Nuclear membrane protein LAP2β mediates transcriptional repression alone and together with its binding partner GCL (germ-cell-less)

An important feature of all eukaryotic cells is organelle and nuclear compartmentalization. The nuclear envelope (NE) separates the nucleus from the cytoplasm, thus allowing several fundamental functions, such as DNA-replication, transcription and RNA-processing, to take place. These functions are highly dependent on nuclear architecture, which is largely determined by the various components of the NE. The NE is composed of outer and inner nuclear membranes, a perinuclear space, nuclear pore complexes (NPCs) and a nuclear lamina. The inner nuclear membrane and nuclear lamina contain unique proteins, including the integral nuclear membrane proteins (IMPs) p58/lamin B receptor (LBR) (Worman et al., 1990; Worman et al., 1988), LAP1 (lamina associated polypeptide) (Martin et al., 1995), LAP2 (Berger et al., 1996; Furukawa et al., 1995; Harris et al., 1994), Emerin (Manilal et al., 1996), p34 (Simos and Georgatos, 1994), p18 (Simos et al., 1996), MAN1 (Lin et al., 2000) and Nurim (Rolls et al., 1999). Three IMPs, LAP2β , Emerin and MAN1 share a 70% identical 43-residue LEM box, which was recently suggested to serve as a novel chromatin-binding domain (Lin et al., 2000). The peripheral nuclear membrane proteins include type A and type B lamins (Fisher et al., 1986; McKeon et al., 1986), otefin (Harel et al., 1989; Padan et al., 1990), and Young Arrest (YA) (Lin and Wolfner, 1991).

LAP2β belongs to the LAP2 (thymopoietin) family of nuclear proteins (Berger et al., 1996; Furukawa et al., 1995; Harris et al., 1994). These proteins are ubiquitously expressed and are highly conserved in mammals (Zevin-Sonkin et al., 1992; Harris et al., 1994; Berger et al., 1995; Theodor et al., 1997; Ishijima et al., 1996). To date, six mouse and human alternatively spliced LAP2 isoforms, designated α, β, γ, δ, ε and ζ were isolated and characterized (Harris et al., 1994; Berger et al., 1996). All of them share an identical N-terminal 186 amino acid domain. The β, γ, δ and ε isoforms also share an identical C terminus that contains a transmembrane domain that enables their insertion into the inner NE and to the perinuclear space. LAP2β has been isolated and characterized as an inner nuclear membrane protein that binds to lamin B and chromosomes in a phosphorylation dependent manner (Foisner and Gerace, 1993). Thus, it was assigned roles in linking chromatin to the NE during interphase and in NE breakdown and reassembly during mitosis (Foisner and Gerace, 1993; Furukawa et al., 1998; Furukawa et al., 1995). The microinjection of the recombinant lamin binding region of LAP2β into mammalian cells inhibited nuclear volume increase and progression into S phase (Yang et al., 1997). In Xenopus laevis the addition of human recombinant LAP2β truncation mutants to cell-free nuclear assembly reactions, severely affected nuclear envelope and lamin assembly (Gant et al., 1999). Both studies clearly indicate a role for LAP2β in lamin dynamics, which are crucial to NE reassembly and nuclear growth. The latter is essential for cell cycle progression from G1 into S phase, possibly linking LAP2β to cell cycle control.

Recently, a protein named BAF (barrier to autointegration factor) was identified as a chromatin binding LAP2β-interacting partner (Furukawa, 1999). This DNA binding protein was initially discovered in retrovirus-infected cells, where it protected viral DNA from self-integration (Lee and Craigie, 1998). LAP2β is suggested to bind BAF through its LEM box, thus predicting that all other LAP2 isoforms, including those in X. laevis (Gant and Wilson, 1997; Lang et al., 1999), as well as the LEM domain proteins emerin and MAN1, bind BAF (Wilson, 2000). Interestingly, a recombinant polypeptide that contains the chromatin-binding region of LAP2β enhanced the efficiency of DNA replication in Xenopus extracts (Gant et al., 1999). This suggests that LAP2β affects chromatin structure and organization, possibly through its binding to BAF or other, yet unidentified, chromatin proteins.

In four independent studies, including the one presented here, the mouse homologue of GCL was isolated and characterized (de La Luna et al., 1999; Kimura et al., 1999; Leatherman et al., 2000). In Drosophila, the maternal GCL protein is required for germ line specification. Drosophila females with reduced GCL function give rise to sterile adult progeny that lack germ cells but are otherwise normal. It was suggested that the Drosophila GCL protein is localized to the NE of those nuclei that later become the nuclei of the germ cell precursors (Jongens et al., 1994). All studies of mouse GCL demonstrated the high conservation, in structure, localization and function, between the mouse and Drosophila GCL proteins. This is best exemplified by the ability of mouse GCL to rescue the Drosophila GCL-null phenotype (Leatherman et al., 2000). However, because the primordial germ cell (PGC) specification process in mouse does not depend on the presence of a maternally inherited germ plasm, as it is in Drosophila, it seems that, in mammalian testis, GCL functions mainly in the process of spermatogenesis.

