Unearthing a novel function of SRSF1 in binding and unfolding of RNA G-quadruplexes.

SRSF1 governs splicing of over 1500 mRNA transcripts. SRSF1 contains two RNA-recognition motifs (RRMs) and a C-terminal Arg/Ser-rich region (RS). It has been thought that SRSF1 RRMs exclusively recognize single-stranded exonic splicing enhancers, while RS lacks RNA-binding specificity. With our success in solving the insolubility problem of SRSF1, we can explore the unknown RNA-binding landscape of SRSF1. We find that SRSF1 RS prefers purine over pyrimidine. Moreover, SRSF1 binds to the G-quadruplex (GQ) from the ARPC2 mRNA, with both RRMs and RS being crucial. Our binding assays show that the traditional RNA-binding sites on the RRM tandem and the Arg in RS are responsible for GQ binding. Interestingly, our FRET and circular dichroism data reveal that SRSF1 unfolds the ARPC2 GQ, with RS leading unfolding and RRMs aiding. Our saturation transfer difference NMR results discover that Arg residues in SRSF1 RS interact with the guanine base but not other nucleobases, underscoring the uniqueness of the Arg/guanine interaction. Our luciferase assays confirm that SRSF1 can alleviate the inhibitory effect of GQ on gene expression in the cell. Given the prevalence of RNA GQ and SR proteins, our findings unveil unexplored SR protein functions with broad implications in RNA splicing and translation.

"Boundary residues" between the folded RNA recognition motif and disordered RGG domains are critical for FUS-RNA binding.

Fused in sarcoma (FUS) is an abundant RNA-binding protein, which drives phase separation of cellular condensates and plays multiple roles in RNA regulation. The RNA-binding ability of FUS protein is crucial to its cellular function. Here, our molecular simulation study on the FUS-RNA complex provides atomic resolution insights into the observations from biochemical studies and also illuminates our understanding of molecular driving forces that mediate the structure, stability, and interaction of the RNA recognition motif (RRM) and RGG domains of FUS with a stem-loop junction RNA. We observe clear cooperativity and division of labor among the ordered (RRM) and disordered domains (RGG1 and RGG2) of FUS that leads to an organized and tighter RNA binding. Irrespective of the length of RGG2, the RGG2-RNA interaction is confined to the stem-loop junction and the proximal stem regions. On the other hand, the RGG1 interactions are primarily with the longer RNA stem. We find that the C terminus of RRM, which make up the "boundary residues" that connect the folded RRM with the long disordered RGG2 stretch of the protein, plays a critical role in FUS-RNA binding. Our study provides high-resolution molecular insights into the FUS-RNA interactions and forms the basis for understanding the molecular origins of full-length FUS interaction with RNA.

ALS-linked TDP-43 mutations interfere with the recruitment of RNA recognition motifs to G-quadruplex RNA.

TDP-43 is a major pathological protein in sporadic and familial amyotrophic lateral sclerosis (ALS) and mediates mRNA fate. TDP-43 dysfunction leads to causes progressive degeneration of motor neurons, the details of which remain elusive. Elucidation of the molecular mechanisms of RNA binding could enhance our understanding of this devastating disease. We observed the involvement of the glycine-rich (GR) region of TDP-43 in the initial recognition and binding of G-quadruplex (G4)-RNA in conjunction with its RNA recognition motifs (RRM). We performed a molecular dissection of these intramolecular RNA-binding modules in this study. We confirmed that the ALS-linked mutations in the GR region lead to alteration in the G4 structure. In contrast, amino acid substitutions in the GR region alter the protein structure but do not void the interaction with G4-RNA. Based on these observations, we concluded that the structural distortion of G4 caused by these mutations interferes with RRM recruitment and leads to TDP-43 dysfunction. This intramolecular organization between RRM and GR regions modulates the overall G4-binding properties.

Modular architecture and functional annotation of human RNA-binding proteins containing RNA recognition motif.

