INTRODUCTION

Top Previous Next References

Posttranscriptional cleavage of mRNA precursor is an essential step in mRNA maturation. Following cleavage, most eukaryotic mRNAs, with the exception of replication-dependent histone transcripts in some organisms, acquire a poly(A) tract at their 3' ends. The process of 3'-end formation promotes transcription termination (101) and transport of the mRNA from the nucleus (215). The poly(A) tail, most probably by providing a binding site for poly(A) binding protein (105), also enhances the translation and stability of mRNA (149, 368, 399, 488).

Defects in mRNA 3'-end formation can profoundly alter cell viability, growth, and development. The essential nature of the yeast genes encoding components of the polyadenylation pathway emphasizes the importance of this process. In metazoan cells, in vivo depletion of one of the cleavage proteins, CstF-64, causes cell cycle arrest and ultimately apoptotic cell death (451). A failure to correctly modify the metazoan poly(A) polymerase during the cell cycle is thought to cause a lower growth rate and cell accumulation in the G₀-G₁ phase (516). The appearance of short GCG repeats in the gene encoding the PAB II polyadenylation factor is associated with oculopharyngeal muscular dystrophy (59). The formation of mRNA 3' ends is a key regulatory step in the expression of many genes, and in some cases aberrant polyadenylation leads to disease. In humans, such defects cause thalassemias (203, 345) and a lysosomal storage disorder (161). Inappropriate polyadenylation may also contribute to the abnormal processing of the EAAT2 glutamate transporter transcripts observed in the brains of patients with sporadic amyotrophic lateral sclerosis (262). In this disease, the loss of functional EAAT2 correlates with motor neuron degeneration. Research into the fundamental mechanism of mRNA 3'-end formation and its regulation should lead to a better understanding of its crucial role in normal cell growth and development.

The past few years have brought astounding progress in our understanding of the biochemistry of mRNA 3'-end formation, its regulation, and its interaction with other aspects of mRNA synthesis. The factors which comprise the basic polyadenylation machinery have been identified, and the coding sequence of many, if not most, of the protein subunits has become available. The molecular mechanism by which several regulatory elements stimulate or inhibit polyadenylation has been dissected in exquisite detail, and the intimate involvement of splicing factors at these sites has been made clear. In addition, much information has accumulated on how the basic polyadenylation machinery is regulated to control the choice of poly(A) site or activity of the poly(A) polymerase. Finally, the coupling of transcription and mRNA 3'-end formation has been convincingly demonstrated in a variety of ways.

We have tried to provide sufficient background information that the reader can evaluate the developments in our understanding of mRNA 3'-end formation, primarily over the last 5 years. Due to space constraints, we are not able to give a more thorough historical account, and so we have focused on a limited number of examples to illustrate the paradigms emerging in the field. We ask the readers to refer to several recent reviews on constitutive and regulated polyadenylation for additional details (101, 139, 238, 473, 475a). Cytoplasmic polyadenylation and the role of the poly(A) tail in translation is covered in a review by Richter elsewhere in this issue (384a).

CLEAVAGE/POLYADENYLATION PATHWAY

Top Previous Next References

RNA Sequences Which Specify Cleavage and Polyadenylation

Mammalian polyadenylation signals. In mammalian cells, three elements define the core polyadenylation signalthe highly conserved hexanucleotide AAUAAA found 10 to 30 nucleotides upstream of the cleavage site, a less highly conserved U-rich or GU-rich element located downstream of the cleavage site, and the cleavage site itself, which becomes the point of poly(A) addition and is thus generally referred to as the poly(A) site (Fig. 1). Additional sequences outside of this core recruit regulatory factors or maintain the core signal in an open and accessible structure.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Schematic representation of poly(A) signals in animals, the yeast S. cerevisiae, and plants. USE, auxiliary upstream enhancer; DSE, downstream element; EE, efficiency element; PE, positioning element; FUE, far-upstream element; NUE, near-upstream element; nt, nucleotides. Adapted from reference 387.

