Eukaryotic RNA Processing

Published on 28/02/2015 by admin

Last modified 22/04/2025

Print this page

This article have been viewed 3152 times

CHAPTER 16 Eukaryotic RNA Processing^*

In all organisms, the genetic information is encoded in the sequence of the DNA. However, to be used, this information must be copied, or transcribed, into the related polymer, RNA. Eukaryotes synthesize many different types of RNA, but none of these RNAs is simply transcribed as a finished product. The mature, functional forms of all eukaryotic RNA species are generated by posttranscriptional processing, and these processing reactions are the major topic of this chapter.

The major RNAs can be assigned to three major classes: (1) The cytoplasmic messenger RNAs (mRNAs) and their nuclear precursors (pre-mRNAs) carry the information that is used to specify the sequence, and therefore the structure, of all proteins in the cell. (2) Other RNAs do not encode protein but function directly, playing major roles in several metabolic pathways, including protein synthesis. These include the ribosomal RNAs (rRNAs) and transfer RNAs (tRNAs), which are the key components of the protein synthesis machinery; the small nuclear RNAs (snRNAs), which form the core of the pre-mRNA splicing system; and the small nucleolar RNAs (snoRNAs), which are important factors in ribosome biogenesis. These RNAs are generally much longer-lived than are the mRNAs and therefore often are referred to as stable or nonprotein coding RNAs (ncRNAs). (3) The third and most recently identified class of RNA comprises several structurally related groups of very small (21 to 25 nucleotides) RNA species that play important roles in regulating gene expression. Base pairing between endogenous micro-RNAs (miRNAs) and target mRNAs in the cytoplasm represses their translation into protein. The packaging of DNA into a nontranscribed form termed heterochromatin (see Fig. 13-9) is promoted by a class of nuclear, small heterochromatic RNAs (shRNAs). Finally, the introduction of small double-stranded RNAs into many cell types and organisms results in cleavage of the target mRNA and consequent silencing of gene expression. This phenomenon is described as RNA interference (RNAi), and the RNAs are referred to as small interfering RNAs (siRNAs).

Synthesis of mRNAs

An overview of mRNA synthesis and degradation is shown in Figure 16-1.

Figure 16-1 Synthesis and degradation of eukaryotic mRNAs. Nascent mRNA transcripts are transcribed by RNA polymerase II. Formation of the 5′ cap structure and cleavage and polyadenylation of the 3′ end of the mRNA both occur cotranscriptionally and involve factors that are recruited by the C-terminal domain (CTD) of the transcribing polymerase (see Fig. 15-4). The termination of transcription requires both the recognition of the site of polyadenylation and the activity of the 5′-exonuclease Rat1, which degrades the nascent RNA transcripts. Rat1 binds to the polymerase CTD via Rtt103. Pre-mRNA splicing can either be cotranscriptional or occur shortly after transcript release, and recruitment of splicing factors is not strongly dependent on the CTD. In human cells, the spliceosome deposits the exon-junction complex (EJC) around 24 nucleotides upstream of the site of splicing. Several steps in nuclear mRNA maturation also are subject to surveillance. In yeast, nuclear pre-mRNAs can be either 3′ degraded by the nuclear exosome complex or decapped and 5′ degraded by the exonuclease Rat1. Nuclear decapping requires the Lsm2–8 complex and is probably performed by the Dcp1/2 decapping complex. Once in the cytoplasm, the mRNA is translated into proteins and undergoes degradation. Several different mRNA degradation pathways have been identified. A, Nonsense-mediated decay (NMD). If the EJCs all lie within or very close to the ORF, they will be displaced by the translating ribosomes. However, if an EJC lies beyond the end of the ORF, it will remain on the translated mRNA. This is taken as evidence that translation has terminated prematurely and triggers the NMD pathway. Recognition of the EJC requires the Upf1/2/3 surveillance complex, which also interacts with the ribosomes as they terminate translation. In yeast, NMD triggers both rapid decapping and 5′ degradation, without prior deadenylation, and 3′ degradation by the exosome. B, General mRNA turnover. During translation, most mRNAs undergo progressive poly(A) tail shortening. Loss of the poly(A) tail leads to rapid degradation. As in the nucleus, cytoplasmic mRNAs can be degraded from either the 5′ or the 3′ end. 5′ degradation occurs largely in a specialized cytoplasmic region termed the P body in yeast or cytoplasmic foci in human cells. Here, the mRNAs are decapped by the Dcp1/2 heterodimer and then degraded by the cytoplasmic 5′-exonuclease Xrn1. Both activities are strongly stimulated by the cytoplasmic Lsm1–7 complex. Alternatively, deadenylated mRNAs can be 3′ degraded by the cytoplasmic exosome. C, ARE-mediated decay. In this pathway, specific A+U rich elements (AREs) are recognized by ARE-binding proteins (ARE-BP) in the nucleus. These are transported to the cytoplasm in association with the mRNA and recruit the cytoplasmic exosome to rapidly degrade the RNA. D, Nonstop decay. If the mRNA lacks a translation termination codon, the first translating ribosome will stall and be trapped at the 3′ end of the RNA. The Ski7 protein, which is associated with the cytoplasmic exosome complex, is believed to release the stalled ribosome and target the RNA for 3′ degradation by the exosome. Note that this legend provides detail beyond the text.

mRNA Capping and Polyadenylation

Two distinguishing features set mRNA apart from other RNAs: a 5′ cap structure and a 3′ poly(A) tail. Both of these elements help to protect the mRNA against degradation and act synergistically to promote translation in the cytoplasm.

The mRNA cap is an unusual structure. It consists of an inverted 7-methylguanosine residue, which is joined onto the body of the mRNA by a 5′-triphosphate-5′ linkage (Fig. 16-2). Cap addition involves three enzymatic activities: A 5′ RNA triphosphatase cleaves the 5′ triphosphate on the nascent transcript to a diphosphate; RNA guanylyltransferase forms a covalent enzyme–GMP complex and then caps the RNA by transferring this to the diphosphate; and RNA (guanine-7) methyltransferase covalently alters the guanosine base by methylation, generating m⁷G. In addition, the first encoded nucleotides are frequently modified by methylation of the 2′ hydroxyl position on the ribose group, but the functional significance of these internal modifications is currently unclear.

Figure 16-2 mRNAs have a distinctive 5′ cap structure. A, The 5′ ends of mRNAs are blocked by an inverted guanosine residue that is attached to the body of the mRNA by a 5′–5′ triphosphate linkage. The N7 position of the guanosine is methylated (red). The first encoded nucleotide of the mRNA (Nuc 1) is also methylated on the 2′-hydroxyl of the ribose ring. The second nucleotide (Nuc 2) may also be methylated. B, Capping of mRNAs is a multistep process.

During 3′ processing, the nascent pre-mRNA is cleaved by an endonuclease, and a tail of adenosine residues is added by poly(A) polymerase. Around 200 to 250 A residues are added to mRNAs in human cells, and around 70 to 90 are added in yeast. Cleavage and polyadenylation are performed by a large complex containing approximately 20 proteins that recognizes sequences in the mRNA, of which the best defined is a highly conserved AAUAAA motif located upstream of the site of polyadenylation (Fig. 16-3).

Figure 16-3 Signals for pre-mRNA polyadenylation. A, Poly(A) tails are added to pre-mRNAs following transcription. After pol II transcribes the protein-coding region of the mRNA, it encounters two sequence elements: AAUAAA and a GU-rich element. These act as signals for the assembly of a large 3′ processing complex that cleaves the nascent pre-mRNA, releasing it from the transcription complex, and adds a tail of up to 200 adenosine residues. B, The poly A signal is highly conserved in vertebrates.