RNA-binding proteins (RBPs) are structurally and functionally diverse macromolecules with significant involvement in several post-transcriptional gene regulatory processes and human diseases. RNA recognition motif (RRM) is one of the most abundant RNA-binding domains in human RBPs. The unique modular architecture of each RBP containing RRM is crucial for its diverse target recognition and function. Genome-wide study of these structurally conserved and functionally diverse domains can enhance our understanding of their functional implications. In this study, modular architecture of RRM containing RBPs in human proteome is identified and systematically analysed. We observe that 30% of human RBPs with RNA-binding function contain RRM in single or multiple repeats or with other domains with maximum of six repeats. Zinc-fingers are the most frequently co-occurring domain partner of RRMs. Human RRM containing RBPs mostly belong to RNA metabolism class of proteins and are significantly enriched in two functional pathways including spliceosome and mRNA surveillance. Various human diseases are associated with 18% of the RRM containing RBPs. Single RRM containing RBPs are highly enriched in disorder regions. Gene ontology (GO) molecular functions including poly(A), poly(U) and miRNA binding are highly depleted in RBPs with single RRM, indicating the significance of modular nature of RRMs in specific function. The current study reports all the possible domain architectures of RRM containing human RBPs and their functional enrichment. The idea of domain architecture, and how they confer specificity and new functionalities to RBPs, can help in re-designing of modular RRM containing RBPs with re-engineered function.

Pumping the brakes: A noncanonical RNA-binding domain in FMRP stalls elongating ribosomes.

Loss of function of the RNA-binding protein FMRP causes fragile X syndrome, the most common inherited form of intellectual disability and autism spectrum disorders. FMRP is suggested to modulate synaptic plasticity by regulating the synthesis of proteins involved in neuronal and synaptic function; however, the mechanism underlying FMRP mRNA targeting specificity remains unclear. Intriguing recent work published in JBC by Scarpitti and colleagues identifies and characterizes a noncanonical RNA-binding domain that is required for FMRP-mediated translation regulation, shedding light on FMRP function.

Spontaneous binding of single-stranded RNAs to RRM proteins visualized by unbiased atomistic simulations with a rescaled RNA force field.

Recognition of single-stranded RNA (ssRNA) by RNA recognition motif (RRM) domains is an important class of protein-RNA interactions. Many such complexes were characterized using nuclear magnetic resonance (NMR) and/or X-ray crystallography techniques, revealing ensemble-averaged pictures of the bound states. However, it is becoming widely accepted that better understanding of protein-RNA interactions would be obtained from ensemble descriptions. Indeed, earlier molecular dynamics simulations of bound states indicated visible dynamics at the RNA-RRM interfaces. Here, we report the first atomistic simulation study of spontaneous binding of short RNA sequences to RRM domains of HuR and SRSF1 proteins. Using a millisecond-scale aggregate ensemble of unbiased simulations, we were able to observe a few dozen binding events. HuR RRM3 utilizes a pre-binding state to navigate the RNA sequence to its partially disordered bound state and then to dynamically scan its different binding registers. SRSF1 RRM2 binding is more straightforward but still multiple-pathway. The present study necessitated development of a goal-specific force field modification, scaling down the intramolecular van der Waals interactions of the RNA which also improves description of the RNA-RRM bound state. Our study opens up a new avenue for large-scale atomistic investigations of binding landscapes of protein-RNA complexes, and future perspectives of such research are discussed.

The solution structure of Dead End bound to AU-rich RNA reveals an unusual mode of tandem RRM-RNA recognition required for mRNA regulation.

Dead End (DND1) is an RNA-binding protein essential for germline development through its role in post-transcriptional gene regulation. The molecular mechanisms behind selection and regulation of its targets are unknown. Here, we present the solution structure of DND1's tandem RNA Recognition Motifs (RRMs) bound to AU-rich RNA. The structure reveals how an NYAYUNN element is specifically recognized, reconciling seemingly contradictory sequence motifs discovered in recent genome-wide studies. RRM1 acts as a main binding platform, including atypical extensions to the canonical RRM fold. RRM2 acts cooperatively with RRM1, capping the RNA using an unusual binding pocket, leading to an unusual mode of tandem RRM-RNA recognition. We show that the consensus motif is sufficient to mediate upregulation of a reporter gene in human cells and that this process depends not only on RNA binding by the RRMs, but also on DND1's double-stranded RNA binding domain (dsRBD), which is dispensable for binding of a subset of targets in cellulo. Our results point to a model where DND1 target selection is mediated by a non-canonical mode of AU-rich RNA recognition by the tandem RRMs and a role for the dsRBD in the recruitment of effector complexes responsible for target regulation.

Identification and Characterization of an RRM-Containing, RNA Binding Protein in Acinetobacter baumannii.

