Isolation and characterization of the Xenopus HIVEP gene family

Authors


T. Pieler, Abt. Entwicklungsbiochemie, Universität Göttingen, Justus-von-Liebig Weg 11, 37077 Göttingen, Germany.
Fax: + 49 551 3914614, Tel.: +49 551 395683,
E-mail: tpieler@gwdg.de

Abstract

The HIVEP gene family encodes for very large sequence-specific DNA binding proteins containing multiple zinc fingers. Three mammalian paralogous genes have been identified, HIVEP1, -2 and -3, as well as the closely related Drosophila gene, Schnurri. These genes have been found to directly participate in the transcriptional regulation of a variety of genes. Mammalian HIVEP members have been implicated in signaling by TNF-α and in the positive selection of thymocytes, while Schnurri has been shown to be an essential component of the TGF-β signaling pathway. In this study, we describe the isolation of Xenopus HIVEP1, as well as partial cDNAs of HIVEP2 and -3. Analysis of the temporal and spatial expression of the XHIVEP transcripts during early embryogenesis revealed ubiquitous expression of the transcripts. Assays using Xenopus oocytes mapped XHIVEP1 domains that are responsible for nuclear export and import activity. The DNA binding specificity of XHIVEP was characterized using a PCR-mediated selection and gel mobility shift assays.

Abbreviations
BMP

bone morphogenetic protein

BRE

BMP-4 response element

Dpp

decapentaplegic

NLS

nuclear localization signal

Shn

Schnurri gene from Drosophila

TGF-β

transforming growth factor-beta

ZAS

zinc finger, acidic, serine/threonine-rich

Znf

zinc finger

The HIVEP family of zinc finger proteins regulates a diverse array of developmental and biological processes through direct DNA binding, as well as interaction with other transcription factors and components of signal transduction pathways [1,2]. Representative members include three human genes: HIVEP1 (also called ZAS1/Shn1/MBP1/PRDII-BF1) [3–6], HIVEP2 (ZAS2/Shn2/Mbp2) [7,8] and HIVEP3 (ZAS3/Shn3) [7,9], as well as the corresponding mouse homologues αACRYBP1[10,11], MIBP1[12] and KRC[13]. Schnurri (Shn), a distantly related ortholog from Drosophila, which is most closely related to HIVEP1, has also been isolated and characterized [14–16].

Typically, the large zinc finger (Znf) DNA binding proteins have a molecular mass greater than 250 kDa and contain two ZAS domains (N and C) that are widely separated in the primary sequence [2,9]. Each ZAS domain harbors a pair of DNA binding C2H2 type zinc fingers followed by an acidic domain located in close proximity to a serine/threonine-rich sequence. Mammalian members of the HIVEP family have been implicated in transcriptional regulation via direct binding to cis-regulatory elements of several genes, including p53[17], IRF-1[17], c-myc[12], αA-crystallin[11], human immunodeficiency virus type1 long-terminal repeat[4], somatostatin receptor type II[18] and the metastasis-associated gene S100A4/mts1[19].

The HIVEP family also has cellular regulatory activities not associated with DNA binding. KRC was shown to regulate the response of the TNF receptor to proinflammatory stimuli via the interaction with the adapter TRAF2 [1]. In addition, knockout studies in mouse have demonstrated that Shn2 plays a pivotal role in the positive selection of thymocytes [20,21]. However, the molecular mechanism for this observation remains undefined.

Drosophila Shn is the most functionally characterized HIVEP member and has been shown to be essential for signaling by the TGF-β superfamily ligand, decapentaplegic (dpp), during anterior–posterior patterning of the wing [22]. Shn mutants mimic a large number of dpp loss-of-function phenotypes and mutations in the Dpp-receptors tkv and punt[15,16]. Cells that lack Shn do not respond to ectopic Dpp [14,15,23]. In response to Dpp, Shn was found to form a complex with Mad and Medea, the intracellular transducers of Dpp signaling [23,24]. Taken together, these results suggest that Shn acts as a Mad/Medea coactivator for Dpp-responsive genes. However, genetic studies have demonstrated that the primary function of Shn is to repress the transcription of brinker (brk), which serves as a repressor for many Dpp-target genes [25,26]. Shn may also cooperate with Mad/Medea to regulate additional Dpp-responsive target genes [23]. A Dpp-regulated silencer element has been identified that controls the expression of brk[27]. This silencer is regulated directly by a complex consisting of Mad/Medea and Shn. While the fundamental aspects of TGF-β signaling are highly conserved and the requirement of this pathway in embryonic patterning in both invertebrates and vertebrates is well established, a role for vertebrate Shn related transcription factors in TGF-β signaling is currently unknown. Moreover, it is also unclear whether vertebrate HIVEPs regulate cellular events through the repression of brk transcription, as vertebrate brk homologs have not yet been identified.

