Nucleolin, a Novel Partner for the Myb Transcription Factor Family That Regulates Their Activity*

To unravel the mechanisms of action of transcriptional regulation by the Myb family of transcription factors, we have set out to isolate their protein partners. We identify nucleolin as one of the nuclear polypeptides that interact specifically with the A-Myb and c-Myb, but not B-Myb DNA-binding domains. We show unambiguously that this interaction is direct and takes place in vivo, as demonstrated by co-immunoprecipitation of the endogenously and exogenously expressed proteins. The minimal DNA-binding domain containing only the R2R3 c-Myb repeats is sufficient for nucleolin binding. Computer analysis of the R2R3 three-dimensional structure, as well as extensive mutational analysis within this region, reveals that the Arg161 residue, present in c-Myb and A-Myb, but not B-Myb, is crucial for this interaction. We show that the interaction of nucleolin with Myb is functional because co-transfection of nucleolin down-regulates Myb transcriptional activity. Nucleolin is a multifunctional phosphoprotein present in both nucleoplasm and more abundantly in the nucleolus and shows helicase and chromatin decondensing activities. This is the first demonstration of nucleolin binding to a transcription factor.

Over the last few years, increasing evidence has emerged that transcription factors operate as multicomponent complexes containing one or more transcription factors, adaptor proteins linking them to the general transcription machinery, histone acetyl transferases, and deacetylases that contribute to chromatin decondensation or assembly, respectively (1)(2)(3). Thus the identification of new protein partners that directly interact with specific transcription factors should allow further definition of the mechanisms of transcription activation.
The mammalian Myb family of oncoproteins is composed of three members, A-Myb, B-Myb, and c-Myb, all of which are transcription factors (reviewed in Refs. 4 and 5). c-Myb is expressed mostly in hematopoietic cells and regulates their proliferation and differentiation (4 -6). Its truncated viral equivalent, v-Myb, is transforming and causes leukemias in chickens (7,8). A-Myb is expressed in few cell types, B lymphocytes, spermatocytes, some neural cells, and mammary epithelium, and regulates the proliferation and/or differentiation of these cells in vivo (9 -14). B-Myb is ubiquitous and its expression closely correlates with proliferation in all cell types studied (reviewed in Refs. 5 and 15). B-Myb is phylogenetically the most divergent among the three Myb proteins (16). Unlike c-Myb and A-Myb, it has usually been found to be a weak or inactive transcription factor on most natural promoters tested and rather to inhibit c-Myb or v-Myb activity (17)(18)(19)(20)(21).
The different functional domains of all three Myb proteins have been characterized. The DNA-binding domain (DBD) 1 at the N terminus is the most conserved region (4,16). The central portion of A-Myb and c-Myb forms the transactivation domain, whereas the C-terminal half has a negative regulatory function on both DNA binding and transcription (22)(23)(24). The definition of distinct functional domains has led to the search for molecules that interact with them and regulate Myb function. In particular, the CBP/p300 acetylases have been demonstrated to interact directly and synergize with both c-Myb and A-Myb (22,(25)(26)(27). The Myb domain responsible for binding to CBP/ p300 is in both cases the transactivation domain, which is conserved in these two proteins. The c-Myb DBD on the other hand has been reported to bind to several proteins in vitro: the coactivator p100 (28,29), the c-Maf transcription factor (30), and the peptidyl-prolyl isomerase enzyme Cyp40 (31). Although these interactions appear to modify Myb function in vitro and/or in vivo, direct binding of these proteins in vivo has not been demonstrated.
Recently, we have shown that the DBD of A-Myb has a regulatory function on the protein transcriptional activity in addition to that of binding to DNA. In addition, we have shown that the A-Myb DBD binds to several nuclear proteins (32).
Here we characterize one of the proteins that bind to the A-Myb DBD, nucleolin and show that this protein binds only to the DBD of the A-Myb and c-Myb but not B-Myb transcription factors. Furthermore we demonstrate that nucleolin affects Myb transcriptional activity. These data add a novel functional partner for a family of transcription factors, with potential activity in the regulation of chromatin assembly (33).