Acinetobacter baumannii is a Gram-negative pathogen, known to acquire resistance to antibiotics used in the clinic. The RNA-binding proteome of this bacterium is poorly characterized, in particular for what concerns the proteins containing RNA Recognition Motif (RRM). Here, we browsed the A. baumannii proteome for homologous proteins to the human HuR(ELAVL1), an RNA binding protein containing three RRMs. We identified a unique locus that we called AB-Elavl, coding for a protein with a single RRM with an average of 34% identity to the first HuR RRM. We also widen the research to the genomes of all the bacteria, finding 227 entries in 12 bacterial phyla. Notably we observed a partial evolutionary divergence between the RNP1 and RNP2 conserved regions present in the prokaryotes in comparison to the metazoan consensus sequence. We checked the expression at the transcript and protein level, cloned the gene and expressed the recombinant protein. The X-ray and NMR structural characterization of the recombinant AB-Elavl revealed that the protein maintained the typical ß1α1ß2ß3α2ß4 and three-dimensional organization of eukaryotic RRMs. The biochemical analyses showed that, although the RNP1 and RNP2 show differences, it can bind to AU-rich regions like the human HuR, but with less specificity and lower affinity. Therefore, we identified an RRM-containing RNA-binding protein actually expressed in A. baumannii.

Unusual RNA binding of FUS RRM studied by molecular dynamics simulation and enhanced sampling method.

Amyotrophic lateral sclerosis (ALS) and frontotemporal lobe degeneration (FTLD) are two inter-related intractable diseases of motor neuron degeneration. Fused in sarcoma (FUS) is found in cytoplasmic accumulation of ALS and FTLD patients, which readily link the protein with the diseases. The RNA recognition motif (RRM) of FUS has the canonical α-ß folds along with an unusual lysine-rich loop (KK-loop) between α1 and ß2. This KK-loop is highly conserved among FET family proteins. Another contrasting feature of FUS RRM is the absence of critical binding residues, which are otherwise highly conserved in canonical RRMs. These residues in FUS RRM are Thr286, Glu336, Thr338, and Ser367, which are substitutions of lysine, phenylalanine, phenylalanine, and lysine, respectively, in other RRMs. Considering the importance of FUS in RNA regulation and metabolism, and its implication in ALS and FTLD, it is important to elucidate the underlying molecular mechanism of RNA recognition. In this study, we have performed molecular dynamics simulation with enhanced sampling to understand the conformational dynamics of noncanonical FUS RRM and its binding with RNA. We studied two sets of mutations: one with alanine mutation of KK-loop and another with KK-loop mutations along with critical binding residues mutated back to their canonical form. We find that concerted movement of KK-loop and loop between ß2 and ß3 facilitates the folding of the partner RNA, indicating an induced-fit mechanism of RNA binding. Flexibility of the RRM is highly restricted upon mutating the lysine residues of the KK-loop, resulting in weaker binding with the RNA. Our results also suggest that absence of the canonical residues in FUS RRM along with the KK-loop is equally important in regulating its binding dynamics. This study provides a significant structural insight into the binding of FUS RRM with its cognate RNA, which may further help in designing potential drugs targeting noncanonical RNA recognition.

Cyp33 binds AU-rich RNA motifs via an extended interface that competitively disrupts the gene repressive Cyp33-MLL1 interaction in vitro.

Cyp33 is an essential human cyclophilin prolyl isomerase that plays myriad roles in splicing and chromatin remodeling. In addition to a canonical cyclophilin (Cyp) domain, Cyp33 contains an RNA-recognition motif (RRM) domain, and RNA-binding triggers proline isomerase activity. One prominent role for Cyp33 is through a direct interaction with the mixed lineage leukemia protein 1 (MLL1, also known as KMT2A) complex, which is a histone methyltransferase that serves as a global regulator of human transcription. MLL activity is regulated by Cyp33, which isomerizes a key proline in the linker between the PHD3 and Bromo domains of MLL1, acting as a switch between gene activation and repression. The direct interaction between MLL1 and Cyp33 is critical, as deletion of the MLL1-PHD3 domain responsible for this interaction results in oncogenesis. The Cyp33 RRM is central to these activities, as it binds both the PHD3 domain and RNA. To better understand how RNA binding drives the action of Cyp33, we performed RNA-SELEX against full-length Cyp33 accompanied by deep sequencing. We have identified an enriched Cyp33 binding motif (AAUAAUAA) broadly represented in the cellular RNA pool as well as tightly binding RNA aptamers with affinities comparable and competitive with the Cyp33 MLL1-PHD3 interaction. RNA binding extends beyond the canonical RRM domain, but not to the Cyp domain, suggesting an indirect mechanism of interaction. NMR chemical shift mapping confirms an overlapping, but not identical, interface on Cyp33 for RNA and PHD3 binding. This finding suggests RNA can disrupt the gene repressive Cyp33-MLL1 complex providing another layer of regulation for chromatin remodeling by MLL1.

