Self Improvement

A Novel Paired Domain DNA Recognition Motif Can Mediate Pax2 Repression of Gene Transcription

Description
The paired domain (PD) is an evolutionarily conserved DNA-binding domain encoded by the Pax gene family of developmental regulators. The Pax proteins are transcription factors and are involved in a variety of processes such as brain development,
Published
of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  A Novel Paired Domain DNA Recognition Motif CanMediate Pax2 Repression of Gene Transcription Bjarte Håvik, Erlend Ragnhildstveit, James B. Lorens, 1 Kari Sælemyr,Oddbjørn Fauske, Lill K. Knudsen, and Anders Fjose 2  Department of Molecular Biology, University of Bergen, P.O. Box 7800, N-5020 Bergen, Norway Received November 12, 1999 The paired domain (PD) is an evolutionarily con-served DNA-binding domain encoded by the  Pax  genefamily of developmental regulators. The Pax proteinsare transcription factors and are involved in a varietyof processes such as brain development, patterning of the central nervous system (CNS), and B-cell develop-ment. In this report we demonstrate that the zebrafishPax2 PD can interact with a novel type of DNA se-quences  in vitro,  the triple-A motif, consisting of aheptameric nucleotide sequence G/CAAACA/TC withan invariant core of three adjacent adenosines. Thisrecognition sequence was found to be conserved inknown natural Pax5 repressor elements involved incontrolling the expression of the  p53  and  J-chain genes. By identifying similar high affinity binding sites in potential target genes of the Pax2 protein,including the  pax2  gene itself, we obtained furtherevidence that the triple-A sites are biologically signif-icant. The putative natural target sites also provide abasis for defining an extended consensus recognitionsequence. In addition, we observed in transformationassays a direct correlation between Pax2 repressoractivity and the presence of triple-A sites. The resultssuggest that a transcriptional regulatory function of Pax proteins can be modulated by PD binding to dif-ferentcategoriesoftargetsequences.  © 1999 Academic Press The  pax  gene family encodes transcription factorswith important roles in development and oncogenesis(1–3). The highly conserved DNA binding motif, thepaired domain (PD), is the most characteristic featureof the Pax proteins which also have different combina-tions of other structures (1, 3). One additional elementis the paired-class homeodomain (HD) that indepen-dently and cooperatively with the PD can contribute toincrease the repertoire of DNA binding specificities forsome of the Pax proteins (4–6). Illustrating the impor-tance of the PD, mutations within this domain in var-ious Pax proteins are known to produce severe devel-opmental disorders in mammals (7, 8).The mammalian Pax family consists of nine mem-bers which can be divided into four groups based upontheir sequence similarity: Pax1/9, Pax2/5/8, Pax3/7 andPax4/6 (1, 3). The Pax2, Pax5 and Pax8 proteins havevery similar properties both with respect to DNA se-quence recognition and transcriptional competence (9–12). A unique structural feature of this group is thateach member contains only the N-terminal half of apaired-class HD (13, 14). Therefore, their DNA binding specificities must rely exclusively on the properties of their highly conserved PDs. Despite significant struc-tural differences between the various Pax groups it hasbeen shown that most PDs recognize similar DNA binding sequences (4, 9, 15–17).Biochemical and crystallographic analyses haveclearly demonstrated a subdivision of the PD into twosubdomains with distinct DNA binding properties (9,18, 19). Both subdomains, PAI (N-terminal) and RED(C-terminal), contain helix-turn-helix (HTH) motifsthat can make important contributions to the DNA binding specificity (19, 20). While the PAI domainseems to be the most critical subdomain as it always isable to bind to DNA (9, 18, 21), the role of the REDdomain is different in various Pax proteins. For thePax2/5/8 and Pax6 proteins