INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

During embryogenesis, many different cell types develop from one fertilized egg. Cell type specificity emerges as a result of differential expression of regulatory genes. Notably, cell-specific sets of active and inactive genes determine the cell's identity. To preserve the identity of the cell, it is important that these specific expression patterns be maintained and stably inherited by daughter cells in a cell-type-specific manner. Therefore, the maintenance of cell type specificity needs to be regulated by a cellular memory system. In Drosophila, for instance, the products of the Polycomb-group (PcG) genes are required for stable repression of gene activity. PcG proteins are evolutionarily conserved, being involved in the inheritably stable repression of homeotic gene expression both in Drosophila and in vertebrates (8, 14, 16, 24, 27).

It has been observed that in Drosophila, different PcG proteins, including Polycomb (Pc), Polyhomeotic (Ph), and Posterior sex combs (Psc), bind in overlapping patterns on polytene chromosomes (18, 36). Based on this observation, it has been proposed that PcG proteins repress gene activity via the formation of multimeric protein complexes. With the genetic yeast two-hybrid system, it is possible to search for direct protein-protein interactions in order to determine the identities of PcG complex components. In this way, several vertebrate PcG homologs have been found to interact. The human homologs of Ph, HPH1 and HPH2, have been found to interact with each other and with BMI1, the vertebrate homolog of the Drosophila PcG protein Psc (9). A human Pc homolog, HPC2, interacts with a RING finger protein, RING1 (21). It has further been found that Pc and Ph coimmunoprecipitate in Drosophila (6). The human HPH1, HPH2, BMI1, HPC2, and RING1 proteins also coimmunoprecipitate, and they colocalize in distinct nuclear domains of mammalian cell lines, termed PcG domains (9, 21). Similar biochemical interactions between homologs of Pc, Ph, Psc, and RING1 have been identified in mice and in Xenopus embryos (1, 10, 19, 21).

Expression analyses of several vertebrate PcG proteins reveal that they are differentially distributed in tissues and cell lines and that the expression of certain PcG proteins in these tissues is dependent on the time of development (4, 9, 15, 19, 21). This finding suggests that different, specific PcG complexes exist with different protein compositions. Direct evidence for the existence of two different vertebrate PcG complexes is gained from the characterization of the vertebrate PcG protein EED. EED coimmunoprecipitates and colocalizes with the mammalian PcG protein Enx1/EZH2 but not with other vertebrate PcG proteins such as HPC2 or BMI1. These findings indicate the existence of different, specific vertebrate PcG complexes that may contribute to specificity for target genes and possibly for different tissues (26, 34).

Recently, we have shown that interference with the function of HPC2 deregulates the expression of the proto-oncogene c-myc. Overexpression of HPC2 results in repression of c-myc. Overexpression of a dominant-negative HPC2 deletion mutant, HPC2, which lacks a conserved C-terminal domain that is crucial for HPC2-mediated gene repression, led to enhanced expression of the c-myc gene in several mammalian cell lines. Concomitantly, overexpression of HPC2 results in cellular transformation and anchorage-independent growth in mammalian cells (22). Although it cannot be concluded whether the effect of HPC2 on c-myc is direct or indirect, these data suggest that one function of the mammalian PcG proteins is to repress the transcription of certain proto-oncogenes. Importantly, HPC2 is not the only PcG member found to be linked with oncogenesis. Two other mammalian PcG proteins, Bmi-1 and mel-18, have also been shown to be involved in tumorigenesis. The mouse PcG gene bmi-1 collaborates with the proto-oncogene c-myc to cause lymphomas (11, 33). Interference with the expression of the mammalian PcG protein mel-18 induces tumors in nude mice (13). These findings indicate that mammalian PcG proteins have oncogenic properties.