Structure of SRSF1 RRM1 bound to RNA reveals an unexpected bimodal mode of interaction and explains its involvement in SMN1 exon7 splicing.

The human prototypical SR protein SRSF1 is an oncoprotein that contains two RRMs and plays a pivotal role in RNA metabolism. We determined the structure of the RRM1 bound to RNA and found that the domain binds preferentially to a CN motif (N is for any nucleotide). Based on this solution structure, we engineered a protein containing a single glutamate to asparagine mutation (E87N), which gains the ability to bind to uridines and thereby activates SMN exon7 inclusion, a strategy that is used to cure spinal muscular atrophy. Finally, we revealed that the flexible inter-RRM linker of SRSF1 allows RRM1 to bind RNA on both sides of RRM2 binding site. Besides revealing an unexpected bimodal mode of interaction of SRSF1 with RNA, which will be of interest to design new therapeutic strategies, this study brings a new perspective on the mode of action of SRSF1 in cells.

Functional and Structural Aspects of La Protein Overexpression in Lung Cancer.

La is an abundant phosphoprotein that protects polymerase III transcripts from 3'-5' exonucleolytic degradation and facilitates their folding. Consisting of the evolutionary conserved La motif (LAM) and two consecutive RNA Recognition Motifs (RRMs), La was also found to bind additional RNA transcripts or RNA domains like internal ribosome entry site (IRES), through sequence-independent binding modes which are poorly understood. Although it has been reported overexpressed in certain cancer types and depletion of its expression sensitizes cancer cells to certain chemotherapeutic agents, its role in cancer remains essentially uncharacterized. Herein, we study the effects of La overexpression in A549 lung adenocarcinoma cells, which leads to increased cell proliferation and motility. Expression profiling of several transcription and translation factors indicated that La overexpression leads to downregulation of global translation through hypophosphorylation of 4E-BPs and upregulation of IRES-mediated translation. Moreover, analysis of La localization after nutrition deprivation of the transfected cells showed a normal distribution in the nucleus and nucleoli. Although the RNA binding capacity of La has been primarily linked to the synergy between the conserved LAM and RRM1 domains which act as a module, we show that recombinant stand-alone LAM can specifically bind a pre-tRNA ligand, based on binding experiments combined with NMR analysis. We propose that LAM RNA binding properties could support the expanding and diverse RNA ligand repertoire of La, thus promoting its modulatory role, both under normal and pathogenic conditions like cancer.

Weak binding to the A2RE RNA rigidifies hnRNPA2 RRMs and reduces liquid-liquid phase separation and aggregation.

hnRNPA2 is a major component of mRNA transport granules in oligodendrocytes and neurons. However, the structural details of how hnRNPA2 binds the A2 recognition element (A2RE) and if this sequence stimulates granule formation by enhancing phase separation of hnRNPA2 has not yet been studied. Using solution NMR and biophysical studies, we find that each of the two individual RRMs retain the domain structure observed in complex with RNA but are not rigidly confined (i.e. they move independently) in solution in the absence of RNA. hnRNPA2 RRMs bind the minimal rA2RE11 weakly but at least, and most likely, two hnRNPA2 molecules are able to simultaneously bind the longer 21mer myelin basic protein A2RE. Upon binding of the RNA, NMR chemical shift deviations are observed in both RRMs, suggesting both play a role in binding the A2RE11. Interestingly, addition of short A2RE RNAs or longer RNAs containing this sequence completely prevents in vitro phase separation of full-length hnRNPA2 and aggregation of the disease-associated mutants. These findings suggest that RRM interactions with specific recognition sequences alone do not account for nucleating granule formation, consistent with models where multivalent protein:RNA and protein:protein contacts form across many sites in granule proteins and long RNA transcripts.

RNA-Binding Specificity of the Human Fragile X Mental Retardation Protein.

