17]. Successful La-dependent maturation of these suppressor tRNAs in the context of the nonsense codon containing ade6-704 allele leads to expression of functional Ade6p. Synthesis of Ade6p in turn relieves the accumulation of a red coloured intermediate in the adenine biosynthetic pathway, yielding a red-white assay for La function in living yeast cells. These experiments revealed that some mutant suppressor tRNA alleles relied on La in a manner that superseded a necessity for simple 3ʹ end protection from the nuclear exosome. Specifically, mutations to the canonical RRM1 RNA binding surface of either S. pombe La or human La resulted in defective La rescue of mutant suppressor tRNAs predicted to form significant misfolds even in the presence of normal UUU-3ʹOH binding. Together with the crystal structure showing an absence of contacts to UUU-3ʹOH, these data indicated that the RRM1 RNA binding surface functions in pre-tRNA maturation in a manner independent from 3ʹ end protection and more related to pre-tRNA folding. Subsequent work validated that RRM1 makes contacts to the main body of the tRNA and that mutations to RRM1 result in defective RNA chaperone activity in vitro [18,24]. Contacts between La and the main body of the tRNA were then further corroborated with experiments mapping direct La contacts to the tRNA D-loop using PAR-CLIP, albeit the region of La responsible for these contacts was not identified [68]. These data confirmed a function for one of the expected nucleic acid binding surfaces of the La module: the β-sheet surface of RRM1 makes UUU-3ʹOH independent contacts to pre-tRNAs and promotes the native folding of La targets. Regions of RRM1 have also been linked to binding of other UUU-3ʹOH lacking La targets such as viral RNAs [38,69], pointing to a broader role for RRM1 in UUU-3ʹOH independent binding.

RRM2 of human La: sequence independent binding of structured RNAs

Together with the discovery of a role for RRM1 in the binding of pre-tRNAs via UUU-3ʹOH independent contacts, a role for RRM2 in the binding of structured RNAs has also been proposed in the absence of sequence specificity. La is hypothesized to be an important factor in the stabilization of the Hepatitis B (HBV) viral RNA, where RRM1 and RRM2 co-operate in binding in the absence of contribution from the La motif [40,70]. Notably, deletion of a short amino acid stretch in RRM2 (as well as in RRM1) resulted in substantial drops in affinity for an extended stem-loop derived from the HBV viral RNA [70]. Other experiments linked RRM2 mediated RNA binding to a hairpin derived from the IRES of Hepatitis C (HCV), another viral RNA target of La, where it was demonstrated that the La module (La motif + RRM1) of human La was substantially impaired in hairpin binding relative to the analogous La construct including RRM2 [21].

One consideration as to how RRM2 might engage RNA ligands concerns the apparent complete occlusion of the canonical RRM RNA binding surface by the unique α3 helix (Figure 2). Clues as to how this might occur have come from the study of the RRM2 domain from the La-related proteins p65 (from Tetrahymena) and human LARP7 (hLARP7) [62,63,71]. Similar to genuine La proteins, these factors bind pol III transcripts, namely the telomerase RNA (TER) in Tetrahymena (which is pol III transcribed in ciliates) and the 7SK RNA in humans. NMR and high-resolution crystal structures of the RRM2 of p65 revealed a very similar topology to the RRM2 of human La [62], and a co-crystal of this domain bound to a minimal TER substrate indicated that the disordered region immediately C-terminal to the RRM2 α3 helix becomes ordered into an α-helical extension upon RNA binding, giving rise to a proposed new, atypical RNA binding mode for RRMs which was coined xRRM [72]. A similar binding mode was proposed for the RRM2 of human La, and subsequent work showed that the RRM2 of hLARP7 uses a related binding mode when engaging the relevant target site on the 7SK RNA, although the degree to which α3 extends upon RNA binding was shorter [63,71]. Thus in both p65 and hLARP7, the C-terminal end of α3 (in cooperation with an adjoining surface on β2 of the RRM, termed RNP3) of the domain plays a substantial role in RNA binding.