RED seems only to have anaccessory function in DNA binding (4, 9). The under-standing and importance of this interaction was re-cently clarified by the crystal structure of the Pax6PD-DNA complex (20).One Pax gene that has provided important informa-tion about the process of brain development is ze-brafish  pax2  (13). This Pax gene is expressed in severalregions of developing embryos including the mid-hindbrain border (MHB) where it participate withother members of the same Pax group to establish aborder between the two parts (10, 11, 22). The func- 1 Present address: RIGEL Inc., 240 East Grand Avenue, South SanFrancisco, CA 94080. 2 To whom correspondence should be addressed. Fax: 47-55589683.E-mail: anders.fjose@mbi.uib.no.Biochemical and Biophysical Research Communications  266,  532–541 (1999) Article ID bbrc.1999.1854, available online at http://www.idealibrary.com on5320006-291X/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.  tional importance of the zebrafish Pax2 protein in thisregion was demonstrated by its loss of function mutant no isthmus  that consequently lacks the corresponding region of the brain (23). The development of the MHBregion is also believed to include vertebrate homo-logues of the Drosophila  wingless  and  engrailed  geneproducts, that participate with Pax proteins in an evo-lutionarily conserved regulatory network (2, 24, 25).Studies of other Pax proteins have highlighted theimportance of the PD in regulating the behavior of theentire Pax protein. For Pax5 the ability to repress oractivate two specific target genes was shown to bedependent upon the nucleotide sequence of the partic-ular PD DNA binding sites (26). Moreover, the PD inPax3 has been demonstrated to regulate the potency of its own transactivational domain (27). Conversely,structural constraints outside the PD have been shownto affect the PD DNA-binding potential of the quailPax6 homologue (28, 29). These observations suggest afunctional interaction between the PD and other partsof the protein that requires further investigation.The Pax proteins are known to function both asactivators and repressors of transcription and thisproperty is critical for a successful developmental pro-cess, in which the PD seems to be an important keyregulator (26). In order to clarify the role of the PD inthis process, we have studied the DNA-binding poten-tial of the zebrafish Pax2 PD in more detail. We showin this report that the PD can interact with a novelcategory of DNA sequences. Moreover, transcriptionalassays provided evidence for a functional role of theserecognition sequences in mediating Pax2 repression of gene transcription. MATERIALS AND METHODS Systematic evolution of ligands by exponential enrichment(SELEX) assay.  The selection was performed as described by Wil-son  et al.  (5) with the following modifications. The binding reactionswere at 4°C with 50 mM KCl (SELEX 50-4), or at room temperaturewith three different salt concentrations (SELEX 50, 50 mM KCl;SELEX 100, 100 mM KCl or SELEX 200, 200 mM KCl) in 10 mMTris-HCl, pH 7.5, 0.5 mM DTT, 100 ng herring sperm DNA and 2   g BSA in a total of 50   l and 100 ng of random sequence oligonucleo-tides (see below) with molar excess of the GST-Pax2 PD fusionprotein included. Several rounds of selection/amplification were car-ried out for 200 mM KCl, 5 rounds; 100 mM KCl, 8 rounds; 50 mMKCl, 8 rounds. The last PCR reaction products were blunted, shotguncloned in pGEM7zf(  ) and individual clones were sequenced. Therandom sequence library was prepared with an oligonucleotide witha core of 25 N, flanked by two PCR primer annealing sites (MM13R/ F): GTGAATTGTCGAC(N) 25  AAGCTTGGCGTA.  Electrophoretic mobility shift assays (EMSAs).  