PTB (polypyrimidine-tract-binding protein) is a ubiquitous RNA-binding protein. It was originally identified as a protein with a role in splicing but it is now known to function in a large number of diverse cellular processes including polyadenylation, mRNA stability and translation initiation. Specificity of PTB function is achieved by a combination of changes in the cellular localization of this protein (its ability to shuttle from the nucleus to the cytoplasm is tightly controlled) and its interaction with additional proteins. These differences in location and trans-acting factor requirements account for the fact that PTB acts both as a suppressor of splicing and an activator of translation. In the latter case, the role of PTB in translation has been studied extensively and it appears that this protein is required for an alternative form of translation initiation that is mediated by a large RNA structural element termed an IRES (internal ribosome entry site) that allows the synthesis of picornaviral proteins and cellular proteins that function to control cell growth and cell death. In the present review, we discuss how PTB regulates these disparate processes.
PTB (polypyrimidine-tract-binding protein), also referred to as hnRNP I [heterogeneous nuclear RNP (ribonucleoprotein) I], preferentially binds polypyrimidine-rich stretches . This protein is able to shuttle between the nucleus and the cytoplasm [2–4], a process that is tightly regulated since the diverse cellular functions which require PTB are dependent on its location (and on its associations with other factors). These processes include splicing , polyadenylation [5,6], mRNA stability  and translation initiation . The protein contains four RNA-binding domains, all of which are capable of interacting with the RNA. It has been proposed that PTB acts as an RNA chaperone, and by restructuring RNA to promote or inhibit the binding of other factors, this protein is then able to act as both a repressor and activator of RNA metabolism [9–11].
PTB is a 57 kDa protein consisting of four RRMs (RNA recognition motifs) (Figure 1) . Although each RRM has its own slightly different consensus RNA-binding sequence, all bind a short pyrimidine-rich sequence with a preference for sequences contained within a longer pyrimidine tract and containing cytosines . RRM3 and RRM4 interact extensively and form a compact globular structure, while the overall protein has an elongated structure as a result of a reasonably linear arrangement of the individual domains [13–15]. By binding to different sites on the same RNA molecule, particularly to sequences that may be separated in the primary sequence, PTB can lead to substantial restructuring of the RNA substrate and the introduction of RNA loops [9,15]. Such conformational changes are thought to be critical in enabling the ribosomal recruitment in IRES (internal ribosome entry site)-mediated translation initiation and may also be a means of preventing the binding or interaction of the required factors for splicing [9,15].
Structural model of PTB and its interaction with RNA
PTB isoforms and homologues
There are three main isoforms of PTB that are the result of alternative splicing [18,19]. PTB1 is the shortest isoform consisting of 531 amino acids, while PTB2 and PTB4 contain an additional 19 or 26 amino acids between RRM2 and RRM3 respectively. This is the result of the inclusion of PTB exon 9, which has two alternative 3′ splice sites (Figure 2). These isoforms are expressed at different levels according to cell type  and, despite their similarity, have been shown to have different activities in splicing and on IRES activity . A fourth isoform has also been reported, which results from the alternative splicing of exons 2–10 and produces a protein lacking RRM1 and RRM2 .
Alternative splicing of PTB
In addition, related genes encode proteins that display 70–80% homology to PTB and these are expressed in a tissue-specific manner. nPTB [neural PTB; also referred to as brPTB (brain-enriched PTB)] is expressed in adult brain, muscle and testis [22–24], whereas ROD1 (regulator of differentiation 1) is restricted to haemopoietic cells . As with PTB, alternative splicing of nPTB can lead to a number of possible isoforms that are expressed in a tissue-specific manner .
The expression of PTB and its homologues is tightly regulated through alternative splicing events. PTB autoregulates its own expression by binding to its mRNA and repressing splicing of exon 11, causing nonsense-mediated decay of the resultant frameshifted mRNA . Similarly, PTB regulates nonsense-mediated decay of nPTB transcripts and both PTB and nPTB promote the non-productive splicing of ROD1 .