(i) AAUAAA motif. The consensus sequence AAUAAA was initially revealed by a comparison of nucleotide sequences preceding the poly(A) sites in several mRNAs (377) and has since been found in almost all polyadenylated mRNAs of animal cells (473). Extensive mutagenesis studies and the analysis of naturally occurring mutations have conclusively established that this hexanucleotide is essential for both cleavage and poly(A) addition (reviewed in references 286, 472, and 487). The sequence AAUAAA is one of the most highly conserved sequence elements known (374). The most frequent variant is AUUAAA, whose activity is comparable to that of the canonical sequence. Mutations of any other nucleotide strongly inhibit processing (422, 489, 496), although in some genes, these forms are the only hexanucleotide-like sequences upstream of the poly(A) site. Poly(A) sites with these variants, as well as those that have no discernible upstream AAUAAA element, i.e., no sequence differing by less than 2 nucleotides from the consensus, are usually associated with alternative (208, 421, 508) or tissue-specific (76, 476) polyadenylation.

(ii) Downstream elements. The second element of the core polyadenylation signal is within approximately 30 nucleotides downstream of the poly(A) site. This downstream element (DSE) is more diffuse and poorly conserved, and two main types have been described, a U-rich element and a GU-rich element. The U-rich element is a short run of U residues (94, 162). The GU-rich type has the consensus YGUGUUYY (Y = pyrimidine) and has been found downstream of the poly(A) site in two-thirds of the 70 genes surveyed in a 1985 study (302). A polyadenylation signal may have only one DSE (94, 300) or may have both a U-rich element and a GU-rich element working together synergistically (162). However, some DSEs contain no matches to either the U- or GU-rich motif (431). Point mutations or small deletions in the DSE have only weak effects, and larger deletions are required to abolish function, which is in agreement with the idea that the DSE is poorly defined and possibly redundant (510). Nevertheless, the proximity of the DSE to the poly(A) site can affect the cleavage site position (280, 294) and the efficiency of cleavage (162, 300).

(iii) Poly(A) site. The selection of the cleavage site is determined mainly by the distance between the upstream AAUAAA sequence and the DSE(s) (80). The local sequence surrounding the cleavage site is not conserved, although adenosine is found at the cleavage site of 70% of vertebrate mRNAs (422). Thus, the first nucleotide of the poly(A) tail in most mRNAs is probably template encoded, although this has been proven experimentally in vitro for only two poly(A) sites (317, 422). A study involving saturation mutagenesis demonstrated that the order of preference for the cleavage site nucleotide follows the order of A > U > C  G (80). The pentultimate nucleotide is most often a C residue (in 59% of all genes analyzed) (422). Thus, a CA dinucleotide defines the poly(A) site for most genes.

(iv) Auxiliary sequences. Other sequence elements can modulate the efficiency of 3' processing in a positive or negative fashion. The molecular mechanisms by which some of these operate are described in later sections, and only the auxiliary sequences are discussed here. One class of enhancer sequence is located upstream of the AAUAAA element (USE) (Fig. 1), and has been found primarily in viral poly(A) sites, such as adenovirus E3 (43), adenovirus L1 (126), adenovirus L3 (372, 373), adenovirus L4 (431), Epstein-Barr virus DNA polymerase (427), simian virus 40 (SV40) late (74, 277, 410), ground squirrel hepatitis virus (392), and retroviruses such as human immunodeficiency virus type 1 (HIV-1) (65, 87, 127, 128, 466). These USEs are often U rich, but a consensus sequence has not emerged (Table 1).

Yeast polyadenylation signals. Signals which direct mRNA 3'-end formation in the yeast Saccharomyces cerevisiae are somewhat different from those used in higher eukaryotes in both sequence and organization. Yeast polyadenylation signals are less highly conserved than are poly(A) signals in higher eukaryotes and are unexpectedly complicated. At least three elements are needed to make up a minimal yeast mRNA 3'-end region: (i) the UA-rich efficiency element and related sequences, functioning to activate the positioning element; (ii) the A-rich positioning element, which directs the position of the cleavage site; and (iii) the actual site of polyadenylation, PyA_n (Fig. 1) (reviewed in reference 182). A recent analysis of 1,352 pre-mRNA 3'-end processing sites, corresponding to 861 different genes, has confirmed this organization of signal sequences (170).