Links between mRNA Processing and Transcription

The processes of cap addition and 3′ cleavage and polyadenylation are both linked to transcription of the mRNA by RNA polymerase II and occur cotranscriptionally on the nascent RNA (Fig. 16-1). The C-terminal domain (CTD) of the largest subunit of RNA polymerase II (RNA pol II) consists of many copies of a seven-amino-acid repeat (YSPTSPS), which undergo reversible modification by phosphorylation (see Fig. 15-4). A pronounced change in the CTD phosphorylation pattern coincides with the release of the polymerase from initiation mode into processive elongation mode. Immediately following transcription initiation, the repeats are largely phosphorylated on the serine residue at position 5. This modification is lost, while serine 2 phosphorylation increases, as the polymerase moves along the transcript. Capping of the 5′ end of the mRNA occurs by the time the transcript is approximately 25 to 30 nucleotides long, and the capping enzyme interacts with the serine 5 phosphorylated CTD. This and other interactions with the polymerase result in strong allosteric activation of capping activity. In contrast, the cleavage and polyadenylation factors involved in 3′ end processing are recruited by interaction with the CTD phosphorylated at serine 2.

The termination of transcription by RNA polymerase II is dependent on RNA processing. Termination requires recognition of the poly(A) site by the cleavage and polyadenylation factors. These are carried with the transcribing polymerase, and their offloading might make the polymerase competent for termination. Cleavage of the nascent transcript also allows the entry of a 5′ exonuclease—an enzyme that can degrade RNA from the 5′ end in a 3′ direction. This enzyme, which is called Rat1 in yeast and Xrn2 in humans, then chases after the transcribing polymerase, degrading the newly transcribed RNA strand as it goes. When the exonuclease catches the polymerase, it stimulates termination of transcription. This is referred to as the Torpedo model for transcription termination.

Human β-globin mRNA precursors contain an additional cleavage site (termed the cotranscriptional cleavage site) downstream of the site of polyadenylation. The cotranscriptional cleavage site RNA sequence has intrinsic self-cleavage activity in the absence of proteins. Such an RNA is referred to as a self-cleaving ribozyme. This cleavage provides an entry site for the Xrn2 nuclease, allowing more efficient termination.

Regulated 3′ End Formation on Histone mRNAs

A different 3′ end processing system is seen for mRNAs encoding the major, replication-dependent histone proteins. These are highly expressed only during DNA replication, when they must package the newly synthesized DNA. A sequence in the 3′ untranslated region (3′ UTR) of these mRNAs is recognized by base pairing to a small RNA: the U7 snRNA. In addition, a specific stem-loop structure is recognized by a protein that is referred to as the stem-loop binding protein. Endonuclease cleavage generates the mature 3′ end of the mRNA, which is not polyadenylated but is protected by the stem-loop binding protein. The efficiency of histone mRNA synthesis is increased during DNA replication at least in part by increased abundance of stem-loop binding protein. Minor histone variants that are synthesized throughout the cell cycle are polyadenylated like other mRNAs.

Pre-mRNA Splicing

Important experiments in the 1950s and 1960s established that genes were collinear with their protein products. It therefore came as a considerable surprise when, in the late 1970s, it emerged that genes in animals and plants frequently had numerous strikingly large inserts whose sequence was not included in the mature mRNA or the protein product. It turns out that most human pre-mRNAs undergo splicing reactions, in which specific regions are cut out and the remaining RNA is covalently rejoined. The regions that will form the mRNA are termed exons, and the bits that are cut out (and are normally degraded) are called introns. In unicellular eukaryotes, introns are generally a few hundred nucleotides in length or shorter. In metazoans, however, they are often several kilobases in length, and pre-mRNAs can contain many introns. It is therefore remarkable that all of the sites can be precisely identified and spliced.

Signals for Splicing

The signals in the pre-mRNA that identify the introns and exons are recognized by a combination of proteins and a group of small RNAs called the small nuclear RNAs (snRNAs). The snRNAs function in complexes with proteins in small nuclear ribonucleoprotein (snRNP) particles. Splicing occurs in a large complex termed the spliceosome, within which the pre-mRNA assembles together with five snRNAs (U1, U2, U4, U5, and U6) and around 100 different proteins. Particularly important protein-splicing factors are members of a large group of SR-proteins—so named because they contain domains rich in serine-arginine dipeptides.

Three conserved sequences within introns play key roles in their accurate recognition by the splicing machinery (Fig. 16-4). These lie immediately adjacent to the 5′ splice site and the 3′ splice site and surrounding an internal region that will form the intron branch point during the splicing reaction. The U1 and U6 snRNAs have sequences that are complementary to the 5′ splice site, while U2 is complementary to the branch point region.

Figure 16-4 Signals and mechanism of pre-mrna splicing. The precursors to most mRNAs in humans and other eukaryotes contain regions (introns) that will not form part of the mature mRNA and do not encode protein products. During pre-mRNA splicing, the introns are removed and the flanking regions (exons) are ligated. A, Introns contain three conserved sequence elements that are recognized during splicing. These lie at the 5′ and 3′ splice sites and surrounding the branch-point adenosine within the intron. Numbers indicate the degree of conservation at each position in mammalian pre-mRNAs. The branch point sequence is much more highly conserved between different pre-mRNAs in yeast. The region between the branch point and the 3′ splice site frequently contains a run of pyrimidine residues, which is referred to as the polypyrimidine tract. B, Pre-mRNA splicing involves two catalytic steps. An attack by the branch-point adenosine on the 5′ splice site releases the 5′ exon and intron as a circularized molecule (referred to as the intron lariat) joined to the 3′ exon. In the second step, the 3′ end of the 5′ exon attacks the 3′ splice site releasing the joined exons and the free intron lariat. The lariat is subsequently linearized (debranched) and degraded.

While the spliceosome will finally bring together the sequences at each end of the intron, it is believed that the splicing machinery initially recognizes the exons in a reaction termed exon definition. This makes sense because mRNA exons are generally quite small—up to a few hundred nucleotides in length—whereas the introns can be many kilobases long.

No sequences in the exons are strictly required for splicing, but there are important stimulatory elements termed exonic splicing enhancers (ESEs), which generally bind members of the SR-protein family. The ESEs have two major functions: They stimulate the use of the flanking 5′ and 3′ splice sites, promoting exon definition, and they prevent the exon in which they are located from being included in an intron. This latter function is particularly important in ensuring that all introns are spliced out without the splicing machinery skipping from the 5′ end of one intron to the 3′ end of a downstream intron.

The Pre-mRNA Splicing Reaction

The splicing reaction proceeds in two steps (Fig. 16-4). In the first, the 5′–3′ phosphate linkage that joins the 5′ exon to the first nucleotide of the intron—at the 5′ splice site—is attacked and broken. This reaction leaves the 5′ end of the intron attached to the adenosine residue via an unusual 5′–2′ phosphate linkage. Since this adenosine remains attached to the flanking nucleotides by conventional 5′ and 3′ phosphodiester bonds, this creates a circular molecule with a tail that includes the 3′ exon. This structure is termed the intron lariat, and the adenosine to which the 5′ end of the intron is attached is termed the branch point, because it has a branched structure. In the second step of splicing, the free 3′ hydroxyl on the 5′ exon is used to attack and break the linkage between the last nucleotide of the intron and the 3′ exon—at the 3′ splice site. This leaves the 5′ and 3′ exons joined by a conventional 5′–3′ linkage and releases the intron as a lariat. This is linearized by the debranching enzyme and is probably rapidly degraded from both ends by exonucleases.

