Targeting the Y/CCAAT box in cancer: YB-1 (YBX1) or NF-Y?

Abstract

The Y box is an important sequence motif found in promoters and enhancers containing a CCAAT box – one of the few elements enriched in promoters of large sets of genes overexpressed in cancer. The search for the transcription factor(s) acting on it led to the biochemical purification of the nuclear factor Y (NF-Y) heterotrimer, and to the cloning – through the screening of expression libraries – of Y box-binding protein 1 (YB-1), an oncogene, overexpressed in aggressive tumors and associated with drug resistance. These two factors have been associated with Y/CCAAT-dependent activation of numerous growth-related genes, notably multidrug resistance protein 1. We review two decades of data indicating that NF-Y ultimately acts on Y/CCAAT in cancer cells, a notion recently confirmed by genome-wide data. Other features of YB-1, such as post-transcriptional control of mRNA biology, render it important in cancer biology.

Facts

• Precise sequence of the CCAAT element.

• The nuclear factor Y (NF-Y) transcription factor (TF) binds to the CCAAT box.

• Lack of solid evidence of DNA-binding specificity of the Y box-binding protein 1 (YB-1) oncogene.

Open Questions

• How NF-Y turns on ‘cancer' genes with CCAAT boxes?

• What is the exact role of YB-1 in the regulation of gene expression?

• Is there an interplay between NF-Y and YB-1?

Y and CCAAT: Two Names, One Entity

The Y box – consensus CTGATTGGT/CT/C – was identified three decades ago as a DNA element conserved in promoters of major histocompatibility complex (MHC) class II genes.¹ In vitro transcription, transfections and transgenic mice experiments with Y box-mutated promoters showed its crucial role, along with neighboring conserved sequences, for the coordinated and tissue-specific expression of these genes.² In reality, the Y box contains an inverted CCAAT sequence – ATTGG underlined above – which had previously been identified in the globin and ovalbumin promoters along with the TATA box.^{3, 4} Importantly, the CCAAT/ATTGG pentanucleotide was shown to be required for transcriptional activation (TA).^{5, 6, 7, 8, 9, 10, 11} In essence, the Y and CCAAT boxes are functionally equivalent. Thereafter, the importance and widespread distribution of the Y/CCAAT box, as precisely defined by the initial genetic and biochemical experiments, has been substantiated by unbiased genomic studies. The exact sequence and frequency of Y/CCAAT in promoters was assessed using different bioinformatic tools: several labs reported the identification of Y/CCAAT as over-represented in human promoters and enhancers by searching with available matrices,^{12, 13, 14, 15, 16, 17, 18, 19} and two studies searching for unbiased ‘words' enriched within promoters identified Y/CCAAT and precise flanking motifs.^{20, 21}

Y/CCAAT is Enriched in Promoters of ‘Cancer' Genes

Analysis of transcriptome profiles during cellular transformation identified the Y/CCAAT box as over-represented in promoters of genes overexpressed in diverse types of cancers, breast, colon, thyroid, prostate and leukemias.^{22, 23, 24, 25, 26, 27, 28, 29} Treating cells with cytotoxic drugs, or overexpressing growth suppressors, led to the downregulation of genes with CCAAT in their promoters.^{30, 31} However, it is important to remark that these exercises were performed with matrices included in TRANSFAC and JASPAR; hence, they could be highly biased, as the lists of transcription factor-binding sites (TFBS) present in databases certainly do not recapitulate all possible TF-binding sequences. More tellingly, de novo motif discovery methods that allow unbiased identification of sequence logos showed that promoters of genes, specifically overexpressed in tumors, are significantly enriched in Y/CCAAT elements.^{32, 33, 34} This indicates that Y/CCAAT is of importance in the overexpression of cancer genes in tumors. Therefore, the TFs recognizing this element are likely relevant for the process of cellular transformation.

The search for the TF(s) binding to Y/CCAAT started in the late 1980s, leading to the apparent identification of more than one activity.³⁵ We will not discuss here C/EBP (CCAAT enhancer-binding proteins) and CTF/NF1 (CCAAT TF/nuclear factor 1), two bona fide sequence-specific TFs originally associated with CCAAT binding, as they were later unambiguously shown to have different sequence specificities.^{36, 37, 38} Two additional proteins have been ‘battling' over the Y/CCAAT ground for over two decades, NF-Y and YB-1. We review here the work of the past 20 years pertinent to the specific role of the two factors in Y/CCAAT activation.