Both Drosophila and mouse GCL belong to the BTB/POZ domain containing proteins. This evolutionarily conserved protein-protein interaction domain is generally found at the N terminus of either actin binding or, more commonly, nuclear DNA binding proteins (Albagli et al., 1995). It is suggested that, by mediating protein binding in large aggregates, the BTB/POZ domain serves to organize higher order macromolecular complexes involved in nuclear events such as chromatin folding (Albagli et al., 1995). In Drosophila melanogaster this domain is found in transcription factors that are required for developmental processes such as pole cell formation in the embryo (Albagli et al., 1995). Nearly all mammalian proteins that contain the BTB/POZ domains are zinc finger proteins that are involved in transcriptional regulation. The two best known examples of these are BCL6 and PLZF. These two proteins were shown to act as transcriptional repressors and their aberrant expression is connected to the development of hematopoietic malignancies (Deweindt et al., 1995; Dong et al., 1996).

The mGCL protein was also isolated by de La Luna and colleagues (de La Luna et al., 1999) as a DP-interacting protein (DIP). In human osteosarcoma U20S cells, mGCL/DIP was suggested to cause the translocation of the E2F-DP heterodimer to the NE, the reduction of its transcriptional activity and the accumulation of cells in the G1 phase of the cell cycle. In this study, we describe the isolation of mGCL as a LAP2β-binding protein, the co-localization of both proteins to the NE and their repression of the E2F-DP transcriptional activity. The possible mechanisms of this transcription regulation are discussed.

The two-hybrid system screening was performed using as the bait amino acids 219-328 of LAP2β (LAP2β specific region) fused to the DNA-binding domain of GAL4 in PAS2. The Saccharomyces cerevisiae strain Y190, which carries integrated LacZ and HIS reporters under the control of the Gal4 responsive promoter, was transformed with the bait and mouse placental cDNA library fused to the GAL4 activation domain. Double transformants were selected on an appropriate selective medium and were screened by filter assay for the induction of β-galactosidase using standard procedures. 37 positive colonies were obtained. Plasmids containing the prey sequences were rescued and checked by back transformation with the bait into the yeast. The positive clones were sequenced on both strands using automated sequencing. A clone designated 7.3, which showed high homology to the Drosophila GCL was used for subsequent analysis.

In order to clone the full-length cDNA of mGCL, adult mouse testis and mouse thymus cDNA libraries were screened using various cDNA segments of the 7.3 clone. Several positive clones were isolated. The sequence of the isolated clones was determined in both strands by automated sequencing. Two of the isolated clones, designated 6.1 and 1.8, containing the full-length mGCL and cloned in the pBluescript vector, were further used for subcloning. Database searches and sequence comparisons were done using the BLAST program (Altschul et al., 1990) provided by the National Center for Biotechnology Information.

PAS2 has been previously described (Bai and Elledge, 1997). pAS2-LAP2β-SR was constructed by inserting a NdeI-BamHI digested PCR product corresponding to amino acids 219-328 of LAP2β in frame with the Gal4 DNA binding domain.

GST-mGCL and GST-LAP2β were constructed by inserting EcoRI (mGCL) and SalI (LAP2β) digested cDNA fragments corresponding to the full length proteins into the pGEX vector (Pharmacia Biotech).

The following plasmids have been described previously: pcDNA3-HA-E2F5 (Lindeman et al., 1997), pCMV-Rb (Qin et al., 1992), E2F-Luciferase (Krek et al., 1993) and pcDNA3.1-HA (gift of M. Walker, Weizmann Institute, Israel). pcDNA-HA-LAP2β and pcDNA-HA-LAP2ζ were constructed by inserting full-length mouse LAP2β and LAP2ζ in frame into pcDNA3.1-HA. pcDNA3-mGCL was constructed by inserting a BamHI-EcoRV fragment containing the full length mGCL derived from the 6.1 pBluescript vector into pcDNA3 (invitrogen).

pcDNA3.HA-DP3α used in the luciferase assay was cloned by first performing PCR on cDNA derived from mouse kidneys using primer oligonucleotides derived from the pl-2 murine DP-3 cDNA sequence (Ormondroyd et al., 1995). Four different PCR products coding the various spliced products of DP3 were cloned into the pCR2.1 TA cloning vector (Invitrogen) and sequenced. The clone coding the DP3α sequence (Ormondroyd et al., 1995) was then subcloned in frame into the pcDNA3.1HA vector.