As well as regulation at the mRNA processing level, it is likely that other post-transcriptional regulatory mechanisms are also involved in regulating nPTB, since its mRNA and protein levels do not always appear to correlate. In some cells, high levels of nPTB mRNA do not result in synthesis of the nPTB protein, whereas as in other situations, such as the up-regulation of nPTB on PTB knockdown (it has been shown that decreasing levels of PTB1 induces nPTB expression, ), the increase in full-length mRNA is not sufficient to account for the large increase in nPTB protein . In addition, limited translation of nPTB mRNA could result from poor codon utilization as many of its codons have an A or U at the third position, which is unfavourable for mammalian expression .
PTB and nPTB bind to the same RNA sequence elements although with some differences in affinity. It has been suggested that the two proteins are functionally redundant, since very few changes in protein expression were observed on siRNA (small interfering RNA) knockdown of PTB unless nPTB was also targeted . Although this seems plausible considering the high sequence homology between the proteins, the tissue-specific expression and cross-regulation suggests otherwise. Differences between the two homologues in splicing activity  and mRNA stabilization , as well as distinct roles in IRES-mediated translation initiation, have been observed [31,32]. In particular, a shift to the neuronal splicing programme that is responsible for directing neuronal differentiation is in part mediated by a switch from PTB to nPTB expression [29,33].
PTB as a repressor of splicing
Alternative splicing is a common mechanism for regulating gene expression in eukaryotes, enabling several mRNAs to be generated from a single gene and therefore the synthesis of a diverse set of proteins. PTB binds to pyrimidine-rich elements in the RNA and thereby mediates splicing repression in a long list of alternatively spliced pre-mRNAs. In some cases, a single PTB protein is sufficient to mediate splicing repression, but in most of the cases a number of binding sites are present [34,35].
The most straightforward model for the repressive effect of PTB in splicing is suggested by the overlapping of PTB-binding sites with those for splicing factors such as U2AF and U2 snRNP (U2 small nuclear RNP), or with other regulatory sequences (reviewed in ). In this scenario, PTB sterically impedes factors involved in splicing to prevent correct splicing of the affected exons. Indeed, as mentioned earlier, high levels of PTB cause skipping of exon 11 of PTB's own mRNA, leading to nonsense-mediated decay and the down-regulation of PTB gene expression. Binding sites near the 3′-end of exon 11 are thought to be responsible for this effect . A variation on this theme is seen in the mouse IgM gene where the fate of two exons (M1 and M2) is determined by competing splicing regulatory elements in exon M2: an enhancer and an inhibitor. The splicing inhibitor requires PTB binding to repress splicing and, via an unknown mechanism, the enhancer is required to disrupt the inhibitor–PTB interaction and promote splicing .
In addition to repressing the inclusion of exons in the mature mRNA, PTB has also been shown to positively regulate the inclusion of an alternative exon in the calcitonin [CT/CGRP (calcitonin/calcitonin gene-related peptide)] gene . Here, PTB binds to a polypyrimidine splicing enhancer just downstream of exon 4 to increase exon 4 and increases its inclusion in non-neuronal cells. This enhancer contains a pseudoexon, complete with branch point, polypyrimidine tract and 5′ splice site, but no internal exon sequence. PTB and U2AF compete for the same binding site, thus binding of U2AF to the enhancer counteracts the effect of PTB and reduces exon 4 inclusion. However, PTB binding to the enhancer is also required for maximal polyadenylation of the exon 4-containing mRNA, implying a role in interaction with the polyadenylation machinery, and consequent effect on splicing efficiency .