NF-Y and CCAAT

NF-Y was originally shown to bind to the Y box of the MHC class II Ea promoter using electrophoretic mobility shift assays (EMSAs).^{35, 39} Later, it became clear that it was identical to CBF (CCAAT-binding factor) shown to interact with collagen promoters,⁶ CP1 (CCAAT protein 1) binding to globin promoters⁷ and EFI binding to the Rous sarcoma virus long terminal repeat (RSV LTR).⁴⁰ It was soon realized that this activity was ubiquitously expressed, composed of multiple subunits and conserved in yeast, where it is called HAP complex.⁴¹ The complex was biochemically purified using conventional affinity purification with oligomerized Y/CCAAT oligos. Initially, two subunits were characterized.^{42, 43, 44, 45} Eventually, NF-Y was shown to be a heterotrimer composed of NF-YA, NF-YB and NF-YC (Figure 1), whose genes are found in all eukaryotes (they are termed HAP2/3/5 in yeast). NF-YB and NF-YC have histone fold domains (HFDs) similar to core histones H2A/H2B, composed of three α-helices separated by short loop/strand regions;⁴⁶ dimerization of HFDs is required for association with NF-YA, which provides sequence specificity to the complex.

Figure 1.

Scheme of NF-Y subunits and YB-1

Several lines of evidence indicate that NF-Y activates transcription through sequence-specific binding to Y/CCAAT:

• Saturation mutagenesis studies on different Y/CCAAT boxes, using in vitro EMSAs, clearly showed that NF-Y binding is absolutely dependent on each of the five core nucleotides and pointed at important flanking nucleotides – 2 bp at the 5′ end and 3 bp at the 3′ end – as, indeed, found in the original Y box.^{6, 7, 35, 39, 40} Unbiased SELEX assays further confirmed the specificity of NF-Y for a 10 bp stretch.⁴⁷

• In vitro transcription and transfections of promoters mutated in the NF-Y-binding sites showed a perfect correlation between the decrease or abolition of NF-Y binding and the decrease of functional activity.^{6, 7, 8, 9, 10} In vitro transcription assays with purified NF-Y, antibodies and recombinant proteins showed an effect of NF-Y on transcriptional initiation, and re-initiation, in various promoters.^{6, 7, 11, 48, 49, 50, 51}

• The development of specific antibodies, used in supershift EMSAs in vitro⁴⁸ and in chromatin immunoprecipitation (ChIP) assays in vivo,⁵² enabled different labs to verify that the band observed in EMSAs, and the protein bound in cells, was indeed NF-Y. The initial in vitro experiments lead to the definition of a first NF-Y positional sequence frequency matrix (PSFM),⁵³ which was soon incorporated into the TRANSFAC and JASPAR databases (Figure 2). Note that the bioinformatic analyses of motifs in promoters of cancer genes mentioned above also retrieved the NF-Y logo.

• The use of dominant-negative NF-YA vectors⁵⁴ and, more recently, the inactivation of NF-Y subunits by small interfering RNA (siRNA) or short hairpin RNA (shRNA) interference allowed the in vivo confirmation that a CCAAT promoter is regulated by NF-Y (reviewed in Dolfini et al.⁵⁵).

• Genomic analysis by ChIP-on-Chip^{56, 57, 58, 59} and ChIP-Seq⁶⁰ confirmed that NF-Y binds to Y/CCAAT in vivo. These experiments further refined the NF-Y PSFM (Figure 2). In summary, a very robust set of data leads to the accepted conclusion that NF-Y regulates gene expression through specific binding of the Y/CCAAT box.

Figure 2.

Y/CCAAT ‘evolution' over time

YB-1 and CCAAT?

YB-1 was identified through screening of a phage expression library using a multimerized Y box oligo from the MHC class II DRa promoter.⁶¹ DRa is the human homolog of the mouse Ea, which was the starting point for the biochemical identification of NF-Y.³⁵ A similar strategy was reported in the cloning of rat EFIa, using a CCAAT oligo from the RSV LTR,⁶² Xenopus FRG-1⁶³ and chicken YB-1.⁶⁴ This technique identifies phages producing a single polypeptide, and it could not have been used for NF-Y, as all three of its subunits are required for DNA binding. YB-1 was shown to be a protein with a known nucleic acid-binding domain, termed CSD (cold-shock domain), which is highly conserved in eukaryotes and prokaryotes (Figure 1 and Mihailovich et al.⁶⁵).