Unlike p65 and hLARP7, the RRM2 of human La is not associated with the binding of any specific RNA substrate, even though evidence does exist that this region does contribute to RNA binding in the absence of sequence specificity and this may have some similarities to that observed for the xRRMs of p65 and hLARP7. First, NMR chemical shift perturbation analysis showed significant conformational changes in the RRM2 α3 helix upon binding to a Hepatitis C IRES derived stem loop [21]. Later, time-resolved electrospray 
ionization hydrogen−deuterium exchange (TRESI-HDX)
 based experiments confirmed that the α3 helix of RRM2 displays significant changes in solvent accessibility upon RNA binding [19]. Notably, the region of α3 whose solvent accessibility changed the most dramatically mapped to a region analogous to the C-terminal α-helical extension in the RRM2 of p65 and hLARP7, and this region alternated between increased or decreased solvent accessibility depending on whether a single-stranded or hairpin RNA ligand was used. Thus, it seems likely that the RRM2-α3 helix of La proteins plays an important role in the binding of RNA ligands, similar to p65 and hLARP7.

Recent data have expanded the role of conserved basic amino acids in RRM2-α3 previously associated to La intracellular trafficking to more direct function in RNA binding in a manner that is also linked to the α1 helix of RRM1 [73–75]. Namely, mutations to RRM2-α3 and RRM1-α1 influence La binding to a pre-tRNA (containing the 5ʹ leader, the tRNA body and the UUU-3ʹOH containing trailer) in a manner that is consistent with RRM1-α1 and RRM2-α3 making an interdomain contact that is disrupted during pre-tRNA binding (Figure 2 and [75]). Consistent with this, collision induced unfolding/ion mobility mass-spectrometry data indicate that mutations to RRM2-α3 predicted to break this bridge result in a partially unfolded state that mirrors a conformational change that occurs when La binds a UUU-3ʹOH containing target. Mutations to RRM2-α3 also resulted in impaired ability of La to discriminate adenylate substrates from UUU-3ʹOH substrates, suggesting that this intersubunit contact is intact during poly(A) binding (see future section on La binding to poly(A)). While these experiments still lack validation from structural studies, they converge on a model in which an interdomain contact between RRM1-α1 and RRM2-α3 persists during adenylate binding but breaks when La binds to a UUU-3ʹOH containing RNA, such as a pre-tRNA. During pre-tRNA binding, basic amino acids on RRM2-α3 are then made available to make UUU-3ʹOH independent contacts 5ʹ to the UUU-3ʹOH and potentially to pre-tRNA 5ʹ leader sequences which then increases affinity of La for pre-tRNA substrates [75]. Related to this, a highly basic region immediately C-terminal to RRM2-α3 of human La (the short basic motif, SBM) is hypothesized to bind leaders of pre-tRNAs, restricting their access to RNase P and inhibiting pre-tRNA maturation until this inhibitory block is removed by phosphorylation of nuclear human La at S366 by CKII [76]. Future work that includes structural characterization of human RRM2 complexed with RNA targets will be required before the complex ways in which this region participates in RNA binding will be fully understood.

Cooperation of disordered regions C-terminal to RRM1 and RRM2

Several lines of evidence indicate that disordered regions C-terminal to RRM1 in yeast and human La as well as RRM2 of human La participate in RNA binding, as well as function in RNA chaperone activity. Mutations to the unique C-terminal α-helix (α3) of RRM1 (Figure 1, in orange), or removal of the disordered region C-terminal to RRM1-α3 results in impaired RNA chaperone activity associated with the La module (La motif and RRM1) of human La in vitro as well as rescue of tRNA-mediated suppression in vivo [24]. Notably, a similar requirement for RRM1 and regions immediately C-terminal to this domain has also been demonstrated to be important for RNA chaperone activity in the La modules of the human La-related proteins hLARP4, hLARP6 and hLARP7 [77]. Similar to hLa, the La motif for these LARPs are also inactive in such assays, consistent with a potentially conserved partitioning of function (3ʹend binding associated with the La motif and UUU-3ʹOH-independent binding associated with RNA remodeling to RRM1) across the La motif containing superfamily. Other work investigating La from S. cerevisiae (Lhp1p) indicates that disordered regions C-terminal to RRM1 are necessary to rescue mutant strains that require LHP1 for growth; these strains contain mutations to single-copy tRNA genes or to the U6 snRNP protein LSM8 [25]. This same disordered region was also required in order to footprint the anticodon stem of a tRNA substrate in vitro, suggesting that this disordered region makes direct contacts to tRNA substrates. However, RNA chaperone activity is not limited to the La module of human La. The C-terminal half of human La, encompassing RRM2 through a highly charged, disordered region through to the C-terminus, is also active in in vitro assays for RNA chaperone activity in a manner that is controlled by phosphorylation by Akt and that promotes La-dependent translation from the IRES of the cyclin D1 (CCND1) gene [26]. Furthermore, mutants of human La containing progressively greater deletions in the disordered C-terminal region have increasingly impaired non-specific binding of both single-stranded and hairpin containing RNAs, confirming that the disordered CTD enhances affinity for La-RNA ligands [19].