PTB and 3′-end processing
PTB has been implicated in 3′-end processing and in particular mRNA polyadenylation [5,6]. PTB is thought to compete with CstF for binding to the pyrimidine-rich downstream element and thereby inhibits the 3′-end cleavage required for polyadenylation [6,41]. However, there are also indications that PTB can be involved in cleavage activation [6,42]. In addition to the downstream elements described above, pyrimidine-rich USEs (upstream elements) have also been identified that are important for polyadenylation. It has been shown that PTB interacts with these elements but, in this situation, seems to mediate, at least in part, USE-dependent cleavage activation [6,42]. PTB can also affect the choice of polyadenylation site. Alternative polyadenylation leads to mRNAs with variable 3′-ends, or proteins with different C-termini, and auxiliary cis-acting RNA elements have been suggested to promote the use of suboptimal polyadenylation signals . In the case of PTB, this regulation appears to be linked to its function as a splicing repressor. Binding of PTB to an intronic pseudoexon in CT/CGRP pre-mRNA results in inclusion of an alternative 3′-terminal exon containing an alternative polyadenylation signal . This alternative polyadenylation signal is in close proximity to a second PTB-binding site, suggesting a possible dual role for PTB, interrupting the productive recognition of the enhancer pseudoexon by splicing factors and positively affecting polyadenylation . A similar mechanism is utilized in the tissue-specific 3′-end processing of the β-tropomyosin gene whereby PTB regulates both the splicing of exon 9A9′ and the cleavage/polyadenylation of this 3′-terminal exon .
PTB shuttles rapidly between the nucleus and cytoplasm [4,43–45]. The translocation of this protein to the cytoplasm, in general, occurs under conditions of cell stress including viral infection [46–48], apoptosis  and exposure of cells to genotoxic agents .
PTB contains an NES (nuclear export signal) and an NLS (nuclear localization signal); they are distinct, yet both are located in the N-terminus of the protein . Within the NES is a conserved serine residue, flanked by two sets of basic residues, that is required for efficient nuclear localization. The phosphorylation of this residue by PKA (protein kinase A) results in accumulation of PTB in the cytoplasm . However, changes in the subcellular localization of the protein have also been observed without detectable changes in phosphorylation [44,51].
PTB and RNA transport
PTB has been shown to play a role in mRNA localization . The Xenopus homologue of PTB (VgRBP60) binds to an element (VM1) in the 3′-UTR (untranslated region) of Vg1 mRNA that has been shown to be important for RNA localization . Mutations in VM1 that prevent PTB binding also block Vg1 RNA localization, and PTB and Vg1 RNAs co-localize in the cytoplasm. This suggests a role for PTB in regulating the localization of mRNA, although probably as one of a number of proteins interacting with the mRNA . PTB-mediated mRNA localization is also involved in neurite growth. Thus, on activation of the PKA pathway, PTB is exported from the nucleus to neurite growth terminals where it interacts with α-actin mRNA, localizing it at neurites. This process is essential for cell motility and neuronal axon growth . Finally, PTB also acts as a nuclear export factor for HBV (hepatitis B virus) RNA .
PTB and mRNA stability
There are several examples where mRNA stability is influenced by PTB. Insulin mRNA, the major mRNA in pancreatic β-cells, can be very stable, particularly at high glucose levels, and the observation that PTB mRNA (although not protein) levels increase 5-fold in MIN6 cells in high glucose concentrations led to the speculation that PTB might be important in regulating insulin mRNA stability. The glucose-stimulated binding of PTB to a polypyrimidine-rich sequence (ins-PRS) in the 3′-UTR of rat insulin mRNA has been shown to increase the stability of the insulin mRNA [54,55]. This glucose-induced binding is inhibited by rapamycin, suggesting that mTOR (mammalian target of rapamycin) signalling is involved . Hypoxia can also stimulate PTB binding to the ins-PRS, but this seems to involve a different pathway and is not inhibited by rapamycin . Inhibition of PTB binding to the ins-PRS by mutation of the binding site leads to a significant reduction in the half-life of the mRNA [54,55,57]. The mechanism by which PTB protects insulin mRNA from degradation is not clear, but it has been suggested that destabilizing elements in the 3′-UTR could be masked in β-cells. Translocation of PTB to the cytoplasm leads to stabilization of insulin mRNA and those mRNAs that encode the secretory granule components required for insulin storage [45,51,58,59]. However, other effects of PTB, for example on polyadenylation, cannot at present be excluded from contributing to the increased stability of insulin mRNA.