(i) Efficiency element. Efficiency elements are found at a variable distance upstream of the cleavage site and often contain alternating UA dinucleotides or U-rich stretches. Early comparisons of a number of yeast mRNA 3' ends led to the proposal of a bipartite motif UAG ··· UAUGUA (395). Another sequence, UUUUUAUA, was also identified as an efficiency element in several yeast genes including GCN4, PHO5, and ADH1 (141, 181, 198, 199, 224). A subsequent study reduced these putative signals to a hexanucleotide, UAYRUA, with the sequence UAUAUA working best for mRNA 3'-end formation (223). The U residues at the first and fifth positions are the most critical nucleotides in this sequence. Furthermore, computer analysis has shown that more than half of the approximately 1,000 yeast nuclear genes examined contain UAUAUA sequences in their 3' region (181). Thus, most yeast genes, e.g., GAL7 and MRP2 (1, 356), use UAUAUA as the efficiency element, while other genes, e.g., CYC1 and GCN4, appear to use related sequences, such as UAUUUA, UAUGUA, and UUUUUAUA (141, 181).

(ii) Positioning element. The second element, the positioning element, directs cleavage to a position approximately 20 nucleotides downstream of this sequence. Deletion or mutation of the sequence UUAAGAAC in the 3' region of CYC1 changes the location of the poly(A) site but not the overall mRNA processing level, indicating that it represents a signal distinct from the efficiency element (395). In addition to the sequence UUAAGAAC, several other A-rich sequences have been characterized as positioning elements, with AAUAAA and AAAAAA being the most efficient (183). Related motifs are also functional, except that sequences such as GAUAAA, GAAGAA, and GAAUAA are either inefficient or completely inactive, suggesting that a guanosine residue at the first position has an inhibitory effect on the signal function.

(iii) Poly(A) site. Mapping of the poly(A) sites of several yeast genes has shown that polyadenylation occurs most frequently at a Py(A)_n sequence (Py = pyrimidine) (36, 195). In contrast to animal genes, in which a single poly(A) site is found downstream of AAUAAA, many yeast genes use a cluster of poly(A) sites downstream of the efficiency and positioning elements. When positioning elements are mutated, the poly(A) sites become scattered over a much broader region (17, 194, 396). A recent survey indicates that T-rich motifs are frequently found immediately before and after the poly(A) site, especially in genes with suboptimal efficiency or positioning elements (170). The importance of these flanking sequences has not been tested experimentally.

(iv) Additional properties of yeast polyadenylation signals. Some yeast polyadenylation signals function in both orientations (17, 224, 356). This is in part due to the convergent transcription of closely packed genes. When this is not the case, it is likely that the sequence variability of the yeast signal motifs, the general TA richness of the 3' ends of yeast genes (170), and the primary dependence on the UA-rich element, and not the positioning element, for efficient processing will increase the probability that an adequate signal is fortuitously found in the reverse orientation.

Plant polyadenylation signals. The process of mRNA 3'-end formation in plants is poorly understood, but some information on important cis-acting elements is available (reviewed in references 220 and 387). At least three signals are required: the near-upstream element (NUE), the far-upstream element (FUE), and the cleavage site itself (Fig. 1). NUE is located about 10 to 30 nucleotides upstream of the cleavage site and presents in variant forms, from AAUAAA-like motifs to other related or unrelated sequences. FUE is usually U rich and is found approximately 100 nucleotides upstream of the cleavage site. Similar to other organisms, cleavage often occurs at a PyA dinucleotide. There are multiple cleavage sites in many genes, and use of a particular site is determined predominantly by the position of the NUE.

Comparison of polyadenylation signals in mammals, yeast, and plants. A tripartite signal, composed of an A-rich sequence, a U-rich element, and a PyA cleavage site forms a common minimal polyadenylation signal in all eukaryotes (Fig. 1). In mammals, the hexanucleotide AAUAAA is highly conserved, present in a single copy, and absolutely necessary for 3'-end processing. In yeast, the A-rich motif sometimes serves only to position the poly(A) site and is often duplicated. The second set of sequence elements are U rich or UA rich and work in conjunction with the A-rich sequence. In mammals, these are most often present in single copy downstream of the cleavage site and are essential. When located upstream of the A-rich signal, they are stimulatory. In yeast and plants, this type of signal is often redundant and is usually found upstream of the cleavage site and most often upstream of the A-rich sequence as well. Cleavage occurs preferentially at CA in mammals and PyA in yeast and plants. No downstream signals have been clearly identified in yeast or plants. For the most part, the polyadenylation signal appears to be recognized as a one-dimensional string of nucleotides.