MATERIALS AND METHODS
Cell Lines and Cell Cultures-The human lymphoid cell lines were grown in RPMI 1640 medium (Seromed, Berlin, Germany). The 293derived Phoenix epithelial cell line (34) was grown in Dulbecco's mod-ified Eagle's medium containing 4.5 g/liter glucose. All media were supplemented with 10% fetal calf serum (Hyclone, Steril System, Logan, UT), glutamine (Life Technologies, Inc.), and 50 g/ml gentamycin (Life Technologies, Inc.).
Plasmid Constructions-The pGST-A-Myb-DBD construct has been described previously (32). pGST-B-Myb-DBD was made by polymerase chain reaction amplification using Pfu enzyme (Stratagene). The DNA fragment encoding aa 1-212 was amplified and inserted in the EcoRI site of pGex-2TK (Amersham Pharmacia Biotech). The pGST-c-Myb-DBD (aa 2-212) was made by inserting the blunted NcoI/HindIII fragment from pHM1 (35) into the SmaI site of the pGex-3x vector. For kinase labeling of GST-A-Myb-DBD, the p1500 plasmid (36) was digested with NdeI and blunted, and the 0.65-kilobase fragment encoding aa 1-217 was inserted into the BamHI site of the pGex-2TK vector. The pGST-c-Myb(R2R3) (aa 93-212) was constructed by digesting pGST-c-Myb-DBD with SacI and BamHI, blunting, and religation of the vector fragment. The c-Myb R2R3 point mutants in the pGex-3x vector background were generated by polymerase chain reaction; the sequences over the mutations were as follows: CAGAAAGTGATAGAGCTTGTA-AAGAAA for R102K/Q108K, GGTACGAAACAGTGG for P112/R114Q, and AAGGTACTG for R116V. All three mutant constructs were made in the background of c-Myb aa 93-212 fused to GST in the pGex-3x vector. The same R116V mutation was also introduced into the pSC background (c-Myb cDNA in the pSG5 vector) by polymerase chain reaction-based mutagenesis. The pNFor4 plasmid carrying the complete human cDNA in a cytomegalovirus expression vector was a kind gift of Dr. N. Maizels (Yale University, New Haven, CT) (37). For Myc-tagged nucleolin, the HindIII fragment encoding aa 1-539 of nucleolin was inserted into the EcoRV site of pCDNA3.1(Ϫ)/Myc-hisA (Invitrogen). All constructions were verified by sequencing.
Microsequencing-Nuclear extracts from 20-liter Namalwa cultures were bound to GST-A-Myb-DBD columns, run in a 5% SDS-PAGE gel, and blotted to polyvinylidene difluoride membrane. Coomassie Brilliant Blue-stained proteins were excised from the blots and N-terminally sequenced by Edman degradation on a pulsed liquid phase 477A/120A protein sequencer (Perkin-Elmer, AB Division, Foster City, CA) using N-methylpiperidine as a coupling base.
Western Analysis-Total cell lysates or glutathione-agarose bound proteins were boiled in SDS loading buffer and run in an 8% SDS-PAGE gel. The blotting and Western conditions were as described previously using the ECL detection system. The CC-98 mAb specific for human nucleolin was a kind gift from Dr. N. H. Yeh (Insitute of Microbiology and Immunology, Taipei, Taiwan, China) (38).
Far Western-Nuclear extracts from 4 ϫ 10 6 cells were run in an 8% SDS-PAGE gel and blotted onto a nitrocellulose membrane (Schleicher & Schuell). The membrane was incubated in 20 mM HEPES, pH 7.9, 5 mM Mg 2 Cl, 1 mM EDTA, 150 mM NaCl, 1 mM dithiothreitol, 2.5% nonfat milk at 4°C overnight. Binding was carried out in 20 mM HEPES, pH 7.9, 5 mM Mg 2 Cl, 1 mM EDTA, 150 mM NaCl, 0.5% nonfat milk, 0.1% Nonidet P-40 using 250.000 cpm/ml of radiolabeled GST-A-Myb DBD protein in presence of a 10-fold excess of unlabeled purified GST. After 5 h of incubation at 4°C, the membrane was washed in binding buffer in absence of recombinant protein, dried, and autoradiographed.