Presently, we describe the isolation of one complete and two partial cDNAs corresponding to three different HIVEP related genes in Xenopus (XHIVEP1, -2 and -3). The Xenopus XHIVEPs are characterized with respect to temporal and spatial expression, nuclear import/export activity and DNA binding specificity.

Materials and methods

Isolation and cloning of Xenopus XHIVEP1, -2 and -3

Screening of amplified cDNA libraries was performed by PCR screening as described previously [28]. Approximately 1.9 × 106 plaque-forming units were screened. PCR was performed in a final volume of 22.5 µL with 2.5 µL of phage lysate as template, using the Gene Amp PCR Kit (Perkin Elmer). Degenerate oligonucleotides initially used as primers were created by comparing the ZAS-C Znf from different members of the HIVEP family: (upper primer 5′-AARTAYATHTGYGARGARTGYGGIATHCG-3′ and lower 5′-CAYTTYTTCATTRGIGCYTTIGYYTTCATRTG-3) resulting in the amplification of a 173 nucleotide product. Individual positive clones were identified by serial dilution of the positive phage fractions. The initial XHIVEP1, -2 and -3 clones contained 2.3, 5.9 and 3.2 kb of cDNA, respectively, in the pBKCMV vector.

The full length XHIVEP1 sequence was obtained by a combination of additional phage screening and RT-PCR amplification affording five partially overlapping cDNA fragments of XHIVEP1. In the first amplification, a degenerate primer (Shn amino acids 1552–1560) and a XHIVEP1 gene specific primer set were used (5′-GARGAYTGYTTYGCNCCNAARTAYCA-3′ and 5′-TCCACGGATGTACACATAC-3′) to amplify a 1.5 kb product from stage 34–38 Xenopus cDNA. In the second amplification, the degenerate primer (HIVEP1 amino acids 971–980) and a XHIVEP1 gene specific primer set derived from the additional sequence obtained in the first amplification (5′-GARAAYTTYGARAAYCAYAARAARTTYTAYTG-3′ and 5′-AGTTCTAATGCTATGTTTGGATGC-3′) afforded a product of 1.7 kb. Additional screening of a Xenopus cDNA phage library with primers derived from XHIVEP1 (5′-TACTGGGGCATTAGAACAACCTT-3′ and 5′-GACATTTCACTTCCACTCTTTCTTG-3′) resulted in the identification of two partially overlapping clones containing 3.5 kb and 3.9 kb of the 5′ sequence of XHIVEP1. PCR-amplified deletion mutants for transport experiments were subcloned into pCS2+NLS-MT vector [29].

Semi-quantitative RT-PCR analysis

Total RNA from embryos and tissues was isolated by phenol/chloroform extraction and LiCl precipitation [30]. The Qiagen RNeasy Kit was used for RNA isolation from dissected gastrula stage embryos. All RNA samples were treated with DNAse I (Boehringer Mannheim) and checked by PCR for DNA contamination. RT-PCR was carried out using the Gene Amp RNA PCR kit (Perkin Elmer), and 1 µCi of [32P]dCTP[αP] was included in each PCR. One-tenth of the PCR products were separated on 6% polyacrylamide gels under denaturing conditions and analyzed using a PhosphorImager (Molecular Dynamics). Primers and conditions used for RT-PCR were as follows: XHIVEP1, 5′-ATCCAGAGGCAGAAGCAG-3′ and 5′-CTGCATTCAGAGTAAGCC-3′, 60 °C, 29 cycles; XHIVEP2, 5′-AAGCAGAGGAATGCAGTAG-3′ and 5′-AATGTCTTTCTCTCCATGG-3′, 60 °C, 29 cycles; XHIVEP3, 5′-GCAGCACTATCCCTGCTAAG-3′ and 5′-TCCCTCGTCCACGGCCTCTTACAT-3′, 60 °C, 29 cycles. Further oligonucleotides: Histone H4, 5′-CGGGATAACATTCAGGGTATCACT-3′ and 5′-ATCCATGGCGGTAACTGTCTTCCT-3′, 60 °C, 22 cycles; Xbra, 5′-GGATCGTTATCACCTCTG-3′ and 5′-GTGTAGTCTGTAGCAGCA-3′, 60 °C, 28 cycles; Gsc, 5′-ACAACTGGAAGCACTGGA-3′ and 5′-TCTTATTCCAGAGGAACC-3′, 60 °C, 28 cycles; XWnt8, 5′-TGTGGCCGGGTCTGAACTTATTTT-3′ and 5′-GTCATCTCCGGTGGCCTCTGTTCT-3′, 60 °C, 28 cycles.