Previously we found that the human RING1 protein interacts with HPC2 and is associated with the human PcG protein complex (21). However, little is known about the function of RING1. Here, we analyzed the functions of RING1 in more detail. Using directed two-hybrid and in vitro protein-protein analyses, we found that RING1 is able to interact with multiple human PcG proteins. We also overexpressed RING1 in mammalian cells and analyzed the differences in gene expression patterns. We found that overexpression of RING1 repressed the gene activity of En-2, a mammalian homolog of engrailed, a well-characterized Drosophila PcG target gene. Overexpression of RING1 further deregulated the expression of the proto-oncogenes c-jun and c-fos. Concomitant with the changes in the expression levels of these oncogenes, cellular transformations and the formation of tumors in athymic mice were induced. Our data suggest that RING1 interacts with multiple human PcG proteins and that overexpression of RING1 leads to oncogenic transformations by the deregulation of specific oncogenes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Construction of the pAS3 two-hybrid vector. Using the pAS2 two-hybrid vector (Clontech), we obtained a GAL4 DNA binding domain (DBD) fusion protein in which the GAL4 DBD is positioned at the N terminus of the protein. To generate C-terminally positioned GAL4 DBD fusion proteins, we constructed a new two-hybrid vector in which the GAL4 DBD is placed downstream of the polylinker to create pAS3, a C-terminal fusion protein that is very similar to pAS2. To construct the pAS3 vector, the ADH promoter and the GAL4 DBD domain from pAS2 were recloned. By PCR, we derived the GAL4 DBD fragment, amino acids (aa) 1 to 147, and the ADH promoter, using pAS2 as a template. The GAL4 DBD and the ADH promoter fragments were cloned in pBluescript in a two-step ligation, creating an ADH promoter-polylinker-GAL4 DBD cassette, which has been entirely sequenced. The pAS2 vector was digested with SacI/SalI, releasing the original ADH promoter, GAL4 DBD, polylinker, and CYH2 selection gene and replacing them by the new ADH promoter-polylinker-GAL4 DBD cassette. The 7.5-kb pAS3 vector has the same properties as the pAS2 vector but lacks the CYH2 selection gene.

Analysis of interacting proteins with the two-hybrid system. Indicated fragments of the cDNAs encoding RING1, BMI1, HPC2, HPH1, HPH2, Enx1, and EED were derived via PCR (Expand; Boehringer). The fragments were subcloned into the pAS2, pAS3, and pGAD10 (GAL4 transactivation domain [TAD]) vectors. The fragments were sequenced over their entire lengths. The resulting plasmids were cotransformed into Saccharomyces cerevisiae Y190. The transformants were plated on medium lacking leucine, tryptophan, and histidine, with or without 30 mM 3-amino-1,2,4-triazole (3-AT). Interactions were scored negative if they failed to grow in the presence of 30 mM 3-AT. Under these nonselective conditions, negative interactions were -galactosidase negative. Positive interactions meet the two criteria of growing in the presence of 30 mM 3-AT and testing -galactosidase positive. To exclude the possibility that the negative interactors did not produce either one of the fusion proteins, we Western blotted equal amounts of protein and incubated the blots with monoclonal antibodies that specifically recognize the GAL4 DBD or TAD protein (Clontech, Palo Alto, Calif.). All positive and negative interactors expressed both GAL4 DBD fusions and the GAL4 TAD fusions at approximately the same levels (data not shown).

Construction of GST fusion proteins, protein preparation, and in vitro binding assay. A 1,131-bp fragment of the RING1 cDNA which encompasses the entire coding region and corresponds to aa 1 to 377 was cloned into pGEX-2TK, thus creating glutathione S-transferase (GST)-RING1. A 990-bp fragment of the bmi-1 cDNA (a gift from M. van Lohuizen) which covers the entire coding sequence and corresponds to aa 1 to 324 was cloned into pGEX-2TK, thus creating GST-Bmi-1. Expression of the GST fusion proteins was induced for 3 h at 30°C with 0.4 mM isopropyl--D-thiogalactopyranoside (IPTG) as instructed by the manufacturer (Pharmacia) (29). The cells were pelleted, resuspended in binding buffer (phosphate-buffered saline containing 1 mM EDTA, 1 mM dithiothreitol, 2 mM phenylmethylsulfonyl fluoride, leupeptin [10 µg/ml], benzamidine [10 µg/ml], trypsin inhibitor [10 µg/ml], and aprotinin [10 µg/ml]) and sonicated. Triton X-100 was added to a final concentration of 1% (vol/vol), and the lysate was incubated for 30 min on ice. Cell debris was removed by centrifugation for 10 min at 14,000 × g, the supernatant was added to glutathione-Sepharose 4B, and the mixture was incubated for 30 min at 4°C. The beads were collected by centrifugation and washed extensively with binding buffer. Capped synthetic HPC2, RING1, and bmi-1 mRNAs were made by in vitro transcription and translated at 20 µg/ml in a rabbit reticulocyte lysate in the presence of [³⁵S]methionine (19). A 10-µl slurry of GST fusion protein (immobilized to glutathione-Sepharose) was preincubated for 30 min on ice in a final volume of 200 µl of binding buffer containing 0.5% Nonidet P-40 and 1 mg of bovine serum albumin per ml. Subsequently, 3 µl of the reticulocyte lysate was added to the mixture and incubated for 30 min at 4°C with rotation. The beads were washed five times with 1 ml of ice-cold binding buffer. The complexes were separated on sodium dodecyl sulfate-polyacrylamide gels, which were subjected to fluorography.