Human lung carcinoma H1299 cells were grown in RPMI supplemented with 10% foetal calf serum (FCS), glutamine, penicillin and streptomycin. CHO cells were grown in DMEM supplemented with 10% FCS, glutamine, penicillin and streptomycin. RIN (rat insulinoma) cells were grown in M199 supplemented with 10% FCS, glutamine, penicillin and streptomycin.

Transfections were performed using the calcium-phosphate method. Cells were plated 24 hours before transfection at 10⁵ cells per well in a six-well plate (transcription assays) or 10⁶ cells per 10 cm plate (protein expression). Glycerol shock, washes and refeeding were performed after 18 hours in the presence of the precipitate. The cells were harvested ∼40 hours after transfection. DNA amounts were kept constant by adding pcDNA3 when required. In the transcription assays pcDNA-β-galactosidase was used as an internal control for the transfection efficiency. Luciferase and β-galactosidase activities were measured in duplicate plates for each point.

RIN cells were harvested in PBS, washed twice, pelleted and resuspended in 1× SDS. Pancreatic tissue was derived from an adult male mouse. The tissue was washed with ice-cold PBS, minced in ice-cold lysis buffer (20 mM Tris pH 7.8, 1% NP-40, 150 mM NaCl, 50 mM NaF, 0.2%SDS, 2 mM EDTA, 10% glycerol, 0.5 mM DTT, protease inhibitor cocktail (complete; Roche)) and pelleted at 4°C for 15 minutes. Protein concentration of the supernatants was determined (BCA, PIERCE) and equal amounts of proteins were subjected to SDS-PAGE.

For the solubility assay, ∼10⁷ RIN cells were harvested in ice-cold PBS, washed, pelleted and resuspended in 400 μl hypotonic buffer (10 mM HEPES pH 7.9, 10 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF and protease inhibitor cocktail (complete; Roche)) for 15 minutes on ice. After that, 25 μl of 10% NP-40 were added and the cells were strongly vortexed for 10 seconds and pelleted at 16,000 g for 30 seconds. The supernatant designated cyt. was resuspended in 5× SDS sample buffer. The pelleted nuclei were washed, resuspended in an ice-cold lysis buffer (20 mM HEPES pH 7.9, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF and protease inhibitor cocktail (complete; Roche)) and 8 M urea, or 250 mM NaCl and 1% Triton X-100, or 500 mM NaCl and 1% Triton X-100, vortexed vigorously for 15 minutes at 4°C and pelleted. Supernatants (designated NS) were resuspended in 5× SDS sample buffer. Pellets (designated NP) were resuspended in 1× SDS sample buffer.

Proteins were separated on 12% SDS-PAGE, transferred to nitrocellulose (Schleicher & Schuell) and detected using the Western Blot Chemiluminescence Reagent Plus (NEN). The following primary antibodies were used: rabbit anti-mGCL polyclonal antibodies raised against a 12 amino acid peptide derived from the C terminus of mGCL (dilution 1:500) and anti-LAP2β monoclonal antibodies (clone 6G11, dilution 1:10000, generous gift of G. Goldstein, NJ, USA).

Cells that were grown on coverslips were treated as follows: fixation in ice-cold methanol for 5 minutes and then ice-cold acetone for another 5 minutes. After the fixation, the cells were washed with TBS (100mM Tris-HCl pH 7.5, 150 mM NaCl). Blocking was done using 5% skim milk in TBS containing 0.1% Tween 20 (TBS-T) for 15 minutes. Incubations with primary and secondary antibodies were in the blocking solution for 30 minutes each. Between and after the incubation with the antibodies were washed using TBS-T. The coverslips were mounted in immunofluore (ICN) and cells photographed with a confocal microscope.

The primary antibodies were used as follows: anti-LAP2β was used in a 1:100 dilution and anti-mGCL was used in a 1:50 dilution. Cy2- and Cy3-conjugated goat anti-mouse, donkey anti-rabbit and goat anti-rat antibodies (Jackson Laboratories) were used as secondary antibodies in a 1:100 dilution.

Full length LAP2β and mGCL were expressed as Glutathione-S-transferase (GST) fusion proteins in the pGEX bacterial expression system. Bacterial DH5α cells expressing GST or both GST fusion proteins were grown overnight at 37°C and diluted 1:100. The cells were then grown to O.D 0.6 at 30°C (LAP2β and mGCL) or 37°C (GST). For GST expression, 0.1 mM IPTG was added for an additional 3 hours. Cells were harvested, sonicated and the recombinant proteins were extracted from the bacteria at 4°C using 1% Triton X-100 and 50 mM EDTA in the presence of protease inhibitor cocktail (complete; Roche). The final volume of cell lysate that contained the recombinant proteins was 1/25 of the starting culture. The expressed proteins were detected by western blot analysis using either specific anti LAP2β and anti-mGCL antibodies or monoclonal antibodies against GST (clone B-11, Santa Cruz). The [³⁵S]-labelled LAP2β, LAP2ζ and mGCL proteins were synthesized in vitro using the TNT T7 quick system (Promega) in the presence of [³⁵S]-labelled methionine.