Intriguingly, two cloning manuscripts published in the same period were at odds with the interpretation that YB-1 was a classic sequence-specific TF: (i) YB-1 was identified in screenings of expression libraries with a completely unrelated oligo, the W box, also from the DRa promoter, but sharing no sequence similarity to Y/CCAAT;⁶⁶ these authors first provided evidence that YB-1 binds well to single-stranded (ssDNA) and, surprisingly, to abasic DNA. The chicken YB-1, subsequently cloned, also showed DNA binding with little dependence on the presence of a Y/CCAAT box.⁶⁴ (ii) The Xenopus FRG-1 gene was cloned with expression libraries⁶⁷probed with antibodies directed against p54/p56,⁶⁸ one of the subunits of the mRNA-binding complex biochemically identified in the 1970s, and widely studied for its role in mRNA translation.⁶⁸ Thereafter, research on YB-1 proceeded, by and large, in two parallel fields: its role in the control of mRNA biology (splicing, stability, translation), in which it has taken a center-stage position (Bouvet and Wolffe⁶⁹; reviewed in Evdokimova et al.⁷⁰), and its role in the control of transcriptional initiation, which has been more controversial.

NF-Y is the Sequence-Specific CCAAT Factor

We list below a comparison of features supporting NF-Y and YB-1 as bona fide sequence-specific TFs.

DNA affinity and sequence specificity

When recombinant YB-1 was tested in in vitro binding assays, which are more sensitive and specific than the Southwestern blots used in the initial cloning papers, the protein was shown to bind to RNA and DNA oligos.⁷¹ With SELEX and Chip methods, it was established that the strings of nucleic acids preferred by YB-1 are GGGG (ssDNA), CACC/T (double-stranded DNA (dsDNA)) and AACAUC (RNA). None of these motifs resemble to a canonical Y/CCAAT box (Bouvet et al.,⁷² Zasedateleva et al.⁷³ and Ray et al.⁷⁴; reviewed by Eliseeva et al.⁷¹). In vitro, YB-1 prefers RNA and ssDNA – K_D in the order of 10⁻⁹ – with respect to dsDNA. The appetite of NF-Y for DNA is much higher, with a K_D of 10⁻¹¹.^{7, 39} As for the specificity, single-nucleotide substitutions within the CCAAT and flanking nucleotides of Y/CCAAT drop affinity by one/two log levels.^{6, 10, 35} Typically, nanograms of NF-Y give a robust EMSA shift,^{47, 75} as opposed to micrograms of recombinant YB-1.⁷⁶ Methylation interference and orthophenanthroline footprinting confirmed that the bases contacted by NF-Y are centered on Y/CCAAT.^{6, 7, 35, 39, 40, 47} The only such data available for YB-1 is relative to a site in the MMP-2 promoter – CTGCTGGGCAAG – which lacks a Y/CCAAT sequence.⁷⁷

In summary, in vitro biochemical analyses indicate that the affinity and specificity of NF-Y for Y/CCAAT is superior to that of YB-1.

Mechanisms of TA

Sequence-specific TFs are known to be modular proteins, composed of a minimum of two domains: a DNA-binding domain (DBD) and a TA domain. Different types of DBDs and TAs have been described. NF-Y has two large Q-rich TAs over150 amino acids in length, in the NF-YA and NF-YC subunits. These TAs function when fused to heterologous DBDs, such as those of yeast GAL4 or bacterial LexA.^{49, 78, 79} Importantly, removal of the NF-YA (CBF-B) Q-rich domain generates a dominant-negative mutant, which affects the activity of CCAAT promoters upon transfection.⁸⁰ Co-transfections of the NF-Y trimer with TFs binding the loci neighboring CCAAT boxes in mammalian and Drosophila cells synergistically activate transcription of CCAAT promoters.^{52, 81, 82} Finally, transfections of recombinant proteins in which NF-YA was linked to a cell-penetrating TAT peptide activate endogenous CCAAT genes but not CCAAT-less units.^{83, 84} All these features are quite standard for TFs: sequence specificity, modularity of DBD and TA, and synergistic activation with neighboring TFs.

