Toward a Systematic Understanding of mRNA 3′ Untranslated Regions

Abstract

Messenger RNAs (mRNAs) contain prominent untranslated regions (UTRs) that are increasingly recognized to play roles in mRNA processing, transport, stability, and translation. 3′ UTRs are believed to harbor recognition sites for a diverse set of RNA-binding proteins that regulate gene expression as well as most active microRNA target sites. Although the roles of 3′ UTRs in the normal and diseased lung have not yet been studied extensively, available evidence suggests important roles for 3′ UTRs in lung development, inflammation, asthma, pulmonary fibrosis, and cancer. Systematic, genome-wide approaches are beginning to catalog functional elements within 3′ UTRs and identify the proteins and microRNAs that interact with these elements. Application of new data sets and experimental approaches should provide powerful insights into how 3′ UTR-mediated regulatory events contribute to disease and may inspire novel therapeutic approaches.

The transcriptome (the full set of RNA transcripts) is a necessary intermediate between the genome and the proteome. In addition to translated (or coding) regions, messenger RNAs (mRNAs) contain 5′ and 3′ untranslated regions (UTRs). A rapidly growing body of evidence indicates that these noncoding mRNA sequences have diverse roles in regulating RNA processing, transport, stability, and translation. A comprehensive systems biology approach to gene regulation will require that we systematically identify active sequences within UTRs (cis elements), identify the cellular trans factors that interact with these cis elements, and determine the functional consequences of these interactions. Here we review available information about human 3′ UTRs, known classes of 3′ UTR cis elements, evidence that supports the importance of 3′ UTRs in lung disease, and recent progress toward transcriptome-wide analysis of 3′ UTRs. The 5′ UTR, which plays a major role in regulating translation, has been reviewed elsewhere (1).

3′ UTRs ARE PROMINENT FEATURES OF mRNAS

The 3′ UTR begins at the stop codon. Polyadenylation signals (typically AAUAAA or a similar sequence) within immature mRNA 3′ UTRs are recognized by an enormous complex of approximately 85 proteins that directs cleavage of the mRNA and addition of a poly(A) tail of approximately 250 nucleotides (nt) (2). Most human genes have more than one polyadenylation signal, and usage of alternative polyadenylation sites is one source of variability in 3′ UTR length (3). For some genes, alternative splicing (with use of an alternative final exon) produces variation in both 3′ UTR length and sequence (4). For example, subjects with systemic lupus erythematosus had increased levels of an alternatively spliced form of the T cell receptor ζ transcript that was less stable than the predominant form seen in healthy control subjects (5). Our analysis of a large set of well-characterized human transcripts (Figure 1) indicates that the length of the 3′ UTR varies widely, but on average 3′ UTRs (not including the poly[A] tail) are 1,278 nt, or 36% of the total length of each mRNA. In contrast, yeast 3′ UTRs average approximately 100 nt and very few are greater than 400 nt (6), suggesting that 3′ UTRs serve additional regulatory functions in complex multicellular organisms. Approximately 15 to 20% of human 3′ UTR sequence was found to be conserved in four other vertebrate species (7). Some conserved regions contain known or predicted cis regulatory elements, but others do not, suggesting that additional classes of cis elements remain to be discovered. The existence of large amounts of nonconserved or poorly conserved 3′ UTR sequence (7) also suggests that 3′ UTRs contribute to differences in gene regulation between species.

Figure 1.

3′ Untranslated regions (UTRs) are prominent features of human messenger RNAs (mRNAs). We analyzed all RefSeq mRNA sequences available from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/projects/RefSeq/, accessed on July 27, 2010) and determined lengths of (A) the 5′ UTR, (B) the coding region, and (C) the 3′ UTR for the 15,295 human RefSeq mRNA sequences that included annotations indicating positions of start and stop codons and one or more polyadenylation sites. For 5,672 of these transcripts, two or more polyadenylation sites were annotated and 3′ UTR lengths were calculated for each of these variants. Mean lengths were 243 nucleotides (nt) for the 5′ UTR, 1,766 nt for the coding region, and 1,278 nt for the 3′ UTR.