Fragile X syndrome is the most common form of inherited intellectual disability and is caused by a deficiency of the fragile X mental retardation protein (FMRP) in neurons. FMRP regulates the translation of numerous mRNAs within dendritic synapses, but how FMRP recognizes these target mRNAs remains unknown. FMRP has KH0, KH1, KH2, and RGG domains, which are thought to bind to specific RNA recognition elements (RREs). Several studies used high-throughput methods to identify various RREs in mRNAs that FMRP may bind to in vivo. However, there is little overlap in the mRNA targets identified by each study, suggesting that the RNA-binding specificity of FMRP is still unknown. To determine the specificity of FMRP for the RREs, we performed quantitative in vitroRNA binding studies with various constructs of human FMRP. Unexpectedly, our studies show that the KH domains do not bind to the previously identified RREs. To further investigate the RNA-binding specificity of FMRP, we developed a new method called Motif Identification by Analysis of Simple sequences (MIDAS) to identify single-stranded RNA sequences bound by KH domains. We find that the FMRP KH0, KH1, and KH2 domains bind weakly to the single-stranded RNA sequences suggesting that they may have evolved to bind more complex RNA structures. Additionally, we find that the RGG motif of human FMRP binds with a high affinity to an RNAG-quadruplex structure that lacks single-stranded loops, double-stranded stems, or junctions.

The RNA-Binding Protein Rasputin/G3BP Enhances the Stability and Translation of Its Target mRNAs.

G3BP RNA-binding proteins are important components of stress granules (SGs). Here, we analyze the role of the Drosophila G3BP Rasputin (RIN) in unstressed cells, where RIN is not SG associated. Immunoprecipitation followed by microarray analysis identifies over 550 mRNAs that copurify with RIN. The mRNAs found in SGs are long and translationally silent. In contrast, we find that RIN-bound mRNAs, which encode core components of the transcription, splicing, and translation machinery, are short, stable, and highly translated. We show that RIN is associated with polysomes and provide evidence for a direct role for RIN and its human homologs in stabilizing and upregulating the translation of their target mRNAs. We propose that when cells are stressed, the resulting incorporation of RIN/G3BPs into SGs sequesters them away from their short target mRNAs. This would downregulate the expression of these transcripts, even though they are not incorporated into stress granules.

Early Metastable Assembly during the Stress-Induced Formation of Worm-like Amyloid Fibrils of Nucleic Acid Binding Domains of TDP-43.

TDP-43 protein travels between the cytosol and the nucleus to perform its nucleic acid binding functions through its two tandem RNA recognition motif domains (TDP-43tRRM). When exposed to various environmental stresses, it forms abnormal aggregates in the cytosol of neurons, which are the hallmarks of amyotrophic lateral sclerosis and other TDP-43 proteinopathies. However, the nature of early structural changes upon stress sensing and the consequent steps during the course of aggregation are not well understood. In this study, we show that under low-pH conditions, mimicking starvation stress, TDP-43tRRM undergoes a conformational opening reaction linked to the protonation of buried ionizable residues and grows into a metastable oligomeric assembly (called the "low-pH form" or the "L form"). In the L form, the protein molecules have disrupted tertiary structure, solvent-exposed hydrophobic patches, and mobile side chains but the native-like secondary structure remains intact. The L form structure is held by weak interactions and has a steep dependence on ionic strength. In the presence of as little as 15 mM KCl, it fully misfolds and further oligomerizes to form a ß-sheet rich "ß form" in at least two distinct steps. The ß form has an ordered, stable structure that resembles worm-like amyloid fibrils. The unstructured regions of the protein gain structure during L ⇌ ß conversion. Our results suggest that TDP-43tRRM could function as a stress sensor and support a recent model in which stress sensing during neurodegeneration occurs by assembly of proteins into metastable assemblies that are precursors to the solid aggregates.

RRM but not the Asp/Glu domain of hnRNP C1/C2 is required for splicing regulation of Ron exon 11 pre-mRNA.

The Ron proto-oncogene is a human receptor for macrophage-stimulating protein (MSP). The exclusion of exon 11 in alternative splicing generates ΔRON protein that is constitutively activated. Heterogenous ribonucleaoprotein (hnRNP) C1/C2 is one of the most abundant proteins in cells. In this manuscript, we showed that both hnRNP C1 and C2 promoted exon 11 inclusion of Ron pre-mRNA and that hnRNP C1 and hnRNP C2 functioned independently but not cooperatively. Moreover, hnRNP C1 stimulated exon 11 splicing through intron 10 activation but not through intron 11 splicing. Furthermore, we showed that, whereas the RRM domain was required for hnRNP C1 function, the Asp/Glu domain was not. In conclusion, hnRNP C1/C2 promoted exon 11 splicing independently by stimulating intron 10 splicing through RRM but not through the Asp/Glu domain. [BMB Reports 2019; 52(11): 641-646].