Microinjection of Xenopus oocytes and analysis of nuclear transport

[35S]Methionine radiolabelled proteins were expressed from cDNAs using the coupled transcription/translation (TNT) system (Promega). In vitro translation products were analyzed by SDS/PAGE and phosphoimaging (Molecular Dynamics). Preparation of oocytes and microinjection assays were performed as described in [31]. Immunoprecipitation was performed as described [32]. Phosphatase treatment of immunopellets was performed with 100 U of λ-phosphatase (NEB) per pellet for 1 h in the appropriate buffer.

In vitro protein preparation

The Znf pair derived from the XHIVEP2 ZAS-N domain (234 bp, amino acids GGFK…KCLE) was cloned in-frame with the N-terminal His-tag of the pRSET vector (Invitrogen). Hexa-His-tagged ZAS-N Znf was expressed in Escherichia coli BL21, induced with CE3 lysogen according to manufacturer's instructions (Stratagene). The fusion protein was purified under native conditions using Ni/nitrilotriacetic acid/agarose (Qiagen) according to the manufacturer's protocol. The purified protein was quantified by the Bradford method.

Electrophoretic mobility shift assays

DNA duplexes were labeled on the upper strands with [32P]ATP[γP] and T4 polynucleotide kinase. The labeled oligomers were annealed by heating to 90 °C an equimolar mixture of the upper and lower strands in reaction buffer and cooling slowly to ambient temperature (1 h). Sequences of the upper strand of the duplexes are listed below. Sites of mutation are underlined: wt, 5′-AGAGAGAATGAGAGGCTTCCCAATAGC-3′; mut1, 5′-AGAGAGAATGATAGGCTTCACAATAGC-3′; mut2, 5′-AGAGAGAATGATAGGCTTCCCAATAGC-3′; mut3, 5′-AGAGAGAATGAGAGGCTTCACAATAGC-3′.

Binding reactions were performed in a total volume of 50 µL containing 50 mm Tris/HCl, pH 8.0, 30 mm KCl, 10 mm MgCl2, 30 µm ZnCl2, 1 mm dithiothreitol, 10% glycerol, 1 µg poly(dI-dC) and 100 µg BSA. The hexa-His-tagged ZAS-N Znf concentration used was 84 or 214 ng. The reactions were allowed to proceed for 30 min at 4 °C and analyzed on a 12% native polyacrylamide gel containing 0.5× Tris-borate buffer (run at 300 V at 4 °C).

In vitro selection

PCR-based site selection was performed essentially as described [33] with bacterially expressed ZAS-N Znf and a 16 nucleotide degenerate DNA duplex. Binding reactions were performed as described above. After seven rounds of binding, recovery of shifted DNA and PCR, the targets were cloned and sequenced.

Results

Isolation of HIVEP genes in Xenopus

Xenopus embryonic tailbud stage head and tailtip cDNA libraries [34] were screened for HIVEP related genes in a PCR-approach using degenerate primers deduced from the second Znf pair within Drosophila Shn. Three cDNAs of Xenopus HIVEP related genes XHIVEP1, -2 and -3 were isolated. Overlapping clones covering 8578 bp of XHIVEP1 cDNA were obtained by RT-PCR on total embryonic RNA using combinations of degenerate and specific primers and by rescreening the cDNA libraries. The partial clones of XHIVEP2 and -3 covered 5.9 and 3.2 kb of the respective 3′ ends and included 3′-UTRs and poly(A)-tails.