For interaction with the in vitro translated products, GST or GST fusion proteins were diluted in 1 ml PBS and incubated by shaking for 5 hours at room temperature with 50 μl of reduced glutathione-Sepharose beads (Pharmacia). Protein-bound beads were washed three times with 1 ml PBS three times, diluted in 0.5 ml PBS and incubated with 10 μl of the in vitro translated products for 1 hour at room temperature. The beads were then washed five times with 1 ml PBS and pelleted. Bound proteins were eluted by boiling the beads in 50 μl of 2× SDS sample buffer, separated on SDS-PAGE and detected by autoradiography ([³⁵S]-labelled proteins) or western blot analysis (GST or GST fusion proteins).

Interspecific backcross progeny were generated by mating (C57BL/6J×Mus spretus) F₁ females and C57BL/6J males as described (Copeland and Jenkins, 1991). A total of 205 N₂ mice were used to map the Mgcl locus (see text for details). DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, Southern blot transfer and hybridization were performed essentially as described (Jenkins et al., 1982). All blots were prepared with Hybond-N⁺ nylon membrane (Amersham). The probe, an ∼2.0 kb EcoRI fragment of mouse cDNA was labelled with [α³²P] dCTP using a nick translation labelling kit (Boehringer Mannheim); washing was done to a final stringency of 1.0× SSCP, 0.1% SDS, 65°C. Fragments of 8.3 kb, 7.0 kb, 5.9 kb, 3.7 kb, 3.2 kb, 2.6 kb and 1.3 kb were detected in ScaI digested C57BL/6J DNA, and fragments of 10.5 kb, 7.0 kb, 5.4 kb, 3.7 kb, 2.8 kb and 1.3 kb were detected in ScaI digested M. spretus DNA. The presence or absence of the 10.5 kb, 5.4 kb and 2.8 kb ScaI M.-spretus-specific fragments, which co-segregated, was followed in backcross mice. A description of the probes and restriction-fragment-length polymorphisms (RFLPs) for the loci linked to Mgcl, including Catna2, Mad and Gpr73, has been reported previously (Uchida et al., 1994; Parker et al., 2000). Recombination distances were calculated using Map Manager version 2.6.5. Gene order was determined by minimizing the number of recombination events required to explain the allele distribution patterns.

In order better to understand the functions of LAP2β, we used the yeast two-hybrid system approach to identify LAP2β interacting proteins. The GAL4 DNA binding domain (DBD) fused to amino acids 219-328 of mouse LAP2β, which compose the LAP2β specific region (SR), was chosen as the bait protein (Fig. 1A). This domain is partially or fully spliced out in other LAP2 isoforms (Berger et. al., 1996, Harris et al., 1994) and partially overlaps the nuclear lamina and NE targeting domain of LAP2β (amino acids 298-370) (Furukawa et al., 1998). We therefore considered it likely to function as a protein interaction domain. A screen of a transactivation domain (TA) tagged library made from mouse placenta revealed a 2-kb positive clone designated 7.3. This cDNA encodes an open reading frame (ORF) of 386 amino acids that was able to bind the GAL4 DBD fused SR. The binding of SR to the protein encoded by the 7.3 clone was specific because no interaction was revealed between the 7.3-TA fusion protein and unrelated GAL4-DBD fused proteins such as cyclin B and MDM2. In addition, neither the LAP2β-DBD fusion protein nor the 7.3-TA fusion protein was able to form blue colonies in a β-galactosidase activity assay when transformed on their own (Fig. 1B).

Blast search against the various DNA and protein databases revealed that the 386 amino acid polypeptide encoded by the 7.3 clone is highly homologous to the C-terminal region of the Drosophila GCL protein (Jongens et. al., 1992). In Drosophila, GCL is required for embryonic germ line formation both in males and females. Drosophila females with reduced GCL function give rise to sterile adult progeny that lack germ cells but are otherwise normal. Using the 7.3 clone as a probe, we screened mouse thymic and testis cDNA libraries and cloned the full-length cDNA of this protein. The full-length protein, which is 524 amino acids long, shares 37% identity and 47% similarity with the Drosophila GCL protein. We therefore named this LAP2β-binding protein mouse germ-cell-less (mGCL). A database and literature search revealed that mGCL was recently cloned independently by three other groups (de la Luna et al., 1999; Kimura et al., 1999; Leatherman et al., 2000). The protein sequence of mGCL and its homology to the Drosophila GCL are presented in Fig. 1C.