Immunoprecipitation-The anti-A-Myb polyclonal rabbit antiserum raised against the bacterially expressed and affinity purified fragment spanning aa 218 -482, which shows poor homology to other Myb proteins, has been described previously (11). For c-Myb and B-Myb immunoprecipitation, commercial antibodies raised against the C terminus of the proteins were used (Santa Cruz). 40 ϫ 10 6 BJAB cell nuclear extracts were incubated with 40 l of protein G-Sepharose Fast Flow beads (Amersham Pharmacia Biotech) without or with precoupled antibody (50 g each) for 1 h at 4°C. The beads were washed in phos-phate-buffered saline three times, boiled in SDS sample buffer, and loaded onto 8% SDS-PAGE gel.
Transient Transfections and CAT Assays-The Phoenix epithelial cells were transfected by the Ca 2ϩ phosphate precipitation method, as described previously (34,36). The KHK-CAT plasmid carrying eight Myb-binding sites upstream from the CAT gene was used as reporter (36). Transfections were carried out in 35-mm dishes using a total of 2-2.7 g of DNA, including 0.2-1.5 g of Myb, 0.5 g of reporter plasmid, 0.1 g of ␤-galactosidase or RSV-luciferase constructs for normalization, and 0.03-1 g of pNFor4 plasmid encoding human nucleolin. Equal amounts of total protein, determined with the Bio-Rad Protein Assay, were used for ␤-galactosidase, luciferase, and CAT assays. CAT assays were performed using a phase extraction protocol (39). The CAT activity values were normalized to the ␤-galactosidase activity. Luciferase activity was measured with the Luciferase Assay System (Promega), and light emission was determined with a Berthold Lumat LB 9507 Luminometer.
Computer Analysis-The three-dimensional model of DNA bound B-Myb R2R3 was constructed using c-Myb as structural scaffold (40) (Brookhaven Protein Data Bank accession code 1MSE) and the Proexplore software package. As a first step, the side chain coordinates were generated using a rotamer library and a table of rotamer-rotamer interaction energies was calculated. In the second step, the intramolecular energy was minimized using simulated annealing Monte Carlo technique. Model structure refinement was performed with the Amber 4.1 software package. First the coordinates of the backbone atoms were kept fixed until a RMS (Route Mean Square) of 0.1 Kcal/mol, using steepest descent algorithm. Full optimization of the model was then performed for 10,000 other steps using conjugate gradient algorithm. The solvent-accessible surfaces of Myb R2R3 were calculated using Chem3D Pro package (Cambridge Soft), with a solvent radius of 1.4 Å on a PC-Pentium 200, subtracting the DNA for clarity. We therefore performed a preparative pulldown assay with GST-A-Myb-DBD and suceeded in purifying and microsequencing the p95 and p85 proteins. The first 19 amino acids deduced for p95 corresponded exactly to amino acids 2-20 of human nucleolin (VLAKAGKNQGDPKKMAPP) (Swiss-Prot accession number P19338), whereas p85 corresponded to the larger subunit of Ku antigen (VRXSGNKAAV-VLCMDVGFTM) (Swiss-Prot accession number P13010). To confirm that nucleolin and Ku antigen can bind to the Myb DBD and that this interaction is specific, we performed pulldown assays with GST-A-Myb, GST-c-Myb, and GST-B-Myb DBD as well as a control GST-Max DBD and analyzed the bound products in Western blots. As shown in Fig. 1B, nucleolin did indeed bind to A-Myb and c-Myb DBD, but not to B-Myb or Max DBD. On the other hand Ku antigen bound to all DBD tested, including that of the unrelated Max protein. As expected neither nucleolin nor Ku antigen bound to the GST-TA control containing the A-Myb transactivation domain. These data suggested that nucleolin binds specifically to the c-Myb and A-Myb DBD and that Ku antigen on the other hand may recognize contaminating DNA associated with the DNA-binding domains of all proteins. To confirm still further the specificity of binding of nucleolin to Myb, independently from DNA, we have performed the pull-down assays in the presence of ethidium bromide, which permits identification of DNA-independent protein interactions (41). As shown in Fig. 1C, nucleolin binding was equally strong in the presence or absence of ethidium bromide. On the contrary, the interaction with Ku antigen was abolished already at the lowest concentration of ethidium bromide, showing that Ku antigen binds to DNA. That nucleolin binding to c-Myb and A-Myb takes place independently of DNA was confirmed by performing pull-down assays in presence of DNase-treated A-Myb DBD and cell lysates (data not shown).