EMSAs were per-formed with [   32 P]ATP kinase labeled target DNA in 10 mM Tris-HCl, pH 7.5, 100 mM KCl, 0.5 mM DTT, 0.05% NP-40, 2.5% ficoll,10–100 molar excess poly(dI:dC) to target DNA and incubated withpurified GST fusion proteins in a total of 20   l. After incubation atroom temperature for 20 minutes, protein-DNA complexes were sep-arated on a 5% polyacrylamide gel in 0.5  TBE buffer at 10 V/cm. The  pax2  5  UTR DNA probe was generated by PCR, using the following [   32 P]ATP kinase labeled primer sets: 5   primer, GATCGAATTC-CCATTCTTGAA; 3  primer, 5  -GCGTAATCAGTCGTTGAG. The PDinteraction region of oligonucleotides used in EMSA are listed in thetext. Complete details can be obtained on request. Cloning, expression and purification of GST-PD fusion proteins. GST fusion proteins encoding amino acids 1–154 (Pax2 PD), 1–104(Pax2 PAI), and 75–154 (Pax2 RED) of the zebrafish Pax2 PD werePCR cloned and inserted into the BamHI/EcoRI site of pGEX-2T.Expression and affinity purification of the fusion proteins were per-formed according to Smith and Johnson (30).  Footprinting assays.  Missing base footprinting were conductedasdescribedbyOrding   etal. (31),underbindingconditionsoptimizedin EMSAs. DMS protection footprinting assays were performed byadding 1   l DMS (Merck) to 50   l binding-reactions (containing   10 6 cpm labeled probe) for the optimized time and stopped by adding NaAc and 2-mercaptoethanol to final concentrations of 0.3 M and1.4%, respectively. Free and bound DNA were subsequently separatedon a 5% polyacrylamide gel in 0.5  TBE buffer at 10 V/cm and treatedas for the other footprinting assays. Interference footprinting of B1-A and B3-A was conducted for monomeric PD interactions to the fol-lowing oligonucleotides (one strand shown): B1-A, CCCCTCAGA-CTGTTCAGACCAGAAACGCTTCTCCACATCGAGGAATT; B3-A,TCTGCTCCAGTGTCCTCTGGTGAAACTGCAGTGTAAATT-CCTCGATG. DMS protection assay was performed on a 82 bp frag-ment generated by annealing of the two oligonucleotides followed byKlenow treatment. Generation of the B4 cell line.  The  pax2  expressing cells (B4 cells)were generated by using a retroviral vector. Briefly, the entire ze-brafish  pax2  cDNA (13) was cloned into the pBabe-puro vector (32)and retrovirus generated by transient transfection (33). NIH-3T3cells were infected in the presence of 5   g/ml polybrene and selectedin 1   g/ml puromycin. Individual subclones were isolated by limiting dilution cloning and assessed for  pax2  expression by EMSA. Construction of reporter plasmids.  pGL-B was constructed bysubstitutingthe5  UTRofthepGL3-Controlreportervector(HindIII/ NcoI fragment) from Promega, with a HindIII/NcoI fragment of thezebrafish  pax2 cDNA(nt120–625).InpGL-Bprothesame  pax2 DNA fragment was inserted in the multiple cloning site of pGL3-Promoter. Positive recombinants were identified by sequencing. Cotransfection assays.  NIH-3T3 cells were transfected by Lipo-fectamine essentially as described by Life Technologies. Cells weregrown in DMEM/10% fetal calf serum (FCS; 35 mm wells) to 60%confluence. The transfection mix (2   g DNA and 10   g Lipofectaminein a total of 200   l DMEM, incubated for 30 minutes) was carefullyadded to NIH-3T3 cells covered by 800   l DMEM/10% FCS andincubated for 6 h before adding DMEM/FCS to 4 ml. Medium waschanged after 24 h. After additional 24 h, the cells were harvestedand assayed for luciferase as described for the Promega Luciferase Assay System. Tranfections were carried out with a CMV-LacZ ex-pression plasmid as an internal control in addition to the reporterconstructs. All luciferase values were related to the transfectionefficiency measured by the LacZ activity. Site-directed mutagenesis.  The triple-A cores of B1-A and B3-A inpGL-B were changed to ACT and CAG (underlined below), respec-tively, according to the protocol of Weiner  et al.  (34). The mutationswere generated by the following synthetic oligonucleotide:5  -AGACCAGACTCGCTTCTCCAC-ATCGAGGAATTTACACT-GCAGCTGCACCAGAGGACA-3  . Subsequent verification by DNA sequencing identified the two described constructs, pGL-Bmut1 andpGL-Bmut2. RESULTS AND DISCUSSION  Identification of optimal PD DNA ligands.  