Far upstream element binding protein 1: a commander of transcription, translation and beyond

Abstract

The far upstream binding protein 1 (FBP1) was first identified as a DNA-binding protein that regulates c-Myc gene transcription through binding to the far upstream element (FUSE) in the promoter region 1.5 kb upstream of the transcription start site. FBP1 collaborates with TFIIH and additional transcription factors for optimal transcription of the c-Myc gene. In recent years, mounting evidence suggests that FBP1 acts as an RNA-binding protein and regulates mRNA translation or stability of genes, such as GAP43, p27^Kip and nucleophosmin. During retroviral infection, FBP1 binds to and mediates replication of RNA from Hepatitis C and Enterovirus 71. As a nuclear protein, FBP1 may translocate to the cytoplasm in apoptotic cells. The interaction of FBP1 with p38/JTV-1 results in FBP1 ubiquitination and degradation by the proteasomes. Transcriptional and post-transcriptional regulations by FBP1 contribute to cell proliferation, migration or cell death. FBP1 association with carcinogenesis has been reported in c-Myc dependent or independent manner. This review summarizes biochemical features of FBP1, its mechanism of action, FBP family members and the involvement of FBP1 in carcinogenesis.

INTRODUCTION

The far upstream element binding protein 1 (FBP1) was first discovered as a transcriptional regulator of the proto-oncogene c-Myc. As a transcription factor, c-Myc controls the expression of about 10% of cellular genes, including those essential for cell proliferation, differentiation and apoptosis.^1–4 A large volume of studies have led to the thorough characterization of signaling pathways and transcription factors regulating the expression of the c-Myc gene. Among the list of the regulators of c-Myc transcription is FBP1, a single stranded DNA-binding protein initially identified as the binding partner for the far upstream element (FUSE), located ~1.5 k base pairs (bp) upstream of the transcription start site in the c-Myc promoter. It was found that the FUSE sequence specifically binds to one protein in undifferentiated but not differentiated cells,⁵ leading to the discovery of FBP1 protein in 1994.⁶ In addition to FBP1 binding, FUSE was later identified as a target of FBP-interacting repressor (FIR). The interplay between FBP1, FIR and FUSE influences the timing and maximal level of c-Myc gene expression.⁷

A number of recent reports have indicated that FBP1 is also an RNA-binding protein. RNA binding accounts for a role of FBP1 in translation or stabilization of several cellular and viral mRNA species.^8–11 These activities, together with c-Myc transcriptional control, place FBP1 at an eminent position in carcinogenesis.^12–15 Aberrant expression of FBP1, mutations in the FBP1 gene or alternative splicing of its repressor FIR have been found in a variety of malignant tissues.^16–19 In addition to its role in carcinogenesis, FBP1 is important for the development of certain organs, such as the lung²⁰ and neural system.^9,21 FBP1 binding to viral RNA contributes to several viral diseases, including hepatitis C, and hand-foot-mouth disease. Despite the prominent role of FBP1 in transcription, mRNA stability and translation, FBP1 remains a mystery to many.

IDENTIFICATION OF FBP1

FBP1 was first cloned from a cDNA library generated from the undifferentiated human monoblastic line U937.⁶ Such an approach was undertaken in an effort to understand how c-Myc transcription was downregulated in U937 and the human promonomyelocytic leukemia HL60 cells upon induction of differentiation.^22,23 Since its discovery, FBP1 has been found to be expressed in a variety of human cancer cell lines, including breast (MCF7 and MDA), colon (Colo 320DM, DLD1, HT29, LS180, SW48, SW480, SW620C and WiDr), prostate (PC3), cervix (HeLa), Burkitt’s lymphoma (Raji) and T-cell leukemia (Jurkat).^24,25

FBP1 binding to FUSE is required for maximal transcription of c-Myc gene. In a chloramphenicol acetyltransferase reporter assay with a 3.2-kb c-Myc promoter sequence fused to the chloramphenicol acetyltransferase coding sequence, elimination of FUSE resulted in an 80% reduction in c-Myc transcription.⁵ Several lines of evidence have indicated that FBP1 interaction with FUSE is critical for optimal c-Myc transcription. Avigan et al.⁵ reported that a DNA probe containing the FUSE sequence was recognized by a protein in the lysate of undifferentiated, but not differentiated, HL60 cells in an Electrophoretic Mobility Shift Assay (EMSA). Following identification of the protein as FBP1, subsequent studies have documented the presence of FBP1 mRNA and protein in undifferentiated HL60 cells, which decline significantly after induction of differentiation with dimethyl sulfoxide and 12-Otetradecanohlphorbol 13-acetate.^6,24 This expression pattern is consistent with the reduction of FUSE-binding activity and c-Myc transcription, and consequently commitment to cell differentiation.^6,24 Overexpression of the FBP1 gene led to a fivefold increase in FUSE-mediated transcription as measured with a c-Myc promoter chimeric chloramphenicol acetyltransferase reporter. This increase was specific for the interaction of FBP1 with the FUSE sequence, as FBP1 failed to stimulate the reporter downstream of a c-Myc promoter sequence containing mutated FUSE.⁶

The physical interaction of FBP1 with FUSE has been demonstrated with copious lines of evidence in vitro and in vivo as summarized in Table 1. Electrophoretic mobility shift assays (EMSAs) showed that recombinant FBP1 or FBP1 protein purified from HL60 cells bound to the oligonucleotides containing the FUSE sequence.⁶ Detailed analyses revealed that FBP1 bound to non-coding strands of FUSE in vitro and in vivo, as FBP1 protected this strand, but not the coding strand, from potassium permanganate modification of unpaired thymine and subsequent cleavage by piperidine.^6,26 The discovery of FBP1 set a new milestone for understanding of transcriptional regulation of the c-Myc gene.

Table 1.