Nucleolin Binds Specifically to the A-Myb and c-Myb, but Not B-Myb, DNA-binding Domains in Vitro-
To demonstrate that nucleolin binding is direct and does not require intervening proteins, a new construct was made encoding the GST-A-Myb DBD containing a phosphorylation site for heart muscle kinase. The purified GST-A-Myb-DBD or GST alone control proteins were labeled in vitro with 32 P-ATP and heart muscle kinase and used in Far Western assays on total cell extracts from Namalwa cells ( Fig. 2A, lanes 1 and 2). One lane of cell extract run in parallel was cut and used in standard Western blot to identify precisely the location of the nucleolin band ( Fig. 2A, lane 3). The results clearly show that the labeled GST-A-Myb-DBD, but not labeled GST alone, can bind to a protein with exactly the same molecular mass as nucleolin. These data demonstrate that the A-Myb DBD binds to nucleolin directly. In addition, the A-Myb-DBD also binds most evidently to other proteins with apparent molecular masses of 110, 80, and 60 kDa ( Fig. 2A).
Finally the binding characteristics of Myb DBD to nucleolin were investigated by performing the assay in different buffer conditions. As shown in Fig. 2B, the binding of nucleolin to A-Myb is strong and resists 300 mM NaCl conditions and is reduced but not abolished in 500 mM NaCl. On the other hand, as little as 0.1% SDS completely abolishes binding (Fig. 2B,  lane 4). These data suggest that the A-Myb-nucleolin interaction involves both hydrophobic and ionic bonds. These results strongly emphasize the specificity of binding of nucleolin to specific domains of A-Myb and c-Myb but not to the DBD of other transcription factors such as B-Myb or Max.

Binding of Nucleolin to A-Myb and c-Myb Takes Place in
Vivo-To demonstrate that binding between nucleolin and A-Myb or c-Myb takes place in vivo, we have performed coprecipitation experiments. The different Myb proteins were immunoprecipitated with specific antibodies, and the immunoprecipitates were tested for the presence of nucleolin by Western blotting. That the antibodies had indeed specifically immunoprecipitated the respective Myb proteins was also verified by stripping the blots and reprobing them with the appropriate anti-Myb antibodies. As shown in Fig. 3A (lanes 1-3), during the first round of immunoprecipitation (IP), anti-A-Myb antibody specifically brought down A-Myb itself as well as nucleolin. Preimmune serum or empty beads were inactive. As expected, when lysate, which had already undergone one round of immunoprecipitation was again incubated with the same amount of fresh anti-A-Myb serum (second round of IP, lane 6), little further A-Myb or nucleolin could be precipitated. This demonstrates that nucleolin was precipitated specifically with A-Myb and was not due to some spurious activity of the anti-A-Myb antiserum, because nucleolin is a much more abundant protein than A-Myb and clearly cannot be exhausted after only one round of immunoprecipitation with anti-A-Myb antiserum. In accordance with the pull-down assays, anti-c-Myb, but not anti-B-Myb antibodies, directed against the C terminus of these proteins, was able to co-precipitate nucleolin (Fig. 3A,  lanes 8 and 9). Incubation of the same blot with anti-B-Myb and anti-c-Myb antibodies confirmed that indeed both proteins had been specifically precipitated (Fig. 3A).
To confirm the specific interaction between nucleolin and Myb in vivo, we also performed the reverse experiment. A nucleolin construct with a Myc tag was prepared and transfected together with c-Myb. As shown in Fig. 3B (lanes 2 and 3), immunoprecipitating the tagged nucleolin with anti-Myc antibodies brought down co-transfected c-Myb only in cells in which the tagged nucleolin was present. As a positive control, precipitation with anti-Myb brought down c-Myb as expected (Fig.  3B, lane 1). Of note is that the tagged nucleolin construct lacked the last 171 C-terminal aa of nucleolin, showing that this terminal region is not involved in the interaction. We have thus rigorously demonstrated that nucleolin and c-Myb/A-Myb interact in vivo by co-precipitating either partners and both endogenous and exogenous proteins.