The initial steps in splicing are the recognition of the 5′ splice site by the U1 snRNA and the binding of U2 snRNA to the branch-point region, assisted by SR-proteins (Fig. 16-5). Base pairing between U2 and the pre-mRNA leaves a single adenosine bulged out of a helix and available for interaction with the 5′ splice site. The U4 and U6 snRNAs then join the spliceosome as a base-paired duplex, within a large complex that also contains the U5 snRNA. The U4 and U6 base pairing is opened, and the liberated U6 sequences displace U1 at the 5′ splice site. They also bind to U2—bringing the 5′ splice site and branch point into close proximity. At this point, the first enzymatic step of splicing occurs. This reaction is believed to be directly catalyzed by the intricate structure of the snRNA/pre-mRNA interactions rather than by the protein components of the spliceosome. The 5′ splice site is attacked and broken by the ribose 2′ hydroxyl group of the adenosine residue that is bulged out of the U2-intron duplex. The U5 snRNA and its associated proteins are responsible for holding onto the now free 5′ exon and correctly aligning it with the 3′ exon for the second catalytic step of splicing.

Figure 16-5 Small nuclear rnas play key role in pre-mrna splicing. Although shown as RNAs, the snRNAs function in large RNA-protein complexes termed snRNPs. Despite this fact, the major steps in both intron recognition and catalysis are believed to be performed by the snRNAs. The 5′ splice site and intron branch point are recognized by base pairing to the U1 and U2 snRNAs, respectively. The U5 snRNA enters the spliceosome in a complex with U4 and U6, which are tightly base-paired. U5 forms contacts with both the 5′ and 3′ exons. U4 releases U6, which base-pairs to U2 and then displaces U1 in binding to the 5′ splice site. Within this very complex RNA structure, the 2′ hydroxyl group on the branch point adenosine, which is bulged out of the duplex between U2 and the pre-mRNA, attacks the phosphate group at the junction between the 5′ exon and the intron. In a transesterification reaction, the phosphate backbone is broken at the 5′ splice site. The 5′ exon is released with a 3′ OH group, and the 5′ phosphate of the intron is transferred onto the 2′ position of the ribose on the branch point adenosine, creating the intron lariat structure. U5 retains the 5′ exon and aligns it for a second transesterification reaction, during which the 3′ hydroxyl on the 5′ exon attacks the 3′ splice site, joining the exons and releasing the intron lariat.

Both catalytic steps in splicing are technically termed transesterification reactions, because nucleotides are linked by phosphodiester bonds, and the new bond is made at the same time as the old bond is broken. For this reason, the splicing reactions do not, in principle, require any input of energy. However, the assembly and subsequent disassembly of the spliceosome require numerous ATPases. Most of these belong to a family of proteins that are generally termed RNA helicases. These are believed to use the energy of ATP hydrolysis to catalyze structural rearrangements within the assembling and disassembly spliceosome.

AT-AC Introns

The large majority of human mRNA splice sites have a GU dinucleotide at the 5′ splice site and AG at the 3′ splice site (Fig. 16-4). However, a minor group of introns contain different consensus splicing signals and are termed AT-AC (pronounced “attack”) introns because of the identities of the nucleotides located at the 5′ and 3′ splice sites. The splicing of the AT-AC introns involves a distinct set of snRNAs—U11, U12, U4_ATAC, and U6_ATAC—which replace U1, U2, U4, and U6, respectively. Only U5 is common to both spliceosomes. However, the underlying splicing mechanism is believed to be the same for both classes of intron.

Alternative Splicing

A surprising finding from the human genomic sequencing project was the relatively low number of predicted protein-coding genes, currently estimated at around 30,000. This result has caused increased interest in the phenomenon of alternative splicing, which allows the production of more than one mRNA, and therefore more than one protein product, from a single gene. Several general forms of alternative splicing are commonly found. Exons can be excluded from the mRNAs, or introns can be included. Some genes have arrays of multiple alternative exons, only one of which is included in each mRNA. In addition, the use of alternative splice sites can generate longer or shorter forms of individual exons (Fig. 16-6).

Figure 16-6 alternative splicing can generate multiple different proteins from a single gene. Here are some of the possible mRNA and protein products of a gene whose pre-mRNA is subject to alternative splicing. Left, Examples show the other effects of skipping one or more internal exons, which produces a set of related proteins with different combinations of “modules.” Right, Examples show the effects of alternative splice sites. In the case shown, the use of alternative 3′ splice sites redefines the 5′ end of the downstream exon. This can lead to the inclusion of additional amino acids in the protein product. Use of an alternative splice site can also cause the exon to be read in a different reading frame (green asterisk), changing the amino acid sequence. If the alternative reading frame contains a translation stop codon (red asterisk), a truncated protein will be produced, and the mRNA will generally be targeted for rapid degradation by the NMD pathway (Fig. 16-1).

Current estimates for the proportion of human genes that are subject to alternative splicing range from 30% to 75%. In some cases, this could potentially give rise to a very large number of different protein isoforms. In other cases, alternatively spliced proteins can have antagonistic functions, such as transcription activation versus transcription repression. For the vast majority of human genes, no information is available on the relative activities of different spliced isoforms. Compounding the difficulty in understanding is the fact that many genes show tissue-specific splicing. Thus, a gene could be transcribed in, say, both the liver and brain but generate products with substantially different functions in each tissue. In addition to generating protein diversity, alternative splicing can generate mRNAs with premature translation termination codons—“nonsense” co-dons. These are subject to rapid degradation by the nonsense-mediated decay (NMD) surveillance pathway (see later). Switching splicing into a pathway that generates an NMD target is therefore a means of downregulating gene expression.

It is likely that alterations in the activities of many different factors can lead to the preferential use of alternative splice sites. In at least some cases, changes in the abundance of a general splicing factor generates tissue-specific patterns of splicing. Modulation of the activities of exonic splicing enhancers is also important in regulating alternative splicing.

Localization of Pre-mRNA Splicing

The location of the splicing reaction within the nucleus was long a contentious topic. The snRNAs can be detected dispersed in the nucleoplasm but concentrate in small structures referred to as nuclear speckles or interchromatin granules, as well as in discrete larger structures known as Cajal bodies (see Fig. 14-2). It is now widely accepted that most splicing is performed by the dispersed snRNA population and can occur either cotranscriptionally or immediately following transcript release. Consistent with this, there is evidence that the recruitment of some splicing factors is promoted by association with the CTD of the transcribing polymerase. The speckles are likely to represent sites at which splicing factors are stockpiled ready for use. The Cajal bodies, in contrast, represent sites of maturation in which the snRNAs undergo site-specific nucleotide modification and perhaps assembly with specific proteins.

Editing of mRNAs

The term RNA editing in humans refers to covalent modifications that are made to individual nucleotides, which alter the base-pairing potential. Since the process of translation involves base pairing between mRNA and tRNAs, editing of the mRNA can have the effect of changing the amino acid that is incorporated and therefore the function of the protein. Like alternative splicing, this increases the diversity of protein products that can be synthesized from the genome.