Regulation of the Hepatitis B virus replication and gene expression by the multi-functional protein TARDBP.

Hepatitis B virus (HBV) infects the liver and is a key risk factor for hepatocellular carcinoma. Identification of host factors that support viral replication is important to understand mechanisms of viral replication and to develop new therapeutic strategies. We identified TARDBP as a host factor that regulates HBV. Silencing or knocking out the protein in HBV infected cells severely impaired the production of viral replicative intermediates, mRNAs, proteins, and virions, whereas ectopic expression of TARDBP rescued production of these products. Mechanistically, we found that the protein binds to the HBV core promoter, as shown by chromatin precipitation as well as mutagenesis and protein-DNA interaction assays. Using LC-MS/MS analysis, we also found that TARDBP binds to a number of other proteins known to support the HBV life cycle, including NPM1, PARP1, Hsp90, HNRNPC, SFPQ, PTBP1, HNRNPK, and PUF60. Interestingly, given its key role as a regulator of RNA splicing, we found that TARDBP has an inhibitory role on pregenomic RNA splicing, which might help the virus to export its non-canonical RNAs from the nucleus without being subjected to unwanted splicing, even though mRNA nuclear export is normally closely tied to RNA splicing. Taken together, our results demonstrate that TARDBP is involved in multiple steps of HBV replication via binding to both HBV DNA and RNA. The protein's broad interactome suggests that TARDBP may function as part of a RNA-binding scaffold involved in HBV replication and that the interaction between these proteins might be a target for development of anti-HBV drugs.

A telomerase subunit homolog La protein from Trypanosoma brucei plays an essential role in ribosomal biogenesis.

The autoantigen La protein is an important component of telomerase and a predominantly nuclear phosphoprotein. As a telomerase subunit, La protein associates with the telomerase ribonucleoprotein and influences telomere length. In the reverse transcription, La protein stimulates enzymatic activity and increases repeated addition processivity of telomerase. As nuclear phosphoprotein, La protein binds the 3' poly(U)-rich elements of nascent RNA polymerase III transcripts to facilitate its correct folding and maturation. In this work, we identified a La protein homolog (TbLa) from Trypanosoma brucei (T. brucei). We revealed that TbLa interacts with ribosome-associated protein P34/P37, 40S ribosomal protein SA, and 60S ribosomal subunit L5 in T. brucei. In the interactions between TbLa protein and (P34/P37)/L5/SA, RNA recognition motif (RRM) domain of TbLa was indicated to make the major contribution to the processes. We determined the solution structure of TbLa RRM domain. NMR chemical shift perturbations revealed that the positively charged RNA-binding pocket of TbLa RRM domain is responsible for its interaction with ribosomal and ribosome-associated proteins P37/L5/SA. Furthermore, depletion of TbLa affected the maturation process of 5S rRNA and ribosomal assembly, suggesting TbLa protein might play a significant role in the ribosomal biogenesis pathway in T. brucei. Taken together, our results provide a novel insight and structural basis for better understanding the roles of TbLa and RRM domain in ribosomal biogenesis in T. brucei. DATABASE: Structural data are available in the PDB under the accession number 5ZUH.

Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM.

Human antigen R (HuR) is a key regulator of cellular mRNAs containing adenylate/uridylate-rich elements (AU-rich elements; AREs). These are a major class of cis elements within 3' untranslated regions, targeting these mRNAs for rapid degradation. HuR contains three RNA recognition motifs (RRMs): a tandem RRM1 and 2, followed by a flexible linker and a C-terminal RRM3. While RRM1 and 2 are structurally characterized, little is known about RRM3. Here we present a 1.9-Å-resolution crystal structure of RRM3 bound to different ARE motifs. This structure together with biophysical methods and cell-culture assays revealed the mechanism of RRM3 ARE recognition and dimerization. While multiple RNA motifs can be bound, recognition of the canonical AUUUA pentameric motif is possible by binding to two registers. Additionally, RRM3 forms homodimers to increase its RNA binding affinity. Finally, although HuR stabilizes ARE-containing RNAs, we found that RRM3 counteracts this effect, as shown in a cell-based ARE reporter assay and by qPCR with native HuR mRNA targets containing multiple AUUUA motifs, possibly by competing with RRM12.