Evidence for FBP–FUSE interactions in vitro and in vivo

Approach	Findings	References
In vitro
EMSA	FUSE activity in undifferentiated but not differentiated HL60 cells	5
EMSA	FBP1 purified from HL60 cells or T cells or HeLa bound to noncoding FUSE sequence directly	6,24,41
EMSA	GST-FBP1 protein bound to the non-coding FUSE sequence	6,25,26,47
LM-PCR	GST-FBP1 protein bound to supercoiled but not linear FUSE	26
NMR	FBP1 KH3 and KH4 domain bound to c-Myc 29-mer of FUSE (− 1525 to 1553 of c-Myc Promoter)	39
EMSA	FBP1 binding to c-Myc 29-mer of FUSE sequence required interaction with FIR	44
LM-PCR	DNA binding by FBP requires sustained superhelical stress	7
CAT reporter	A protein in undifferentiated but not differentiated cells binds to FUSE element	5
In vivo
Northern and western blot;   Immunohistochemistry	Presence of FBP1 in undifferentiated but not differentiated HL60 cells, corresponding to c-Myc expression	6,24
CAT reporter	Ectopic expression of FBP1 increases the wild type, but not mutated FUSE reporter activity	6,30
CAT reporter	Expression of Gal4-FBP C-terminus in HeLa cells promoted FUSE reporter activity	25,27
CAT reporter	FIR inhibits FBP1-activated transcription of CAT reporter	41
CAT reporter	XPB and XPD mutations repress FBP1-activated transcription of CAT reporter	31
LM-PCR	Decreased sensitivity to KMnO4 in the noncoding strand of FUSE	6
CAT reporter	Dominant negative FBPcd domain represses FUSE reporter activity	30
ChIP	FBP1 binds to FUSE due to serum stimulation in Hs68 primary fibroblasts	7

Abbreviations: CAT, chloramphenicol acetyltransferase; ChIP, chromatin immunoprecipitation; FBP1, far upstream binding protein 1; FUSE, far upstream element.

FBP FAMILY MEMBERS

The human FBP1 gene (NM_003902.3), also known as FUBP1 (far upstream binding protein 1), encodes a 644-amino acid protein with three well-defined domains: an amphipathic helix N-terminal domain, a tyrosine-rich C-terminal transactivation domain and a DNA-binding central domain (Figure 1).^6,27 The N-terminal domain represses the activity of the C-terminal domain,²⁷ suggesting that FBP1 protein folds into an enclosed conformation in an inactive state and undergoes a conformational change upon activation. The DNA-binding central domain contains four K-homology (KH) motifs, the conserved α-helices with a hydrophobic core first identified from heterogeneous nuclear ribonucleoprotein K.²⁸ Each of the tandem KH domains is followed by an adjacent amphipathic helix (Figure 1). KH3 together with the KH4 domain is sufficient for FUSE binding in vitro, as the GXXG motif in KH3 or KH4 domain bound to a single stranded 5′-dTTTT or 5′-dATTC sequence of FUSE in vitro, respectively.²⁹ When the central domain, that is, FBPcd, was expressed ectopically, it acted as a dominant negative competitor for endogenous FBP1.³⁰

Figure 1.

Schematic structure of human FBP1 protein. The N-terminus is composed of amino acid residues 1–106. The central domain spans amino acid residues 107–447 and is responsible for DNA binding. The C-terminus contains amino acid residues 448–644, including three YM motifs. The well-defined NLS sequences and localizations at the N-terminus and the Central domain are illustrated. Two isoleucines I378 and I379, important for FBP1 localization, are underlined. The caspase cleavage site, DQPD, is marked with asterisks. The domains are not drawn proportional.

The transactivation domain at the C-terminus (448–644 aminoacid residues) contains three tyrosine-rich motifs (YM). Such a structural feature is important for physical interaction of FBP1 with TFIIH, a basal transcription factor.³¹ TFIIH, a multimeric protein containing 10 subunits, exhibits p89/XPB/ERCC3 and p80/XPD/ERCC2 DNA helicase activities.³² FBP1 binding stimulates p89/XBP/ERCC3 3′-5′ helicase activity of TFIIH, which mediates c-Myc promoter melting.³¹ Interestingly, FBP1 itself has an ATP-dependent 3′-5′ helicase activity and is known as DNA Helicase V. Such helicase activity allows FBP1 to unwind DNA–DNA and RNA– RNA duplexes in vitro.^33,34 This helicase activity is likely mediated by the Argine-Glycine-Glycine (RGG) motifs at the C-terminus, as this motif is conserved among several proteins exhibiting helicase activity, for example, nucleolin and Ras-GAP SH3 domain-binding protein (G3BP/helicase VIII).³⁵ The helicase activity of FBP1 may have a role in unwinding the DNA duplex around FUSE to sustain c-Myc transcription.³⁴

The FBP1 gene is conserved in eukaryotes, sharing over 90% sequence homology among mammalian species. Screening of the human skeletal muscle cDNA libraries led to the discovery of two homologs: FBP2, also known as KHSRP for K-homology splicing regulator protein and FBP3.²⁵ In Caenorhabditis elegans, a protein homologous to FBP1 has been found, which is considered as the fourth member of FBP family.²⁵ The FBP1 gene is located at chromosome 1p31.1, whereas FBP2 and FBP3 are located at 19p13.3 and 9q34.11, respectively.³⁶ These three genes share a high degree of sequence homology and exhibit common structural features.²⁵ Their expression patterns were similar in most cell lines tested, for example, K562 chronic myelogenous leukemia cells, Colo320 colon cancer cells and Jurkat T-cell leukemia cells.²⁵ In experimental animals such as mice, the expression patterns of FBP1 and FBP3 are nearly identical in various tissues, high in the thymus and pancreas, but low in the ovary, liver, heart and skin. FBP2 expression is relatively low in all tissues except the heart and kidney. During development, FBP1 expression is high in the fetal brain but low in adult brain, whereas FBP3 expression is the opposite (http://www.genecards.org).

Sequence homology, common structural features and similarities in tissue distribution suggest functional overlaps among FBP family members. However, although the RGG motifs are present in all three FBPs, helicase activity was only detected in FBP1.³⁴ Whereas FBP1 is best known as a DNA-binding protein regulating c-Myc gene transcription, FBP2 is better known for RNA binding and therefore has been mostly studied for its role in the stability of mRNA.³⁷ Knocking down FBP2 resulted in an increased level of FBP1 mRNA and protein, pointing to FBP2 as a negative regulator of FBP1 expression.¹⁷ On the other hand, knocking down of FBP1, but not FBP2, showed a noticeable decrease of Hep3B cell proliferation.¹⁷ In contrast, loss of FBP2, but not FBP1, led to a marked inhibition of cell migration.¹⁷

FBP3 has been less studied in comparison with FBP1 or FBP2. It seems that FBP3 may cooperate with FBP2 for RNA binding. This is exemplified by FBP3 binding to the 3′-untranslated region (UTR) of thrombin mRNA, thereby stabilizing the mRNA species in cooperation with FBP2 in HUH7 hepatoma cells.³⁸ The RNA-binding activity of the FBP2/FBP3 complex appears to be regulated by p38 MAP kinase signaling.³⁸ Therefore, despite highly homologous sequences and similarities in tissue distribution, the activities of FBP family members do not always overlap.