A GenBank search revealed highest homology of the three deduced proteins, XHIVEP1, -2 and -3, with the mammalian zinc finger proteins HIVEP1, -2 and -3, respectively (Fig. 1). Similar to other vertebrate HIVEP related proteins, the XHIVEP proteins lack the C-terminal Znf triad found in Drosophila Shn, but exhibit between 75 and 92% homology to Znf pairs within the two ZAS domains of Shn (Fig. 2). Compared to the corresponding vertebrate proteins, the Xenopus ZAS Znf DNA binding domains and the isolated Znf has between 96 and 100% identity (Fig. 2). The regions outside these domains exhibit lower sequence identities in a comparison of the three vertebrate proteins (40–60%), although regions of higher sequence conservation are distributed over the proteins, including several serine-rich stretches [9].

Figure 1.

Structural organization of the XHIVEP proteins. Black bars indicate the position of the ZAS C2H2 zinc fingers (ZAS-N and ZAS-C) and the isolated C2HC zinc finger (I-Znf). The ZAS zinc finger DNA binding domains are followed by acidic domains indicated by white bars. The Znf triad unique to Schnurri is indicated in the figure by an asterisk. The level of sequence conservation between the respective domains within Xenopus proteins and their human orthologs are indicated as percentages. Within the Drosophila protein, the relative positions of the oligomerization domain and a Mad interaction domain are indicated. dm, Drosophila melanogaster; hs, Homo sapiens; xl, Xenopus laevis; aa, amino acids.

Figure 2.

Sequence comparison of conserved zinc fingers within HIVEP type proteins. The predicted primary sequences of the Xenopus (xl) HIVEP ZAS domain zinc fingers (ZAS-N and ZAS-C) and the isolated zinc finger (I-Znf) are shown in comparison with the corresponding domains from human (hs), mouse (mm), rat (rn) and fly (dm). Amino acids involved in complex formation with zinc ions are marked in bold and amino acids within zinc fingers that are likely to be involved in DNA binding are boxed. Identical amino acids are represented by dashes.

The 8578 bp cDNA sequence of XHIVEP1 contains 242 bp of the 5′-UTR, an open reading frame of 7734 bp and 602 bp of the 3′-UTR. The deduced 2578 amino acid protein has two C2H2 type Znf containing ZAS domains and a single C2HC type Znf (Fig. 3). The reported start and stop codons, as well as the Znf sequences of XHIVEP1, correspond to those of mammalian HIVEP1. Overall amino acid sequence identity between XHIVEP1 and the corresponding human and mouse sequences is 50% and 70%, respectively. XHIVEP1 is likely to be post-translationally modified, as 10% of all amino acids constitute putative target sites for a wide array of different Ser-, Thr- and Tyr-kinases (http://www.expasy.org).

Figure 3.

Amino acid sequence of the XHIVEP1 protein. The protein sequence of XHIVEP1 was predicted from five overlapping cDNA fragments (GenBank Accession number AY363297). Zinc finger domains are shaded in gray and acidic-rich regions are boxed in black. The serine-rich regions are underlined with dashed lines. Putative nuclear localization signals are underlined in black and the putative nuclear export signal is underlined with a dotted line.

Expression of Xenopus HIVEP transcripts

To determine if the different XHIVEP genes are differentially expressed, their temporal mRNA expression patterns were analyzed by semiquantitative RT-PCR analysis using total RNA isolated from various stages of Xenopus embryos. As shown in Fig. 4A, XHIVEP1, -2 and -3 transcripts are maternal and continue to be detected at similar levels until the onset of gastrulation (egg until stage 10). Throughout late gastrula and neurula stages (stages 11–20), expression decreases temporarily and increases again at stage 24. In adult tissue, XHIVEP transcripts were detected at comparable levels in all tissues examined (Fig. 4B).

Figure 4.

Temporal and spatial expression of XHIVEP mRNAs. Semiquantitative RT-PCR analysis was performed with RNA isolated from staged embryos (A), adult organs and tissues (B) and dissected regions of stage 10 embryos (C) making use of primers specific for either XHIVEP1, XHIVEP2, XHIVEP3, Wnt8, Gsc, Xbra or Histone H4. Abbreviations: a, animal pole; bl, bladder; br, brain; d, dorsal; E, embryo; e, egg; ey, eye; fa, fat tissue; gu, gut; he, heart; in, intestines; ki, kidney; l, lateral; li, liver; lu, lung; mu, muscle; ov, ovary; ph, pharynx; -RT, without reverse transcriptase; sc, spinal chord; sk, skin; sp, spleen; te, testis; v, ventral.