Thehigh degree of sequence diversity between different PDbinding sites suggests that they hold information that  Vol. 266, No. 2, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS533  extend beyond their function in merely providing aprotein docking site on DNA. Consistent with such aproposal two different monomeric PD binding siteshave been found to function as Pax activator and re-pressor elements  in vivo,  respectively (26). Hence, if PD binding sites contain the information that governthe activator and repressor functions of Pax proteins,or at least provide one way for such a regulation, itsuggests that the PD can recognize and distinguishbetween different DNA motifs. This is however notdemonstrated to date and only one PD DNA consensussequence is known although some natural sites sharemarginal similarity with it. Hence we propose that thefull DNA-binding potential of the PD has not yet beenelucidated and that additional recognition motifs maybe identified by further analyzing its DNA-binding abilities under alternative buffer conditions  in vitro. Using this approach we have identified novel DNA binding properties for the zebrafish Pax2 PD.To isolate DNA ligands that interact with the Pax2PD, the PD was fused to the glutathione S-transferase(GST) protein and allowed to select ligands from poolswith DNA sequence randomized oligonucleotides(SELEX assay). Four different selection buffers weretested (SELEX 200, 100, 50 and 50-4; see Materialsand Methods). Most interestingly several PD recogni-tion motifs that shared the core DNA sequence AAAG/C frequently occurred among the DNA ligandsselected at the highest stringency conditions (SELEX 200 buffer with 200 mM KCl and room temperature). A major fraction of these sequences also shared one ex-tended motif G/CAAACA/TC (Fig. 1). Gradually reduc-ing the salt concentration, the frequency of the triple-A containing motif was halved in the selection product of SELEX 100 and finally not detectable in SELEX 50 (B.Håvik and A. Fjose, unpubl. res.). Moreover, the lowerstringency settings identified two other PD recognitionmotifs, of which one was similar to the consensus se-quence known for other PDs (15, 16; B. Håvik and A.Fjose, unpubl. res.). These observations demonstratethat the salt concentration is of major importance inSELEX assays and it may be used as a tool to reveal awider range of DNA-binding specificities.  A triple-A DNA motif is conserved in natural Paxresponse elements.  To investigate if the  in vitro  se-lected triple-A motif has any biological significance wehave looked for related DNA-sequences in known nat-ural response elements of Pax proteins. This approachwas also taken to avoid further characterization of   invitro  selected sites which may only partially have thefeatures of natural target sequences. Hence it has beennoted by others that the PDs recognize shorter DNA motifs when selecting DNA ligands from sequence ran-domized oligonucleotide pools than from chromatin-fragment precipitation (35). This property has beenproposed to reflect a tendency of PAI to dominate overRED in the  in vitro  selections (36). Among the relatively limited number of known Paxprotein target sequences, we found convincing matchesonly between the triple-A DNA motif and two naturalPD binding sites which both are natural Pax5 (BSAP)repressor elements. The AAAG/C core motif is presentin these two sites that are located in the 5  UTR of the  p53  gene (named 53-A in this report) and the  J-chain (JC) gene promoter, respectively (Fig. 2A; 26, 37). TheJC site has previously been shown to be functionallyinterchangeable with the Pax5 activator site CD19-2 FIG. 1.  