R2R3 of c-Myb Is Sufficient, and Arg 161 Is Required for Nucleolin Binding-Given that Myb proteins are highly conserved within their DNA-binding domains, we set out to determine more precisely which region and which aa are involved in binding to nucleolin. We first deleted further the c-Myb and A-Myb DBD at the N terminus and tested them in pull-down assays. As shown in Fig. 4A, a GST fusion construct carrying only the R2 and R3 repeats of c-Myb (aa 91-212) was still able to bind strongly to nucleolin (Fig. 4A, lane 1). On the other hand, a fusion protein containing the c-Myb N terminus including only R1 (aa 1-89) did not bind (Fig. 4A, lane 2). Similarly, an R2R3 construct of A-Myb did bind to nucleolin (data not shown). We therefore analyzed the aa differences between human A-Myb/c-Myb and B-Myb R2R3. Most aa differences are conservative, with the exceptions of aa 108, 112, and 114 of R2 and aa 161 of R3 (c-Myb numbering). We therefore decided to mutagenize pairwise four clustered residues in R2 (aa 102 and 108; aa 112 and 114) and singly aa 161 of R3 from their c-Myb to their B-Myb configuration. Thus Arg 102 and Gln 108 in c-Myb R2R3 were mutagenized to Lys in one construct (102/108). In the second mutant Pro 112 and Arg 114 were changed to Thr and Gln, respectively (112/114). Finally Arg 161 was mutagenized singly to Val in the last construct. In all cases the background was c-Myb R2R3 fused to GST. All mutants were then tested in pull-down assays. As shown in Fig. 4A (lanes 3-5), the point mutants were still able to bind to nucleolin except the 161 mutant, which was completely inactive. That all proteins were made in equivalent quantities and bound equally well to DNA was verified in electrophoretic mobility shift assay using a standard labeled Myb-binding site (mim-A oligonucleotide; Fig.  4A) (36). Furthermore quantitative bandshift analysis at different concentration of mim-A DNA probe revealed that the affinity of the different mutant proteins was equivalent to that of c-Myb wild type. The calculated K d for the c-Myb R2R3 161 mutant was 0.07 nM, and for wild type it was 0.09 nM (data not shown). These values are in agreement with those previously described for the c-Myb DBD (42). Thus these data demonstrate that the Arg residue in position 161 of c-Myb (conserved in position 156 of A-Myb) abolishes specifically nucleolin binding, whereas the other mutated aa in R2 do not. Furthermore we also tested a GST-v-Myb DBD fusion protein, which carries four point mutations relative to c-Myb (I91R, L106H, V117D, and I181V) (43). The GST-v-Myb DBD bound nucleolin-like c-Myb wild type (data not shown), showing that these further mutations in R2R3 are not crucial for the interaction. Furthermore these data suggest that the oncogenic form of c-Myb, v-Myb, can bind to nucleolin.
To verify that the R161V mutation also affected in vivo binding of nucleolin to c-Myb, we transfected both constructs with the tagged nucleolin and performed a co-precipitation experiment. As shown in Fig. 4B, c-Myb wild type was efficiently co-precipitated with anti-Myc tag antibody (lane 2),  Fig. 5, the red indicating hydrophobic residues and blue residues indicating hydrophilic ones. The location of all mutated residues and of the groove that makes direct contact with DNA are indicated. It is interesting to note that Arg 161 is located close to a patch of hydrophobic aa residues (Fig. 5). Both the patch of hydrophobic aa and Arg 161 are located on the surface of the R3 portion of the structure and on the opposite side relative to the groove that makes direct contacts with DNA and is at the back of the structure. On the other hand, the other mutated and nonconserved residues that did not affect binding (Arg 102 , Pro 112 , and Arg 114 ) are clearly located on the R2 side of the structure, also opposite to the DNA-binding pocket. We therefore hypothesize that the binding site for nucleolin may involve both the Arg 161 residue (through ionic interactions) and the near hydrophobic patch, but not R2. To compare the threedimensional structure of c-Myb with that of B-Myb, which does not bind to nucleolin, we performed molecular modelling to deduce the DNA-bound form of B-Myb R2R3. As shown in Fig.  5, the B-Myb R2R3 surface is very similar to that of c-Myb and in particular maintains the hydrophobic patch of R3. However, the crucial hydrophilic Arg 161 residue next to it is replaced by a hydrophobic valine (Val 152 ; Fig. 5).