Attempts to analyze the spatial expression of XHIVEP mRNAs by whole mount in situ analysis on Xenopus embryos revealed only a weak expression suggesting low abundance of the transcripts (data not shown). As Schnurri has been implicated in TGF-β signaling, we further investigated the spatial expression of the XHIVEPs during gastrulation, when these signals play an essential role in patterning of the mesoderm. Early gastrula stage embryos were dissected and total RNA isolated from pools of seven defined regions shown in Fig. 4C. Semiquantitative RT-PCR analysis revealed that the three XHIVEP transcripts are ubiquitously present at this stage. In comparison, the mesodermal marker genes XWnt8, Xbra and Gsc showed the expected restricted expression patterns (Fig. 4C) [35].

Mapping of Xenopus HIVEP1 import and export domains

Several members of the HIVEP family have been shown to be nuclear transcriptional regulators [6,23,24]. In order to analyze the in vivo subcellular localization of the 300 kDa XHIVEP1 protein, we used the Xenopus oocyte system. Myc-tagged fragments of the XHIVEP1 protein were translated in vitro in the presence of [35S]methionine and the radiolabelled protein fragments were microinjected into Xenopus oocytes (stage V and VI). The XHIVEP1 fragments (F1–F5) that were used are shown schematically in Fig. 5A. To evaluate import and export activity, the protein fragments were injected into the cytoplasm or the nucleus, respectively. At different time points, nuclear and cytoplasmic fractions were prepared, the labeled proteins immunoprecipitated, resolved by SDS/PAGE and visualized by autoradiography.

Figure 5.

Delineation of nuclear import and export domains within XHIVEP1. (A) Schematic representation of the full length XHIVEP1 and the Myc-tagged (MT) deletion mutants. Black and white boxes indicate the position of the Znf domains and the serine-rich stripe, respectively. The relative positions of putative nuclear localization signals (NLSs) are indicated by an asterisk and the nuclear export signal by a plus symbol. The amino acid residues contained in each fragment are indicated to the right of each mutant. (B) To map nuclear transport regulatory domains within XHIVEP1, 35S-labeled XHIVEP1 deletion mutants were produced in vitro and microinjected into the nucleus or the cytoplasm of Xenopus oocytes. Immediately, or after an incubation of 6 or 24 h, nuclear (N) and cytoplasmic (C) fractions were manually separated and analyzed for XHIVEP1 protein content by immunoprecipitation and SDS/PAGE. (C) Sequence comparison of five putative NLS sequences of the SV-40 type within the XHIVEP1 protein. Consensus sequences for the NLS are shaded and basic amino acids are indicated in bold. NLS1–5 contain classical NLS characterized by a K(K/R)X(K/R) consensus sequence. The bipartite sequence contains two adjacent basic amino acids followed by a spacer containing 10 amino acids and at least three basic residues in the subsequent five positions (NLS6 and 7). Position of the terminal amino acid for each of the depicted sequences is indicated to the right of each sequence. (D) Amino acid sequence of a hydrophobic putative nuclear export signal sequence within XHIVEP1. Hydrophobic amino acids are indicated in bold. The position of the terminal amino acid for the depicted sequence is indicated to the right.

To evaluate for import activity, the labeled proteins were injected into the cytoplasm (Fig. 5B, Import). As shown in the control at time 0 h, labeled protein fragments are detected only in the cytoplasm, demonstrating appropriate targeting. F4 and F5 were maintained exclusively in the cytoplasm, even after 24 h. In contrast, the amino-terminal protein F1 was strongly imported to the nucleus. The internal fragments F2 and F3 were also imported to the nucleus, albeit weakly, with both fragments detected predominately in the cytoplasm even after 24 h. Moreover, the F3 protein band appeared as a blurred band after isolation from the Xenopus oocyte, suggesting post-translational modification such as phosphorylation of the fragment. Correspondingly, treatment of the immunoprecipitated proteins with λ-phosphatase prior to loading on the gel resolved the blur into a sharp band.

To identify fragments containing nuclear export activity, the labeled proteins were injected into the nucleus (Fig. 5B, Export). The F2 and F3 proteins, which exhibited weak import activity, were not exported from the nucleus. The strongly imported N-terminal F1 protein displayed weak export activity. In contrast, the C-terminal F4 and F5 were strongly exported from the nucleus, with 50% of the labeled protein becoming cytoplasmic after 6 h and exclusively located in the cytoplasm after 24 h.