YB-1 has two domains flanking the central CSD: the A/P domain, and the C-terminal domain, in none of which is a typical TA apparent; most importantly, there are no data showing that these domains can function as TAs, in hybrids, or within the intact protein. The inclusion of YB-1 among TFs is based on changes in the expression of many genes by overexpression or functional inactivation of YB-1 by siRNA or shRNA.⁷¹ However, it should be remembered that northern blots, RT-PCR, qRT-PCR and microarray profiles all measure mRNA steady-state levels and not RNA Pol II transcription rates. In general, a troublesome issue in the manuscripts showing an effect of YB-1 in transcription is that they do not take into account its known role in mRNA biology: one needs to be sure that the reported changes in mRNA levels upon overexpression or inactivation are really due to transcriptional initiation events, rather than to post-transcriptional effects on mRNA stability. This consideration is particularly relevant for experiments measuring enzymatic activities of reporter genes, such as chloramphenicol acetyltransferase or Luciferase.

Nuclear run-on assays can provide evidence for a role in transcriptional initiation events, but to the best of our knowledge, no such experiments have been described for overexpression or inactivation of YB-1, in contrast to the situation for NF-Y-dependent transcriptional units.^{85, 86, 87, 88, 89, 90} Analogously, many in vitro transcription assays suggest a role for NF-Y in the formation of the pre-initiation complex and in re-loading of the RNA Pol II machinery.^{6, 7, 11, 48, 49, 50, 51} The only report showing in vitro transcription data on YB-1 used the heat-shock 70 kDa protein (HSP70) and thymidine kinase (TK) promoters with an unpurified bacterially produced protein.⁶³ It is worth noting that the HSP70 CCAAT element was subsequently investigated as a canonical NF-Y target by the same group.^{91, 92}

As to the mechanistic details of TA, NF-Y has been shown to (i) promote the DNA binding of neighboring activators, (ii) make contacts with multiple general transcription factors, including TBP (TATA-binding protein) and TAFs (TBP-associated factors), (iii) mediate recruitment of Pol II and (iv) bind multiple coactivators – p300, PCAF, MLL – and be post-translationally modified by some of them.⁹³ The role of YB-1 in activation appears to be related to ssDNA binding in promoter regions (Stein et al.⁹⁴; reviewed in Eliseeva et al.⁷¹).

The targets

YB-1 has been shown to regulate a number of genes. Among them, the multidrug resistance protein 1 (MDR1) promoter was analyzed in detail, due to the paramount importance that this gene has in the mechanisms of drug resistance. Overexpression of this efflux pump as a result of TA is a normal response to the treatment of cells to many chemicals, including cytotoxic drugs. Upon administration of anticancer compounds, cancer cells often acquire resistance to pharmacologic doses of drugs through the overexpression of MDR1, rendering antitumor regimens ineffective. The transcriptional control of MDR1 has been the object of many studies, as preventing its overexpression could be highly desirable.⁹⁵ The MDR1 promoter contains a crucial Y/CCAAT element, and a number of contradictory reports concerning the identity of the activator were published. Several papers suggested that YB-1 activates MDR1 through the Y/CCAAT box.^{96, 97, 98, 99} The identification of YB-1 as the CCAAT activator relied on EMSAs challenged with anti-YB-1 antibodies and on overexpression of YB-1 and inactivation by an antisense YB-1 transcript. Other investigators later showed that overexpression of YB-1 has no effect on MDR1 transcription,¹⁰⁰ and that inactivation by different siRNAs and shRNAs, which completely obliterates YB-1 expression, have no effect on MDR1 basal, or activated expression.¹⁰¹ The MDR1 Y/CCAAT is a perfect NF-Y site – CTGATTGGCT – located in the NF-Y canonical position, at −70 from the TSS. Indeed, several labs showed that NF-Y is the activator, both under basal and under multiple inducing conditions.^{102, 103, 104, 105, 106, 107, 108} In EMSAs performed with the Y/CCAAT box, these authors detected NF-Y as the only DNA-binding complex. Specifically, Scotto's lab showed that a NF-YA dominant-negative mutant affects MDR1 expression.¹⁰⁴ Finally, in vivo ChIPs reported NF-Y binding to MDR1.¹⁰⁹