A major recurring theme in the mechanisms by which the La-related proteins and the La-modules in particular engage diverse RNA targets is the high degree of plasticity by which these factors use variations on folds (particularly in RRM1) as well as alterations in the length and orientation of the linker bridging the La motif and RRM1 to bind RNAs in distinct ways [78,79]. Notably, this linker is hypothesized to be substantially shorter in LARP4 family members and in hLARP6 [47,79], which may consequently restrict both the degree of flexibility between the La motif and RRM1 as well as the diversity of RNA binding modes available to these factors, at least relative to human La where the linker is extended and flexible in the absence of ligand.

Recognition of poly(A) tails by the LAM and RRM1

La is predominantly nuclear where it is associated with the binding of nascent RNA pol III transcripts. La also shuttles to the cytoplasm, and can substantially relocalize to this compartment under a variety of conditions including viral infections, apoptosis and varying types of cellular stress [38,80–86]. In the cytoplasm, La is associated with binding coding mRNAs in yeast [37], Xenopus oocytes [87,88] and human cells [42]. La promotes the translation of a number of cellular mRNAs containing IRESs [89–93] and the uORF containing mRNA for MDM2 [35], while varying studies have linked La to the enhancement or inhibition of translation of 5ʹ terminal oligopyrimidine (5ʹTOP) motif containing mRNAs [36,87,88,94,95]. Despite this long-standing link between La and mRNA translation, it has previously been unclear how La might target coding RNAs in the cytoplasm.

Links between the La superfamily and the poly(A) tail emerged when it was shown that the human La-related protein hLARP4 binds poly(A) in a length dependent manner, in which hLARP4 bound A15 approximately 200X more tightly than A10 or U15 [96]. Length dependent binding of poly(A) was subsequently described for the human La-related protein hLARP1 [97], with these and other experiments linking Drosophila LARP1, hLARP1, hLARP4 or hLARP4b with cytoplasmic poly-A binding protein (PABP) [97–100]. Interestingly, these studies describing binding of hLARP4 and hLARP1 also showed that human La could bind lengthy poly(A) sequence (hLa to A20 in [96], hLa affinity purified by A60 in Supplementary Data [97],), even though previous work had rigorously demonstrated that human La has minimal affinity for adenylates in the context of shorter lengths [13,17,20]. The investigation of La-related proteins from the LARP1, LARP4 and LARP6 families in a variety of model organisms (worms, flies, plants) has lead to a broad association of these factors with translating ribosomes as well as RNA stability [100–104], with links to the poly(A) tail directly or in combination with contacts to PABP (for a recent comprehensive review of the La-related proteins, their function and RNA binding targets, see [60]).

Recently, evidence for a direct interaction between La and poly(A) tails has been described [42], and similar to hLARP1 and hLARP4, this binding is length dependent. Consistent with a conserved mode of interaction with the La-related proteins, it relies on the La motif and RRM1, even though there is evidence that the mechanism of binding is at least partially distinct from the La module dependent mode of binding described for UUU-3ʹOH. Specifically, while loss of either the La motif or RRM1 is sufficient to cause near complete loss of uridylate recognition, deletion of either of these domains causes substantial but not complete loss of binding for poly(A), consistent with a more additive mechanism of adenylate binding and with the adenylate binding length dependence. Furthermore, while hLa variants carrying mutations to amino acids known to make direct contacts to UUU-3ʹOH cause near complete loss of binding to uridylate tailed substrates, these same mutants have similar binding to A20. On the other hand, variants carrying mutations to the winged-helix face of the La motif (which was previously demonstrated to be devoid of contacts to UUU-3ʹOH) have a more substantial effect on A20 binding than to a U10 substrate. These mutants still bind A20 with higher affinity relative to deletion of the entire La motif, suggesting that these mutants still harbor other contacts on the La motif (and RRM1) used to bind A20, nevertheless these same mutations are sufficient to impair La entry into polysomes or to immunoprecipitate La directly linked to poly(A) tails as measured by UV-crosslinking immunoprecipitation coupled with limited RNase digestion. Elucidation of all contacts required for adenylate binding will likely require high-resolution structural studies. Other unanswered questions with respect to La binding to poly(A) includes the degree of conservation for this function with La proteins from other eukaryotes, as well as the extent of overlap between how La binds adenylates relative to the La-related proteins. Another important unanswered question involves the possibility that La might use other RNA-binding modes (i.e. RRM2 and the CTD) to contact other parts of an mRNA while the La module engages the poly(A) tail. Since disruption of binding to the poly(A) tail can impair La-associated promotion of translation at an IRES [42], it is tempting to speculate that La is recruited to mRNAs through contacts between the poly(A) tail and the La module so as to promote remodeling of RNA structure elsewhere on the RNA target using one or more of its other RNA binding modes.