Multiple protein factors are required for the formation of mRNA 3' ends and are generally conserved between yeast and mammals. The in vitro processing assays developed in mammalian cells by Moore and Sharp (315) and in yeast cells by Butler and Platt (68) have provided a successful approach for isolating these activities via fractionation of cell extracts and cloning the relevant genes and cDNAs from protein sequence. In mammals, cleavage/polyadenylation specificity factor (CPSF), cleavage-stimulatory factor (CstF), cleavage factors I_m and II_m (CF I_m and CF II_m), RNA polymerase II (pol II) and poly(A) polymerase (PAP) are involved in the cleavage step and CPSF, PAP, and poly(A)-binding protein II (PAB II) are involved in polyadenylation (Fig. 2). In yeast, cleavage requires CF IA CF IB, and CF II and polyadenylation uses CF IA, CF IB, polyadenylation factor I (PF I), Pab1, and Pap1 (Fig. 2). Yeast genetics and application of the two-hybrid screen in yeast cells has led to the discovery of some of the important yeast genes, and the recent completion of the S. cerevisiae genomic sequence has allowed the quick identification of several yeast proteins as homologues of mammalian CPSF subunits. In this section, we describe what is known about these factors (summarized in Tables 2 and 3 for mammal and yeast cell factors, respectively). At the end, we present a model for the complexes involved in each step of the reaction.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   mRNA precursors are processed at their 3' ends in a two-step reaction. A primary transcript is cleaved endonucleolytically at the poly(A) site, and this is followed by the addition of adenylate residues to the 3' end of the upstream fragment to form a poly(A) tail. The factors responsible for each step of the reaction in mammals and in the yeast S. cerevisiae are indicated on each side.

(i) Cleavage/polyadenylation specificity factor. In the late 1980s, it was discovered that AAUAAA-dependent processing required a factor termed cleavage and polyadenylation factor (CPF), specificity factor (SF), or polyadenylation factor 2 (PF2), and now called CPSF (96, 168, 453, 454). CPSF is required for both the cleavage and poly(A) addition reactions and, consistent with this function, recognizes AAUAAA, a signal also essential for both reactions. Gel retardation experiments showed that CPSF specifically binds AAUAAA-containing RNAs (32, 45, 237). RNA modification interference assays indicated that all six nucleotides of AAUAAA are necessary for binding (237) and that RNAs as short as 10 nucleotides can be bound specifically (491). CPSF thus appears to recognize only the AAUAAA sequence, independent of any secondary structure. The binding of purified CPSF is very weak but can be greatly enhanced by a cooperative interaction with CstF bound to the downstream signal sequence (167, 280, 485, 499). CPSF purified from calf thymus or HeLa cells is a large protein complex containing subunits of 160, 100, 70, and 30 kDa, referred to as CPSF-160, CPSF-100, CPSF-70, and CPSF-30, respectively (45, 328).

(ii) Cleavage stimulation factor. CstF is necessary for cleavage but not for poly(A) addition (168, 453), although it can stimulate poly(A) addition on substrates with a CstF binding site upstream of the AAUAAA hexanucleotide (319). Purification of CstF from HeLa cells showed that it consists of three polypeptides of 77, 64, and 50 kDa (167, 452). cDNAs for all three have been sequenced (447, 448, 450). CstF-77 and its yeast and Drosophila homologs have eight repeats very similar to the tetratricopeptide repeat (TPR)-like motifs found in the yeast Prp39 and Prp42 U1 small nuclear ribonucleoproteins (snRNP) (303, 438). The repeats in the CstF-77 family of proteins lack the highly conserved alanine and glycine residues of a TPR, and this new motif has been termed HAT (half a TPR) (370). Like TPR repeats, HAT repeats may mediate protein-protein interactions (101, 370). Consistent with this sequence feature, CstF-77 was shown to be the middle subunit bridging CstF-64 and CstF-55, with three of them arranged in a linear fashion (448). Its direct interaction with CPSF-160 probably contributes to the mutual stabilization of the CPSF-CstF-RNA complex (329). CstF-77 is homologous to the Drosophila suppressor of forked [su(f)] protein (448). Mutations in su(f) can enhance or suppress the effects of transposon insertion, probably through changes in polyadenylation.