Recent experiments reported interactions of YB-1 with APE1, a protein originally characterized for its role in base excision repair (BER), and shown to be important in the specific system to recruit Pol II by association with p300 and YB-1. APE1 is also known as redox effector factor 1 (Ref1), as it affects the redox status of, among other proteins, many TFs. APE1/Ref1 and YB-1 directly interact,¹¹⁰ they bind the MDR1 core promoter and removal of APE1 leads to decreased YB-1, Pol II and p300 promoter association in ChIP assays.¹¹¹ This led the authors to propose that the YB-1/CCAAT interaction mediates the recruitment of the APE1/p300/Pol II complex onto the promoter. The redox potential of NF-Y affects directly its DNA-binding capacity, acting on three conserved cysteines of the NF-YB HFD, and Ref1 is an important regulator.¹¹² Therefore, the above data are consistent with an alternative explanation in which APE1/Ref1 acts on NF-Y/CCAAT interactions to activate MDR1 transcription.

In general, we find it extremely surprising that in the reports pointing to YB-1 as the MDR1 CCAAT activator, NF-Y was either not observed or not recognized as such in EMSAs, given its superior affinity for the Y/CCAAT sequence. Technical considerations are perhaps unlikely to account for this, as NF-Y is readily observed in nuclear extracts of all growing cells, regardless of extraction protocols employed, incubation and EMSA conditions. In fact, it has been detected in EMSAs in hundreds of manuscripts reporting CCAAT binding in disparate promoters (Dolfini et al.⁵⁵ and references therein). Essentially, the MDR1 system is apparently a rare exception to the rule that a Y/CCAAT binding complex detected in vitro contains NF-Y. This confusion has been replicated with the related MRP2 promoter, suggested to be either activated by NF-Y¹¹³ or YB-1.¹¹⁴

The same argument applies to other Y/CCAAT promoters, which YB-1 was shown to activate, including DNA Pola,¹¹⁵ cyclin A, cyclin B1,¹¹⁶ where many reports have also demonstrated an NF-Y dependence (Farina et al.¹¹⁷; reviewed in Gurtner et al.¹¹⁸). For other YB-1 targets, such as EGFR, PIK3CA, MET and CD44,^{119, 120, 121, 122} the Y/CCAAT, as it has been characterized genetically and bioinformatically, is absent.

Focusing on the global picture, transcription profiling analysis after functional inactivation is available both for YB-1¹²³ and NF-Y.¹²⁴ In YB-1-regulated genes, an abundance of E2F sites was reported: we re-examined these data with bioinformatic tools and indeed confirm this point, but neither the Y/CCAAT logo nor related variants are over-represented (Figure 3). On the other hand, inactivation of single NF-Y subunits led to different phenotypes and sets of regulated genes,^{124, 125} with the enrichment of the Y/CCAAT logo in downregulated genes a common feature (Figure 3).

Figure 3.

Enriched TFBS in YB-1- or NF-Y-regulated genes. Data of gene expression profilings of different cell lines, reported by Lasham et al.,¹²³ were analyzed by pscan (upper panels). In the lower panels, a similar analysis is reported on gene expression profiles of HCT116 cells inactivated with shNF-YA or shNF-YB (Benatti et al.¹²⁴)

The genomic loci

The genome-wide identification of TFBSs through ChIP-on-Chip and, more recently, ChIP-Seq has vastly advanced our understanding. For NF-Y, ChIP-on-Chip studies performed on CpG islands, promoters and oligo tiling arrays led to the conclusion that the protein is bound not only to promoters but also to many enhancers. The presence of Y/CCAAT in these regions was consistent, although not all loci contained the pentanucleotide. ChIP-Seq experiments performed in the framework of the ENCODE Project (Wang et al.⁶⁰) provide far higher precision. The data are clearcut: almost all peaks do contain the Y/CCAAT consensus, essentially identical to the original NF-Y matrix (Figure 2). Moreover, the CCAAT-less sites are variation of one nucleotide in the core sequence that also harbor optimal flanking sequences. The technique is so spectacularly powerful and precise that it is possible to discriminate the exact area bound by NF-YA – the CCAAT pentanucleotide – from the immediately flanking nucleotides bound by NF-YB, in perfect accordance with the in vitro biochemical data.