Further considerations for mechanisms of La-RNA binding

With such a wide diversity of RNA ligands associated with La, attempts to arrive at a unified model for La-RNA target binding and La function have been challenging. However, decades of research into this conserved, abundant factor with critical roles in RNA processing and gene expression have uncovered several important themes that underpin their principal functionality in RNA binding:

i) Coupled use of both sequence-specific and non-specific RNA contacts. Early work demonstrating La binding to the 3ʹ trailer of RNA pol III processing intermediates immediately pointed to the UUU-3ʹOH motif as a sequence-specific RNA binding site for La [13,14]. However, it is also known that La engages pre-tRNAs via both the trailer and the main body of the pre-tRNA, as human La will bind a UUU-3ʹOH containing pre-tRNA with approximately 10X greater affinity than the same trailer sequence in isolation [18]. These UUU-3ʹOH independent pre-tRNA contacts have been linked to La function as an RNA chaperone, leading to a model in which the sequence specific UUU-3ʹOH motif engaged by conserved amino acids on the LAM recruits non-specific La associated RNA chaperone activity linked to RRM1 and the C-terminal domain to the appropriate La targets [27]. A similar model has been hypothesized for La function on mRNAs, in which the sequence-specific poly(A) binding mode associated with the LAM and RRM1 might recruit RNA chaperone activity associated with the CTD (or other regions of RRM1, similar to La binding to pre-tRNAs) to non-coding regions that might benefit from La-associated remodeling, such as IRESs [42]. Thus, in both of these hypothesized scenarios, a sequence-specific RNA binding mode is used to recruit a non-specific RNA binding mode to an RNA chaperone substrate.

One question that has not been extensively addressed is whether some sort of RNA secondary structure or other RNA motif contributes to the sequence-independent RNA binding modes of La. One important study addressing this issue showed that in order to bind a hairpin in Domain IV of the HCV IRES with high affinity, human La required both a hairpin structure and a single-stranded extension, however, neither the precise sequence of these nor cleavage of the apical loop of the stem had substantial effects on binding affinity [21]. Interaction of La with this substrate as well as with the bodies of pre-tRNAs has also been shown to be salt-sensitive, arguing for an important contribution through electrostatic contacts to the phosphodiester backbone [18,21]. These data suggest that La binding to RNAs other than via the UUU-3ʹOH or poly(A) binding mode is generally sensitive to shape and secondary structure but that this mode of binding is still highly promiscuous. It is notable, however, that human La has been crosslinked in the immediate vicinity of the AUG of the HCV IRES [33,105], and with high affinity to a specific stem-loop in the HBV RNA [40], suggesting that La can indeed discriminate RNA targets in absence of apparent sequence-specific binding.

Another study with relevance to this question queried whether Sla1p preferentially engages folded versus misfolded pre-tRNA substrates in vivo and in vitro, using the dimethylation of G26 by the tRNA methyltransferase Trm1p as a switch for a physiologically relevant misfolded pre-tRNA substrate with established links to La function [27,65]. Reminiscent of previous work on the HCV Domain IV stem-loop, Sla1p did not preferentially engage hypo-modified pre-tRNAs, suggesting that La does not preferentially engage misfolded substrates even as they rely on La for productive maturation. The question of hidden target preference by La might thus benefit from unbiased screens for high affinity ligands, especially ones that incorporate elements of secondary structure [106], and/or transcriptome wide mapping of La target sites upstream of uridylate or adenylate tails.