FBP1–FUSE–FIR AXIS FOR TRANSCRIPTIONAL REGULATION OF C-MYC

The FUSE is an AT-rich element located at −1.5 kb in the promoter of the c-Myc gene.⁵ The core sequence of c-Myc FUSE, 5′-TATATTCCCTCGGGATTTTTTATTTTGTG-3′, is located in the noncoding strand of DNA and recognized by the KH3 and KH4 domains in vitro.³⁹ A Systematic Evolution of Ligands by Exponential Enrichment (SELEX) assay indicated that FBP1 bound to FUSE with low stringency. In this study, in vitro binding of a series of synthetic 29-mers containing partial FUSE sequence and FBP1 contiguous pairs of KH domains showed the optimal sequence for full-length FBP1 binding as: TTGTa(N)_4/5 TYGTa(N)_4/5TYGTa(N)_4/5KTGY (Y = T or C, and K = T or G), where each KH domain recognized a tetrad with a weak preference of A at the fifth position as indicated in the lower case.⁴⁰ The binding of FBP1 to FUSE is necessary, but not sufficient, for c-Myc transcription as the FUSE element alone, even in multiple copies, cannot induce c-Myc transcription. Instead, FBP1 binding to FUSE coordinates additional cis-elements and their corresponding transcription factors to promote optimal transcription of the c-Myc gene.⁴¹

An elegant study by Liu et al.⁴¹ has shown that FBP1 binds to the supercoiled non-coding strand of FUSE and at the same time interacts with the basal transcription factor TFIIH at the transcription start site, causing formation of a promoter loop. This promoter loop brings FBP1 from a relatively far distance to be within proximity of the TFIIH and RNA polymerase II complex. As described in the section above, binding of FBP1 results in activation of TFIIH helicase and subsequently enhances transcription of c-Myc gene.

TFIIH and the central domain of FBP1 interact with a protein in common: FBP interacting repressor (FIR, NM_014281). FIR is an alternatively spliced variant of the 60-kDa poly-U-binding factor (PUF60, NM_078480). As a homolog of the U2 auxiliary factor (U2AF), PUF60 is an essential splicing factor.^42,43 Compared with PUF60, FIR lacks 17 amino acids in the N-terminus due to truncation of exon 5. Because of extensive sequence homology and potential functional overlap, the term FIR is often used interchangeably with PUF60. However, whether the additional 17 amino acids in PUF60 cause functional deviation from FIR remains unclear. It is possible that PUF60 shares the inhibitory role of FIR on c-Myc expression, as both proteins contain exon 2, which encodes the domain critical for regulating c-Myc expression in vitro and in vivo.¹⁸ Nevertheless, FIR is a 542-amino-acid protein containing two RNA recognition motifs at the central domain.^31,41,44

The interaction of FIR with FBP1 results in FIR binding to FUSE as a homodimer. FIR binds to FUSE through its U2AF homology motif at the C-terminus and two central RNA recognition motifs.^43,45 This interaction may facilitate the N-terminus of FIR to contact TFIIH, thereby inhibiting helicase activity of TFIIH.^31,41 It appears that FIR inhibits FBP1 mediated c-Myc transcription in a TFIIH helicasedependent manner,⁴⁶ as such inhibition is lost in cells derived from Xeroderma Pigmentosum group B (XP B) and XP D patients, in which TFIIH is defective in helicase activity.³¹ When FIR is recruited to the promoter loop, it counteracts TFIIH helicase and silences c-Myc transcription.^7,46,47

It should be noted that FBP1 does not regulate the basal transcription of c-Myc gene. The basal transcription of c-Myc gene by RNA polymerase II usually pauses at the promoter-proximal region within 50 bp downstream of P2 promoter.⁴⁸ FBP1 promotes the maximum level of c-Myc transcription by relieving this pause of RNA polymerase II during serum stimulation. A delicate tripartite FBP-FUSE-FIR interaction illustrates how FBP1 fine tunes c-Myc transcription.⁴⁵ During transcriptional initiation, counter-rotation of DNA by the basal transcriptional machinery generates a torsional stress and causes the double-stranded DNA helix around FUSE to melt into two separate strands.⁶ When FBP1 recognizes and binds to the non-coding strand of FUSE, it brings together TFIIH through promoter looping and stimulates p89/XBP/ ERCC3 3′–5′ helicase activity of TFIIH,^26,31 which in turn resumes RNA polymerase II activity on the c-Myc promoter. As a result, FBP1 binding to FUSE and interaction with TFIIH leads to an increase of c-Myc transcription above the basal level during serum stimulation. When FIR is recruited to interact with FBP1 on FUSE, FIR causes an inhibition of helicase activity of TFIIH, thereby brings back serum-induced c-Myc peak expression to the basal level.^31,41

Levens⁴⁹ has proposed a three-phase paradigm for c-Myc transcription involving FBP1 (Figure 2). In Phase I, conventional transcription factors promote transcription via the RNA polymerase II, which is latent and parks in the proximal promoter region, while the FUSE region DNA is exposed owing to chromatin remodeling. In Phase II, FUSE-binding proteins, FBP1 and FBP3, recognize and bind to FUSE. Initially, FBP3 binds to FUSE, but is then replaced by FBP1. Once FBP1 binds to FUSE, it interacts with TFIIH to stimulate the helicase activity of TFIIH. Owing to the interactions between FBP1, FUSE and TFIIH, the DNA bends and forms a promoter loop, allowing c-Myc transcription to resume. In the final phase, FIR displaces FBP1 from FUSE binding, preventing full scale c-Myc transcription.

Figure 2.