INTERACTIONS BETWEEN 3′ UTRs AND RNA BINDING PROTEINS REGULATE mRNA PROCESSING, LOCALIZATION, STABILITY, AND TRANSLATION

Interactions between RNAs and proteins lead to formation of ribonucleoprotein (RNP) complexes (8). Many RNA-binding proteins have been identified and in some cases consensus RNA sequence motifs or structures required for protein binding have been defined (8, 9). Although RNA-binding proteins interact with all portions of mRNAs (5′ UTRs, coding sequences, introns, and 3′ UTRs), certain RNA-binding proteins interact preferentially with 3′ UTRs (Table 1). This relates at least in part to the prevalence of specific recognition sequences in 3′ UTRs. Individual 3′ UTRs may contain multiple protein binding sites, which may allow for synergistic, antagonistic, or other combinatorial regulatory interactions (10). Systems biology approaches similar to those used to understand combinatorial interactions within DNA promoters (11) will be required to fully understand how 3′ UTR signals are integrated.

TABLE I.

SELECTED 3′ UNTRANSLATED REGION FUNCTIONAL ELEMENTS

Cis Element	Motif	Trans Factor	Major Function	References
Polyadenylation signal	AAUAAA or similar	Cleavage/polyadenylation specificity factor complex	Marks site for addition of poly(A) tail	2
Zipcode	54-nt sequence	Zipcode binding protein-1	Localizes β-actin mRNA in cytoplasm	14
AU-rich element (ARE)	One or more AUUUA repeats	ARE-binding proteins	Control of mRNA stability	17
miRNA target	Complementary to miRNA sequence	miRNA in RISC complex	Inhibition of mRNA stability and/or translation	24, 42, 44

Definition of abbreviations: miRNA, micro RNA; mRNA, messenger RNA; nt, nucleotide; RISC, RNA-induced silencing complexes.

RNP complex formation is critical for RNA processing and localization (8, 9). The best-known example of a 3′ UTR cis element with a role in mRNA processing is the polyadenylation signal, which is present in virtually all eukaryotic mRNAs (with the exception of histone mRNAs) (12). Studies in neurons and other polarized cells have led to the identification of 3′ UTR sequences important in targeting mRNAs to specific regions within the cytoplasm, allowing for localized protein synthesis (13). For example, the RNA binding protein zipcode binding protein-1 binds a 54-nucleotide “zipcode” sequence within the 3′ UTR of β-actin mRNA and directs the mRNA to the leading edge of fibroblasts and other cells where actin filament assembly is required (14).

Interactions between 3′ UTRs and RNA-binding proteins can affect protein production by altering mRNA stability and translation (15, 16). One of the best-studied 3′UTR cis-regulatory elements is the AU-rich element (ARE) (17). AREs contain one or more repeats of an AUUUA motif or other AU-rich sequences. ARE-binding proteins, such as AUF1/hnRNPD, HuR, TIA-1, KSRP, and tristetraprolin, recognize these sequences leading to changes in stability of mRNA transcripts (15, 16, 18). Some ARE-binding proteins tend to increase RNA stability, whereas others decrease stability, and effects of a given ARE can depend on cellular context (18). For example, cytokine-induced changes in binding of KSRP and HuR to an ARE in the 3′ UTR of human inducible nitric oxide synthase (iNOS) mRNA were found to contribute to regulation of iNOS, because KSRP destabilized this mRNA, whereas HuR had a stabilizing effect (19). Mechanisms underlying ARE-mediated changes in mRNA stability are complex (20) and incompletely understood. Certain AREs promote recruitment of mRNA to cytoplasmic processing bodies (P bodies) where enzymes involved mRNA decay are concentrated (21). Evidence suggests that AU-rich elements can affect protein production by altering mRNA stability, by sequestering mRNAs from ribosomes (e.g., by sequestration in P bodies), and by recruiting proteins that directly affect translation efficiency (22, 23). Sequences that are recognized by microRNAs (miRNAs) and other small (20–30 nt) RNAs, including small interfering RNAs (siRNAs) and Piwi-interacting RNAs (piRNAs) (24), also tend to be found largely in 3′ UTRs. These small RNAs associate with members of the Argonaute family of small RNA-binding proteins within larger RNA-induced silencing complexes (RISCs) (25). Base-pairing interactions between a small RNA in the RISC and a partially complementary sequence in the 3′ UTR lead to specific interactions between mRNAs and RISCs (24). RISCs are heterogeneous, but may contain Argonaute family members and other proteins that can cleave small RNA:mRNA heteroduplexes, block translational initiation or elongation, recruit poly(A) deadenylating and mRNA decapping complexes, and direct mRNAs to P bodies (24). In addition to AREs and small RNA/RISC targets, many other 3′ UTR cis elements can also affect mRNA stability (13).