ChIP-on-Chip and ChIP-Seq experiments for YB-1 were also reported recently.^{121, 126} We analyzed the data for >30 000 ChIP-Seq peaks in three cancer cell types and could not identify a Y/CCAAT sequence, either searching for known TFBS or with de novo motif discovery tools.¹²⁷ This is uncommon for a sequence-specific TF, whose logo is usually recognizable within the bound peaks, but not unheard of:⁶⁰ in many cases, one can identify either a new logo or one of those characterized for other TFs. Of the 35 YB-1-targeted genes previously analyzed by various means, including reporter assays (summarized in Eliseeva et al.⁷¹), 5 are associated with YB-1 peaks in BT747 cells, 15 in HR5 and 13 in HR6 cells: collectively, 18 genes are targeted (Figure 4). A minority (seven) of the YB-1 peaks is in promoters and even fewer are at the exact sites described in the functional studies. Obviously, these results can be heavily influenced by the dissimilar cellular contexts used in the functional and location assays. However, a similar NF-Y ChIP-Seq analysis performed on the same 35 genes, also from disparate cells, showed positivity for 19 units, with all but 4 peaks being present in promoters: this leads to the somewhat paradoxical notion that, in unbiased experiments, NF-Y targets more ‘bona fide' YB-1 sites in vivo than YB-1 itself.

Figure 4.

Presence of YB-1 and NF-Y peaks in YB-1-regulated genes. YB-1-regulated genes, as summarized by Eliseeva et al.,⁷¹ were analyzed for the presence of YB-1 peaks in the YB-1 ChIP-Seq data reported by Astanehe et al.,¹²⁶ and for NF-Y peaks present in ENCODE data of K562, HeLa-S3 and GM12878 cells (Wang et al.⁶⁰ and Fleming and Struhl, submitted for publication)

Cellular localization

Intuitively, TFs exert their function in the nucleus, where their genomic targets are located. In this respect, NF-Y is prototypical, with NF-YA and NF-YB being found exclusively in the nucleus, NF-YC being in part in the cytoplasm and traveling to the nucleus with the NF-YB HFD partner.¹²⁸ YB-1 is fundamentally cytoplasmic in normal cells, and nuclear in transformed cells, or after specific stimuli. Many inducible TFs are found in the cytoplasm or membrane bound, and are transferred to the nucleus only after a specific stimulus. To explain the activation of G1/S cell-cycle-regulated promoters, such as DNA Pol α and cyclin A by YB-1, it was proposed that there is a transient nuclearization of the protein at the G1/S boundary.¹¹⁶ Even so, it not clear how YB-1 could activate transcription of the many proposed target genes that are constitutively expressed, or of the G2/M-specific cyclin B, in normal cells.

Protein structure considerations

The modality of CCAAT recognition by NF-Y is now fully understood, as the crystal structure of the complex bound to a CCAAT oligo has been solved.¹²⁹ NF-Y contacts DNA in a non-sequence-specific manner via the HFD subunits, which bind the double helix in a way that is essentially identical to H2A/H2B within nucleosomes, with sequence specificity imparted by minor groove binding to the CCAAT, via an α helix (A2) and a novel motif of NF-YA. This modality of DNA recognition is unprecedented among TFs. Overall, a high number of amino-acid residues (46) contact the DNA over a 25–28 bp area, which helps to explain the extraordinary affinity of the trimer for DNA. The domain of YB-1 required for nucleic acid binding is the CSD, composed of RNP1 and RNP2: the structure of this domain is known, both from X-ray crystallography and NMR studies,⁶² but the interactions with RNA or DNA are not detailed and modeling exercises could not provide compelling reasons why YB-1 should bind to double-stranded Y/CCAAT with any type of specificity.

Altogether, from the most reductionistic in vitro assays to in vivo approaches, a large set of data strongly argues against YB-1 being a direct regulator of transcriptional initiation by binding to the Y/CCAAT sequence, or variations of it. Because of the widespread and profound influence played by YB-1 on mRNA biology, likely impacting on results obtained in overexpression or functional inactivation experiments, it is even debatable as to whether it has any role in direct promotion of transcriptional initiation events. On the contrary, NF-Y's credentials as a paradigmatic sequence-specific TF are impeccable.

The – Apparent – Paradox of Y/CCAAT, NF-Y and YB-1 in Cancer

To the best of our knowledge, NF-Y subunits are neither consistently mutated – as for p53 – nor generally overexpressed – as for MYC – in human tumors. However, changes in the expression of NF-Y subunits, often NF-YA, have indeed been reported,⁹³ and this phenomenon should be studied in a quantitatively credible and statistically significant manner. Moreover, the links of NF-Y with activation of cancer pathways mediated by mutant p53 and E2Fs are well established.¹³⁰ More recently, strong connections emerged from ChIP-Seq experiments with JNKs,¹³¹ the PRAME oncogene,¹³² and a specific group of oncogenic TFs in ENCODE data (Fleming and Struhl, submitted for publication).