DNA binding specificity of Xenopus HIVEP

Schnurri proteins are known to have DNA binding activity; therefore, preferential DNA binding sites of a Znf pair derived from the XHIVEP2 ZAS-N domain was determined in a PCR based in vitro site selection assay [33]. As the amino acids that confer DNA binding specificity are conserved among the XHIVEPs (Fig. 2), it is therefore anticipated that they have similar DNA binding activities. A duplex DNA library, containing 16 base pairs of degenerate sequence flanked by known sequences that contained restriction sites and served as primer binding sites, was used as a substrate for the bacterially expressed ZAS-N Znf in electrophoretic mobility shift assays. The protein–DNA complexes were recovered from the gel and used in successive rounds of amplification and selection.

After seven rounds, the selected DNA duplexes were subcloned and 49 clones were sequenced (Fig. 6A). In all of the clones analyzed, a CCC trinucleotide was present. Many sequences also contained a TG or TT dinucleotide immediately upstream from the invariant CCC sequence and displayed a preference for GC-rich sequences upstream of this motif. The isolated pool was also enriched in sequences having a GAGA or GACCG. These motifs were often overlapping and the GAGA and GACCG sequences were located with a variable distance of 6–8 nucleotides and 3–4 nucleotides, respectively, upstream of the invariant CCC trinucleotide. In addition, one sequence was represented five times (Fig. 6A).

Figure 6.

Sequence-specific binding by XHIVEP. (A) Comparison of target sites for the Znf pair derived from the XHIVEP2 ZAS-N domain, as determined by PCR site selection using a DNA duplex degenerate in 16 positions, flanked by sequences for PCR amplification. After seven cycles, the DNA sequences were cloned. In total, 49 clones were sequenced and aligned in reference to the CCC trinucleotide that was found in all sequences (left). The bars indicate the frequency of the nucleotides at each position. On the right, the abundance of specific sequences upstream of the CCC is indicated. In addition, one sequence that was identified five times is shown. (B) Specific binding of ZAS-N Znf to the BMP-4 response element of the Xvent-2B promoter. Nucleotides of the Xvent-2B promoter protected by ZAS-N Znf in DNase I footprinting experiments are underlined in the wild type (wt) duplex and the GAGA and CCC sequences are boxed in gray. The nucleotides that were mutated are indicated by unfilled boxes. DNA electrophoretic mobility shift analysis comparing ZAS-N Znf binding to a wild type 27 bp duplex spanning the protected region of the Xvent-2B promoter and with that of the same duplex containing a mutation in either the GAGA box or CCC trinucleotide motifs are shown on the right.

The finding that HIVEP members directly interact with members of the Smad family, led us to investigate TGF-β responsive elements for Shn binding sites [21,23,24]. One well characterized TGF-β responsive promoter is that of Xvent-2B[36,37]. In vitro DNase I footprinting experiments demonstrated that ZAS-N Znf protected the region between −280 and −260 located at the 5′-end of the bone morphogenetic protein (BMP)-4 response element (BRE) (data not shown). This BRE has previously been shown to contain Smad1 and Smad4 binding sequences and is sufficient to drive expression in the early Xenopus embryo in a similar manner to that of the endogenous gene [36,37]. The protected region within the characterized BRE of the Xvent-2B promoter (Fig. 6B) resembles preferred sequences identified by in vitro selection. This sequence has, at its core the invariant CCC and an upstream GAGA box. To evaluate the contribution of these elements to ZAS-N Znf binding, mutations were created in either the GAGA box or the trinucleotide CCC sequences in a 27-mer duplex spanning the protected region. The binding of the ZAS-N Znf to the mutated and the corresponding wild type duplexes was evaluated in electrophoretic mobility shift experiments (Fig. 6B). While ZAS-N Znf bound strongly to the wild type duplex, binding was completely abolished in the duplex containing mutations in both the GAGA and the CCC sequences (Fig. 6B, compare lanes 2 and 3 with 5 and 6). Mutation of the GAGA motif only slightly altered the ZAS-N Znf binding compared with the wild type duplex (Fig. 6B; compare lanes 2 and 3 with 8 and 9). In contrast, the mutation of the CCC alone significantly disrupted binding demonstrating the essential contribution of this motif for binding (Fig. 6B; lanes 11 and 12).

Discussion

The central components of the TGF-β pathway, including ligands, receptors and intracellular signaling molecules, are highly conserved. In vertebrates as well as in insects, TGF-β signaling is crucial during early patterning of the embryonic mesoderm. The finding that in Drosophila, the large nuclear multizinc finger transcription factor Schnurri, related to the vertebrate HIVEP family, functions to interpret the intracellular signaling of Dpp, prompted us to analyze a functional conservation of Schnurri related proteins in vertebrates [26].