Western blot analysis of RING1. Expression of the RING1 protein was analyzed in cell lysates of RING1 stably transfected Rat1a (Rat1a/RING1) and control cell lines. For RING1 detection, the blots were incubated with a 1:1,000 dilution of affinity-purified rabbit anti-RING1 antibodies (21). Equal amounts of proteins were loaded, as measured by the bicinchoninic acid method (30) and as visualized by Coomassie staining of a gel.

Atlas cDNA expression array. Rat1a cells overexpressing wild-type RING1 or pcDNA3 vector, which were used in the soft agar growth assay, were grown, and stably transfected lines were selected by culturing the cells in Dulbecco's minimal essential medium supplemented with 10% newborn calf serum containing 500 µg of Geneticin (G418; Gibco) per ml for 2 weeks. Surviving cells were clonally expanded in medium containing 250 µg of G418 per ml for 2 to 4 weeks. Individual cell clones were selected and cultured in individual dishes. After five passages, poly(A)⁺ RNA was isolated and subjected to differential display using the commercial mouse Atlas expression arrays (Clontech). We also blotted poly(A)⁺ RNA of the selected Rat1a/RING1 clones and control cells and hybridized the blots with probes for GAPDH, c-jun, c-fos, c-myc, and En-2. Isolation of RNA and Northern analysis were performed according to standard procedures. The blots were hybridized with [-³²P]dATP-labeled DNA probes, and the blots were autoradiographed with intensifying screens at 70°C, using preflashed X-ray films.

Soft agar growth assay. Cell lines were analyzed for anchorage-independent growth as described previously (20, 28, 31). Rat1a cells were transfected by the calcium phosphate transfection procedure with full-length RING1, an N-terminal part of RING1 (RING1 aa 1 to 203), and a C-terminal part of RING1 (RING1 aa 154 to 377), all cloned in the pcDNA3 vector. As a positive control, c-myc cDNA cloned in the pRcCMV vector and the C-terminal deletion mutant of HPC2 (HPC2) (22) were transfected. As a negative control, the pcDNA3 vector alone was transfected. The cells were subjected to Geneticin (G418; 500 µg/ml) selection. Cells were cultured for 14 days. The clones were trypsinized, and cells were counted. Then 5 × 10⁴ cells in 5 ml of 10% Dulbecco's modified Eagle's medium containing 0.4% (wt/vol) agarose were seeded in 5-cm-diameter petri dishes which contain 1% (wt/vol) agarose. Plates were inspected 21 to 28 days after seeding of the cells, and colonies were counted. The entire procedure, including transfection of cDNAs, was performed in triplicate.

Metastasis in athymic mice. For this study, we used athymic nude (nmri/nude) mice that at the time of injection were 4 to 6 weeks of age. All mice were maintained in microisolator cages under HEPA-filtered laminar air. NIH 3T3 cells were transfected with pcDNA3-RING1 and pcDNA3 via calcium phosphate transfection and allowed to grow for 1 week in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 250 µg of geneticin (G-418) per ml. Cells were prepared for injection only from cultures in logarithmic growth at the time of harvest. The cells were briefly treated with 0.025% trypsin and 0.1% EDTA in salt solution. The cells were quickly removed from trypsin by centrifugation, resuspended in saline, and injected within 1 h in 0.2 ml in the body cavity with a 26-gauge needle. The mice were maintained under aseptic barrier conditions until the end of the experiment. After 6 weeks, the animals were analyzed for tumors at the surface and in sections of tissues.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Multiple interactions between RING1 and PcG proteins in the two-hybrid system. Previously, we used the yeast two-hybrid system to identify proteins that interact with components of the multimeric PcG complex. We found that RING1, a previously identified protein with unknown function, interacts with the vertebrate Pc homologs XPc and HPC2 (21). It has been determined that the evolutionarily conserved C-terminal domain of the Pc homologs is the domain of RING1 interaction. The region within the RING1 protein which is responsible for the interaction with HPC2 has not been mapped in detail. However, RING1 contains a well-characterized zinc binding domain, the RING finger, which is not involved in the interaction with HPC2 (21). It has been argued that the RING finger is a domain involved in mediating protein-protein interactions (7). Therefore, it is feasible that RING1 interacts with other proteins besides HPC2. The fact that RING1 is part of a multimeric protein complex suggests that RING1 indeed may interact with more than one protein.