The stability of VEGF (vascular endothelial growth factor) mRNA has also been linked to its association with PTB , although in this case PTB is in a complex containing CSD (cold shock domain) Y-box proteins. These complexes bind to conserved sites in the 5′- and 3′-UTRs of the VEGF mRNA and increase its stability. It has been suggested that the CSD–PTB complexes may act by stabilizing structures that are required for the increased stability of the VEGF mRNA , although it is also possible that they protect the mRNA by excluding factors that would otherwise lead to its degradation.
PTB in viral translation and replication
Most of the translation initiation is mediated via an m7Gppp cap structure at the 5′-end of the mRNA, which is recognized by the eukaryotic initiation factor complex, eIF4F (eukaryotic initiation factor 4F). The 40S ribosomal subunit interacts with the eIF4F complex and scans along the RNA until it reaches the AUG start codon . In contrast, the RNAs found in the family of Picornoviridae [a large family of human and animal RNA viruses, which includes PV (poliovirus), HRV (human rhino virus), EMCV (encephalomyocarditis virus), HAV (hepatitis A virus) and FMDV (foot and mouth disease virus)] lack a cap structure and, instead, the recruitment of the ribosome occurs internally at highly structured elements known as IRESs. Interaction may occur directly between the IRES and the 40S ribosomal subunit but is mediated by canonical trans-acting factors and ITAFs (IRES trans-acting factors) such as PTB. This protein has been shown to be required in internal ribosome entry by a number of viral IRESs [63–66].
It has been shown that PTB interacts at a number of sites within the picornaviral IRESs, each consisting of short polypyrimidine tracts with the consensus sequence CUUU . Both binding sites can be bridged by a single PTB molecule, an arrangement that supports a role for PTB as an RNA chaperone . It has been proposed that PTB may stabilize or alter the IRES structure to enable the recruitment of the ribosome and to position it correctly at the start codon . PTB binding to PV, EMCV and TMEV (Theiler's murine encephalomyelitis virus) IRESs similarly affects the overall structure of the IRES [69,70]. Tissue-specific differences in the expression of ITAFs may be responsible for determining the activity of viral IRESs and affecting virulence in different cell types [32,69,74]. There is evidence suggesting that differential expression of PTB homologues is important in the neurovirulence of TMEV and PV . For example, nPTB and PTB both bind to the TMEV IRES and stimulate 48S complex formation  and mutations of the binding site decrease binding and IRES activity of both proteins; however, the stimulatory effect of nPTB is reduced to a greater extent .
PTB has been shown to bind to HCV IRES RNA [71,72], but it is not required for IRES activity in vivo . However, it is thought to be involved in the replication of this virus as the phosphorylated form of PTB is present in HCV replicon cells and is found to co-localize with the HCV replication complex [47,48,75]. PTB may therefore provide a potential target for anti-HCV therapeutic agents .
PTB as an ITAF for cellular IRESs
Under conditions that inhibit cap-dependent translation initiation (e.g. cell stress, apoptosis and viral infection), the expression of certain proteins is maintained by IRES-mediated translation [77,78]. PTB has been proposed to be a general ITAF since most of the cellular IRESs studied require this protein for function . In many cases, it has been shown that PTB has a role in the recruitment of the ribosome. For example, the BAG-1 (Bcl-2-associated athanogene 1) IRES is stimulated by the combined action of PTB and PCBP1 [poly(rC)-binding protein 1] in a two-step process (Figure 3). PCBP1 binds, remodelling the RNA structure and facilitating the binding of PTB that mediates ribosome recruitment . A similar mechanism also occurs in the IRES of Apaf-1 (apoptotic protease-activating factor 1). In this case, the initial binding of the ITAF UNR (upstream of N-Ras) causes a conformational change to allow PTB binding . Interestingly, Apaf-1 IRES preferentially binds nPTB and is therefore more active in cell lines where this protein is more highly expressed, providing a further example of tissue-specific regulation by PTB/nPTB .