Slightly confusingly, the term editing is also used for quite different mechanisms that insert and delete nucleotides from RNAs in some single-celled eukaryotes. The best-characterized example is in the mitochondria of trypanosomes, which are protozoans that cause major human diseases, including African trypanosomiasis, Chagas’ disease, and leishmaniasis. Uracil residues are added and, less frequently, deleted from the mitochondrial mRNAs at many sites. These changes are specified by a large number of small guide RNAs. This form of editing is not known to occur in higher eukaryotes.

C-to-U Editing

Deamination of cytosine to uracil is performed by an editing complex, sometimes referred to as the editosome, which includes the deaminase Apobec-1 (Fig. 16-7). Only a small number of nuclear-encoded targets have been identified to date, and in these, editing generates translation termination codons, producing shorter forms of the encoded proteins. The best-characterized example of C-to-U RNA editing involves the mRNA encoding intestinal apolipoprotein B (ApoB), where CAA-to-UAA editing in the loop of a specific stem-loop structure generates a stop codon. The truncated protein, ApoB48, has an important role in lipoprotein metabolism. In other cases editing may generate mRNAs that are targets for NMD (see later), leading to down-regulation of protein expression.

Figure 16-7 rna editing changes nucleotide base pairing. The coding potential of an mRNA can be altered by deamination. In C-to-U editing, the amino group at position 4 of the cytosine base is replaced with a carbonyl group, creating uracil. In A-to-I editing, replacement of the amino group at position 2 of adenosine creates inosine, which base-pairs with C residues rather than with U.

A-to-I Editing

The enzyme ADAR (adenosine deaminase acting on RNA) can convert adenine residues to inosine by deamination of the base (Fig. 16-7). Inosine acts like guanosine and base-pairs with cytosine rather than uracil, potentially altering the protein encoded by the mRNA. Most of the transcripts that are edited by ADAR encode receptors of the central nervous system, and RNA editing is required to create the full receptor repertoire. The amino acid substitutions that result from editing of the mRNAs can greatly alter the properties of ion channels, and aberrant editing occurs in various disorders ranging from epilepsy to malignant brain gliomas. ADAR binds as a dimer to imperfect double-stranded RNA duplexes, which are formed between the target site and sequences in a flanking intron. Editing is generally not 100% efficient, so heterogeneous populations of proteins are generated.

Cytoplasmic Polyadenylation

The early steps of embryogenesis in metazoans occur before transcription of the genome commences. All the mRNAs that are present in early embryos were therefore inherited from the mother. These “maternal messages” are frequently translationally inactive, at least in part because they lack a poly(A) tail. They can be activated for translation by polyadenylation in the cytoplasm. Cytoplasmic polyadenylation events are critical for many developmental decisions in oocytes and embryos. In addition, regulated cytoplasmic polyadenylation at synapses controls local translation in neuronal cells. This involves a family of cytoplasmic polymerases that are distinct, and their association with substrates and activity are both regulated by specific RNA-binding proteins.

mRNA Degradation and Surveillance

The Exosome Complex

The exosome is a protein complex composed of multiple different 3′ to 5′ exonucleases. Nuclear and cytoplasmic forms of the complex share 10 common components, some of which have proven or predicted exonuclease activity. The nuclear complex is associated with an additional 3′ to 5′ exonuclease (Rrp6 in yeast, PM-Scl100 in humans), whereas the cytoplasmic complex is associated with a GTPase (Ski7) that is homologous to translation factors. The nuclear exosome participates in RNA maturation, notably in the processing of the 5.8S rRNA, but its major functions are probably in the surveillance and degradation of nuclear RNAs. The cytoplasmic exosome functions in several different mRNA turnover pathways.

Degradation of mRNA

Most analyses of the regulation of gene expression have concentrated on changes in the levels of mRNA transcription. However, the rate at which mRNAs are de-graded is also important, influencing both the total amount of protein synthesized and the timing of protein synthesis following a transcription event. mRNAs are frequently described as having half-lives, but this is generally quite misleading. Degradation is not stochastic, and it is probably better to think of mRNA lifetimes. There are enormous variations in the lifetimes of different human mRNAs—from a very few minutes to many days—which have a large impact on protein expression levels.

Different pathways of mRNA degradation can be classified as (1) the default pathway (i.e., when we do not yet know of any specific activator or repressor of degradation), (2) regulated degradation pathways that respond to developmental or other signals, and (3) surveillance pathways that identify and rapidly degrade aberrant mRNAs or pre-mRNAs. A theme emerging from studies of all mRNA decay pathways is that RNA-binding proteins, which associate with the newly transcribed precursor in the nucleus, can be retained when the mRNA is exported to the cytoplasm. These proteins maintain a record of the nuclear history of the RNA that can be “read” by the cytoplasmic degradation machinery, and this plays a key role in determining the cytoplasmic fate of the mRNA.

A key step in the timing of degradation of most mRNA is the slow, stepwise removal of the poly(A) tail by enzymes called deadenylases. The intact poly(A) tail is bound by multiple copies of the poly(A)-binding protein (PABP), at a stoichiometry of around one molecule per 10 to 20 A residues. Surprisingly, PABP antagonizes 5′ cap removal, probably via interactions with the translation initiation factor eIF4G, which in turn stabilizes the cap-binding protein eIF4E. These interactions effectively circularize the mRNA and strongly stimulate translation initiation (see Fig. 17-9). When the tail becomes too short for the last PABP molecule to bind, these interactions are lost. The cap can then be removed by a decapping complex, which cleaves the triphosphate linkage to the body of the mRNA, releasing m⁷GDP. Cap removal allows rapid 5′ to 3′ degradation of the mRNA by the 5′ exonuclease Xrn1. In addition, loss of the PABP/poly(A) complex allows 3′-degradation of the mRNA by the cytoplasmic exosome.

ARE-Mediated Degradation

The degradation of many mRNA species in human cells is triggered by the presence of sequence motifs referred to as A+U-rich elements (AREs) (Fig. 16-1C). These are generally located in the 3′ UTR of the mRNA, where bound proteins will not be displaced by the translating ribosomes. This pathway plays an important regulatory role in gene expression, as it targets for rapid turnover mRNAs that encode proteins such as cytokines, growth factors, oncogenes, and cell-cycle regulators, for which limited and transient expression is important. Computational analyses indicate that up to 8% of human mRNAs carry AREs, and there is evidence that alterations in the activity of this pathway are associated with both developmental decisions and cancer. ARE-binding proteins associate with the nuclear pre-mRNAs and are exported to the cytoplasm, where they can either activate or repress ARE-mediated decay. Some ARE-binding proteins that activate degradation function by directly recruiting the exosome complex to degrade the mRNA from the 3′ end.

Surveillance of mRNAs

Nonsense-Mediated Decay

The surveillance of mRNA integrity is important because defective molecules can encode truncated proteins, which are frequently toxic to the cell. The presence of a premature translation termination signal (or nonsense codon) strongly destabilizes mRNA via the nonsense-mediated decay (NMD) pathway (Fig. 16-1A). In human cells, termination codons are identified as being located in a premature position by reference to the sites of pre-mRNA splicing. Normal termination codons are within, or very close to, the 3′ exon, so no former splice sites lie far downstream. If any former splice site is located more than about 50 nucleotides downstream of the site of translation termination, the mRNA is targeted for degradation. The sites of former splicing events can be identified in the spliced mRNA product, because the spliceosome deposits a specific protein complex on the mRNA during the splicing reaction (Fig. 16-1). This is called the exon-junction complex (EJC), and it binds to the 5′ exon sequence ˜24 nucleotides upstream of the splice site. Several of the EJC components remain associated with the mRNA following its export to the cytoplasm. In normal mRNAs, the EJCs will all be displaced by the first translating ribosome, so if one (or more) remains on the mRNA, then translation has terminated too soon.