(iii) Cleavage factors I_m and II_m. CF I_m and CF II_m are required only for cleavage. (The designations CF I and CF II have been used for both the mammalian and yeast systems, but the factors are not homologous. The subscript "m" is included to differentiate the mammalian factor from the yeast one.) CF I_m has been purified to near homogeneity. Three polypeptides of 25, 59, and 68 kDa and possibly a fourth one of 72 kDa copurify with CF I_m activity (388, 389). The three smaller subunits can be UV cross-linked to RNA substrates (388). By gel retardation assays, purified CF I_m has higher affinity for RNAs containing polyadenylation signal sequences than for unrelated RNAs. Furthermore, CF I_m increases the stability of the CPSF-RNA complex, suggesting that this factor may also interact with CPSF and contribute to the overall stability of the 3'-end-processing complex (388). cDNAs encoding the 25- and 68-kDa subunits have been recently isolated (389). While CF I_m-25 has no known motifs, CF I_m-68 contains three distinct domains: an amino-terminal RNP-type RBD, a proline-rich region in the middle, and a carboxyl-terminal section consisting of alternating residues of opposite charge, with arginine residues alternating with glutamate, aspartate, and serine residues. This domain organization is strongly reminiscent of that found in the superfamily of RS-rich splicing factors. Most interestingly, recombinant CF I_m-25 and CF I_m-68 can be assembled in vitro and can replace purified CF I_m in cleavage assays. Preliminary sequence data from studies of the CF I_m-59 polypeptide reveal that it is similar to CF I_m-68, suggesting either that it is a degradation product of CF I_m-68 or that CF I_m exists as heterodimers of CF I_m-25-68 or CF I_m-25-59 (389). Analysis of the kinetics of the cleavage reaction indicates that interaction of CF I_m with RNA substrate may be an early step in the assembly of the 3'-end-processing complex, which facilitates the recruitment of other processing factors (389).

(iv) RNA polymerase II. Pol II, through the conserved carboxyl-terminal domain (CTD) of its largest subunit, has properties which make it an authentic cleavage factor. The CTD is found in a hyperphosphorylated form in elongating transcriptional complexes (186). An understanding of a role for the CTD in the cleavage of pre-mRNAs evolved from a study of the function of creatine phosphate (CP) in this reaction (204). In the early in vitro work on cleavage and polyadenylation, ATP was shown to be necessary for cleavage of precursor mRNAs (316), and therefore CP was included to regenerate ATP through creatine phosphokinase activity in the extracts. However, other phosphocompounds can substitute for CP in the cleavage reaction, leading to the proposal that CP was acting as a mimic for a phosphoprotein (204, 205). This hypothesis was confirmed when a synthetic CTD, even without phosphorylation, or purified pol II was shown to direct the cleavage of polyadenylation precursor in the absence of CP or ATP (205). Pol II does not appear to be involved in the poly(A) addition step. The interaction of the CTD with CPSF and CstF (299) may stabilize the cleavage complex or, as proposed by Hirose and Manley (205), have allosteric effects, such as those found for CTD and the capping enzyme (443). This involvement of Pol II in cleavage is consistent with earlier studies showing that protein-encoding mRNAs transcribed by Pol I and Pol III were for the most part not polyadenylated (257, 429, 432) and with the recent demonstration of the colocalization of CstF and phosphorylated pol II in vivo (439). Other consequences of the coupling of transcription and mRNA 3'-end formation are explored in a later section.

(v) Poly(A) polymerase. An activity which added adenosine residues to the 3' ends of RNAs was discovered in the early 1960s, and at the time it was a reaction of unknown significance (reviewed in reference 135). It is now well established that the this activity, PAP, plays a key role in the 3'-end formation of mRNA in eukaryotic cells. The first PAP, purified to homogeneity from calf thymus, was a degradation product of 57 to 60 kDa (475). Cloning and expression of the bovine PAP cDNAs have identified at least two isoforms of PAP which are generated by alternative splicing (380, 475). The longest forms, PAP I (77 kDa) and PAP II (82 kDa), differ only at their C termini and are enzymatically active (380, 515). PAP II may be the predominant full-length species, since most cDNAs isolated from other animal cells encode PAP II-related isoforms (30, 460, 475). Several other short forms of PAP (PAPs III, V, and VI) encode truncated proteins that are enzymatically inactive, and their function is unknown (380, 515).