The Interaction of Nucleolin with Myb Is Functional-To investigate the functional significance of binding of A-Myb and c-Myb to nucleolin, we co-transfected nucleolin with either wild type c-Myb or the nucleolin binding defective 161 mutant and a standard Myb responsive reporter (the KHK-CAT construct) (36). As shown in Fig. 6A, the addition of increasing doses of nucleolin to c-Myb wild type led to a rapid inhibition of transcriptional activity, inhibition being virtually complete at 0.3-1 g/ml nucleolin. As expected, the transcriptional activity of the 161 mutant was not affected at lower doses of nucleolin and only marginally inhibited at higher doses. In addition, transfected nucleolin did not affect transcription of basal KHK-CAT activity in absence of Myb or of the control RSV-luciferase construct (of note is that in Fig. 6A, a value of 1 was arbitrarily given for the KHK-CAT or RSV-luciferase activities in absence of nucleolin, and the other values in presence of nucleolin were calculated in proportion to it).
Perhaps surprisingly, we observed that the transcriptional activity of the c-Myb 161 mutant in absence of co-transfected nucleolin was significantly lower than that of wild type c-Myb by about 50% (Fig. 6A). Because c-Myb is known to have a bell-shaped dose response curve because of squelching effects at higher doses, we compared the activities of the c-Myb wild type and 161 mutant at several doses in absence of exogenously added nucleolin. We observed a peak transcriptional activity of c-Myb wild type at 1 g, whereas the 161 mutant activity remained low at all concentrations (Fig. 6B). We have not yet been able to establish whether the lower activity of c-Myb 161 mutant is due to a lack of interaction of this protein with endogenous nucleolin or with other regulatory proteins that bind to c-Myb DBD (28 -31) or both. We conclude that the interaction of nucleolin with Myb is functional and that exogenous nucleolin down-regulates Myb activity. DISCUSSION In this report, we identify a new functional partner protein for the Myb family of transcription factors. We demonstrate that the DBD of A-Myb and c-Myb, but not B-Myb, bind to nucleolin both in vitro and in vivo. We show in an unambiguous way that binding is direct and is independent of the presence of contaminating DNA. Out of nine different aa tested by mutational analysis, we identify the Arg residue at position 161 of c-Myb as a key aa for nucleolin binding and suggest that the contact site includes Arg 161 and a hydrophobic patch on the surface of R2R3, opposite the DNA-binding site. We demonstrate in co-transfection assays that nucleolin can down-regulate Myb activity.
Nucleolin is an abundant protein, mostly present in the nucleolus but also detected diffusely in the nucleus (45)(46)(47), particularly in proliferating cells (38). Its major function relates to its presence in the nucleolus, where it has been shown to regulate ribosomal RNA transcription and processing (48,49). However, it is thought to be a multifunctional protein and has been implicated in a number of additional processes that take place in the nucleus: RNA polymerase II dependent transcription (50 -52), immunoglobulin switching (37,(53)(54)(55), the attachment of genomic DNA to the nuclear matrix (56), and the decondensation of chromatin (Ref. 57; reviewed in Ref. 33). Thus binding of Myb to nucleolin is likely to relate to the nuclear rather than nucleolar functions of nucleolin because Myb is not localized in the nucleolus. Interestingly, the nucleolar protein Arf has recently been shown to be involved in regulating nuclear p53 activity through sequestering of Mdm2 (58). Thus novel interactions between nucleolar proteins and transcription factors are emerging. The previous report of the functional interaction of c-Myb with another nucleolar protein, p160, is intriguing in this context (29).
Several biochemical functions have been identified for nucleolin: the central portion of the molecule contains four RNA-binding domains and is responsible for binding to RNA (49). The acidic N-terminal portion is thought to mediate binding to DNA and to histone H1 (37,57) and contains arrays of consensus phosphorylation sites. Finally nucleolin has been shown to have helicase activity, which appears to reside in the RG-rich C-terminal portion of the protein (59). It can unwind DNA-DNA and DNA-RNA as well as RNA-RNA templates. The capacity of nucleolin to bind to histone H1 is thought to contribute to its reported chromatin decondensing function (57).