The main feature of the mGCL protein sequence is the existence of the evolutionarily conserved protein-protein interaction BTB/POZ domain in its N terminus (amino acids 89-198) (Fig. 1C). However, although nearly all BTB/POZ proteins isolated so far contain either a DNA or an actin binding domain in their C terminus, both the Drosophila and mouse GCL do not contain any known motif in their C terminus. Other features of mGCL include two putative nuclear localization signal (NLS) motifs at its N terminus (residues 48-52 and 82-87) (Kalderon et al., 1984) and a possible tyrosine phosphorylation site (amino acid 405) that are also conserved in the Drosophila GCL homologue (Fig. 1C). Hydrophobicity and motif analyses of mGCL do not reveal any hydrophobic region suggestive of a transmembrane domain as found in its binding partner LAP2β (data not shown).

Because the bait used in our two-hybrid assay consisted only of LAP2β specific region and the prey contained only the C terminus of mGCL, we carried on to prove that the full length proteins physically interact with each other. In order to do so, we expressed the full-length mGCL and LAP2β as GST fusion proteins in the pGEX bacterial expression system. Both fusion proteins were toxic, insoluble and expressed at very low levels in bacteria. Thus, the expressed GST-mGCL or GST-LAP2β protein could be detected only with specific antibodies raised against mGCL (Fig. 2, western blot lanes 3 and 6), LAP2β (lane 9) and GST (data not shown). We then performed GST pull down experiments using in vitro translated [³⁵S]-labelled mGCL, LAP2β and LAP2ζ, which is an alternatively spliced product of LAP2β, missing its lamin and mGCL binding region but containing its chromatin binding domain, thus serving as a natural deletion mutant of LAP2β. As shown in Fig. 2, [³⁵S]-labelled LAP2β specifically bound GST-mGCL (lane 3) but did not bind GST alone. As expected, [³⁵S]-labelled LAP2ζ did not bind either GST or GST-mGCL (lanes 5 and 6, respectively). In the opposite orientation, [³⁵S]-labelled mGCL specifically bound GST-LAP2β (lane 9) but did not bind GST alone (lane 8).

The mouse chromosomal location of Mgcl was determined by interspecific backcross analysis using progeny derived from matings of ((C57BL/6J×M. spretus) F₁×C57BL/6J) mice. This interspecific backcross mapping panel has been typed for over 3000 loci that are well distributed among all the autosomes as well as the X chromosome (Copeland and Jenkins, 1991). C57BL/6J and M. spretus DNAs were digested with several enzymes and analysed by Southern blot hybridization for informative RFLPs using a mouse Mgcl cDNA probe. The 10.5 kb, 5.4 kb and 2.8 kb ScaI M. spretus RFLPs (see Materials and Methods) were used to follow the segregation of the Mgcl locus in backcross mice. The mapping results indicated that Mgcl is located in the central region of mouse chromosome 6 linked to Catna2, Mad and Gpr73. Although 88 mice were analysed for every marker and are shown in the segregation analysis (Fig. 3), up to 160 mice were typed for some pairs of markers. Each locus was analysed in pairwise combinations for recombination frequencies using the additional data. The ratios of the total number of mice exhibiting recombinant chromosomes to the total number of mice analysed for each pair of loci and the most likely gene order are: centromere-Catna2-(5/160)-Mad-(0/126)-Mgcl-(1/105)-Gpr73. The recombination frequencies (expressed as genetic distances in centiMorgans (cM) ± the standard error) are: Catna2-3.1±1.4-(Mad, Mgcl)-1.0±1.0-Gpr73. No recombinants were detected between Mad and Mgcl in 126 animals typed in common, suggesting that the two loci are within 2.4 cM of each other (upper 95% confidence limit).