View this table: [in this window] [in a new window]   TABLE 1.   -Galactosidase activities of RING1 interactions in the two-hybrid system

RING1 interacts with multiple PcG proteins in vitro. Using the two-hybrid system, we investigated the protein-protein interactions of RING1 and found that RING1 is able to interact with multiple PcG proteins (Table 1). However, different RING1 fusion proteins were used, and it was found that the RING1 protein interactions depend on whether the GAL4 DBD is fused to the N-terminal or C-terminal part of the RING1 protein (Table 1). To rule out the possibility of artifactual positive two-hybrid interactions and confirm the RING1 protein interactions, we performed an independent in vitro protein-protein interaction analysis, the GST pull-down assay.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Association of RING1 with itself, Bmi-1, and HPC2 and of Bmi-1 with itself in vitro. GST-RING1 fusion protein, immobilized on glutathione-Sepharose, interacted with in vitro-translated, [35S]methionine-labeled HPC2 or RING1. [35S]methionine-labeled HPC2 (lane 1) was incubated with GST-Sepharose alone (lane 2) and with GST-RING1 (lane 3). [35S]methionine-labeled RING1 (lane 4) was incubated with GST-Sepharose alone (lane 5) and with GST-RING1 (lane 6). GST-RING1 and GST-Bmi-1 fusion proteins, immobilized on glutathione-Sepharose, interacted with in vitro-translated, [35S]methionine-labeled Bmi-1. [35S]methionine-labeled Bmi-1 (lane 7) was incubated with GST-Sepharose alone (lane 8), with GST-RING1 (lane 9), and with GST-Bmi-1 (lane 10). All proteins used in the assay are full length. Molecular masses are indicated in kilodaltons. The input (lanes 1, 4, and 7) was 10% of the amount incubated with the GST fusion proteins.

RING1 contains two different domains involved in RING1-RING1 interaction. Using the yeast two-hybrid system and the in vitro GST pull-down assay, we found that RING1 is able to interact with itself. We performed a deletion analysis, using the yeast two-hybrid system, to determine the domains within the RING1 protein that are required for RING1-RING1 protein interaction. We found that RING1 contains two regions that are able to associate (Fig. 2). Both the N-terminal region (aa 1 to 205) and the C-terminal region (aa 214 to 377) of the RING1 protein interact with full-length RING1 (aa 1 to 377) (Fig. 2A). The N-terminal region contains the RING finger (aa 1 to 65). Mapping the two interaction domains further, we find that the N-terminal region of RING1 (aa 1 to 205) interacts strongly with the same N-terminal region (aa 1 to 205) but not with the C-terminal half of the RING1 protein (aa 214 to 377) (Fig. 2B). In determining whether the RING finger is involved in mediating this interaction, we made two deletion mutants. One RING1 deletion mutant (aa 1 to 80) still contains the N-terminally located RING finger domain (aa 1 to 65), and the other mutant contains the remaining N-terminal region (aa 80 to 200). We found that both regions are able to interact with RING1 (aa 1 to 205), but not as strongly as the intact N-terminal regions interact with each other (Fig. 2B).

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Mapping of homodimerization domains of RING1. (A) Full-length RING1 (aa 1 to 377) was fused to the GAL4 DBD, which in all constructs shown is located at the C-terminal end of RING1. The plasmids were cotransformed with different portions of RING1, which were fused to the GAL4 TAD, which is located at the N terminus of RING1. Interactions were positive (+) when the transformants were able to grow on selective medium lacking histidine and when they were also -galactosidase positive. Relative strength of the interactions is a qualitative indication based on the time needed for blue coloring (++, within 30 min; +, between 30 and 120 min) and the size of the colonies. (B) N-terminal portions of RING1 fused to the GAL4 DBD were tested for interactions with N- and C-terminal portions of RING1. These constructs are fused to the GAL4 TAD. (C) The C-terminal portion of RING1 fused to the GAL4 DBD was tested for interaction with C-terminal portions of RING1 fused to the GAL4 TAD. (D) Schematic representation of the two RING1-RING1 protein interaction domains. The RING finger domain of the RING1 protein is indicated as a hatched black box.