In a homology screen, we identified three Xenopus laevis HIVEP related cDNAs, XHIVEP1, -2 and -3, which show high similarity with the corresponding mammalian HIVEP genes. The overall structure of the three Xenopus proteins with their respective orthologs from vertebrates is well conserved (Fig. 1), while sequence conservation outside the Znf domains and in a number of other regions is much lower, even among the mammalian orthologous proteins. HIVEP and the Drosophila Schnurri proteins contain two pairs of C2H2 Znf and, with the exception of the HIVEP2 family, a conserved C2HC-type Znf. In addition, Drosophila Schnurri contains a conserved carboxyl-terminal Znf triplet that is not found in the vertebrate members. Sequence conservation between vertebrate HIVEP and the Drosophila Schnurri proteins is generally low with the exception of the two ZAS domains, which are highly conserved (Fig. 2). An additional stretch of 31 amino acids in XHIVEP1, located between the ZAS-N and isolated zinc fingers (amino acids 703–733), is also weakly conserved between HIVEP1/2 and Drosophila Schnurri proteins. A larger protein fragment of Schnurri that contains this sequence element was shown to form homo-oligomers in vitro[24]. Our data indicate that the HIVEP/Shn protein family has retained remarkable conservation in their overall structure as well as in the sequence of specific domains in different vertebrate species.

Accumulating experimental evidence supports that the HIVEP proteins are nuclear transcription factors. Drosophila Schnurri was localized in the nucleus after transfection of COS cells [23,24], and the endogenous human HIVEP (PRDII-BF1) protein was detected in the nucleus of MG63 cells [6]. Visual inspection of the full length XHIVEP1 protein revealed the presence of five classical nuclear localization signal sequences (NLS1–5) of the SV40 type with the basic core sequence K(K/R)X(K/R) [38] (Fig. 5A,C). All of the classical NLSs, with the exception of NLS1, are conserved between mammalian and Xenopus HIVEP1 proteins. Also found within the XHIVEP1 sequence, are two bipartite NLS motifs that are not present in the corresponding mammalian XHIVEP1 proteins (NLS6 and 7). This NLS motif is characterized by a stretch of DNA containing two adjacent basic amino acids (K or R) followed by a spacer of 10 residues and at least three basic residues in the five subsequent positions [39].

Using labeled XHIVEP1 protein fragments in nuclear import and export assays in the Xenopus oocyte, we were able to gain further insights into the regulation of HIVEP1 subcellular localization. While an amino-terminal fragment (F1) containing four putative NLSs was strongly imported into the nucleus, the internal fragments F2 and F3, which harbored one and two putative NLSs, respectively, were only weakly imported (Fig. 5B). Interestingly, NLS4 is located adjacent to a serine-rich sequence element that may have caused phosphorylation of protein fragment F3 in the oocyte (Fig. 5B). The close proximity of the NLS to the serine-rich region suggests that it may be regulated by phosphorylation. Site-directed mutagenesis of the putative NLS should unambiguously identify the motifs that are responsible for XHIVEP1 nuclear localization. It is however, apparent from the deletion studies that multiple motifs are capable of localizing XHIVEP1 to the nucleus.

Experiments in which the protein fragments were injected into the nucleus revealed that two overlapping fragments of the carboxyl terminus of XHIVEP1 (F4 and F5) were strongly exported from the nucleus. Nuclear export signals are frequently composed of hydrophobic leucine-rich sequences [40–42]. Within the carboxyl terminus of XHIVEP1 (F4 and F5), a hydrophobic stretch of 21 amino acids length could be identified that contains a high content of leucine and isoleucine residues (Fig. 5D). This region is also conserved in the mouse and human HIVEP1 proteins.

The presence of import as well as export activity, located at opposite ends of HIVEP1, could enable the protein to undergo nucleocytoplasmic shuttling. Post-translational modification at numerous phosphorylation sites may also regulate the localization of the protein.