Polyclonal antibodies were raised against a synthetic peptide derived from mGCL C-terminus. These antibodies specifically recognized ectopic mGCL expressed in H1299 cells but did not detect an endogenous protein in these cells (Fig. 4A, lanes 1 and 2). They were used to search adult tissues and cell lines expressing GCL. As seen in Fig. 4A, the antibodies detected a strong band of the expected 60 kDa in a rat insulinoma (RIN) cell line derived from pancreatic β cells and in mouse pancreas (lanes 3 and 4, respectively). Interestingly, in RIN cells, the anti-GCL antibodies detected a slower migrating band in addition to the 60 kDa protein (Fig. 4, lane 3; Fig. 5, upper panel). This could represent a post-translational modified GCL. Western blot analysis using anti-LAP2β antibodies revealed that LAP2β was also strongly expressed both in RIN cells and pancreas (lanes 5 and 6, respectively). We next examined the cellular localization of the endogenous GCL using confocal analysis of RIN cells that were stained with the anti-GCL antibodies. As seen in Fig. 4B, endogenous GCL was localized to the NE, similar to the staining of its Drosophila GCL homologue (Jongens et al., 1994). In addition, an intranuclear speckled pattern was observed that resembles the staining pattern of other PTB/POZ containing proteins such as BCL6 and PLZF (Dhordain et al., 1995; Dong et al., 1996). Confocal analysis of RIN cells that were stained using both anti-LAP2β and anti-mGCL antibodies revealed the co-localization of both proteins to the NE in all cells (Fig. 4C, upper panel). However, no co-localization was observed in the intranuclear speckles (Fig. 4C, lower panel). This indicates that these speckles are not invagination of the NE but might represent some kind of nuclear bodies similar to the promyelocytic leukaemia oncogenic domains (Dyck et al., 1994) or RNA splicing bodies (Spector et al., 1991).

LAP2β is an integral membrane protein of the NE that is inserted in the membrane via its transmembrane domain found in its C terminus (Fig. 1). By contrast, no sequence suggestive of a transmembrane domain is found in mGCL. RIN cells expressing endogenous GCL and LAP2β were biochemically extracted with either 8 M urea or a mixture of nonionic detergent plus salt (Fig. 5). GCL fractionated with the nuclear pellet and was completely extracted by 8 M urea, indicating that it was not inserted into the nuclear membrane. Using the same conditions, the faster migrating band of LAP2β was retained in the nuclear pellet, whereas the slower migrating band, which might represent a mitotic, phosphorylated form of this protein, was found in the soluble fraction. A similar distribution of LAP2β under these conditions was previously demonstrated (Dechat et al, 1998). The same results were obtained for both mGCL and LAP2β when nuclei were extracted with 1 M NaCl in the absence of Triton X-100 (data not shown). Additionally, when 1% Triton X-100 and either 250 mM or 500 mM NaCl was used, LAP2β was completely solubilized as expected, whereas approximately half of GCL co-fractionated with it. The fact that part of GCL was not solubilized using high salt conditions suggest that it associates with large complexes or aggregates in the nucleus and that it might interact tightly with at least one additional partner other than LAP2β. This might correlate with the intranuclear speckled localization of GCL, to which LAP2β is not co-localized (Fig. 4C).

These results support our hypothesis that, although GCL is specifically associated with LAP2β at the NE, it is not an integral membrane protein but is rather a nuclear matrix protein, which might associate with complex nuclear structures containing lamins or chromatin, in addition to its previously reported binding to the E2F-DP transcription factor.

Recently, GCL was independently isolated as a specific DIP (de la Luna et al., 1999). It was shown that co-expression of mGCL with the E2F5-DP3a heterodimer caused the translocation of this transcription factor to the NE and a significant reduction in its transcriptional activity (there). We hypothesize that LAP2β is the protein responsible for the anchoring of mGCL to the NE. We therefore examined whether LAP2β also affected the transcriptional activity of the E2F-DP3 complex. This was performed by transfecting H1299 cells with various combinations of the E2F5, DP3α, LAP2β and mGCL proteins, and studying their effect on the activity of a luciferase reporter gene under the control of a minimal promoter containing three E2F binding sites in tandem. As can be seen in Fig. 6, both E2F5 and DP3α activated the promoter on their own (lanes 2 and 5), albeit at different potencies: E2F5 was usually two to three times more potent then DP3α. Co-expression of mGCL with either DP3α or E2F5 caused the reduction of DP3α’s activity, as expected, but, interestingly, also reduced the activity of E2F5 (lanes 6 and 3, respectively). Similarly, co-expression of LAP2β with either DP3α or E2F5 caused a reduction of their transcriptional activity (lanes 7 and 4, respectively). When E2F5 and DP3α were co-expressed, their activity as a complex was slightly more than additive (lane 8). Co-expression of the E2F5-DP3α heterodimer plus either mGCL (lane 9) or LAP2β (lane 10), caused a reduction of 50-60% of the transcription activity, with LAP2β reproducibly causing more repression than mGCL (lanes 9 and 10, respectively). By contrast, addition of LAP2ζ, which does not bind mGCL (Fig. 2), did not affect the transcriptional activity of the heterodimer (lane 11). Co-expression of the E2F5-DP3α heterodimer with both LAP2β and mGCL resulted in the reduction of transcriptional activity to nearly background level (lane 12) and was equal to that of pRb (lane 13) transfected at the same DNA concentration. These results suggest that LAP2β and mGCL might be potent inhibitors of gene expression for genes regulated by the E2F-DP complex and possibly other transcriptional regulators.

