RNA components embedded in mRNAs

The UTRs of protein-coding RNAs are known to contain many regulatory elements, some of which are specific sequences that are bound by protein effectors. However, some are locally structured elements that can be considered a form of nested RNA. In every known case, these elements affect the translation of their host transcripts, either directly through interaction with the translational machinery, or indirectly by altering the composition or location of the mRNA.

Internal ribosome entry site

The start codon, AUG, at the 5 end of an mRNA defines the site of translation initiation, where the protein-coding message begins. The upstream sequence is by definition an untranslated region. AUG also codes for the amino acid methionine, and generally occurrence of that particular three-nucleotide combination is not rare; thus initiating translation from the correct AUG is essential for producing a correct protein sequence. Normally the 5cap directs the initiation complex to the correct site. However, the presence of an internal ribosome entry site (IRES) in the 5 UTR of an mRNA facilitates cap-independent translation initiation. IRESs were first identified in viral transcripts [83] as a way to cause host cells to preferentially translate viral RNA when coupled with inhibition of cap-binding proteins necessary for normal cap-

dependent translation. Subsequently, IRESs were found in cellular mRNAs spanning several species including human, Drosophila, and yeast, and may exist as a way to enhance translation of endogenous transcripts under cellular conditions that affect normal translation initiation [84]. There is no consensus IRES sequence or structure, suggesting either difficult-to-detect higher-order structural similarity or a diversity of mechanisms for achieving similar function.

RNA localization motifs

The better-understood mode of regulating protein location in eukaryotes occurs via explicit transport of a protein from the site of translation (the ribosome, usually positioned on the endoplasmic reticulum, an organelle responsible for protein traf- ficking throughout the cell) to another destination distal to the nucleus – the plasma membrane for example. However, protein localization can also be mediated prior to translation, by localizing the mRNAs themselves to the correct subcellular com- partment, where proteins can subsequently be translated. The advantages to this mode of localization include a finer level of regulation, since local stimuli can directly affect protein production, rather than triggering a more time-consuming cascade of signaling; the reduced cost of localizing a single mRNA molecule that can produce multiple proteins on site, compared to localizing several proteins independently; and the effective sequestering of protein products in a particular compartment to prevent off-target effects where the protein activity is not desired [85].

Localization of mRNAs is thought to involve specific sequence and structural sig- nals contained predominantly in the UTRs that are bound by carrier complexes and shuttled to their destination, though few of these signals have been well characterized compared to the number of transcripts believed to be localized. The best known ex- amples are involved in developmental pattern formation in Drosophila embryos – the

bicoid mRNA contains a large,∼ 600 base pair region in its 3 UTR that was shown to be necessary for localization of the transcript to the anterior pole of Drosophila

oocytes [86]. This region consists of several localization elements, many of which fold into hairpin structures that each confer specificity for the different stages of localiza- tion. In budding yeast, the mating-type determiner ASH1 is selectively localized to the daughter cell during cell division via a cluster of four small stem-loop structures, each of which was shown to be independently sufficient to confer localization of the transcript to the bud tip, but to have enhanced efficacy in combination [87]. Localiza- tion motifs have also been identified forXenopus vg1 transcript, chickenβ-actin, and

Camk2a and Map2 in rodent neurons, where these (along with potentially hundreds

other transcripts [88]) are localized to dendrites.

Selenocysteine insertion sequence

The selenocysteine insertion sequence (SECIS) element is a motif in the 5 UTR that mediates the introduction of a non-standard amino acid, selenocysteine, into a protein sequence [89]. The presence of a SECIS element in the transcript causes recruitment of a specialized selenocysteine-carrying tRNA, which binds to the UGA codon during translation; UGA is normally read as a stop codon. SECIS elements are common among both eukaryotic and bacterial transcripts that encode a class of proteins called selenoproteins, but despite similarity in size (∼ 60 nts) and shape (hairpin), the different SECIS signals have distinct sequence [90]

Iron response element

Cells are often responsive to different concentrations of small molecules and ions. Transcripts that function in iron metabolism pathways contain hairpin motifs called iron response elements (IRE), which are bound by iron-regulatory proteins (IRPs)

that affect the translation and stability of the transcripts. In ferritin transcripts, IREs located in the 5 UTR mediate translation inhibition when cellular levels of iron are low, and the iron-storage functionality of ferritin protein are not needed. In transferrin receptor transcripts, IREs occur in the 3 UTR in a cluster, and binding of these by IRPs in low iron states causes stabilization of the transcript, facilitating active translation of receptor proteins for iron uptake [91]. IREs are structurally conserved among many different transcripts, as a bulged hairpin with a specific six- nucleotide loop.

Riboswitches

In contrast to IREs, which respond to iron concentration with the aid of protein com- plexes, riboswitches are a class of autonomous RNA aptamers, which cause transcrip- tional modulation upon binding of specific metabolites [92]. The canonical ribozyme is located in the 5 UTR of an mRNA and consists of two domains. The first is the metabolite-binding aptamer domain, which is highly specialized to bind specifi- cally to one particular metabolite, such as nucleotides or coenzyme B12. The second domain is the expression platform, the effector of transcriptional or translational con- trol. Binding of the metabolite to the aptamer domain induces a conformational change that causes activation of the expression platform. In the case of the bacte- rial coenzyme-B12 riboswitch, the expression platform has two distinct effects. Upon binding of the coenzyme-B12 molecule to a partly transcribed mRNA, the conforma- tional change induces the formation of a terminator stem that causes premature tran- scription termination before a functional mRNA can be created; under low coenzyme- B12 concentrations, the unbound aptamer allows the expression platform to form an anti-terminator that inhibits formation of the terminator stem, so that the mRNA is transcribed normally [93]. The second effect is on fully transcribed mRNAs, where the

conformational change blocks access to the ribosome binding site and inhibits transla- tion [94]. Another mechanism for control is found in theglmS gene in Gram-positive bacteria, which contains a riboswitch that is a self-cleaving ribozyme; activation un- der conditions of high concentration of glucosamine-6-phosphate, the product whose formation is catalyzed by GlmS, causes cleavage and subsequent degradation of the

glmS mRNA [95].

Although most riboswitches have been found in bacteria, the thiamine pyrophos- phate (TPP) riboswitch, which is sensitive to cellular thiamine levels, is found in plants and fungi as well, though interestingly in plants the riboswitch resides in the 3 UTR [96].