At the mid-blastula transition, TGF-β ligands, their receptors and Smad mRNAs are ubiquitously expressed, and their expression patterns are refined during gastrulation in those regions where the corresponding pathways are active [44,45]. We found XHIVEP mRNAs to be expressed maternally and maintained until the onset of gastrulation, at which point they are distributed equally throughout the embryo. The corresponding proteins can therefore be expected to be present at the right time and place to function as mediators of TGF-β signaling during mesoderm patterning events. Consistent with a function of HIVEP members in regulating TFG-β signaling is the finding that both vertebrate and invertebrate proteins can associate with the Smads [21,23,24]. While we were not able to obtain reproducible in vivo interaction data between XHIVEP1 and Smad proteins, we observed an in vitro interaction with 35S-labeled XHIVEP1 and bacterially expressed GST-Smads (data not shown).

The DNA binding specificity of the XHIVEPs was evaluated in a PCR site selection experiment using a Znf pair derived from the XHIVEP2 ZAS-N domain (ZAS-N Znf). All 49 clones that were sequenced contained a CCC trinucleotide. We also observed a preference for a TG or TT dinucleotide immediately upstream of the invariant CCC sequence. The isolated pool was also enriched in sequences having a GACCG or GAGA motif at a variable distance from the CCC trinucleotide. Most vertebrate HIVEP proteins and Drosophila Schnurri were also shown to bind GC-rich sequences related to the NFκB related enhancer motifs with the consensus target sequence GGG(N)4−5CCC [13]. Such sequences are present in cis-regulatory regions of promoters involved predominantly in immune response, and the HIVEP1 protein has been shown to activate transcription of the human immunodeficiency virus enhancer in human [46] and the αA-crystallin gene in the mouse [11]. However, many of the HIVEP Znfs have also been shown to bind additional unrelated sequences. HIVEP3/KRC Znf has been shown to exhibit dual DNA binding specificity, binding to both the NFκB related enhancer and to the V(D)J recombination signal sequence elements [47–49].

With the intention to search TGF-β responsive promoter elements in Xenopus for XHIVEP binding sites, we could identify an optimal target site for the ZAS-N Znf that closely resembles the known mammalian consensus sites. An element that is similar to the one identified in vitro was found within the BRE of the Xvent-2B promoter and is located adjacent to the immediate BMP-responsive region of the 5′ flanking region of Xvent-2B[37]. DNase I footprint analysis and gel shift assay with the wild type and a mutated duplex confirmed the specific binding of ZAS-N Znf to this sequence and demonstrated the essential nature of the CCC sequence for DNA binding.

The physiological relevance of the interaction of XHIVEP with the Xvent-2B promoter is not known. Luciferase reporter assays with the BRE and the corresponding mutations that disrupt ZAS-N Znf binding demonstrated that during gastrulation the reporter was still responsive to BMP signaling (data not shown). While the mutations were sufficient to disrupt binding of the zinc finger pair, they may not be capable of inhibiting binding of the full length protein. Additionally, it has been shown in Drosophila, that the Dpp-mediated early patterning of the dorsal-ventral axis is independent of Schnurri activity [26].

To analyze the function of XHIVEP transcription factors in BMP signaling in the Xenopus embryo in more detail, we performed in vitro transcription of the full length 300 kDa XHIVEP1 for use in microinjection experiments (data not shown). Unfortunately, premature in vitro transcription termination events at several distinct sites within the 8 kb synthetic mRNA led to the production of predominately truncated mRNAs. Thus, there was an insufficient quantity of full length mRNA transcripts for microinjection experiments. Attempts to eliminate the termination sites by silent mutations in the affected regions were not successful. We have also performed injection experiments with mRNA encoding fusions of ZAS-N Znf to VP16 activator and En repressor domains to analyze the function of XHIVEP in the context of Xenopus embryogenesis (data not shown). However, the interpretation of the in vivo role of XHIVEP was not conclusive as the activator and repressor fusion constructs gave similar effects in various functional assays. Therefore, to gain further understanding of the function of the extremely large XHIVEP1 by over expression in Xenopus embryos, it may be necessary to create specific dominant negative and constitutively active constructs by the generation of deletion mutants containing discrete functional domains of XHIVEP1. Thus, the cloning and characterization of the XHIVEP interacting factors would be of interest and should also provide additional insight into the function of this protein in early development and further elucidate its role in TGF-β signaling in the vertebrate embryo.

Acknowledgements

The authors would like to thank Susanne Loop for assistance in the transport experiments, Dr Sepand Rastegar for performing promoter reporter assays, and acknowledge the technical assistance of Y. Harbs. This work was supported by funds from the Deutsche Forschungsgemeinschaft to T. P. (SFB 523-A1) and W. K. (SFB 497-A1).

Ancillary