3′ UTR IN DEVELOPMENT AND DISEASE

Given the fundamental regulatory role of 3′ UTRs, it is not surprising that 3′ UTRs and associated protein complexes make critical contributions to normal development (26). Levels of many miRNAs change substantially during development of the lung (27, 28) and other organs (29, 30). Mice with targeted deletions of specific miRNAs are beginning to reveal the functions of these miRNAs in vivo. For example, deletion of the muscle-specific miRNA miR1-2, which recognizes targets in the 3′ UTR of the cardiac transcription factor Irx5 mRNA and other cardiac mRNAs, showed a role for this miRNA in cardiac morphogenesis, electrical conduction, and cell cycle control (31). Although little is known about the in vivo effects of specific miRNAs in lung cells, miRNA overexpression experiments together with analyses of effects of concurrent changes in miRNA and mRNA levels suggest that miRNA interactions with miRNA targets in 3′ UTRs also play important regulatory roles in lung development (27, 28).

AREs are frequently found in 3′ UTRs of mRNAs encoding cytokines and these AREs are important in regulating inflammatory and immune responses (32). A striking demonstration of the importance of these elements is provided by studies of mice with targeted deletions of the ARE in the 3′ UTR of tumor necrosis factor (TNF) mRNA (33). These mice have impaired regulation of TNF and develop chronic inflammatory arthritis and inflammatory bowel disease. Recent studies of experimental asthma indicate that many miRNAs, including miR-21 (34), miR-126 (35), and let-7 (36), are involved in establishment of Th2 responses and allergic inflammation. Some relevant target sites in 3′ UTRs have been identified (e.g., the miR-21 target site in IL-12p35 mRNA [34]), but more work will be required to identify the full range of targets responsible for in vivo effects of miRNA inhibition in these asthma models. Alterations in 3′ UTRs or in 3′ UTR interacting proteins and miRNAs have also been implicated in several other lung diseases, including lung cancer (37, 38) and pulmonary fibrosis (39). A small number of rare disease-causing mutations in 3′ UTRs have been identified (40), but the potential contributions of many common 3′ UTR polymorphisms to disease susceptibility remain largely unknown.

SYSTEMS BIOLOGY APPROACHES TO THE 3′ UTR

Although many valuable insights have come from traditional, low-throughput studies of individual 3′ UTRs or 3′ UTR cis elements and the trans-regulatory complexes that interact with them, a systematic understanding of the roles of 3′ UTRs will require the use of multiple transcriptome-wide approaches. Computational approaches (based on sequence analysis) predict the existence of AREs in approximately 5 to 8% of human 3′ UTRs (41) and of one or more miRNA targets in as many as 60% or more of human 3′ UTRs (42). Experimental approaches have been used to define the complement of RNAs that associate with specific RNA-binding proteins (“ribonomics”) (43). Recent improvements in high-throughput sequencing and cross-linking techniques have enabled fine mapping of protein binding sites within the transcriptome (44). High-throughput sequencing has recently been used together with massively parallel DNA synthesis for detailed analysis of DNA regulatory elements (45), and similar approaches are likely to be suitable for analyzing 3′ UTR regulatory elements. These types of transcriptome-wide approaches should refine our understanding of known classes of cis elements and trans-regulatory factors and seem likely to identify novel classes of 3′ UTR regulatory elements.

CONCLUSIONS

A systematic understanding of gene regulation will require that we understand in detail how each level of regulation works and how different levels are integrated. Ample evidence indicates that 3′ UTRs of human mRNAs contain many cis regulatory elements that control RNA processing, localization, stability, and/or translation. These elements are important in normal development and in disease in the lung and other organs. Much more work is required to identify the complete set of 3′ UTR cis elements and the trans-regulatory elements that interact with them and to determine the functional consequences of these interactions. Fortunately, powerful transcriptome-wide computational and experimental methods are now being used to address these issues. Together with lower-throughput reductionist approaches, these transcriptome-wide approaches should move us closer to a systems biology understanding of how 3′ UTRs contribute to gene regulation in health and disease.