There is little doubt that YB-1 expression changes dramatically, at both the mRNA and protein levels, in transformed cells; the correlation is so impressive that in some tumors it is considered a prognostic marker (for a review see Eliseeva et al.⁷¹ and Costessi et al.¹³³). Furthermore, the localization of YB-1 becomes strongly nuclear in tumor cells. The changes are maximal in aggressive tumors resistant to drugs, and in advanced stages of cancer. It is also undisputed that a large set of genetic experiments point to YB-1 as a protein promoting transformation, epithelial–mesenchymal transition and growth of metastatic cancer cells: in essence, YB-1 is a powerful oncogene.^{134, 135, 136}

Largely because of the alleged role of YB-1 on MDR1 transcription, two syllogisms emerged in the literature. First, Y/CCAAT is enriched in cancer genes, YB-1 is a Y/CCAAT binding TF enriched in cancer cells; hence, YB-1 is responsible for the activation of CCAAT cancer genes in cancer cells.¹³⁷ Second, MDR1 expression is induced by and responsible for resistance of cancer cells to cytotoxic drugs, YB-1 is responsible for MDR1 overexpression (and it is overexpressed in cancers); hence, YB-1 mediates cancer resistance by enhancing MDR1 expression. Flaws in the antithesis, as explained above, lead to an at least partially incorrect synthesis.

So how could YB-1 be mediating cancer progression and resistance to drugs, if not by binding directly to the Y/CCAAT boxes of MDR1 or other overexpressed genes? Reviews of available literature point to many possible hypotheses, the most likely of which highlight the control of various aspects of RNA metabolism, such as stability, splicing and translation.^{71, 135} The mechanisms of YB-1 mRNA regulation were studied in reliable reconstituted in vitro systems, and they are now relatively well understood, and consistent with specific binding of the protein to RNA. Interestingly, the preferred target logo, derived from the analysis of ChIP-Seq experiments, is consistent with the RNA-binding features of YB-1, and indeed resembles Kozak sequences.¹²⁷

Thus, we offer an alternative explanation for the many reports of YB-1 binding in ChIP experiments, and indeed ChIP-Seq, in transformed cells:¹²⁶ YB-1 could be transitorily located in an area physically ‘close' to promoters, or other important regulatory regions, where transcriptional initiation decisions are made, but loaded on partially synthesized primary RNAs (Figure 5). One finding consistent with this interpretation is that chromatin association of YB-1 is apparently lost upon treatment with ribonucleases, which destroys pre-mRNAs.¹³⁸ It is well established that coactivators, which do not bind to DNA directly, can be crosslinked efficiently to DNA, resulting in peaks in ChIPs and ChIP-Seq. It is also known that promoter structures and specific TF combinations have an impact on the loading and composition of the mRNA splicing apparatus.^{139, 140} Thus, we propose that rather than being enemies battling over the same DNA sequence, NF-Y and YB-1 take on different tasks, cooperating to alter gene expression in cancer cells: transcriptional initiation through Y/CCAAT sequence-specific binding the former, and post-transcriptional mechanisms through RNA binding the latter.

Figure 5.

Scheme of mechanisms of gene expression control by NF-Y and YB-1

Glossary

NF-Y

nuclear factor Y

YB-1

Y box-binding protein 1

MDR1

multidrug resistance protein 1

TFBS

transcription factor-binding sites

TF

transcription factor

CBF

CCAAT-binding factor

MHC

major histocompatibility complex

EMSA

electrophoretic mobility shift assay

HFDs

histone fold domains

RSV LTR

Rous sarcoma virus long terminal repeat

ChIP

chromatin immunoprecipitation

HSP70

heat-shock 70 kDa protein

PSFM

positional sequence frequency matrix

siRNA

small interfering RNA

shRNA

short hairpin RNA

CSD

cold-shock domain

ssDNA

single-stranded DNA

dsDNA

double-stranded DNA

DBD

DNA-binding domain

TA

transcriptional activation

The authors declare no conflict of interest.