In this study, we have isolated a novel LAP2β binding partner, mGCL, the mouse homologue of the Drosophila germ-cell-less protein (dGCL) and a BTB/POZ domain containing protein. we have shown that mGCL binds LAP2β in vitro and that endogenous GCL co-localizes with LAP2β at the NE. the mGCL protein does not contain any region resembling a transmembrane domain and fractionation studies showed that it is not an integral protein of the NE but rather has the biochemical characteristics of an insoluble nuclear matrix protein, similar to lamins. the high concentrations of both mGCL and LAP2β that were needed for the in vitro binding (Fig. 2) together with their different stability upon salt extraction (Fig. 5) might suggest that their interaction could either be of low affinity or be highly dynamic in the cell. this is the first example of an interaction between an inner nuclear membrane protein and a nuclear peripheral protein other than lamins.

mGCL is a BTB/POZ domain containing protein that was originally cloned as a DIP (de la Luna et al., 1999). DP3 belongs to the DP family of proteins, which are essential heterodimeric partners of the E2F proteins in creating the various functional E2F transcription factors (Ormondroyd et al., 1995; Lam and La Thangue, 1994). These factors have a major role in the co-ordination and integration of early cell-cycle progression and an aberrant regulation of E2F activity is found in most human tumour cells (Lam and La Thangue, 1994). mGCL was shown to interact with the E2F5-DP3 heterodimer, probably through the DP subunit, and to cause its localization to the region of the NE (de la Luna et al., 1999). Co-expression of mGCL with the E2F5-DP3 transcription factor caused a reduction in its transcriptional activity (de la Luna et al., 1999).

LAP2β is an integral protein of the NE, which was shown to bind lamins and chromatin in a cell-cycle-dependent manner. Expression of LAP2β deletion mutants revealed a role for this protein in the expansion of reforming nuclei and an effect on DNA replication efficiency in Xenopus extracts (Foisner and Gerace, 1993; Yang et al., 1997; Gant et al., 1999). This replication effect is hypothesized to be due to LAP2β effect on chromatin structure. A direct role for LAP2β in transcription regulation has, however, not yet been examined or demonstrated.

In our study, we show that, similar to its binding partner (mGCL), LAP2β can also reduce the activity of the E25F-DP3 heterodimer. Although the nuclear lamina was shown to be involved in transcription regulation (Mancini et al, 1994; Imai et al, 1997; Rafiq et al, 1998), we demonstrate for the first time evidence for the direct involvement of an integral NE protein in transcriptional repression. We also show that the co-expression of both LAP2β and mGCL causes a stronger transcriptional repression than that of either mGCL or LAP2β alone, and that this repression equals the effect of the pRb protein, the best known inhibitor of the E2F complex. These results could reveal a novel spatial regulatory pathway for the E2F complex. In this pathway, under certain physiological circumstances, the complex is bound to the nuclear envelope through mGCL and LAP2β and is thus rendered inactive. We propose two possible models for this repression of transcriptional activity.

In the first model, the spatial separation of the E2F-DP heterodimer from its target promoters, by its association to the NE, causes repression of transcriptional activity. This can then be viewed as part of an emerging pattern in which E2F-DP activity is controlled by regulating its intracellular location (de la Luna et al., 1996; Magae et al., 1996; Allen et al., 1997; Lindeman et al., 1997). The second model involves transcriptional repression through modulation of higher order chromatin structure. Several studies in mammals, Drosophila and yeast have demonstrated that the recruitment of euchromatin genes to heterochromatin regions at the nuclear periphery can cause ‘position effect’ transcriptional repression of these genes (Andrulis et al., 1998; Henikoff et al., 1995; Brown et al., 1997). For example, the Ikaros transcriptional regulator, which usually activates lymphocyte specific expression, was found to associate with transcriptionally inactive genes in heterochromatin foci (Brown et al., 1997). To this effect LAP2β and mGCL might act as heterochromatin ‘recruiters’. As such, they might cause the translocation of the E2F-DP heterodimer to the NE while still bound to chromatin through its target promoters. The chromatin is then modulated into a repressive form, thus inactivating transcription of E2F-DP regulated genes. A major group of proteins involved in transformation of chromatin into a repressive form is the histone deacetylases (HDACs) family (Cress and Seto, 2000). Because mGCL contains a BTB/POZ domain, it is interesting that this domain, found in several other proteins, was proposed to mediate transcriptional repression through recruitment of HDACs (Muto et al., 1997; Dhordain et al., 1997; David et al., 1998; Huynh and Bardwell, 1998). It would be interesting to check whether mGCL is indeed associated with HDACs. If demonstrated, this interaction could provide the actual mechanism for chromatin modulation discussed above.