The identification of premature termination codons in yeast and Drosophila does not rely on cues provided by splice sites but probably involves recognition of other nuclear RNA-binding proteins that are retained on the cytoplasmic mRNAs. In all organisms tested, NMD also requires a surveillance complex, which bridges interactions between the terminating ribosome and the “place markers” on the mRNAs.

In yeast and probably in humans, recognition of an mRNA as prematurely terminated activates both 5′ and 3′ degradation. The mRNA can be decapped and 5′-degraded by Xrn1 without prior deadenylation or can be rapidly deadenylated and 3′-degraded by the exosome. In contrast, the degradation of mRNAs targeted by the NMD pathway in Drosophila is initiated by an endonuclease cleavage.

Nonstop Decay

Some mRNAs lack any translation termination codon, because they have been inappropriately polyadenylated, inaccurately spliced, or partially 3′-degraded. Translating ribosomes efficiently stall at the ends of such nonstop mRNAs, and this inhibits the repeated synthesis of truncated proteins (Fig. 16-1D). The cytoplasmic form of the exosome complex is associated with Ski7p, which is homologous to the GTPases that function in translation. The interaction of Ski7p with the stalled ribosome is believed to both release the ribosome and target the mRNA for rapid degradation.

Nuclear RNA Degradation

Analyses of RNA degradation have focused largely on cytoplasmic mRNA turnover, but most RNA synthesized in a eukaryotic cell is actually degraded within the nucleus. Pre-mRNAs are predominantly composed of intronic sequences, and almost all stable RNAs are synthesized as larger precursors that undergo nuclear maturation. In contrast to the role of poly(A) tails in stabilizing mRNAs in the cytoplasm, there is evidence that poly(A) tails can act as destabilizing features during RNA degradation in the nucleus. In yeast, complexes that include nuclear poly(A) polymerases activate the exosome complex during surveillance and degradation of many defective nuclear RNAs, including tRNAs and pre-rRNAs. In Bacteria such as Escherichia coli, poly(A) tails are added to RNAs to make them better substrates for degradation. This has led to the proposal that the original function of polyadenylation was in RNA degradation, and this role is maintained in the eukaryotic nucleus. Following the appearance of the nuclear envelope in early eukaryotes, poly(A) tails took on a distinctly different function in promoting mRNA stability and translation in the cytoplasm.

Synthesis of Stable RNAs

Transfer RNA Synthesis

All tRNAs are processed from precursors (pre-tRNAs) that are extended at their 5′ and 3′ ends (Fig. 16-8). Some pre-tRNAs are polycistronic, with two or more tRNAs excised from the same precursor. In yeast, at least, the genes that encode tRNAs cluster around the surface of the nucleolus, and pre-tRNA processing appears to occur largely within the nucleolus.

Figure 16-8 Mature trnas are generated by processing. A, Transcription by RNA polymerase III generates a pre-tRNA that is 5′ and 3′ extended and may also contain an intron. Cleavage by RNase P generates the mature 5′ end. B, The 3′ end is cleaved by an unidentified nuclease, and the sequence CCA is added by a specific RNA polymerase. This sequence forms a single stranded 3′ end on all tRNAs. C, If an intron is present, it is removed in a splicing reaction that is distinct from pre-mRNA and does not involve small RNA cofactors. The anticodon (green) is generally located 1 nucleotide away from the splice site.

The 5′ end of the mature tRNA is generated by cleavage by the ribozyme endonuclease RNase P, which recognizes structural elements that are common to all tRNAs. The 3′ ends of all mature tRNAs have the sequence Cp-Cp-A_OH, to which the aminoacyl group is covalently attached. However, this CCA sequence is not encoded by the tRNA gene in eukaryotes, although it is encoded by tRNA genes in many Bacteria. Instead, the pre-tRNA is initially trimmed, and the CCA sequence is then added by a specific RNA polymerase that belongs to the same family as the poly(A) polymerases that add tails to mRNAs.

Many pre-tRNAs contain a single, short intron, which is removed by splicing. The enzymology of tRNA splicing is quite different from that of pre-mRNA splicing. The pre-tRNA is cleaved at the 5′ and 3′ splice sites by a tetrameric protein complex containing two endonucleases and two targeting factors. The cleavages leave products with 5′ hydroxyl residues and 2′–3′ cyclic phosphate. A separate tRNA ligase then recognizes these termini and rejoins the exons.

In addition, tRNAs are subject to a bewildering array of covalent nucleotide modifications. Almost 100 different modified nucleotides have been identified in tRNAs, ranging from simple methylation to the addition of very elaborate molecules. All are added without breaking the phosphate backbone of the RNA. The structures of all mature tRNAs are very similar, since each must fit exactly into the A, P, and E sites of the translating ribosome (see Fig. 17-7). It is likely that the modifications help the tRNAs fold into precisely the correct shape. They also aid accurate recognition of different tRNAs by the aminoacyl-tRNA synthases, which are responsible for charging each species of tRNA with the correct amino acid.

Ribosome Synthesis

The synthesis of ribosomes is a major activity of any actively growing cell. Three of the four rRNAs—the 18S, 5.8S, and 25S/28S rRNAs—are cotranscribed by RNA polymerase I as a polycistronic transcript. This pre-rRNA is the only RNA synthesized by RNA polymerase I (RNA pol I) and is transcribed from arrays of the ribosomal DNA (rDNA) repeated in tandem. In humans, approximately 300 to 400 rDNA repeats are present in five clusters (on chromosomes 13, 14, 15, 21, and 22). These sites often are referred to as nucleolar organizer regions, reflecting the fact that nucleoli assemble at these locations in newly formed interphase nuclei. The pre-rRNAs are very actively transcribed and can be visualized as “Christmas trees” in electron micrographs taken following spreading of the chromatin using low-salt conditions and detergent (Fig. 16-9A). The 5S rRNA is independently transcribed by RNA polymerase III. In the majority of eukaryotes, the 5S rRNA genes are present in separate repeat arrays.

Figure 16-9 ribosome synthesis. A, “Christmas trees” of nascent pre-rRNA transcripts. This electron micrograph shows rDNA genes in the process of transcription. Note the numerous molecules of RNA polymerase I along the rDNA, each associated with a pre-rRNA transcript. In the enlarged inset, the terminal balls can be seen on the transcripts. These large pre-rRNA-processing complexes (small subunit processomes) assemble around the binding site for the U3 snoRNA and are required for the early pre-rRNA processing steps. B–C, Roles of the modification guide snoRNAs. The pre-rRNAs undergo extensive covalent modification. Most modification involves methylation of the sugar 2′ hydroxyl group (2′-O-methylation) or pseudouridine ψ formation, at sites that are selected by base pairing with a host of small nucleolar ribonucleoprotein (snoRNP) particles. Human cells contain well over 100 different species of snoRNPs, and each pre-rRNA molecule must transiently associate with every snoRNP. Sites of 2′-O-methylation are selected by base pairing with the box C/D class of snoRNAs, which carry the methyltransferase Nop1/fibrillarin. Sites of pseudouridine formation are selected by base pairing with the box H/ACA class of snoRNAs, which carry the pseudouridine synthase Cbf5/dsykerin. D, Key steps in eukaryotic ribosome synthesis. Following transcription of the pre-rRNAs, most steps in eukaryotic ribosome synthesis take place within the nucleolus. The preribosomes are then released from association with nucleolus structures and are believed to diffuse to the nuclear pore complex (NPC). Passage through the NPC is preceded by structural rearrangements and the release of processing and assembly factors. Further ribosome synthesis factors are released during late structural rearrangements in the cytoplasm that convert the preribosomal particles to the mature ribosomal subunits. During pre-rRNA transcription and processing, many of the approximately 80 ribosomal proteins assemble onto the mature rRNA regions of the pre-RNA. E, The pre-rRNA processing pathway. The pathway is presented for the budding yeast Saccharomyces cerevisiae, but extensive conservation is expected throughout eukaryotes. The mature rRNAs are generated by sequential endonuclease cleavage, with some of the mature rRNA termini generated by exonuclease digestion. Scissors with question marks indicate that the endonuclease responsible is unknown.