ii) La binds different target cohorts depending on its subcellular localization. La shuttles between the nucleus and cytoplasm [82], but predominantly localizes to the nucleus where it is positioned to be one of the first proteins to bind nascent RNA polymerase III transcripts via tandem use of both the UUU-3ʹOH dependent and UUU-3ʹOH independent binding modes [107]. In the case of pre-tRNAs, endonucleolytic processing of the 3ʹ trailer is hypothesized to cause release of both the pre-tRNA body and the cleaved trailer through the resulting lower affinity for each individual ligand, allowing for recycling of La onto new transcripts [18]. Since processing/removal of UUU-3ʹOH occurs in the nucleus prior to nuclear export, this coincidence of localization and processing ensures that the nuclear target cohort of La is largely restricted to nascent pol III transcripts. Evidence consistent with this comes from the use of antibodies specific for human La that has or has not been phosphorylated at S366 by CKII, as the phosphorylated and non-phosphorylated La isoforms localize to the nucleus and cytoplasm, respectively [34]. Specifically, antibodies specific for the S366 phosphorylated isoform (nuclear) co-immunoprecipitate the RNA pol III transcribed pre-tRNAs and the Y RNAs, which are hypothesized to be released by La as their 3ʹends are processed and they are exported to the cytoplasm [108]. On the other hand, antibodies specific for the S366 non-phosphorylated isoform, which localizes to the cytoplasm and nucleolus, immunoprecipitate mRNAs (and 5ʹ TOP mRNAs in particular), from human cells. Also consistent with partitioning of targets based on localization, human La that accumulates in the cytoplasm due to removal of the C-terminal nuclear localization signal (NLS) results in substantially greater association with poly(A) tails relative to full-length, predominantly nuclear La [42]. As La has been shown to redistribute to the cytoplasm during periods of viral infection and cellular stress [38,80–86], and this is associated with enhanced binding of La to cytoplasmic viral and cellular RNA targets, the subcellular localization of La must be taken into consideration with respect to the cohort of RNA targets that it will engage.

iii) La-RNA target binding can be fine-tuned by post-translational modifications. Human La that localizes to the nucleus is phosphorylated at S366 and this affects its binding to pre-tRNAs. When S366 is mutated such that this residue can no longer be phosphorylated, 5ʹ leader processing of La-associated pre-tRNAs is inhibited, suggesting that this modification is necessary to prevent La from restricting access of associated 5ʹ leaders to RNase P [67]. This mutant also shows increased association with a 5ʹTOP mRNA and a resulting inhibitory effect on its translation [94]. Human La can also be phosphorylated by Akt at T389 in the extreme CTD, and this affects the capacity of La to promote translation from the IRES of the cyclin D1 (CCND1) mRNA through RNA chaperone function [26]. In mouse cells, phosphorylation at T301 by Akt results in La export to the cytoplasm and consequent reprogramming of translation of a number of mRNAs [81]. Furthermore, sumoylation of La in the La motif is associated with intracellular La transport in axons [109], while other sumoylation sites between RRM1 and RRM2 are associated with increased binding to 5ʹTOP mRNA motifs in vitro [110]. Since structural studies and La/RNA binding assays typically rely on La purified from recombinant systems, it seems likely that the importance of post-translational modification on La-RNA binding modes is underappreciated.

Concluding remarks

Since the publication of the La module-3ʹUUU-OH crystal structures, there have been significant biochemical and genetic advances in understanding the mechanisms of UUU-3ʹOH independent binding modes and the importance of these in the expression, folding and processing of La targets. With characterized sequence-specific binding modes for UUU-3ʹOH and poly(A) as well as promiscuous recognition of RNA secondary structures, there is the sense that nearly any RNA could be a La target, consistent with the wide breadth of literature describing a diverse cohort of cellular and viral transcripts that rely on La for normal function. In this sense, prioritization should be placed on characterizing the ensemble of RNA targets bound by La in the contexts of living cells, where potential RNA ligands must compete for La binding. Annotation and characterization of altered La target cohorts upon changes in La localization or post-translational modification would thus represent substantial advances in informing how this enigmatic RNA binding protein contributes to the metabolism of La ligands. Recently developed cell-based UV-crosslinking/next-generation sequencing technologies such as iCLIP [111] would thus be well suited to answering these questions.

Acknowledgments

This work is supported by a Open Operating Grant from the Canadian Institutes of Health Research’s Institute of Genetics to MAB.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by an open operating grant from the Canadian Institutes of Health Research's Institute of Genetics.