Steps of c-Myc transcriptional activation by FBP1. FBP1 mediated c-Myc transcriptional activation may be divided into three phases. In phase I, initiation of transcription by conventional transcription factors unwinds double-stranded DNA, exposing single-stranded FUSE, which recruits FBP1. In phase II, through its interaction with TFIIH, FBP1 causes formation of a promoter loop and promotes c-Myc transcription by RNA polymerase II. In phase III, FIR binds to FBP1, forms a dimer and displaces FBP1 from FUSE. The binding of FIR to TFIIH represses TFIIH helicase activity, and thus halts c-Myc transcription.

Although the FBP1–FUSE–FIR axis is well characterized for c-Myc gene transcription, one would predict that FUSE or FUSE-like elements are present in genes beyond c-Myc. In fact, ubiquitin-specific peptidase 29 (USP29), a type-2 ubiquitin carboxylterminal hydrolase important for p53 protein degradation, has been found to contain a 54-bp FUSE located ~2.5 kb upstream of the transcription start site. FBP1 promotes transcription of the USP29 gene through binding to FUSE.⁵⁰ In addition to USP29, it has been reported that four potential FBP1-binding sites are present in the promoter of p21 cyclin-dependent kinase inhibitor.¹⁹ Such presumed FUSE-like sequences confer FBP1 transactivating activity in a luciferase reporter assay.¹⁹ If the human genome is searched, it is likely that additional genes containing FUSE or FUSE-like sequences will be found. Sequencing of FBP1 targeted DNA fragments, such as by chromatin immunoprecipitation coupled DNA sequencing (ChIP-seq) technique, should confirm the result of genome-wide searching and reveal the genes subjected to FBP1 binding and mediated transcriptional regulation.

FBP1 AS AN RNA-BINDING PROTEIN

The KH domain is important for protein binding to single stranded DNA or RNA.²⁸ Each FBP family member contains four KH domains and likely binds to RNA in addition to single-stranded DNA. Indeed, FBP2 has been reported to bind to a number of mRNA species.³⁷ As FBP1 is highly homologous to FBP2 in protein sequence, it is not surprising that FBP1 competes with FBP2 for binding to cellular mRNA or viral RNA. Such interactions contribute to mRNA stabilization, protein translation and viral replication. Table 2 summarizes the genes regulated by FBP1 due to RNA binding.

Table 2.

Summary of FBP1 as a RNA-binding protein

Target	Binding site	Function	Experimental model	Reference
GAP34	3′-UTR	GAP34 mRNA degradation	Rat brain	9
p21	3′-UTR	p21 mRNA degradation	Hep3B cells	19,46
Cox-2	3′-UTR	Unclear	HeLa cells	54,66
p27	IRES in 5′-UTR	Enhances protein translation	MCF7 cells	15
NPM	3′-UTR	Inhibits protein translation	Tsc−/− p53−/− MEFs	61
HCV	3′-UTR	Promotes viral replication	MH14 cells	11
JEV	5′-UTR	Suppresses protein translation/viral replication	HeLa cells NT2 cells	8
EV71	IRES in 5′-UTR	Enhances protein translation/viral replication	SK-N-MC, SF268, RD and Vero cells	61

Abbreviations: FBP1, far upstream binding protein 1; HCV, hepatitis C virus; IRES, internal ribosomal entry site; JEV, Japanese encephalitis virus; NPM, nucleophosmin; UTR, untranslated region.

FBP1 has a role in neural development through post-transcriptional regulation of growth-associated protein 43 (GAP43), a membrane-associated phosphoprotein critical for axonal growth and regeneration.⁵¹ FBP1 binds to a 26-nucleotides pyrimidinerich region within the 3′-UTR of GAP43 mRNA about 300 nucleotides downstream from the stop codon.⁹ This 3′-UTR region has been shown to contribute to the instability of the mRNA species⁵² and it is possible that FBP1 binding promotes GAP43 mRNA degradation.⁹

There are three AU-rich elements (AREs) in the 3′-UTR of p21 mRNA. It seems that the stability of p21 mRNA is subjected to FBP1 regulation through these AREs.⁵³ It has been reported that p21 mRNA increased >fivefold in Hep3B cells when the FBP1 gene was knocked down.¹⁹ This result is apparently incompatible with the finding that the p21 promoter contains multiple FUSE-like elements, which were activated by FBP1 binding in an in vitro reporter assay.¹⁹ This discrepancy points to the possibility that FBP1 suppression of p21 mRNA stability is predominant over its role in p21 gene transcription.

FBP1 has been shown to bind to the AREs in the 3′-UTR of Cox-2 mRNA.⁵⁴ An in vitro binding assay indicated that FBP1 bound to a conserved region 1 (CR1) in the 3′-UTR of Cox-2 mRNA.^54,55 This CR1 region has six AREs with a core sequence of AUUUA. AREs are known to confer the stability of numerous mRNA species including Cox-2 and mutations in CR1 destabilize Cox-2 mRNA. However, CR1 mutants remained capable of binding to FBP1.⁵⁵ FBP1 from Hela cell extracts appeared to show little specificity for Cox-2 ARE.⁵⁴ These call into question about the biological significance of the observed FBP1 binding to CR1 in vitro, if the binding is specific.

A recent study by Zheng and Miskimins¹¹ showed that FBP1 promoted p27^Kip translation via an internal ribosomal entry site (IRES) in the 5′-UTR of p27^Kip mRNA in human breast cancer MCF7 cells. In this study, FBP1 was found to bind specifically to the U-rich region of p27 5′-UTR, a putative IRES, and to promote the recruitment of ribosomes. As a result, ectopic expression of FBP1 protein caused an increased abundance of p27^Kip protein, whereas siRNA knockdown of FBP1 reduced p27^Kip protein levels.¹¹ It seems that FBP1 binds to p27^Kip IRES through its N-terminal and central domains, as deletion of either domain led to a reduction of p27^Kip IRES activity as measured by a bicistronic reporter assay. In contrast, truncation of the C-terminus of FBP1 did not affect the reporter activity of p27^Kip IRES.¹¹ As the central domain contains KH motifs mediating single-stranded DNA or RNA binding, these motifs may be the key for p27^Kip IRES binding.