The Nucleolus

Most steps in ribosome synthesis take place within a specialized nuclear substructure, the nucleolus (see Fig. 14-3). In micrographs, the nucleolus appears to be a very large and stable structure, but kinetic experiments indicate that it is in fact highly dynamic, with most nucleolar proteins rapidly exchanging with nucleoplasmic pools. There is little evidence that signals for the localization of proteins or mature snoRNAs to the nucleolus are distinct from the features that allow them to function there. A current view of the nucleolus is that its assembly is the consequence of many relatively weak and transient interactions between the nucleolar proteins. The result is a self-assembly process that greatly increases the local concentration of ribosome synthesis factors. This is envisaged to promote efficient preribosome assembly and maturation while allowing the rapid and dynamic changes in preribosome composition in-volved in this pathway. Similar mechanisms may generate other subnuclear structures such as Cajal bodies.

The key steps in ribosome synthesis are (1) transcription of the pre-rRNA, (2) covalent modification of the mature rRNA regions of the pre-rRNA, (3) processing of the pre-rRNA to the mature rRNAs, and (4) assembly of the rRNAs with the ribosomal proteins (Fig. 16-9D). During ribosome synthesis, the maturing preribosomes move from their site of transcription in the dense fibrillar component of the nucleolus, through the granular component of the nucleolus. They are then released into the nucleoplasm prior to transport through the nuclear pores to the cytoplasm. Here, the final maturation into functional 40S and 60S ribosomal subunits takes place.

Pre-rRNA Processing

The posttranscriptional steps in ribosome synthesis are very complex, involving approximately 200 proteins and approximately 100 snoRNA species, in addition to the four rRNAs and approximately 80 ribosomal proteins. Ribosome synthesis is best understood in budding yeast, but all available evidence indicates that it is highly conserved throughout eukaryotes. Many pre-rRNA processing enzymes have been identified, although others remain to be found (Fig. 16-9E). A combination of endonuclease cleavages and exonuclease digestion steps generates the mature rRNAs in a complex, multistep processing pathway. The remaining species, 5S rRNA, is independently transcribed and undergoes only 3′-trimming. Notably, all of the nucleases indicated in Figure 16-9E process other RNAs in addition to the pre-rRNAs. It seems very probable that when the enzymes responsible for the remaining processing activities are identified, they too will be found to process other substrates.

Modification of the Pre-rRNA

The rRNAs are subject to covalent nucleotide modification at many sites. Modification takes place on the pre-rRNA, either on the nascent transcript or shortly following transcript release. The majority of modifications are methylation of the 2′-hydroxyl group on the sugar ring (2′-O-methylation) and conversion of uracil to pseudouridine by base rotation. The sites of these modifications are selected by base pairing with two groups of small nucleolar RNAs (snoRNAs). The box C/D snoRNAs direct sites of 2′-O-methylation and carry the methyltransferase (called fibrillarin in humans and Nop1 in yeast) (Fig. 16-9B). The box H/ACA snoRNAs select sites of pseudouridine formation and carry the pseudouridine synthase (called dyskerin in humans and Cbf5 in yeast [Fig. 16-9C]).

A small number of snoRNAs do not direct RNA modification but are required for pre-rRNA processing. The best characterized is the U3 snoRNA, which binds cotranscriptionally to the 5′–external transcription factor (ETS) region of the pre-rRNA. Base pairing between U3 and the pre-rRNA is required for the early processing reactions on the pathway of 18S rRNA synthesis and directs the assembly of a large pre-rRNA processing complex called the small subunit processome. This complex can be visualized as a “terminal knob” in micrographs of spread pre-rRNA transcripts (Fig. 16-9A).

A subset of ribosome synthesis factors interacts with both the rDNA and RNA polymerase I. These interactions might promote both efficient pre-rRNA transcription and recognition of the nascent pre-rRNA. This is reminiscent of the association of mRNA processing factors with RNA polymerase II and suggests that maturation of different classes of RNA and their assembly with specific proteins might be functionally coupled to transcription.

Small Nuclear RNA Maturation

The U1, U2, U4, and U5 snRNAs are encoded by individual genes transcribed by RNA polymerase II (Fig. 16-10C). Like mRNAs, the snRNA precursors undergo cotranscriptional capping with 7-methylguanosine, but they are not polyadenylated. In human cells, the newly synthesized precursors to these snRNAs are then exported to the cytoplasm. Once in the cytoplasm, the snRNAs form complexes with the Sm-proteins. This set of seven different, but closely related, proteins assembles into a heptameric ring structure. Sm-proteins are named after the human autoimmune serum that was initially used in their identification. On their own, the Sm-proteins show low substrate specificity in RNA binding. However, in human cells, the assembly of the snRNAs with the Sm-proteins is highly specific and is mediated by a large protein complex. This complex includes the SMN protein (survival of motor neurons), which is the target of mutations in the relatively common genetic disease spinal muscular atrophy. While in the cytoplasm the snRNAs are further processed; the 3′ end of the RNA is trimmed, and the cap structure undergoes additional methylation to generate 2,2,7-trimethylguanosine. This hypermethylated cap structure is also present on small nucleolar RNAs (see later) and might be important to allow resident nuclear RNAs to be distinguished from mRNA precursors.

Figure 16-10 different patterns of stable rna synthesis by rna polymerase ii. A, Primary transcripts encoding mRNAs generally contain one or more introns, which are removed and degraded to produce the mature mRNA. B, In human cells, the snoRNAs that are involved in rRNA modification are generally synthesized by excision from the introns of highly transcribed protein-coding genes. The snoRNP proteins bind to the snoRNA sequence within the pre-mRNA and protect it from degradation. C, The spliceosomal U1, U2, U4, and U5 snRNAs are transcribed by RNA polymerase II and, like mRNAs, are capped with 7-methylguanosine and are bound by the nuclear cap-binding complex (CBC). The pre-snRNA is exported to the cytoplasm, where it associates with the Sm-protein complex and is 3′ trimmed. The cap is then hypermethylated to 2,2,7-trimethylguanosine, and the RNA-protein complex is reimported into the nucleus. The newly imported snRNPs localize to the Cajal bodies, where the snRNA is covalently modified at sites selected by base pairing to the small Cajal RNAs (scaRNAs), a class of modification guide RNAs. Assembly with specific proteins then generates the mature snRNPs. D, Some snoRNAs, including U3, are individually transcribed by RNA polymerase II. Like the snRNAs, they are initially capped by with 7-methylguanosine and bind CBC. Following association with a set of snoRNA-specific proteins, they undergo cap-trimethylation and 3′ trimming. The snoRNPs then localize to the nucleolus, where they themselves undergo snoRNP-dependent modification and then participate in rRNA processing.