View larger version (15K): [in this window] [in a new window]   FIG. 3.   Schematic diagram of the bovine and yeast poly(A) polymerases. The scale at the top indicates the size in amino acids. The hatched bar indicates the conserved regions. DID and D mark the locations of the three aspartate residues essential for nucleotidyltransferase catalytic activity. Gray bars in different shades in the bovine PAP indicate the C terminus of PAP I and PAP II generated from alternative splicing. NLS, nuclear localization signal; RD, regulatory domain involved in inhibition by U1A and in coupling of splicing and polyadenylation; SpD, specificity domains; C-RBS, carboxyl-terminal RNA-binding site.

(vi) Poly(A)-binding protein II. CPSF and PAP suffice for poly(A) addition to a precleaved RNA substrate. However, rapid elongation and control of poly(A) tail length requires an additional factor, PAB II (44). The 33-kDa PAB II contains a very acidic amino-terminal domain, a very basic carboxyl-terminal domain, and a single RNP domain in its middle region. The protein tends to form oligomers and binds specifically to poly(A) and poly(G) (331, 474). In vitro assays indicate that PAB II binds directly to CPSF-30 (85). A Drosophila homolog is the product of the rox2 gene (60).

Yeast cells. Fractionation of whole-yeast-cell extracts has identified five functionally distinct activities involved in cleavage and polyadenylation (82, 241). CF IA, IB, and II are sufficient for the cleavage reaction, while specific poly(A) addition requires CF IA and CF IB, Pap1, Pab1, and PF I (Fig. 1). All these factors have been purified to near homogeneity, and genes encoding most of the components have been cloned (Table 3). All the genes that have been cloned are essential. While the polyadenylation signals used by mammals and yeast are rather different in consensus sequence and organization, the factors which comprise the cleavage/polyadenylation apparatus in these two organisms exhibit surprising conservation.

(i) Cleavage/polyadenylation factor IA. CF I was originally identified as an activity needed for both the cleavage and poly(A) addition reactions (82), and further purification separated it into two components, CF IA and CF IB (241). CF IA consists of four polypeptides, Rna14 and Rna15 (241, 311), Pcf11 (13), and a 50-kDa polypeptide (241, 369). The first indication of the involvement of Rna14 and Rna15 in poly(A) mRNA metabolism came from the dramatic poly(A) tail shortening seen in strains harboring temperature-sensitive mutations in the RNA14 and RNA15 genes (312). These mutations are synergistically lethal with mutations in the PAP1 gene (311). Extracts from rna14 and rna15 mutants are defective in both cleavage and poly(A) addition, and fraction complementation assays suggested that Rna14 and Rna15 were components of CF I (311). Purification of CF I showed that Rna14 and Rna15 are indeed CF IA subunits (241). These two polypeptides are tightly associated, as indicated by a two-hybrid assay (239) and by coimmunoprecipitation as a heterodimer from whole-cell extract or CF I-containing fractions by antibodies against Rna15 (241). The 76-kDa Rna14 has sequence homology to mammalian CstF-77 (24% identity) (448).

(ii) Cleavage/polyadenylation factor IB. Purified CF IB is a single polypeptide of 73 kDa (241) encoded by the HRP1/NAB4 gene (239). This gene was previously identified as a suppressor of a temperature-sensitive npl3 allele, a gene encoding a protein which is involved in mRNA export (201) and can be cross-linked to nuclear poly(A)⁺ RNA in yeast (308). Hrp1 is structurally related to the mammalian hnRNP A, B, and D proteins (239) and has two RRMs in its middle region, both containing RNP1 and RNP2 sequences (201). The last 50 amino acids of Hrp1 are rich in arginine and glycine, with potential RGG methylation sites. A similar domain of hnRNP A1 can mediate protein-protein interactions (97). Experimental data indicate that the UA-rich polyadenylation signal is the likely binding site for Hrp1. The UV-induced RNA cross-linking of recombinant Hrp1 and the endogenous protein in yeast extracts is greatly enhanced by the presence of this sequence (83, 239). A recent SELEX analysis has also shown that UAUAUA is a high-affinity binding site for Hrp1 (464). The closest counterpart to Hrp1 in the mammalian system, at least in terms of having an amino-terminal RBD and a function in cleavage, is CF I_m-68.