Our results show that LAP2β was able to reduce the transcriptional activity of the E2F5-DP3α heterodimer without the co-expression of mGCL (Fig. 6, lane 10). This effect is probably not exerted through the interaction of LAP2β with endogenous mGCL because no expression of mGCL was revealed in these cells (data not shown). Moreover, LAP2β reduced the activity of E2F5 without the co-expression of DP3 (Fig. 6, lane 4). Thus, we believe that LAP2β might regulate the activity of the E2F-DP complex through interacting with other cellular E2F-associating proteins, or by a more general effect on chromatin structure. Such an effect is proposed to be connected to LAP2β’s recently reported association with BAF, a chromatin and DNA binding protein (Furukawa, 1999).

The mGCL protein was able to reduce the transcriptional activity of the E2F5-DP3 heterodimer without the co-expression of LAP2β and also repressed E2F5 without co-expression of DP3 (Fig. 6, lanes 9 and 3, respectively). The mGCL protein was also shown to possess an intrinsic ability to inhibit transcription, provided that it was brought to DNA by a DNA binding protein such as the GAL4 DBD (de la Luna et al., 1999). This transcriptional regulation could be through the interaction of mGCL with endogenous LAP2β, a ubiquitous protein found in nearly all cells (Zevin-Sonkin et al., 1992; Harris et al., 1994; Berger et al., 1995; Theodor et al., 1997; Ishijima et al., 1996). Alternatively, the activity of HDACs, mediated through the mGCL BTB/POZ domain, as discussed above, might be involved.

Finally, there is evidence that the regulatory pathway proposed here is not specific for the E2F-DP protein but might be relevant to other transcription factors. For example, a similar regulatory mechanism was found to involve the insulin transcription factor PDX/IPF1. In insulinoma cells, it was shown that the NE localization of PDX/IPF1 was connected with repression of its transcriptional activity (Rafiq et al., 1998). It is yet unknown what factors cause the translocation of PDX/IPF1 to the NE but, as we show in Fig. 4, GCL is highly expressed in pancreas and specifically in insulinoma cells. Thus, it would be interesting to check whether LAP2β and GCL are also involved in the regulation of the transcriptional activity of PDX in these cells.

Recent studies from various groups strongly link the NE to human diseases. Loss or defects in the emerin protein cause the X-linked recessive form of Emery-Dreifuss muscular dystrophy (EDMD) (Bione et al., 1994). Mutations in LMNA gene, which encodes lamins A and C, cause three autosomal dominant diseases: a form of EDMD (Bonne et al., 1999), a form of dilated cardiomyopathy (CMDA1) (Fatkin et al., 2000) and Dunnigan-type familial partial lipodystrophy (FPLD) (Cao and Hegele, 2000; Shackleton et al., 2000). In this study, we have mapped the mGCL gene to the central region of chromosome 6 (Fig. 3), a region that is syntenic to human chromosome 2p13-14. Indeed, human GCL (HGCL), recently cloned and characterized by us, was mapped to this region (Nili et al., 2001). Linkage analysis studies mapped the gene responsible for the Alstrom syndrome, a rare autosomal recessive disorder (Alstrom et al., 1959) to the same region (Collin et al., 1997; Collin et al., 1999; Macari et al., 1998). Alstrom syndrome is characterized by retinitis pigmentosa, deafness, obesity, hyperlipidaemia and non-insulin-dependent diabetes mellitus (NIDDM). In some cases, acanthosis nigricans, cardiomyopathy, hepatic dysfunction, progressive chronic nephropathy and male hypogonadism are also observed. We have proposed that HGCL is a candidate gene that is responsible for the Alstrom syndrome (Nili et al., 2001). If further studies searching for GCL mutations in this syndrome do indeed verify this hypothesis, this will be yet another example of the connection of the NE and its related proteins to disease. Interestingly, some of the symptoms in this syndrome, such as the NIDDM and cardiomyopathy, overlap those of CMDA1 and FPLD, which are caused by LMNA mutations. This raises the possibility of a common defective mechanism in these diseases. Based on the ability of both mGCL and LAP2β to regulate transcription, as presented in this study, this mechanism could be the regulation of gene expression, which is normally dependent on correct interactions and concerted interplay between transcription factors, chromatin and proteins of the NE.

We thank D. B. Householder for excellent technical assistance. This research was supported, in part, by the National Cancer Institute, DHHS. G. R. holds the Dora and Gregorio Shapiro chair of Hematologic Malignancies, Sackler School of Medicine, Tel-Aviv University. D. G. holds the Recanati Career Development Chair of Cancer Research, Weizmann Institute of Science. The work performed is part of the PhD thesis of E. N., to be submitted to the Sackler School of Medicine, Tel-Aviv University, Israel.