Identification of a triple-A motif in binding sites selectedby the Pax2 PD. Optimal alignment of 37 sequenced DNA ligandsselected at high stringency conditions by SELEX 200. The overallminimal recognizable consensus motif AAAG/C is shared among 87%of the ligands of which the degenerated subgroup motif (G/ C)AAAC(A/T)C is the most frequent (31%). Four nondegeneratedsubgroups are identified based upon intrinsic nucleotide combina-tions within each ligand (underlined): The three subgroups CAC- AAACTC, ANNAAACA and TAAAGT, are defined by the ligands2–7, 11–15, and 23–26, respectively. In addition, a CCT DNA se-quence in nucleotide position 7–9 of ligand 18–20 consequently re-sults in an extended recognition motif. Nucleotides that correspondto the numbering on the top are shaded. Vol. 266, No. 2, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS534  during B cell development (26). Hence the JC site mostlikely encode information that is interpreted by thePax protein as a repression signal. The repressive na-ture of both the 53-A and JC target sites may be coin-cidental but could also link the triple-A motif to such anegative regulating function. Alternatively, the ability FIG. 2.  Comparisons of zebrafish Pax2 PD recognition sites from putative target genes with respect to DNA sequence and binding properties.(A)EightPax2PDtriple-Asitesfromthepotentialtargetgenes  pax2 (B1-A,B3-A),  eng2 (E1-A,E2-A,E3-A), hlx1 (H1-A,H2-A),and  svp[44]  (S1-A)are listed and aligned to the W1 site of the  wnt1  gene and the known Pax5 recognition sequences in the  p53  and  J-chain  genes. The alignmentrelative to the previous reported Pax2/5/8 consensus sequence (9, 17) is based on the position of the RED half site determined for E3-A and H2-A (see text). The genomic triple-A consensus sequence is derived from bases with a frequency of 50–100% (capital letters) or 40% (lowercase letters). Aconserved invitro selectedsequenceofthetriple-Acategoryisalsoincludedinthealignment(seetext).Thenumbers1–13refertothenucleotidepositions in the PAI contacting half site according to the Prd PD-DNA crystal structure (19). The CA dinucleotide important for RED interactionis underlined. (B) EMSA analysis of W1 (lane 9) and various triple-A sites (lanes 1–8) using Pax2, Pax6 and Pax7 PDs. The binding affinities arerelated to CD19-2Ains (lane 10) and 53-A (lane 11). The specificity for all interactions with Pax-2 PD was verified, using related (CD19-2Ains) andunrelated (P3) competitors in EMSA (data not shown). 100 mM KCl buffer conditions were used, and found optimal for all interactions. For thePax6 and Pax7 PDs only the relevant part of the autoradiograph containing the protein DNA complexes is shown. The molar ratios relative to thePax2PDfusionproteinareshowninparentheses(seetext).(C)AnalysisofDNAbindingpropertiesoftruncatedPax2PDs(seetext).Therelevantpart of the autoradiograph containing the protein-DNA complexes is shown. (D) EMSAs with H2-A and H2-AdelA probes using the Pax2 PD (lane1 and 2) and RED (lane 3 and 4) fusion proteins. Note the reduced affinity for complexes with H2-AdelA. PD/RED-DNA complexes (blackarrowhead) and free probe (open arrowhead) are indicated. Vol. 266, No. 2, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS535  to repress may relay on other sequences in these tworesponse elements. Identification of additional targetsites with triple-A motifs is required to clarify thisaspect and to define other features that may be com-mon to these recognition sequences.  Identification of genomic triple-A DNA sequenceswith optimal interaction to the Pax2 PD.  To solve theissue raised by the two natural targets 53-A and JCwe designed a strategy to identify candidate PD sitesin potential target genes of zebrafish Pax2. In addi-tion to  pax2  itself, several genes that are active inthe developing zebrafish brain can be considered aspotential target genes on the basis of their overlap-ping or juxtaposing expression domains relative to  pax2.  The potential target genes analyzed for hold-ing the presence of related binding sites includedzebrafish  eng2, hlx1, svp[44]   and  wnt1  (38–41).These four genes all have embryonic expression do-mains correlating with an essential site of   pax2  ac-tivity at the midbrain-hindbrain boundary. Furtherexperimental evidence connects pax2 directly or in-directly to the transcriptional regulation of   eng2  and Wnt1  (23, 25, 42). As this search was based uponlimited amounts of DNA sequence, we also requiredthat all identified sites should conform to an optimalPD consensus sequence together with JC and 53-A. Additional requirements included DNA binding as-says to demonstrate high binding affinity and spec-ificity, as well as preference for one of the PD sub-domains PAI or RED (since at least the triple-A motif is shared). Finally it was important to confirm PDinteractions with the target sequences directly byDMS footprinting (Fig. 4; DMS footprints are onlyshown for some sequences). Satisfying all criteria aDNA sequence could be defined as a Pax2 specificrecognition sequence  per ce.  