Nucleophosmin (NPM), a nucleolar phosphoprotein also known as numatrin, is negatively regulated by FBP1 through mRNA binding.¹⁰ NPM, a regulator of cell proliferation and genomic instability due to association with nucleolar ribonucleoproteins and binding to single-stranded DNA or RNA, is considered an oncogene or tumor suppressor depending on the cell type and expression levels.^56–58 Heterozygous deletion of NPM in mice resulted in early onset of lymphoma and an increased frequency of myeloid malignancy.^57,58 Apparently FBP1 binds to the 3′-UTR of NPM mRNA and inhibits NPM mRNA from being translated into the protein by interfering with translation initiation. Knocking down FBP1 by siRNA increased the recruitment of NPM mRNA to the polysomes and elevated translation of NPM protein in Tsc^−/− p53^−/− MEF cells.¹⁰ In contrast, overexpression of FBP1 led to a redistribution of NPM mRNA from its association with polysomes to the mono-ribosomes, hinting at a decrease of NPM protein translation.¹⁰

FBP1 has been reported to mediate replication of several retroviruses in human cells through RNA binding. Hepatitis C virus (HCV) is a major etiological factor of hepatocellular carcinoma in infected individuals. HCV replication in host cells requires the binding of FBP1 to poly(U) tract within the poly(U/UC) region of 3′-UTR of HCV RNA.⁵⁹ Evidently, such binding is important for HCV RNA replication as knocking down FBP1 significantly reduced HCV replication, whereas overexpression of FBP1 had an opposite effect.⁵⁹

In contrast to the positive role of FBP1 in HCV replication, FBP1 suppresses infection of Japanese encephalitis virus (JEV), the leading cause of encephalitis in Asia.⁸ Such suppression results from FBP1 binding to UTRs, primarily the 3′-UTR of JEV RNA.⁸ JEV production was enhanced in FBP1 knockdown HeLa cells and the opposite was observed in cells overexpressing FBP1.⁸ JEV infection results in cytoplasmic localization of FBP1, detectable in the fraction of cells undergoing apoptosis.⁶⁰ However, there is no mechanistic study on how FBP1 affects viral replication and as such the role of FBP1 in viral replication remains controversial.

A better understanding of how FBP1 promotes viral replication came from a recent study on Enterovirus 71 (EV71), the retrovirus causing hand-foot-mouth disease as well as severe neurological complications in some patients.⁶¹ This study showed that the KH3 and KH4 domains of FBP1 interacted with the IRES in the 5′-UTR of EV71 viral core RNA.⁶¹ The IRES promotes translation of the viral proteins in the host cells when global translation is inhibited owing to viral infection.⁶² By competing with inhibitory IRES trans-acting factor (ITAFs), notably FBP2, FBP1 promotes IRES-mediated translation of EV71 proteins.⁶¹ In an in vitro reporter assay using EV71 5′-UTR, addition of recombinant FBP1 protein enhanced IRES-mediated translation by 1.8-fold.⁶¹ Conversely, knockdown of FBP1 leads to a decrease in the EV71 5′-UTR reporter activity and EV71 replication.⁶¹ These studies indicate that translation is a key step in FBP1-mediated viral replication.

SUBCELLULAR LOCALIZATION OF FBP1

As a transcription regulator, FBP1 is primarily located in the nucleus.^24,50,60 However, some of its biological functions such as RNA stabilization and IRES-mediated protein translation require cytoplasmic localization. The nuclear localization of FBP1 protein is driven by three nuclear localization signals (NLSs), one in each of its N-terminal, central and C-terminal domains (Figure 1). The N-terminal domain contains a canonical bipartite NLS located at amino acids 63– 78. When fused with GFP, this NLS is sufficient for driving nuclear localization of GFP.⁶³ The residues 366–386 in the central domain form an atypical NLS. Two isoleucines located in 378 and 379 of the atypical NLS are critical for nuclear transport, as mutations of these two residues to prolines completely abolished the NLS.⁶³ The NLS in the C-terminal domain is also atypical yet less well-defined. It appears that the residues between 576 and 644 contain the NLS, as truncation of this sequence caused a 95% reduction of nuclear localization of FBP1.⁶³ Following FBP1 protein synthesis in the cytosol, the nuclear distribution of FBP1 is probably mediated through importin α5, as a direct interaction of FBP1 with importin α5 has been reported.⁶⁴

Under normal physiological conditions, FBP1 primarily localizes in the nucleus.^24,50,60 With stress such as heat shock, viral infection and oxidative stress either nuclear import of FBP1 has been stopped or FBP1 translocates from the nuclei to the cytoplasm, resulting in cytoplasmic accumulation. In the cytoplasm, FBP1 may be recruited to the stress granules, where mRNA and translational machinery are temporarily sheltered when cells are stressed, such as during JEV infection.⁶⁵ Cytoplasmic translocation of FBP1 was observed in early phase of JEV or EV71 infection, when FBP1 was exported from the nuclei and bound to the core RNA of these viruses in the cytoplasm.^8,61 This cytoplasmic localization of FBP1 is likely essential for mediating de novo viral protein translation.

Cytoplasmic localization of FBP1 has been observed in apoptotic cells. FBP1 was detected in the cytoplasm of 30% of cells treated with TNF-α plus cycloheximide, which induce apoptosis.⁶⁰ Such localization is related to the fact that FBP1 is a substrate of caspase-3 and -7, which cleave FBP1 at the DQPD consensus site located in the N-terminal NLS. FBP-1 may be cleaved by caspase-3 and -7 in vitro and in vivo.^60,66 Caspase-7 cleaves FBP1 but not FBP2 and FBP3.⁶⁰ The DQPD consensus sequence appears to be unique to FBP1, as the corresponding sequence in FBP2 is DQPE and FBP3 lacks a related sequence in the region. Caspase cleavage of FBP1 apparently destroys the N-terminal NLS, leading to the cytoplasmic retention of truncated FBP1.⁶⁰

With JEV infection, cytoplasmic FBP1 was observed in the early stage of infection with high virus titers.⁸ In the later phase of JEV infection when cells were able to produce the virus, FBP1 was exclusively nuclear.⁸ As cytoplasmic localization of FBP1 corresponds to apoptosis, a possible explanation is that FBP1 suppresses JEV replication by inducing apoptosis of infected cells at the onset of high titer infection as a protective mechanism.

The signal responsible for the subcellular relocalization of FBP1 remains unclear and the biological significance of cytosolic FBP1 remains unresolvded with regard to the process of apoptosis. A study with A549 lung carcinoma epithelial cells may have provided some clues on the significance of FBP1 cytoplasmic localization. Treatment of A549 cells with transforming growth factor-β stimulated FBP1 migration from the nucleus to the cytoplasm along with FBP2, hnRNP-L and hnRNP-H1.⁶⁷ As hnRNP-L and hnRNP-H1 often serve as carriers for transporting newly synthesized RNA from the nucleus to the cytosol, this result implies that FBP1 may have a role in the nuclear export of RNAs. In addition, the cytoplasmic localization of FBP1 may facilitate FBP1 participation in selective protein translation during stress.

POST-TRANSLATIONAL REGULATION OF FBP1

As an important regulator of the c-Myc proto-oncogene, the level of FBP1 protein is expected to be tightly regulated. Although it is not fully characterized yet, FBP1 gene expression is likely regulated by specific signaling pathways. The transcription or translation of the FBP1 gene appears to be downstream of the mTOR pathway, because rapamycin treatment of mouse embryonic fibroblast cells resulted in an increased level of FBP1 protein.¹⁰ As rapamycin is known for inhibiting Cap-dependent protein translation, but not Cap-independent,⁶⁸ this result suggests that translation of FBP1 protein itself may be regulated by Capindependent mechanisms, for example, IRES, during rapamycin treatment.

A few chemicals or treatments have been reported to alter the expression of FBP1 protein. Atrazine, a widely used pesticide, induced elevation of FBP1 protein in MCF7 human breast cancer cells.⁶⁹ Polychlorinated biphenyl, a commonly studied environmental toxicant, also caused elevation of FBP1 protein in MCF7 cells.⁶⁹ A similar finding was reported with human microvascular endothelial cells.⁷⁰ Although the mechanism by which these chemicals cause elevation of FBP1 protein is not known, these findings imply a role of FBP1 in cellular response to environmental stress.

Proteolysis has an important role in determining the level of FBP1 protein. FBP1 protein stability is regulated by p38/JTV-1, a protein isolated by chance when Nicolaides et al.⁷¹ were screening a cDNA library for a longer fragment of the Post-Meiotic Segregation 2 gene, mutations of which lead to impaired DNA repair. Later, p38/JTV-1 was identified as a factor associated with a protein complex consisting of several different aminoacyl-tRNA synthetases and therefore was named AIMP2, for aminoacyl-tRNA synthetase-interacting multifunctional protein type 2.⁷² FBP1 interacts with p38/JTV-1 physically and this interaction appears to be specific for the C-terminus of FBP1, as p38/JTV-1 did not bind to FBP2 nor FBP3, nor the N-terminus of FBP1.²⁰ Without altering FBP1 mRNA level or the overall rate of FBP1 protein synthesis, p38/JTV-1 promoted FBP1 degradation via ubiquitination. Inhibition of the 26S proteasomes prevented p38/JTV-1 from inducing FBP1 protein degradation.²⁰

Transforming growth factor-β induces p38/JTV-1 expression in various cancer cell lines and consequently leads to the reduction of FBP1 protein levels.²⁰ An in vivo study showed that knocking out p38/JTV-1 increased the levels of FBP1 protein and c-Myc expression in fetal lungs, leading to a defect in lung differentiation and development of respiratory distress syndrome.²⁰ Interestingly, p38/JTV-1 is a downstream target of the c-Myc transcription factor, representing another layer of the negative feedback loop for finetuning c-Myc gene expression.⁷³

The mechanism by which p38/JTV-1 promotes FBP1 protein degradation remains to be elucidated. FBP1 protein is upregulated in association with Parkinson’s disease,⁷⁴ an illness often linked to mutation or functional loss of the Parkin gene. FBP1 protein elevation was observed in the brain stem and cortex tissues of Parkin knockout mice.⁷⁴ As Parkin is an E3 ubiquitin ligase, it is possible that p38/JTV-1 serves as a scaffold protein that recruits Parkin, which in turn ubiquitinates FBP1, causing proteasomemediated degradation of FBP1 protein.

Ubiquitination of FBP1 at certain positions seems to affect its binding to FUSE. A recent study indicated that FBP1 is subjected to de-ubiquitination by ubiquitin-specific protease 22 (USP22).⁷⁵ Interestingly, USP22 mediated deubiquitination did not appear to change FBP1 protein stability and abundance, as judged by USP22 depletion in HeLa cells. Instead, the accumulation of Lys-63 ubiquitinated FBP1 owing to USP22 depletion, resulted in a reduction of FBP1 binding to FUSE.⁷⁵ It is possible that Lys-63 ubiquitination masks the binding domain of FBP1, or affects FBP1 binding to FUSE indirectly through altering binding partners.

FBP1 AS A CANCER BIOMARKER

It is not surprising that FBP1 is implicated in multiple types of cancers as summarized in Table 3. Although regulating c-Myc gene transcription is an important aspect of FBP1 in carcinogenesis, FBP1 may exhibit functions independent of c-Myc in carcinogenesis, such as facilitating replication of HCV, an etiological factor of liver cancer in many patients,⁵⁹ as well as inhibiting expression of p21 through mRNA binding.¹⁹ Mounting evidence suggests that FBP1 is a new proto-oncogene, as there is a positive correlation between FBP1 expression and the development of tumors. Although not all reports examined the c-Myc level in tumor tissues, Weber et al.⁷⁶ showed that elevation of FBP proteins correlated with c-Myc expression in clear cell renal cancer but not in bladder or prostate cancer. This indicates that aberrant expression of FBP1 may have c-Myc independent functions in carcinogenesis.

Table 3.

Role of FBP1 in human malignancies

Malignancy	Alterations	Downstream target	Reference
Oligodendrogliomas	FBP1 mutation	NA	16
Non-small cell lung cancer	FBP1↑	Stathmin 1 ↑	77
Breast	FBP1↑	NA	69,97
Clear cell renal cancer	FBP1↑	c-Myc↑	76
Liver	FBP1↑	p21↓ p15↓, Cyclin D2 ↑	15,19
Liver	FBP1↑	Stathmin 1 ↑	15,17
Colon	FIR truncation	c-Myc↑	12,18
Bladder	FBP1↑	NA	76
Prostate	FBP1↑	NA	76

Abbreviations: FBP1, far upstream binding protein 1; FIR, FBP-interacting repressor.

The low background expression in normal tissues and high abundance in tumor tissues make FBP1 a potential biomarker for these types of cancers.^18,19,76,77 The first document of FBP1 involvement in cancer was established in Xeroderma Pigmentosum, a hereditary disease with genetic defects in nucleotide-excision repair due to mutations in DNA repair enzymes. Individuals with Xeroderma Pigmentosum are highly susceptible to ultra violet damage and prone to skin cancer.⁷⁸ As described in the section ‘FBP1–FUSE–FIR axis for transcriptional regulation of c-MYC’, FIR inhibition of FBP1 activity was defective in cells from Xeroderma Pigmentosum individuals, resulting in an aberrant expression of the c-Myc gene.³¹

Loss of FIR inhibition of FBP1 is also observed in colorectal cancer. A splicing variant of FIR lacking exon 2 has been found in colorectal cancerous tissue.¹² This truncated variant lacks the N-terminus, and thus is unable to inhibit FBP1 activity. Importantly, this FIR slicing variant appears to be specific for colorectal cancer, as normal neighboring tissues or blood cells from the same patients do not express this variant.^12,18 In addition to its inability to inhibit FBP1, this truncated variant likely acts as a dominant negative mutant and prevents the wild-type FIR from inhibiting FBP1.¹²

The human FBP1 gene has been mapped to chromosome 1p31.1, a region frequently amplified in association with osteosarcomas.⁷⁹ A proteomic study revealed an elevated FBP1 protein level in osteosarcoma cell line U2OS.¹³ A similar approach used by Engidawork et al.⁸⁰ found overexpression of FBP1 in a medulloblastoma cell line, DAOY. FBP1 expression was found to be significantly elevated in 83% of 109 human hepatic carcinoma samples,¹⁹ in consistency with other studies using hepatocarcinoma cell lines and tissues.^15,17,19 However, whether FBP1 elevation correlates with c-Myc expression appears to be controversial. Although Zubaidah et al.¹⁵ showed that both FBP1 and c-Myc were elevated in hepatocarcinoma tissues, other studies did not find such correlation.^17,19 If FBP1 protein level is examined in additional cancerous tissues, it is likely that the list of cancer types with FBP1 elevation will be expanded. FBP1 elevation may be independent of alterations in c-Myc in some tumor tissues.

FBP1 AT THE CELLULAR LEVEL: INHIBITING APOPTOSIS, PROMOTING CELL PROLIFERATION AND ENHANCING CELL MIGRATION

Carcinogenesis is characterized by the hallmarks of ‘self-sufficiency in growth, insensitivity to growth-inhibitory signals, evasion of programmed cell death, unlimited replicative potential, sustained angiogenesis, tissue invasion, and metastasis’.⁸¹ FBP1 has an important role in resistance to apoptosis, stimulating cell proliferation and promoting cell migration, as summarized in Figure 3.

Figure 3.

FBP1 involvement in carcinogenesis. FBP1 is a critical regulator of the proto-oncogene c-Myc, which has an important role in cell proliferation, survival and metastasis. In addition, FBP1 promotes cell proliferation through inhibition of p21 or nucleophosmin, and promotes cell migration by increasing Stathmin 1 and Stathmin 3, independent of c-Myc transcription. FBP1 is also required for HCV replication, the primary cause of hepatocellular carcinoma. FIR inhibits FBP1 activity through protein–protein interaction. There is a negative feedback loop for maintaining c-Myc transcription via upregulation of p38/JTV-1, which promotes FBP1 degradation.

Apoptosis

There is evidence that FBP1 acts as an inhibitor of apoptosis. Whereas FBP1 is found in the cytoplasm of apoptotic cells and such localization correlates to cell death, expression of the C-terminus of human FBP1 suppressed CED4-induced apoptosis in fission yeast S. pombe.¹⁹ Overexpression of FBP1 resulted in inhibition of apoptosis in human colorectal carcinoma RKO cells.¹⁹ Consistent with a role of FBP1 in inhibition of apoptosis, ectopic expression of a mutated form of FBP1, which may no longer be cleaved by caspases due to a mutation of the consensus sequence from DQPD to DQPA, imposed resistance against cytoplasmic localization and apoptosis in MCF-7 breast carcinoma cells.⁶⁰ Knocking down FBP1 enhanced apoptosis of Hep3B cells in response to DNA damage, such as ultra violet irradiation and doxorubicin treatment.¹⁹ Overexpression of FIR, the inhibitor of FBP1, induced apoptosis in HeLa cells.¹⁸ These lines of evidence suggest that hyperactive FBP1 and its nuclear localization somehow trigger a cell survival response.

Cell proliferation

Through regulating c-Myc expression, FBP1 has an important role in cell proliferation. c-Myc is a master switch of cellular proliferation and this topic has been reviewed extensively elsewhere.^1,3,82 Loss of c-Myc caused a 12-fold reduction of cyclin-dependent kinase (Cdk)4/6 activity and cell cycle arrest at the G₁ phase.⁸³ Among a long list of genes under the control of c-Myc are proliferating cell nuclear antigen (PCNA), Cdc2 and Rb-binding protein.⁸⁴ In addition, FBP1 may regulate cell-cycle progression through c-Myc-independent mechanisms. For example, the p21 gene encodes an inhibitor of cyclin-dependent kinases and inhibits cell proliferation in general.⁸⁵ As discussed in section ‘subcellular localization of FBP1’, p21 mRNA is downregulated by FBP1 probably through AREs at the 3′-UTR.¹⁹ Furthermore, FBP1 may also regulate p21 protein abundance through NPM, as FBP1 negatively regulates NPM expression and NPM stabilizes p21 protein through protein–protein interaction.⁸⁶ Such interaction blocks ubiquitination and subsequent degradation of p21 protein. A similar mechanism of NPM-mediated protein stabilization was proposed for p19^ARF, a tumor suppressor inhibiting cell proliferation.^87,88 Therefore, FBP1 activity is expected to result in reduction of p21 and p19^ARF. However, ectopic expression of FBP1 may also induce the cell cycle inhibitor p27^Kip by promoting its translation via IRES in the 5′-UTR.¹¹ As IRES-mediated translation of p27^Kip was reported in nonproliferating cells⁸⁹ and IRES-mediated protein translation usually represents a stress response mechanism, this result suggests that FBP1 may participate in the stress response and exhibit different functions, such as inhibition of cell proliferation rather than facilitating cell proliferation, depending on whether or not cells are stressed.

Metastasis

FBP1 may promote tumor metastasis via the c-Myc transcription factor.^90,91 Overexpression of human pituitary tumor-transforming gene PTTG has been implicated in metastasis of multiple tumors in a c-Myc-dependent manner.⁹² c-Myc also promotes metastasis by transcriptional activation of an epithelial-to-mesenchymal transition-associated gene SNAIL,⁹³ a gene known to be involved in tumor metastasis.⁹⁴ In support of this, knockdown of c-Myc gene with shRNA reduced cell invasion in vitro and metastasis in vivo.⁹⁰

In addition to regulating c-Myc transcription, FBP1 may participate in tumor metastasis through the regulation of Stathmin 1 and Stathmin 3, two microtubule-destabilizing proteins elevated in multiple types of malignancies, including leukemia, lymphoma, breast, lung, ovarian and liver cancers.^{14,77,95–97} Singer et al.⁷⁷ examined the expression of Stathmin 1 and Stathmin 3 in nonsmall cell lung carcinoma and found overexpression of Stathmin 1 or Stathmin 3 in ~80% or 50% of the samples, respectively. Importantly, both proteins were highly expressed at the invasion front, consistent with their roles in cell motility and metastasis. The same study showed that there was a causative relationship between FBP1 and Stathmin 1 or 3, as knockdown of FBP1 in two non-small cell lung carcinoma cell lines caused a reduction of both proteins, while ectopic expression of FBP1 had the opposite effect.⁷⁷ Stathmin 1 was expressed in ~80% of high-grade serous ovarian carcinoma tissues while absent in benign fallopian tube epithelium. These lines of evidence support a role for FBP1 in tumor metastasis via the regulation of Stathmin 1 and Stathmin 3.⁹⁵ Indeed, FBP1 together with FBP2 may promote cancer cell metastasis through coordinated expression of Stathmins.¹⁷

CONCLUSION

FBP1 expression is low in most normal tissues, but is significantly elevated in various cancerous tissues (Table 3). Increasing evidence suggests that FBP1 is an emerging proto-oncogene that has a role in carcinogenesis. The fact that FBP1 controls the transcription of c-Myc and that there is a correlation between FBP1 and c-Myc expression in several cancer types suggests a role of FBP1 in carcinogenesis via c-Myc-dependent manner.

On the other hand, the role of FBP1 in carcinogenesis can be c-Myc independent. At the cellular level, knocking down FBP1 in Hep3B cells failed to eliminate the expression of c-Myc gene.¹⁹ Two studies reported the lack of correlation of FBP-1 with c-Myc expression in hepatocarcinoma tissues.^17,19 c-Myc-independent targets such as USP29, p21, p27, NPM and Stathmins may be important mediators of FBP1 in carcinogenesis.

The importance of FBP1 in carcinogenesis points to the potential of using FBP1 as a biomarker and a therapeutic target for some cancer types. Thus, finding FBP1 inhibitors has practical implications. Huth et al.⁹⁸ have designed several small molecules targeting the interaction between the KH3/4 domains and DNA sequence of FUSE, based on the structural conformation of FBP1 protein from NMR studies. These benzoylanthranilic acid based compounds bind to FBP1 and disrupt FBP1 binding to a DNA probe in vitro.⁹⁸ Although the implication of these compounds for cancer therapy remains to be established with additional assays including in vivo validations, small molecule pharmacological agents may be designed to mimic the function of FIR in blocking the interaction between FBP1 and TFIIH. Such inhibitors are expected to block c-Myc gene expression and ultimately reduce cell proliferation.

As an RNA-binding protein, the sequence or characteristics of its RNA targets has yet to be defined. The binding affinity to RNA versus single-stranded DNA may differ, contributing to altered functionality of FBP1 in response to changing conditions. The most dramatic effects by FBP1 have been observed when cells are stimulated with growth factors, or responding to stress, such as chemical toxicity and viral infection. Liu et al.⁵⁰ indicated that FBP1 is important for oxidative stress responses. Such a scenario involves the stabilization of p53 tumor suppressor,⁵⁰ which is important for cell cycle arrest and apoptosis. Consistent with such negative regulation of cell proliferation, FBP1 upregulates p27^Kip translation through an IRES-mediated mechanism,¹¹ which usually occurs during the selective translation of a subset of genes under stress conditions when global protein synthesis is halted.^62,99 Moreover, FBP1 translocation from the nucleus to the cytoplasm has been observed in apoptotic cells.⁶⁰ These observations suggest that FBP1 may switch its priority from regulating c-Myc gene transcription during cell proliferation to its function of turning on growth inhibitory or apoptotic genes, such as p53 and p27^Kip, thereby inhibiting cell proliferation and promoting apoptosis under stress conditions. Understanding the signaling pathway that transmits stress to altered functionality of FBP1 becomes important for developing new therapeutic approaches through modulation of FBP1.

Cells have adopted multiple layers of feedback mechanisms for regulating FBP1 function. We have discussed p38/JTV-1-, Parkin- and USP22-mediated ubiquitination and degradation of FBP1 protein, in addition to the role of FIR as the negative regulator of FBP1 activity. Whereas the mechanism of elevated expression of FBP1 in cancerous tissues remains to be elucidated, mutations of FBP1 occur in association with carcinogenesis. A recent study showed that the majority of malignant oligodendroglioma tissues displayed a loss of heterozygosity in chromosome 1p where the FBP1 gene resides. Multiple point mutations caused inactivation of gene products encoded by DNA in chromosome 1p, including FBP1.¹⁶ This finding adds to the complexity of the role of FBP1 in carcinogenesis, as loss of FBP1 correlates with carcinogenesis in the case of oligodendroglioma. A long list of questions remains to be answered in order to solve the controversies about FBP1 in carcinogenesis. For example, how is FBP1 transcription shut down upon cell differentiation? What causes FBP1 upregulation in some cancers? Which signaling pathway controls FBP1 expression and subcellular relocalization in cells responding to stress? Does FBP2 or FBP3 have a functional overlap with FBP1 in carcinogenesis? Addressing these questions will pave the path for a better understanding of FBP1 and this gene family in carcinogenesis.