Nucleolin protein expression and phosphorylation is regulated during the cell cycle, being induced during S phase entry (38,53). It is also highly phosphorylated and is the substrate of casein kinase II, PKC-, p34 cdc2 , and cyclin-cdk complexes (46, 60 -62). Preliminary data indicate that nucleolin needs to be phosphorylated to bind efficiently to Myb. 2 c-Myb and A-Myb expression is also regulated during the cell cycle, being present mostly in the S/G 2 /M phases of the cell cycle. Thus both the expression and the phosphorylation of nucleolin by cell cycleregulated kinases may correlate with the highest expression of the Myb proteins to which it also binds. Further work will be required to determine which region of nucleolin is required for Myb binding. We show here that a tagged and truncated nucleolin construct lacking the RG domain still interacts with Myb, excluding a role for this domain in binding to Myb.
Recently, evidence has emerged that transcription factors act as multiprotein complexes. The different members of the complexes may take part in different aspects of transcriptional regulation, such as bridging the specific transcription factors to the general transcription machinery (adaptor proteins) or bridging them to the histone acetylases and deacetylases, which are thought to allow the remodelling of chromatin (1-3). The binding of A-Myb and c-Myb to nucleolin would add another type of interaction, although the exact function of nucleolin in this context remains to be determined. Even though nucleolin has been suggested previously to play a role in regulating transcription (50), this is the first demonstration of its binding to a specific transcription factor and of its ability to regulate its transcriptional activity. The previous report of its interaction with E47 was demonstrated only in vitro and only when E47 was fused to a ␤-galactosidase tag (63).
We observed down-regulation of Myb transcriptional activity by co-transfected nucleolin, suggesting that nucleolin binding inhibits Myb activity. In apparent contrast to the co-transfection data, the c-Myb 161 mutant, which is unable to bind to nucleolin, showed lower transcriptional activity than c-Myb wild type. It is possible that the Arg 161 mutation affects binding of another protein(s) in addition to nucleolin and that this lost interaction is responsible for the lower activity of the mutant. Indeed, several other proteins have been shown to bind to c-Myb DBD: the adaptor molecule p100 binds to the DBD of c-Myb as well as to other transcription factors and may link them to the general transcription machinery (28). Very recently, Leverson and Ness (31) have shown that the peptidylpropyl isomerase Cyp40 can bind to the c-Myb but not v-Myb DBD and that its enzymatic activity could inhibit binding to DNA. Thus further work will be required to determine the mechanism by which nucleolin regulates Myb transcriptional activity, for example through the use of nucleolin mutants defective in specific functions, and is beyond the scope of this article.
Both R2 and R3 form three helices, the third helix being the DNA recognition helix (40,64). The identification of a specific residue of c-Myb, Arg 161 , required for nucleolin binding, as well as the experimental evidence that both hydrophobic and ionic interactions are involved in Myb-nucleolin interaction has led us to propose that the hydrophobic patch in close vicinity to Arg 161 as well as Arg 161 itself are directly contacting nucleolin via hydrophobic and ionic interaction respectively. Arg 161 is in the first helix of R3, and both Arg 161 and the hydrophobic patch are located on the surface of the R2R3 structure on the opposite side with respect to the site of DNA binding, making this region available for protein-protein interactions. Molecular modelling has shown that the corresponding structure of B-Myb has maintained the hydrophobic patch but lost the crucial hydrophilic Arg residue. The data presented are the first that indicate a specific function for R3 in protein-protein interactions. Mutations that had been implicated previously in the functional activity of Myb involved residues mutated in v-Myb relative to c-Myb that reside in the first two helices of R2 (31,43).
To summarize, we have identified and characterized a new protein partner, nucleolin, specific for the c-Myb and A-Myb, but not B-Myb transcription factor. We have identified a single aa residue on the surface of the R3 repeat of c-Myb, Arg 161 , which is crucial for nucleolin binding. Strong evidence is presented that nucleolin regulates Myb transcriptional activity. These data open a new area of investigation on the mechanisms of transcriptional regulation by specific transcription factors.