Once the cap is trimethylated and bound by the Sm-proteins, the snRNAs can be reimported into the nucleus, where they initially localize to discrete subnuclear structures termed Cajal bodies (see Fig. 14-2). Within the Cajal bodies, specific nucleotides in the snRNAs are modified by 2′-O-methylation and pseudouridine formation. The sites of these modifications are selected by base pairing with a group of resident small Cajal body RNAs (scaRNAs), which carry the RNA-modifying enzymes. The scaRNAs closely resemble the snoRNAs except that single scaRNAs can frequently direct both 2′-O-methylation and pseudouridine for-mation.

Maturation of U6 snRNA is quite different from that of the other snRNAs. U6 is transcribed by RNA polymerase III and is not exported to the cytoplasm. Mature U6 retains the 5′ triphosphate and 3′ poly(U) tract that are characteristic of primary transcripts made by RNA pol III (see Chapter 15). However, the 5′ triphosphate is methylated on the g-phosphate (i.e., the position furthest from the nucleotide), while the terminal U of the poly(U) tract carries a 2′ to 3′ cyclic phosphate. Both of these modifications may help to protect the RNA against degradation. U6 does not bind the Sm-proteins but instead associates with a related heptameric ring structure that is comprised of seven Lsm-proteins (“like sm”). Two distinct but related heptameric Lsm-complexes are present in the nucleus and cytoplasm. The nuclear Lsm2-8 complex binds to the U6 snRNA and also participates in the decapping of mRNA precursors that are destined for degradation in the nucleus (Fig. 16-1). In contrast, the Lsm1-7 complex participates in mRNA decapping and 5′ degradation in the cytoplasm. Nucleotides within the U6 snRNA are also modified at positions that are selected by guide RNAs, but this modification occurs in the nucleolus rather than the Cajal body.

Small Nucleolar RNA Maturation

The small nucleolar RNAs (snoRNAs) are all transcribed by RNA polymerase II (except in some plants in which pol III-transcribed snoRNAs can be found). However, the genes encoding snoRNAs can have a surprising variety of different organizations. In human cells, most snoRNAs are excised from the introns of genes that also encode proteins in their exons (Fig. 16-10B). The introns that encode snoRNAs are released by splicing and then linearized by debranching. The mature snoRNA is then generated by controlled exonuclease digestion. In contrast, most characterized snoRNAs in higher plants and several yeast snoRNAs are processed from polycistronic precursors that encode multiple snoRNA species. Individual pre-snoRNAs are liberated by cleavage of the precursor by the double–strand-specific endonuclease RNase III (Rnt1 in yeast) and then trimmed at both the 5′ and 3′ ends. SnoRNAs can also be processed from single transcripts, and these have many features in common with snRNA transcripts. Like snRNAs, these individually transcribed snoRNAs carry trimethylguanosine cap structures (Fig. 16-10D). However, unlike snRNAs, which have a cytoplasmic phase, the maturation of snoRNAs and assembly of snoRNPs take place entirely within the nucleus, most steps probably occurring in the nucleolus.

Synthesis and Function of miRNAs

The terms sRNAs and miRNAs are used to describe recently identified groups of RNAs that are physically similar but have distinct functions and a variety of different names. All are around 22 nucleotides in length and associate with a protein complex called the RNA-induced silencing complex (RISC). Under different circumstances, sRNAs can lead to cleavage of target RNAs, repress translation of mRNAs, or inhibit transcription of target genes via formation of heterochromatin. It seems likely that miRNAs play major roles in regulating global patterns of gene expression in human cells.

Endogenous micro-RNAs (miRNAs) are encoded in the genomes of many eukaryotes, including humans (Fig. 16-11). These are frequently transcribed as polycistronic precursors called pri-miRNAs. Within the pre-miRNA, the precursors to the individual miRNAs (pre-miRNAs) form stem-loop structures. The stems are first cleaved by a nuclear double-strand-specific endonuclease called Drosha, releasing the individual pre-miRNAs. These are then exported to the cytoplasm, where cleavage by a second double-strand-specific endonuclease, Dicer, releases the miRNA in the form of a duplex with characteristic 2-nucleotide 3′ overhangs and 5′ phosphate groups. These duplexes are incorporated into the RISC complex, where one of the strands becomes the functional miRNA. If the target mRNA sequence is incompletely complementary to the miRNA, its translation is repressed (Fig. 16-11). This is likely to be the normal function of most endogenous miRNAs. It has recently been estimated that 30% or more of human mRNAs are targets of miRNA regulation. miRNAs show tissue-specific patterns of expression and dynamic changes in expression during differentiation. Individual miRNAs can modulate the expression of many different mRNAs.

Figure 16-11 mrna maturation. The polycistronic miRNA precursors (termed primary-miRNAs, or pri-miRNAs) are cleaved by the double-strand-specific endonuclease Drosha within the nucleus. The individual pre-miRNAs are then exported to the cytoplasm by the export factor Exportin 5 in complex with Ran-GTP (see Fig. 14-17). Once in the cytoplasm, the pre-miRNAs are cleaved by the double-strand-specific endonuclease Dicer. One strand of the resulting duplex is then incorporated into the RNA-induced silencing complex (RISC) and becomes the functional miRNA. Imperfect duplexes are formed between the miRNA and target mRNAs; this results in the inhibition of the mRNA translation.

If a target RNA sequence is found that is perfectly complementary to the miRNA, it is cleaved by a component of the RISC complex, Ago2 (“Slicer”). Target RNA cleavage occurs within the miRNA: mRNA duplex at a fixed distance (between nucleotides 10 and 11) from the 5′ end of the miRNA, which is specifically bound and used to precisely position the duplex relative to the catalytic site.

This pathway can be exploited in techniques for the specific inactivation of target mRNAs, termed RNA interference (RNAi [Fig. 16-12]). The technique uses exogenously provided RNAs that are generally fully complementary to the target, typically provided as 22-nucleotide RNAs termed small interfering RNAs (siRNAs). In many organisms (e.g., in Drosophila or the nematode Caenorhabditis elegans), RNAi can be performed by introducing long double-stranded RNAs. These are cleaved in vivo by Dicer into 22-base-pair fragments, which are then incorporated into the RISC complex. In mammals, including human cells, long double-stranded RNAs cannot be used for RNAi, as they trigger an antiviral response and cell death. RNAi can, however, be performed in human cells by the introduction of precleaved 22-bp RNA fragments. Alternatively, small hairpin structures can be expressed that resemble endogenous pre-miRNAs and are processed into functional 22-nucleotide siRNAs in vivo. The small size, ease of use, and potent function of siRNAs have made RNAi the method of choice for many analyses of eukaryotic gene function.

Figure 16-12 sirna function in mrna cleavage. In contrast to the endogenous miRNAs, exogenously added siRNAs are generally perfectly complementary to the target RNA, which is then cleaved by the Ago-2 component of the RISC complex. In many organisms (including the nematode worm C. elegans and insects such as Drosophila), long-double stranded RNAs can be used, which are processed to approximately 22-nucleotide duplexes. In human cells, siRNAs are generally introduced as preformed 22-nucleotide duplexes or as stem-loops with structures that resemble endogenous pre-miRNAs. In either case, the siRNAs associate with Dicer, the double-strand RNA-binding protein TRBP, and Argonaut 2 to form the RISC complex. One strand becomes the functional siRNA, while the “passenger” strand is lost from the complex.