(iii) Cleavage factor II. CF II has been purified by taking advantage of its ability to reconstitute the cleavage reaction in the presence of purified CF IA and CF IB (512). It contains four polypeptides, Cft1/Yhh1 (150 kDa) (442, 512), Cft2/Ydh1 (105 kDa), Brr5/Ysh1 (100 kDa), and Pta1 (90 kDa) (513). Cft1/Yhh1 was first identified by its sequence homology to mammalian CPSF-160 (24% identity and 51% similarity) (442). Depletion of extract with antibodies to Cft1/Yhh1 abolished both cleavage and poly(A) addition, consistent with CF II also being part of PF I (see below). However, addition of a CF II-containing fraction restored cleavage activity but not poly(A) addition (442). Cft1/Yhh1-specific antibodies recognize the 150-kDa component of purified CF II (512) and precipitate only four subunits (Cft1/Yhh1, Cft2/Ydh1, Brr5/Ysh1, and Pta1) from partially purified preparations (513).

(iv) Polyadenylation factor I. PF I was originally identified as an activity which supported poly(A) addition but not cleavage (82). A multiprotein complex from yeast containing PF I activity has been recently purified by restoration of polyadenylation activity to extracts with mutated Fip1 subunit of PF I (369). In addition to Fip1, this complex contained Pap1 (see below), Yth1, all four subunits of CF II, and two uncharacterized proteins, Pfs1 (58 kDa) and Pfs2 (53 kDa). The PF I-Pap1-CF II association is also indicated by coimmunoprecipitation experiments with yeast whole-cell extracts, since all CF II components as well as Pap1 and Fip1 were precipitated specifically by antibodies against Pap1 or Fip1 (513). It is not clear whether CF II in extract and cells exists in two forms (free and associated with PF I-specific subunits) or whether it is separated from PF I activity only by chromatography.

(v) Poly(A) polymerase. Yeast Pap1 was the first factor of the yeast 3'-end processing machinery to be purified, and its gene, PAP1, was the first to be identified. PAP1 was cloned both by sequencing of the purified 64-kDa protein and by complementation of a temperature-sensitive pap1 allele (263, 351). The yeast and mammalian Pap1 proteins are 47% identical within the first 400 amino acids, a region thought to comprise the catalytic domain and include the nucleotidyltransferase active site (291) (Fig. 3). The carboxyl-terminal regions are not conserved. Monoclonal antibodies which recognize epitopes in the amino- and carboxyl-terminal regions of Pap1 do not recognize the mammalian enzyme (242). An RNA-binding site at the carboxyl-terminus (C-RBS) is thought to interact in part with the phosphate backbone of polynucleotides and is essential for processive activity (517). At least two other contacts with the RNA substrate exist in addition to C-RBS. While their location on the enzyme is not known, one is thought to recognize the last 3 nucleotides of the RNA primer and to help the enzyme discriminate against deoxyribonucleotide substrates, and another base-specific site is proposed to interact with the primer 12 to 14 nucleotides upstream of the 3' end (517).

(vi) Poly(A)-binding protein I and poly(A) nuclease. Yeast Pab1 is the major RNP associated with the poly(A) tails of mRNA in both the nucleus and cytoplasm (3, 420, 446). Amino acid sequence analysis of this 70-kDa polypeptide revealed four RRM-type RBDs at its amino-terminal region and a proline-rich carboxyl terminus (67). Pab1 is important for translation initiation (456-458) and for deadenylation-dependent mRNA turnover (71). Recent studies have found that it is also involved in mRNA poly(A) tail formation. Pab1 cofractionates with CF IA (310) and interacts with Rna15 in two-hybrid assays and by coimmunoprecipitation (12). Cells bearing a pab1 mutation conferring temperature-sensitive growth show aberrantly long poly(A) tails in vivo (400) and in vitro (12, 310). Addition of anti-Pab1 antibodies to a processing extract results in an elongated poly(A) tail but has no effect on the cleavage reaction (12). In a reconstituted-poly(A) addition assay, Pab1 was further confirmed to function in limiting the length of poly(A) tail but was not required for cleavage (239).