BAG-1 IRES activation by RNA-binding proteins PCBP1 and PTB
The consensus sequence for PTB-mediated ribosome recruitment has been identified as (CCU)n as part of a polypyrimidine tract . AIRESs (artificial IRESs) have been constructed that comprise double-stranded (CCU)n motifs [73,81], and the data show that this sequence alone, as part of a double-stranded RNA stem loop, is sufficient to mediate internal ribosome entry , with ribosome recruitment being favoured when there is a spacer region between the hairpin and the start codon and when there is slight unwinding of the structure induced by base-pair mismatches . This suggests that the role of PTB binding is not only to induce structural changes in IRES elements but also it may alone, or in conjunction with an interacting protein partner, provide a bridge between the IRES and the ribosome.
There are, however, a few examples however where PTB appears to act as a repressor of IRES-mediated translation .
Changes in the availability of PTB, through cytoplasmic relocalization, may also be a critical factor in the regulation of IRES-mediated translation initiation. Such a relocalization has been observed after exposure to the chemotoxic drug vincristine and appears to be a requirement for IRES-mediated translation of BAG-1 . There is also evidence suggesting that an increase in cytoplasmic PTB may be important in IRES activation during apoptosis . PTB is required by a number of IRESs during apoptosis ; moreover, reduced expression of PTB is sufficient to cause inhibition of apoptosis, whereas increased expression causes a small but reproducible increase in the rate of spontaneous apoptosis .
There remains little doubt as to the importance of PTB in a vast number of post-transcription regulation processes. Although the array of different functions may at first seem incongruous, on closer analysis, most of these effects come from just two simple modes of action. Regulation can occur either through direct competition at an RNA-binding site with another factor required for RNA processing, or through structural rearrangement of the RNA, resulting in changes in the RNA–RNA and protein–RNA interactions that can form.
PTB is as important in RNA processing in the nucleus, as it is in RNA translation in the cytoplasm. Since PTB is a shuttling protein, control of its rate of import or export to or from the nucleus can be used to modulate which role dominates.
Recent genome-wide investigations have shown PTB to interact with a large number of mRNAs that encode proteins with a very diverse range of functions (K. Sawicka, M. Bushell, K.A. Spriggs and A.E. Willis, unpublished work; ). It is therefore likely that PTB is involved in the regulation of a large number of cellular functions that are yet to be identified.
Post-Transcriptional Control: A Biochemical Society Focused Meeting held at the University of Manchester, U.K., 26–28 March 2008 as part of the Gene Expression and Analysis Linked Focused Meetings. Organized and Edited by Nicola Gray (MRC Human Genetics Unit, Edinburgh, U.K.), Simon Morley (University of Sussex, U.K.) and Graham Pavitt (Manchester, U.K.).
apoptotic protease-activating factor 1
Bcl-2-associated athanogene 1
cold shock domain
calcitonin/calcitonin gene-related peptide
eukaryotic initiation factor 4F
- hnRNP I
heterogeneous nuclear RNP I
polypyrimidine-rich sequence of insulin mRNA
internal ribosome entry site
IRES trans-acting factor
nuclear export signal
poly(rC)-binding protein 1
phosphoglycerate kinase 2
protein kinase A
regulator of differentiation 1
RNA recognition motif
Theiler's murine encephalomyelitis virus
vascular endothelial growth factor
We thank Stephen Curry (Imperial College) for providing the structure of PTB. K.S. is funded by a studentship from the BBSRC (Biotechnology and Biological Sciences Research Council) and M.B. holds a BBSRC David Phillips Fellowship.