In the nucleus, a closely related system is used to establish transcriptional silencing of RNA synthesis (Fig. 16-13). Although important gaps remain in our understanding, it appears that transcription of a region of the chromosomal DNA on both strands, generating a double-stranded RNA, may be sufficient to induce its silencing. The double-stranded RNA is likely to be cleaved by Dicer and/or Drosha to generate 22-nucleotide fragments, in this case termed small heterochromatic RNAs (shRNAs). These associate with a nuclear complex called RITS (RNA-induced transcriptional silencing [see Fig. 16-13]), which is related to the cytoplasmic RISC complex. These shRNAs identify the corresponding gene, possibly by binding to nascent RNA transcripts and, together with the RITS complex components, recruit a protein methyltransferase. This methylates histone H3 on lysine 9, a hallmark of repressive heterochromatin, which in turn recruits other heterochromatin proteins such as HP1 (see Fig. 13-9). The RITS complex includes an RNA-dependent RNA polymerase, and this may be able to generate new shRNAs, allowing the spreading of the heterochromatin into flanking sequences. The tendency of heterochromatin to spread into the flanking euchromatin has long been recognized and gives rise to the phenomenon of position effect variegation (see Fig. 13-9). In some eukaryotes, the methylated histone H3 can also recruit DNA methyltransferases that modify cytosine residues to 5′-methylcytosine. This reinforces heterochromatin formation and makes it heritable by daughter cells. It is likely that this system is important for the establishment of heterochromatin domains, such as those surrounding the centromeres in higher eukaryotes. It might also function as a defense system against the amplification of transposable elements.

Figure 16-13 shrna function in heterochromatin formation. The targets of miRNAs and siRNAs are cytoplasmic mRNAs. However, sRNAs can also function in the nucleus. Small double-stranded RNAs in the nucleus can associate with the RNA-induced transcriptional silencing (RITS) complex. The sRNA-RITS complex then identifies the genomic site of transcription, possibly by recognition of the nascent transcripts. This leads to the establishment of heterochromatin at this location, via the recruitment of protein methyltransferases that methylate lysine 9 on histone H3, a hallmark of repressive heterochromatin (see Fig. 13-9). In some organisms, this is followed by methylation of the DNA, which makes the repressed heterochromatic state more stable and heritable.

The irony is that it now seems likely that the large-scale organization of the genome in many eukaryotes will involve RNAs that long eluded detection because they are so small.

Ribozymes

Some RNAs have catalytic activity in the absence of proteins. Such RNA enzymes are termed ribozymes. Only nine classes of ribozymes are known, so cells appear to have far fewer ribozymes than protein enzymes, but ribozymes play some key roles.

Group I and Group II Self-splicing Introns

Two classes of introns can catalyze their own excision from precursor RNAs. These ribozymes are referred to as group I and group II self-splicing introns. Both classes of RNA fold into complex structures that catalyze splicing via two-step transesterification pathways (Fig. 16-14).

Figure 16-14 comparison of self-splicing with pre-mrna splicing. Groups I and II introns are catalytic RNAs or ribozymes that are able to excise themselves from precursor RNAs in the absence of proteins. A, The removal of group I introns is mechanistically distinct from nuclear pre-mRNA splicing and commences with the binding of an exogenous guanosine nucleotide (red G) within a pocket created by the intronic RNA structure. This G is used to attack and break the phosphate backbone at the 5′ splice site. Subsequently, the free 3′ end of exon 1 attacks the phosphodiester bond at the 3′ splice site, leading to exon ligation and the release of the linear intron. B, In contrast, the mechanism of splicing group II introns is very similar to pre-mRNA splicing. An adenine residue (A) near the 3′ end of the intron attacks the 5′ splice site, leading to the formation of a lariat intermediate. The subsequent attack of the free 3′ end of exon 1 on the phosphodiester bond at the 3′ splice site leads to exon ligation and the release of the intron lariat (compare to Fig. 16-4). C–D, Parallels can be drawn between structure and mechanism of group II self-splicing introns and pre-mRNA splicing. This suggested the model that group II introns gave rise to the nuclear pre-mRNA splicing system. The snRNAs may be derived from fragments of a group II intron, which developed the ability to function in trans (i.e., on other RNAs) rather than acting only in cis on its own sequence. Specifically, Domain VI of the group II introns functions like the U2-branch point duplex in activating the branch-point adenosine by bulging it out of a helix. Domain V acts like the U2-U6 duplex in bringing this adenosine to the 5′ splice site. Domain III resembles the U5 snRNA in base pairing to both the 5′ and 3′ exons at the splice sites.

The first group I intron was identified in 1981 as a 413-nucleotide fragment that was able excise itself from the pre-rRNA synthesized in the ciliate Tetrahymena. This was a major surprise, since at that time, all known enzymes were proteins. The demonstration that an RNA could function as an enzyme had a major impact on subsequent RNA research. Group I introns are found in the pre-rRNAs of other unicellular eukaryotes, in the mitochondria and chloroplasts of many lower eukaryotes, and in the mitochondria of higher plants.

Group II introns have been found in mitochondria of plants and fungi and in chloroplasts. The splicing mechanism of group II introns strikingly resembles nuclear pre-mRNA splicing (Fig. 16-14C–D). This led to the proposal that the nuclear pre-mRNA splicing system derived from ancestral group II introns. During early eukaryotic evolution, the catalytic center of the group II intron might have become fragmented and separated into the present spliceosomal snRNAs. This would have converted a system that could work only on its own transcript into a system that could process other RNAs, greatly increasing the potential range of spliced RNAs.

RNase P and RNase MRP

Shortly after the identification of the group I intron in Tetrahymena, the RNA component of RNase P was shown also to function as a ribozyme. RNase P is an RNA-protein complex that cleaves pre-tRNAs at the 5′ end of the mature tRNA sequence in all organisms. The bacterial enzyme has one RNA component and one protein, but the RNA can cleave pre-tRNAs in vitro in the absence of the protein. In eukaryotes, RNase P has become more complicated, with one RNA and nine protein components. The eukaryotic RNA has not been shown to be active in the absence of proteins, but it does show structural similarities to the bacterial RNA, and it is assumed to be the catalyst.

Eukaryotes also contain a second RNA-protein enzyme, called RNase MRP, which is closely related to RNase P. The RNA components share common structural features, and the complexes share eight common proteins. RNase MRP cleaves the preribosomal RNA between the small and large subunit rRNAs (Fig. 16-9E). Notably, in many Bacteria, RNase P can cleave the pre-rRNA at a similar position because of the presence of a tRNA within the pre-rRNA transcript. This suggests that RNase MRP arose in an early eukaryote as a specialized form of RNase P, with a specific function in pre-rRNA processing. By analogy to RNase P, cleavage by RNase MRP is predicted to be RNA catalyzed. RNase MRP also functions in mRNA turnover, at least in yeast, initiating the cell-cycle-regulated degradation of a small number of mRNAs.

Large Subunit rRNA

The most important ribozyme is the rRNA component of the large ribosomal subunit, which does not participate in RNA processing but catalyzes peptide bond formation (see Fig. 17-10). During translation elongation, the peptidyl-transferase reaction (the reaction by which amino acid residues are attached to each other to form proteins) is catalyzed by the rRNA itself. The peptidyl-transfer reaction is energetically favorable, and it is currently believed that the catalytic activity derives primarily from the precise spatial positioning of the A-site and P-site tRNAs by the rRNA. The ribosomal proteins act as chaperones in ribosome assembly and as cofactors to increase the efficiency and accuracy of translation.