(vii) Proteins with possible auxiliary or regulatory roles in yeast polyadenylation. Other proteins which are not components of constitutive polyadenylation factors can influence the efficiency or accuracy of processing in yeast. Some of these gene products suggest interactions of polyadenylation with other cell processes or ways of regulating the activity or specificity of the reaction. For example, temperature-sensitive mutations in the essential ESS1/PTF1 gene lead to increased readthrough of certain poly(A) sites and a decrease in the level of total poly(A)⁺ RNA in the cell (187). This gene encodes a peptidylprolyl-cis/trans-isomerase (PPIase), an enzyme thought to function in protein folding. While in vitro work is necessary to localize the defect to transcription termination or cleavage/polyadenylation, the authors noted that PPIases associate with the phosphorylated C-terminal domain of mammalian RNA pol II (55) and with the U4/U6.U5 tri-snRNP splicing factor (209, 459) and are proposed to promote disassociation of the HIV-1 capsid core (56). These findings led to speculation that the yeast PPIase activity might be involved in the assembly of polyadenylation or transcription complexes or their disassembly at termination sites, a topic discussed later in this review.

Mammals. The mutually cooperative interactions of CPSF, CstF, CF I_m, PAP, and PAB II in catalyzing accurate and efficient cleavage and polyadenylation have been well documented (167, 329, 388, 471, 499; for a review, see reference 473). The following model of mRNA 3'-end formation in mammalian cells (Fig. 4A) is derived from these numerous studies. The initiating step in assembly of a functional cleavage/polyadenylation complex is probably the recognition of signals on the precursor by CPSF and CstF in a process assisted by CF I_m. CPSF binds to AAUAAA through CPSF-160 with the help of CPSF-30 and possibly CPSF-100, and CstF binds to the DSE via CstF-64. The individual interactions of CPSF and CstF with their cognate sequences are weak but are stabilized by the cross-factor interaction of CPSF-160 and CstF-77. A final component of the initiation complex is pol II (205). While it is not known when the precise cleavage site is chosen, the CPSF-CstF interactions define the region in which it must lie. The formation of a cleavage-competent complex requires the additional recruitment of CF II_m and PAP. The contacts of CF I_m and CF II_m with the other factors and RNA are not known, but PAP at this point probably interacts with CPSF-160.

View larger version (68K): [in this window] [in a new window]   FIG. 4.   Schematic representation of the mammalian (A) and yeast (B) mRNA 3'-end-processing complexes. (A) The mammalian cleavage complex assembles through a cooperative binding of CPSF at the AAUAAA signal and CstF at the U- or GU-rich sequence. CPSF-160 directly interacts with CstF-77 and PAP. The arrangement of CF Im and CF IIm is not known. After the cleavage step, CPSF and PAP remain bound to the cleaved RNA and elongate the poly(A) tail in the presence of PAB II. (B) CF IA, Hrp1, and CF II are sufficient for the cleavage step in yeast. Poly(A) tail synthesis requires the addition of Pap1, PF I and Pab1. The Pan2/Pan3 deadenylase helps to regulate the poly(A) tail length.

Yeast. Since many of the yeast factors have only recently been characterized, less is known about how they interact with the RNA precursor and with each other. The available data and the homology to mammalian factors do allow us to construct a working model for assembly and processing in this organism (Fig. 4B). By analogy to the mammalian system, CF II and CF IA are probably prime players in the initiation of complex assembly. Their binding sites are not known, but from the structure of the core polyadenylation site in yeast, they are most probably located upstream of the cleavage site and may correspond to the UA-rich efficiency element and the A-rich positioning element. Extrapolating from the similarity of the best A-rich motifs (AAAAAA and AAAUAA) to the CPSF-binding site, we have placed CF II at this position. However, the cross-linking of Cft2 from highly purified CF II is dependent on the UA-rich sequence and the sequence at or downstream of the cleavage site, a finding which is not in accord with this model. Rna15 of CF IA shows some preference for U-rich RNAs, and CF IA may participate in recognition of the UA-rich motif or U tracts flanking the cleavage site. In yeast, Hrp1 (CF IB) appears to stabilize the assembly of the cleavage complex at the authentic poly(A) site (308). Recent findings suggest that it interacts with the UA-rich element as well (83, 239, 464) and, in doing so, can prevent the cross-linking of Cft2/Ydh1. The positioning of factors in the model is obviously provisional, and further experiments must be done to establish the architecture of the processing complex. A possible sequence of events is that CF II initially identifies the poly(A) site by interacting with both the UA-rich and A-rich elements and that the subsequent recruitment of Hrp1 and CF IA results in a reorganization of these contacts. Moreover, the prevalence of adenosine and uridine residues in the yeast 3' untranslated sequence may provide abundant contact points for both Rna15 and Hrp1 if they are generally positioned in the vicinity of the cleavage site by CF II.