By this approach weidentified a total of eight genomic binding sites withthe triple-A motif (Fig. 2A). Five of these putativePax2 target sites are located in the 5  UTRs of   eng2 (E1-A, E2-A, E3-A) and the  pax2  gene itself (B1-A,B2-A; see also Fig. 3A). This location is similar to the53-A recognition site in the  p53  tumor suppressorgene (37). The three remaining sites are distributedwithin the 3  UTRs of   hlx1  (H1-A, H2-A) and  svp[44]  (S1-A). Notably, for  wnt1  the only binding site wehave verified to date (W1; Fig. 2A) is related to thesecond motif group identified at low stringency se-lections. This recognition sequence also shows strong affinity and specificity for the Pax2 PD (Fig. 2B).Together with the 53-A and JC sites, genomictriple-A sites show a significant degree of sequencesimilarity and share a consensus sequence that ex-tends 23 base pairs (Fig. 2A). Nine nucleotides areconserved in 50–100% of the sites and eleven addi-tional positions are identical in 40% of the cases. Thetriple-A motif is located in the most conserved 3   endwhich is the only part with significant similarity to the in vitro  selected sequences.The affinity and specificity of Pax2 PD binding tothe genomic triple-A sites were tested by EMSAs inwhich comparisons were made to previously charac-terized target sites and the PDs of other Pax pro-teins. These comparative analyses were of interestsince the PDs of different Pax subgroups are knownto have optimal binding sites that differ from thecommon consensus at particular position (17, 19).Hence we compared the binding of the eight ze-brafish triple-A sites to PDs from three different Paxsubgroups, Pax2, Pax6 and Pax7 by EMSA (Fig. 2B).The high affinity mouse Pax5 binding sequenceCD19-2Ains (9) and the low affinity Pax5 binding site 53-A (37) were used as references because ze-brafish Pax2 and mouse Pax5 PDs differ in only oneamino acid that presumably has no specific contactwith DNA (20). To standardize the PD concentra-tions all binding reactions were carried out underconditions that gave 50% mobility shift of CD19-2Ains. This was achieved when larger amount of Pax6 (1.4  ) and Pax7 (7.5  ) PDs were used than forPax2. These observations are in agreement with pre-vious estimates and the degree of amino acid diver-sity between the respective PDs (9, 43). For all thetriple-A sites tested the Pax2 PD shows strongerbinding than Pax6 and Pax7 (Fig. 2B and Table 1). TABLE 1 Comparison of the Binding Affinities of the Pax2, Pax6, and Pax7 PDs to Various PD Recognition Sequences E1-A CD19-2Ai W1 H1-A E3-A E2-A H2-A B1-A 53-A S1-A B3-A Pax2 PD 100 13.6  0.3 8.2  0.2 5.6  0.2 4.9  0.2 3.2  0.2 3.2  0.1 1.7  0.1 0.61  0.03 0.45  0.01 0.44  0.02Pax6 PD 2.8  0.5 9.4  0.2 0.070  0.004Pax7 PD 0.25  0.02 1.7  0.1  Note.  The natural triple-A sites are described and aligned in Fig. 2. The high affinity CD19-2Ains and 53-A sites are used as references. All values are related to the high affinity binding between E1-A and the Pax-2 PD given as 100% activity. The values describe percentagerelationships between dissociation constants determined by three independent experiments. 5–10 binding reactions with increasing proteinconcentrations were used to estimate individual dissociation constants by Scatchard plots (data not shown). Relative affinities for somePD-DNA combinations are not listed, as they were too weak to detect or define as specific interactions. Standard deviation equals the  valueindicated. Vol. 266, No. 2, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS536
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks