AND-1, a natural chimeric DNA-binding protein, combines an HMG-box with regulatory WD-repeats

The nucleoplasm is a large subcellular compartement containing a pool of soluble proteins, protein complexes and particles that can be rapidly recruited, in a regulated way, to serve specific nuclear functions such as chromatin formation, replication, transcription, ribonucleoprotein assembly, polyadenylation and splicing. With increasing availability of amino acid sequence information about such nucleoplasmic proteins, it has become clear that they can be grouped according to certain functional roles or functional sequence motifs. Examples of important groups of the latter category of nuclear proteins include those characterized by zinc-binding domains (reviewed by Harrison, 1991; Saurin et al., 1996), leucine zipper (reviewed by Busch and Sassone-Corsi, 1990; Harrison, 1991), WD-repeats (reviewed by van der Voorn and Ploegh, 1992; Neer et al., 1994; Neer and Smith, 1996) and HMG-boxes (reviewed by Johns, 1982; Landsman and Bustin, 1993; Laudet et al., 1993).

In the present study we report the first example of a nu-cleoplasmic protein (M_r 125,000) that is acidic and combines hallmark features of two different regulatory protein families, the DNA-binding HMG-box and the WD-repeat protein families. We show that this protein, which can bind intensely to DNA, is a widespread, probably ubiquitous nucleoplasmic constituent, and we discuss the possibility that its domain dualism reflects dual or even multiple regulatory interactions.

Clawed toads (Xenopus laevis) were purchased from the South African Snake Farm (Fish Hoek, South Africa). Cultured cells derived from Xenopus laevis kidney epithelium (XLKE, line A6; American Type Culture Collection, ATCC, CCL 102) were grown at 27°C in 78.3% Dulbecco’s minimal essential medium (DMEM; ICN Biomedicals, Costa Mesa, CA, USA), supplemented with 8.7% FCS and 13% H₂O (e.g. 90 ml DMEM + 10 ml FCS + 15 ml H₂O). Cultured cells of the human hepatocellular carcinoma line PLC (ATCC, CRL 8024) were grown at 37°C in 90% DMEM supplemented with 10% FCS.

Nuclei of mature (stages IV-VI; Dumont, 1972) Xenopus laevis oocytes were obtained by mass isolation, a large-scale procedure described by Scalenghe et al. (1978) and modified by Kleinschmidt and Franke (1982). The subsequent fractionation of the intact nuclei was performed as described by Hügle et al. (1985), resulting in a ‘low speed pellet’ (LSP; 3,500 g, 15 minutes) containing nucleoli, nuclear envelopes and chromosomes, a ‘high speed pellet’ (HSP; 100,000 g, 1 hour), enriched in preribosomal particles and a ‘high speed supernatant’ (HSS), containing soluble nuclear proteins.

Monoclonal antibodies were raised essentially according to the method of Köhler and Milstein (1975). A 7-week-old female Balb/c mouse was immunized with 100 μg HSP-material from Xenopus laevis oocyte nuclei, prepared as described above. After two booster injections, the spleen cells were harvested and fused with cells of the mouse myeloma line, P3×63-Ag8.653, at a ratio of 3:1 in the presence of 40% PEG 4,000. Antibody-producing hybridoma cell lines were selected essentially as described by Schmidt-Zachmann et al. (1984). Immunoglobulin subclasses were determined by ELISA with subclass-specific antibodies (Medac, Hamburg, Germany) and peroxidase-coupled secondary antibodies (Sigma, Munich, Germany).

One of the mAbs obtained in this preparation was named mAb AND-1/23-5-14. This antibody of the IgG1-subclass was used for the experiments described in the present study. The hybridoma cell line 23-5-14 was also propagated as ascites in Balb/c mice.

Extracts from XLKE-A6 cells were prepared by incubating the cells of one confluent 10 cm cell culture dish (2×10⁷ cells) in 1 ml lysis buffer (100 mM Na-MES pH 6.5, 250 mM NaCl, 0.5 mM MgCl₂, 5 mM EDTA, 0.5% Triton X-100, 0.02% NaN₃, 50 μM Pefa-Block, 20 μM Pepstatin, 20 μM Leupeptin) for 2 minutes at room temperature. The lysed cells were cleared from cellular debris by centrifugation (14,000 g, 10 minutes, 4°C).

Protein fractions were separated by SDS-PAGE according to Thomas and Kornberg (1975). The two-dimensional separation of proteins was carried out by isoelectric focussing (IEF; O’Farrell, 1975) in the first dimension and by SDS-PAGE in the second dimension. The polypeptides were transferred to nitrocellulose and visualized by Poinceau S staining. The nitrocellulose membranes were blocked in TBS containing 0.05% Tween (TBST) and 5% non-fat dry milk for 1 hour and then incubated at room temperature with culture supernatant, diluted 1:2 with TBST for 1 hour. After washing with TBST, bound antibodies were either detected by colour reaction after incubation with alkaline-phosphatase-coupled secondary anti-mouse IgG1 antibodies (Dianova, Hamburg, Germany) or by chemiluminescence using the ECL-system (Amersham, Braunschweig, Germany) with horse radish peroxidase (HRP)-coupled secondary anti-mouse IgG1 antibodies (Dianova). In the case of ECL detection, the incubation with the HRP-coupled antibodies was carried out in the presence of 5% non-fat dry milk in TBST.

For immunofluorescence microscopy studies on cultured cells, the cells grown on coverslips were either fixed with methanol (5 minutes, -20°C) and acetone (30 seconds, -20°C) or with 2% formaldehyde, freshly prepared from paraformaldehyde (in PBS, pH 7.4) for 20 minutes. Following fixation, cells were washed twice with PBS and incubated with mAb AND-1/23-5-14 (ascites fluid, 1:200 diluted in PBS) for 30 minutes. After several PBS washes, the cells were incubated for 15 minutes with Texas Red-labeled goat anti-mouse Ig secondary antibodies (Dianova) diluted 1:100 in PBS. The cells were then washed again with PBS, dehydrated in ethanol, air-dried and mounted with Elvanol (Hoechst, Frankfurt, Germany).

Cryosections (5 μm) through Xenopus laevis ovary, epidermis and other tissues were fixed with 2% formaldehyde. Incubation times of primary and secondary antibodies were as described above. Micrographs were taken with an Axiophot microscope (Zeiss, Oberkochen, Germany).

Analysis of the HSS-fraction by density gradient centrifugation was done essentially as described before (Hügle et al., 1985) with minor modifications. 600 μl of the HSS-fraction were fractionated across a 5%-30% linear sucrose gradient, made in ‘5:1’-medium (10 mM Tris-HCl pH 7.4, 83 mM KCl, 17 mM NaCl, 2 mM MgCl₂, 2.5 mM DTT, 50 μM Pefa-Block). 15 fractions of 0.8 ml each were collected from the top (light) to the bottom (heavy) of the gradient. Marker proteins (BSA, catalase, thyroglobulin; all from Pharmacia, Freiburg, Germany) were separated on parallel gradients.

For gel filtration analysis the lysis supernatant derived from one 10 cm culture dish of confluently grown XLKE-A6 cells was applied on a Superose 12 column (HR 10-30; Pharmacia) at room temperature. For calibration, reference proteins (thyroglobulin, ferritin, catalase, aldolase and BSA; all from Pharmacia) were fractionated together with cellular extracts. Proteins were eluted (48 fractions, 0.2 ml each) in 100 mM Na-MES, pH 6.5, 250 mM NaCl, 0.02% NaN₃, 0.5 mM MgCl₂, 5 mM EDTA. After methanol precipitation, 1/6 of each fraction was analyzed by SDS-PAGE and immunoblotting.

Immunoprecipitations were performed using whole, homogenized Xenopus oocyte nuclei, HSS-material, XLKE-A6 cell lysates and total egg extracts. The latter ones were prepared as described previously by Cordes et al. (1993). Antibodies were first coupled by pre-incuba-tion (1 hour, 4°C) of mAb AND-1/23-5-14 ascites fluid (50 μl, diluted with 450 μl PBS) with 50 μl pre-swollen protein-G-Sepharose (Pharmacia). Protein solutions were also pre-incubated with protein-G-Sepharose (1 hour, 4°C) to avoid unspecific binding during the immunoprecipitation process. During end-over-end rotation (2 hours, 4°C), the pre-cleared protein sample was incubated with the mAb AND-1/23-5-14-Protein G complex. After low-speed centrifugation (800 g, 5 minutes) the resulting supernatant was precipitated with acetone and prepared for SDS-PAGE. The Sepharose beads with the bound immune complexes were washed five times with PBS, then once each with PBS containing 0.1% Triton X-100 and with PBS, and finally boiled in SDS-sample buffer. The solubilized proteins from the precipitate were analyzed by SDS-PAGE, together with the starting material and the reserved supernatant fraction.

Total RNA from Xenopus ovaries was prepared as described by Chomczynski and Sacchi (1987). Poly(A)⁺ RNA was prepared using the mRNA Purification Kit (Pharmacia). For northern blot analysis RNA was separated on a 1% agarose-gel containing 0.6 M formaldehyde, transferred to Biodyne A filters (Pall, Dreieich, Germany), hybridized, washed and visualized essentially as described by Heid et al. (1994). For hybridization, RNA blots were probed with a 1.59 kb ³²P-labeled antisense cRNA derived from EcoRI digestion of pBS AND-1-Xen and prepared using RNA-polymerase T7.

A λ unizap cDNA expression library from total Xenopus laevis ovary (Stratagene, Heidelberg) was screened using mAb AND-1/23-5-14. Positive plaques were identified by ECL-detection (Amersham). Two positives clones were selected, plaque-purified and released from the phages by in vivo excision, according to the manufacturer’s protocol. The two resulting cDNAs cloned in pBluescript were investigated by restriction mapping and preliminary sequence analysis, revealing that the ∼2.6 kb cDNA was identical to the 3’-end of the ∼4 kb cDNA. The entire ∼4 kb cDNA sequence of pBS- and-1-Xen was analyzed by constructing a set of deletion clones using the Double-stranded nested deletion kit (Pharmacia) and subsequent sequencing. The sequence analysis was performed using the HUSAR (Heidelberg Unix Sequence Analysis Resources) software program package.

For transient expression in human PLC cells, the AND-1 cDNA was subcloned as a ApaI/NotI-fragment into the mammalian expression vector pRc/CMV (Invitrogen via ITC Biotechnology, Heidelberg, Germany). Transfections, using 10 μg of DNA per 3.5 cm tissue culture dish, were carried out according to the method described by Chen and Okayama, (1987). Transfected cells were analyzed by immunofluorescence microscopy 24 hours after removal of the DNA-Ca²⁺ precipitate.

The same construct was used for the production of [³⁵S]methion-ine-labeled protein AND-1 in vitro using the TNT™ Coupled Reticulocyte Lysate System (Promega via Boehringer Ingelheim Bioproducts, Heidelberg, Germany).

AND-1 transcripts in different tissues and cell lines from Xenopus and rat were detected by PCR. Two oligonucleotides (AND-1 sense 1, corresponding to nucleotides 1,835-1,857; AND-1 antisense 1, reverse complementary to nucleotides 2,212-2,235) were designed corresponding to high homologous regions within the AND-1 cDNA and an EST sequence from rat adrenal chromaffin PC-12 cells (EST 108147; Lee et al., 1995). RT-cDNA from PC-12 cells and from rat brain and liver were kindly provided by Dr R. Leube (Johannes Gutenberg University, Institute of Anatomy, Mainz, Germany; for preparation of RT-cDNA see Leube et al., 1994). Suitable templates for PCR were also obtained by preparation of phage DNA from λ cDNA expression libraries of Xenopus laevis kidney, liver and ovary (Stratagene) or human fetal brain and epidermis (Clontech, Heidelberg, Germany). The amplification was performed under the following conditions: 5 μl of template was added to the reaction, containing 10 pmol of each primer and 0.2 mM dNTP in 1× reaction buffer. 1 unit Taq-polymerase (Boehringer Mannheim, Germany) was added to a final volume of 50 μl. The following PCR-profile was used: 1× 94°C/3 minutes; 15× (92°C/30 seconds, 55°C/1 minute (-0.5°C/cycle), 72°C/1 minute); 25× (92°C/30 seconds, 50°C/1 minute, 72°C/1 minute (+2 seconds/cycle)); 1× 72°C/5 minutes. The reactions were performed using a Minicycler (MJ research, via Biozym Diagnostics, Hess. Oldendorf, Germany).

The DNA cellulose chromatography was performed essentially as described by Miller et al. (1989). A column was packed with 1 ml cellulose, coupled with native, double-stranded calf thymus DNA (Pharmacia) and pre-swollen in buffer A (15 mM Tris-HCl pH 8.0, 50 mM NaCl, 5 mM EDTA, 1 mM DTT, 15% glycerol, 0.05% NP-40). 2 ml of the HSS-fraction, diluted with 8 ml buffer A, were loaded. The column was washed with 10 ml of loading buffer A and bound proteins were subsequently eluted with successive applications of 2 ml of buffer A containing 100 mM, 150 mM, 200 mM, 250 mM and 1.5 M NaCl, respectively. Fractions of 0.5 ml were collected, proteins were precipitated with methanol and subsequently analyzed by SDS-PAGE.

The AND-1-HMG-box was expressed and purified from bacteria using the Qiaexpress system (Qiagen, Hilden, Germany). The corresponding cDNA was subcloned as a 296 nt PstI-fragment (nt 3,09933,395) in the pQE 30 vector. Following high expression in E. coli strain M15[pREP4], the protein, containing a 6× His affinity tag, was purified on Ni-NTA resin under denaturing conditions. The protein was subsequently dialyzed step by step in buffer BC-0 (20 mM Tris-HCl, pH 7.9, 20% glycerol, 0.2 mM EDTA, 5 mM MgCl₂) containing 6 M, 4 M, 2 M, 1 M and 0 M urea, respectively.

The preparation of the four-way junction DNA and the electrophoretic mobility shift assay was done as described by Bianchi (1988) and Kuhn et al. (1994). 400 and 600 ng of the bacterial-expressed AND-1 HMG-box was used in the assay. 20 ng of the full-length bac-culovirus-expressed mouse UBF1 (kindly provided by Dr Renate Voit, this center, Division of Molecular Biology of the Cell II) containing five HMG-boxes was used as a control.

In the course of our studies on proteins stored in the nucleus (‘germinal vesicle’) of Xenopus laevis oocytes, such as the histone-binding proteins nucleoplasmin (Krohne and Franke, 1980a,b), and N1/N2 (Kleinschmidt and Franke, 1982; Kleinschmidt et al., 1985, 1986), high mobility group protein HMG-A (Kleinschmidt et al., 1983a), a nuclear actin-binding protein (Ankenbauer et al., 1989), the Mg²⁺-ATPase and the protea-some particles (Kleinschmidt et al., 1983b, Hügle et al., 1983; Peters et al., 1990, 1994), we had come across a protein with several unusual properties. One of the murine monoclonal antibodies, mAb AND-1/23-5-14 (IgG1), which was obtained after immunization with high speed pellet (HSP) proteins from isolated oocyte nuclei, reacted in immunoblotting experiments specifically with a polypeptide of M_r ∼125,000 identifiable in whole oocyte nuclei (Fig. 1, lane 1), in the HSP fraction containing particulate nuclear material (Fig. 1, lane 2), in the high speed supernatant (HSS) fraction containing the soluble nuclear proteins (Fig. 1, lane 3) and in total egg extracts (Fig. 1, lane 4). From these experiments we gained the impression that in oocyte nuclei the protein was present in a particulatebound and in a soluble form. To examine the occurrence of this protein in somatic cells, cultured Xenopus laevis kidney epithelial cells (XLKE-A6) were lysed in a buffer containing 0.5% Triton X-100 and 250 mM NaCl, and solubilized proteins were fractionated by centrifugation. Using immunoblotting on SDS-PAGE-separated proteins, the same polypeptide of M_r 125,000 was detected and completely recovered among the soluble proteins (Fig. 1, lanes 5, 6). Due to the minor occurrence of the protein in XLKE-A6 cells (compared to Xenopus oocytes and eggs) lane 5 had to be overloaded. To allow a comparison between the novel protein and other well-characterized nuclear Xenopus proteins, the immunoblot shown in Fig. 1A’ was reprobed with antibodies recognizing the histone-binding protein N₁/N₂ (kindly provided by Dr J. Kleinschmidt, this center, Division of Applied Tumor Virology; see Kleinschmidt et al., 1985, 1986). The result is shown in Fig. 1A”.

After separation of HSS-material by two-dimensional gel electrophoresis it was also possible to compare the immunoreactive protein with protein N1/N2 or nucleoplasmin on Coomassie blue-stained gels (Fig. 1B). It appeared as a faint, but Coomassie-visible spot (marked by an arrow), whereas N1/N2 and nucleoplasmin (Np) were the most abundant proteins of this fraction. Immunoblot analysis of the separated proteins showed that the polypeptide recognized by mAb AND-1/23-5-14 was an acidic protein with a pI of ∼5.5 (arrow, Fig. 1B’).

Sections through frozen Xenopus oocytes were examined by immunofluorescence microscopy with mAb AND-1/23-5-14, and the antigenic protein was detected only in the nucleus (Fig. 2A,A’). The protein was also localized in nuclei of certain other tissues such as epidermis (Fig. 2B,B’) and in nuclei of monolayer cultures of XLKE-A6 cells (Fig. 2C-G), where it appeared in interphase cells with a dispersed, sometimes finely granular distribution throughout the nucleoplasm. During mitosis (prometaphase to anaphase, Fig. 2D-G) the immunostained protein was dispersed over the cytoplasm, showing in metaphase and anaphase a distribution that somewhat accentuated the surfaces of the condensed chromosomes (Fig. 2E,F). In telophase the protein rapidly re-accumulated within the forming daughter nuclei (Fig. 5G). The protein resembled certain sequence-specific transcription factors in some of its properties and distribution during mitosis (cf. Martinez-Balbás et al., 1995).

Because of its major properties we refer to the protein by the acronym AND-1 (acidic nucleoplasmic DNA-binding protein).

The state of soluble protein AND-1 was analyzed by density gradient centrifugation. The protein could be identified in a complex with an S value of ∼7.5 (Fig. 3A,A’), which when analyzed by gel filtration, appeared with an apparent molecular mass (M_app) of 800,000 (Fig. 3B). According to Siegel and Monty (1966) this allows an estimate for the relative molecular mass (M_r) of ∼240,000 for the soluble form.

The complex containing protein AND-1 could also be immunoprecipitated from protein extracts of whole Xenopus oocyte nuclei, the HSS fraction of soluble nuclear proteins, egg extracts and XLKE-A6 cell lysates, but no co-precipitating proteins were detected in near stoichiometric amounts (Fig. 4).

From these data, we conclude that the bulk of the soluble protein AND-1 appears as a homo-oligomer, probably a homodimer.

To isolate the cDNA encoding protein AND-1, a λ cDNA expression library of Xenopus laevis ovary was screened with mAb AND-1/23-5-14. We obtained two clones with inserts of ∼2.7 kb and ∼4 kb. The smaller clone was found to be identical to the 3’-end of the larger cDNA clone. The latter was denoted pBS- and-1-Xen and further analyzed.

Total RNA and poly(A)⁺ RNA from Xenopus ovary tissue were probed in northern blot experiments with a 1.59 kb antisense cRNA derived from pBS- and-1-Xen. We obtained a strong signal corresponding to a mRNA of ∼4 kb (Fig. 5A), indicating that we had isolated a cDNA clone of full or nearly full length.

The completeness of the cDNA clone was demonstrated in two types of expression experiments. First, the ∼4 kb cDNA was subcloned into the mammalian expression vector pRc/CMV and subjected to coupled transcription and translation in vitro. After separation of the [³⁵S]methionine-labeled translation products by one-(Fig. 5B) or two-dimensional gel electrophoresis (Fig. 5C) the major product identified showed SDS-PAGE mobility and an apparent pI that were practically identical to those of the native protein AND-1 (compare Fig. 5B,C with Fig. 1). This was confirmed by co-electrophoresis. After separation of the radiolabeled translation product by twodimensional gel electrophoresis, the native protein AND-1, detected by alkaline phosphatase reaction, and the product of in vitro translation, detected by autoradiography, were found to co-localize in the same polypeptide spot (results not shown).

In transiently transfected cultured human PLC cells the protein encoded by the cDNA clone showed the same nucleo-plasmic localization as the native protein AND-1 (see Fig. 5D,D’).

The nucleotide sequence of the cDNA was determined in both directions (for data, see EMBL data base, accession no. X98884), and the amino acid (aa) sequence deduced from it is shown in Fig. 6A. The clone of 3,978 nucleotides, in agreement with the ∼4 kb size of the mRNA as detected by northern blot analysis (cf. Fig. 5A), contained a 5’-untranslated region of 81 nucleotides, with an in-frame stop codon (nt 17-19) upstream of the first possible start codon (nt 82-84), an open reading frame of 3,381 nt, and a 3’ region of 515 untranslated nucleotides with a polyadenylation signal (AATAAA) 12 nt before the poly(A) tail, of which 22 nt were included in the clone. The open reading frame encoded a polypeptide of 1,127 amino acids with a calculated molecular mass of 124.52 kDa and a pI of 5.27, both in close agreement with the data found for the native protein and indicating that the mature polypeptide has not been considerably altered by modifications that would have resulted in changes of its electrophoretic properties.

Data base searches revealed that the protein was novel, but also that its amino acid sequence contained some striking features (Figs 6, 7): the amino-terminal segment consists of seven WD repeats, a type of repeating unit with a basic number of ∼40 aa that often ends with a tryptophan (W) and an aspartic acid (D) residue (for references see van der Voorn and Ploegh, 1992; Neer et al., 1994). Members of the WD repeat protein family have been shown to be involved in the regulation of diverse fundamental cell functions, including nuclear ones (see Discussion). Of particular interest is the relationship of protein AND-1 to the SepB protein of Aspergillus nidulans, which is known to be involved in chromosome segregation and cytokinesis initiation (Harris and Hamer, 1995). An alignment of the seven WD repeats found in protein AND-1 compared to the corresponding region of the SepB protein is shown in Fig. 7A. The sequence homology extends beyond the mere occurrence of the repeats. Although the primary structure of WD-repeats is not well conserved, they have a conserved secondary structure. A WD-repeat was defined as a β-strand-turn-β-strand-turn-β-strand alternation (Neer et al., 1994). This was also predicted for the WD-repeats found in protein AND-1 (Fig. 7B).

A second conspicuous feature of protein AND-1 is the presence of a single characteristic HMG-box in the carboxyterminal part of the protein (aa positions 1013-1076), indicating that this protein belongs to the HMG-box superfamily of DNA-binding proteins (Fig. 7C) involved, for example, in regulation of chromatin formation, replication and transcription, (Johns, 1982; Jantzen et al., 1990; Griess et al., 1993; Laudet et al., 1993). An HMG-box is defined as three α-helices able to adopt an L-shaped arrangement (Baxevanis et al., 1995; Teo et al., 1995). The secondary structure prediction of the AND-1 HMG box showed the same helical composition (Fig. 7C).

Protein AND-1 also contains some clusters of acidic amino acids (boxed in Fig. 6A), as they have been described in various other major nuclear proteins (e.g. Kleinschmidt et al., 1986; Schmidt-Zachmann et al., 1987; Schmidt-Zachmann and Franke, 1988; for review see Earnshaw, 1987). The overall structural organization of protein AND-1 is schematically shown in Fig. 6B.

We also noticed candidates for nuclear localization signals (NLS; for review see Boulikas, 1993) within the amino acid sequence of the protein (underlined in Fig. 6A). Griess et al. (1993) and Poulat et al. (1995) have proposed a bipartite NLS within each HMG-box motif, based on the NLS sequence requirements discussed by Dingwall and Laskey (1991). In protein AND-1, this might correspond to KKPKx₁₀R (aa 1,013-1,027). Other possible NLS sequences might be KKKKx₁₁RK (aa 599-615), somewhat reversely similar to the bipartite NLS of Xenopus nucleoplasmin (KR_x10KKKK; Robbins et al., 1991) or KKRKR (aa 1,078-1,082), similar to the monopartite NLS of human lamin A (KKRK; Loewinger and McKeon, 1988).

The sequence similarities of protein AND-1 to both the WD-repeat and the HMG-box families are restricted to these motifs, with the single exception of protein SepB of Aspergillus nidulans. This protein of 710 amino acids contains WD-repeats in its amino-terminal part but is 417 amino acids shorter, lacking the entire region homologous to the HMG-box-con-taining portion of protein AND-1 (cf. Fig. 7A), and shows 22.9% identity (50.8% similarity) with protein AND-1.

The DNA-binding capacity of the native protein AND-1 was tested by DNA-affinity chromatography, the classic approach used to enrich DNA-binding proteins (Alberts et al., 1968) and to purify transcription factors (Rosenfeld and Kelly, 1986). The soluble proteins of the oocyte nuclear HSS-fraction were applied to native, double-stranded DNA coupled to cellulose, the DNA-binding proteins were eluted at increasing salt concentrations, and the fractions obtained were analyzed by immunoblotting experiments (Fig. 8A,A’). The protein was found to be bound to DNA quantitatively and with high affinity. It was completely eluted from the column using buffer containing 150 mM NaCl (Fig. 8A’ lane 4). Using the same assay, we could also prove the DNA-binding capacity of the recombinant expressed AND-1 HMG-box (Fig. 8B), indicating that native protein AND-1 present in the HSS-fraction was bound directly to DNA via its HMG-box and not retained through a possible binding partner present in the HSS-fraction (results not shown).

It has been shown for several HMG-box proteins that they also recognize specific DNA conformations like four-way junctions, stem loops, cisplatin-modified or bended DNA (reviewed by Landsman and Bustin, 1993). Therefore, we examined binding of protein AND-1 to four-way junction DNA in an electrophoretic mobility-shift assay (Bianchi, 1988). Unfortunately, the full-length protein AND-1 could not be expressed and purified from bacteria in sufficient amounts, but because it has been shown that a single recombinant HMG-box is sufficient for shifting four-way junction DNA (Ferrari et al., 1992; Kuhn et al., 1994; Teo et al., 1995), we used the purified recombinant HMG-box of protein AND-1 (Fig. 8B) and could indeed show its binding to such DNA structures (Fig. 8C, lanes 3,4). The assay was controlled by the use of mouse UBF1 (c.f. Kuhn et al., 1994).

The mAb to protein AND-1 of Xenopus did not cross-react with any protein of other higher vertebrate species. On the other hand, in data base searches we discovered, in addition to the fungal protein SepB, a 601-nucleotide-long expressed sequence tag from cDNA of rat adrenal neuroendocrine cells ofline PC-12 (EST 108147; Lee et al., 1995) matching the AND-1 cDNA (nt 1,765-2,308) in 72.0% of the positions. Assuming that this EST sequence corresponds to the AND-1 homologue in rat, we examined cDNAs derived from rat PC12 cells, brain and liver, in comparison with Xenopus ovary, kidney and liver, by using PCR with two oligonucleotides corresponding to the AND-1 cDNA sequence. In all cases, the correct PCR-segment of 400 nt could be amplified (data not shown). A similar result was obtained with cDNAs from human brain, epidermis, liver and stomach by PCR, suggesting that mRNA encoding protein AND-1 also occurs in mammalian species in very different cells and tissues.

The large (125 kDa) nucleoplasmic protein AND-1 that we have discovered in oocytes and various other cells of Xenopus laevis is a remarkable case of a naturally occuring chimeric molecule combining hallmark features of two different families of regulatory proteins, the WD-repeat and the HMG-box proteins, suggesting that the DNA-binding protein AND-1 is also involved in protein-protein interactions. A highly related protein has also been identified in various mammals, and a partial cDNA encoding the orthologous human protein has recently been cloned from somatic cells (results not shown). Therefore, we conclude that the gene encoding this protein has been evolutionarily conserved and is widely, if not ubiquitously, expressed in different cell types and under different conditions.

Within the superfamily of HMG-box proteins, protein AND-1 falls into the subfamily of those members which contain only a single box. These proteins recognize, and bind to, DNA with high sequence specificity (Laudet et al., 1993) and rapidly and effectively accumulate in a soluble form dispersed throughout the nucleoplasm (for reviews see Landsmann and Bustin, 1993; Laudet et al., 1993), in contrast to other HMG-box proteins which, in the same cells, distribute over both compartments, the nucleoplasm and the cytoplasm, including the prototype proteins HMG1 and HMG2 (e.g. Bustin and Neihart, 1979; Kleinschmidt et al., 1983a; Einck and Bustin, 1985; Landsmann and Bustin, 1993 and further references therein). Several HMG-box proteins of both subfamilies, wether they bind to DNA in a sequence-specific or -unspecific manner, also recognize non-B-DNA structures like four-way junctions, which can be generated by recombination events and by intrastrand base pairing of inverted repeat sequences. Obviously, this is also true for protein AND-1, since the binding of the recombinant AND-1 HMG-box to such structures has been demonstrated in the electrophoretic mobility shift assay.

Our observations that diffusible homodimer complexes of protein AND-1 accumulate in the interchromatin nucleoplasm and that the protein appears to be absent from nucleoli and dense chromatin, including mitotic chromosomes, argue against a specific function in inactivated genes and in rDNA transcription, in contrast to the HMG-box-containing UBF proteins (see, e.g. Jantzen et al., 1990; Zatsepina et al., 1993) and the single-strand nucleic acid-binding Drosophila protein, DssRP (Hsu et al., 1993). On the other hand, the occurrence of protein AND-1 in many diverse cell types indicates that its function is not exclusive to a given cell type or developmental stage. In this respect, the expression pattern of protein AND-1 is fundamentally different from those of many other HMG-box proteins that occur only in certain cells, such as the sex-determining proteins of the SRY-Sox group (e.g. Gubbay et al., 1990; Sinclair et al., 1990; Laudet et al., 1993; Harley and Goodfellow, 1994), the lymphocyte differentiation-specific transcription factors LEF-1 and TCF-1 (e.g. Travis et al., 1991; Van de Wetering et al., 1991), and the stage-specific POP-1 protein of C. elegans (Lin et al., 1995). The occurence of protein AND-1 in the fraction containing the ‘soluble’ nucle-oplasmic proteins also shows that it is not non-specifically associated with chromatin, but suggests that its reaction from the soluble to the DNA-bound is regulated.

As first shown for the specific binding of the multiple HMG-box-containing protein UBF to one of the RNA polymerase I subunits (Schnapp et al., 1994), HMG-box motifs not only effect DNA recognition and binding but can also determine protein-protein interactions. A special form of such interaction appears to be the homodimer formation shown to involve an HMG-box in UBF (O’Mahony et al., 1992) or in protein HMG1 (Bianchi et al., 1992). We are currently testing whether the dimer formation of protein AND-1 observed in Xenopus oocytes and eggs is also mediated by the single HMG-box present.

Protein-protein interaction as a major reaction and function has also been ascribed to WD-repeats, the other familydefining motif prominent in protein AND-1. Proteins possessing variable numbers of WD-repeats of around 40 amino acids have been found in eukaryotes (Neer et al., 1994) and, recently, also in one prokaryote (Janda et al., 1996) and have been reported to determine the local protein conformation and to contribute to a diversity of nuclear and cytoplasmic functions (reviewed by Neer et al., 1994; Neer and Smith, 1996). Such functions include signal transductions (for review see Simon et al., 1991; Neer, 1995), histone binding, chaperoning and escorting (Kaufman et al., 1995; Edmondson et al., 1996), chromatin replication or control of cell cycle and mitosis (Yochem and Byers, 1987; Kaufman et al., 1995; Muro et al. 1995), chromosome segregation and cytokinesis (Harris and Hamer, 1995), gene transcription and suppression (e.g. Keleher et al., 1992; Dynlacht et al., 1993; Koleske and Young, 1994; Komachi et al., 1994; Paroush et al., 1994; Gutjahr et al., 1995; Sathe and Harte, 1995), mRNA modification (Takagaki and Manley, 1992), nuclear pore complex architecture (Sinios-soglou et al. 1996), and also vesicle formation and fusion (Harrison-Lavoie et al., 1993; Pryer et al., 1993; Stenbeck et al., 1993; Gerich et al., 1995). Moreover, defects in some WD-repeat proteins have been discussed in relation to appearances of specific inheritable developmental disorders (e.g. Henning et al., 1995; Lorain et al., 1996).

Among the members of this family, protein AND-1 has a ‘normal’ number of repeats (7) of a normal length distribution, but stands out by its relatively large size (see, however, similarly-sized polypeptides such as α- and β’-COPs; Gerich et al., 1995; Stenbeck et al., 1993; Harrison-Lavoie et al., 1993) and its generally nucleoplasmic location (for some WD-repeat proteins with nuclear enrichment, dependent on the specific cell cycle or developmental stages; see Gutjahr et al., 1995; Muro et al., 1995). The special relationship of protein AND-1 to protein SepB of Aspergillus nidulans, which extends beyond the WD-repeat region, points to the possibility of a functionally close relationship to this fungal protein, which is known to be involved in the regulation of chromosome segregation and cytokinesis (Harris and Hamer, 1995).

At present, the functional significance of the WD-repeat domain with respect to these diverse functions is still not fully understood. The prevailing general concept that WD-repeats might be directly involved in, and essential for, specific protein-protein interactions, including examples in which they act as assembly-promoting factors or chaperons (Edmondson et al., 1996; Kaufman et al., 1995; Sathe and Harte, 1995; Komachi et al., 1994; Neer et al., 1994; Conklin and Bourne, 1993; Keleher et al., 1992; van der Voorn and Ploegh, 1992) seems to be challenged by demonstrations that in certain proteins the entire domain containing these repeats is not needed for effecting the functionally important binding to another protein (e.g. Paroush et al., 1994). We are currently searching for possible AND-1-binding proteins to learn more about the functional mechanisms involving this complex ‘dual family membership’ nuclear protein.

We thank Sonja Reidenbach, Cilly Kuhn and Michaela Hergt for expert technical assistence, Jutta Müller-Osterholt for preparation of photographs and Peter Eichhorn for excellent care of the animals. We gratefully acknowledge Dr Renate Voit (this center, Division of Molecular Biology of the Cell II) for valuable discussions and for the help with the electrophoretic mobility shift assay, and Rudolf F. Zirwes for valuable discussions and for kindly providing XLKE-A6 cell-extracts fractionated by gel filtration. This work has been supported in parts by the Deutsche Forschungsgemeinschaft (grant Schm 862/2-3 to M.S.-Z.).