Mapping of the RING1 interaction domain for HPC2. Previously, RING1 was found to interact in the two-hybrid system with the evolutionarily conserved C-terminal box of vertebrate Pc homologs such as HPC2 (21). The domain within the RING1 protein that interacts with HPC2 has not been mapped precisely. We were interested in analyzing whether the RING1-RING1 and RING1-HPC2 protein interaction domains are identical, since we found that one of the interaction regions of RING1 (C-terminal region) is similar to the region of interaction with HPC2 (Fig. 2). The RING1 two-hybrid clone that we identified as interacting with HPC2 encompasses aa 214 to 377 (Fig. 3B). We made further deletions from this fragment and found that we could narrow down the interaction region only to aa 230 to 377 of RING1 (Fig. 3B). Smaller fragments of RING1 (aa 200 to 270 and aa 270 to 370) which were found to be involved in the self-binding of RING1 did not interact with HPC2. Yet a different fragment of RING1, ranging from aa 230 to 320, also appeared to be insufficient for the association with HPC2.

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Mapping of interaction domains between RING1 and HPC2. (A) Indicated portions of HPC2 were fused to the GAL4 DBD, which in all constructs shown is located at the N-terminal end of HPC2. The plasmids were cotransformed with full-length RING1 (aa 1 to 377), which is fused to the GAL4 TAD. The GAL4 TAD is located at the N terminus of RING1. (B) Full-length HPC2 (aa 1 to 558) fused to the GAL4 DBD was tested for interaction with various C-terminal region of RING1 fused to the GAL4 TAD. (C) Schematic representation of the interaction domains of HPC2 and RING1. The HPC2 protein contains a chromodomain and a C box, which are indicated as grey and black dotted boxes, respectively. The RING finger domain of the RING1 protein is indicated as a hatched black box.

The RING finger proteins RING1 and BMI1 interact with each other. In analyzing the two-hybrid grid and performing the in vitro GST pull-down assay, we detected that RING1 and BMI1 are able to interact physically (Table 1 and Fig. 1). Further two-hybrid deletion analyses (Fig. 4) of different BMI1 regions show that the N-terminal region of BMI1 (aa 1 to 136) containing the RING finger motif is the region of interaction with RING1. The central and C-terminal regions of BMI1 (aa 114 to 326 [Fig. 4A]) containing the putative helix-turn-helix-turn-helix-turn motif is not required for the interaction. This finding suggests that the RING finger of BMI1 is the domain of interaction with RING1. However, the RING finger domain of BMI1 (aa 1 to 80) itself is not sufficient for the interaction (Fig. 4A). It seems, therefore, that both the RING finger and a region of the adjacent C-terminal region of BMI1 are needed for association with RING1.

View larger version (28K): [in this window] [in a new window]   FIG. 4.   Mapping of interaction domains between RING1 and BMI1. (A) Full-length RING1 (aa 1 to 377) was fused to the GAL4 DBD, which in all constructs shown is located at the N-terminal end of RING1. The plasmid was cotransformed with the indicated portions of BMI1, which is fused to the GAL4 TAD. In all constructs, the GAL4 TAD is located at the N-terminus of BMI1. (B) The indicated portions of RING1 were fused to the GAL4 DBD and tested for interaction with full-length BMI1 fused to the GAL4 TAD. (C) Schematic representation of the interaction domains of RING1 and BMI1. The RING finger domain of the RING1 protein is indicated as a hatched black box. The BMI1 protein has a RING finger domain and a helix-turn-helix-turn-helix-turn (H-T-H-T-H-T) domain, indicated as grey and striped boxes, respectively.

BMI1 is able to interact with itself. Studying the protein-protein interactions of RING1, we found that RING1 is able to interact with itself. The ability to oligomerize has been detected for several other PcG proteins (9, 10, 19). Therefore, we studied whether BMI1 is also able to interact with itself and found that indeed BMI1 interacts with itself in vitro (Fig. 1). Next, two-hybrid deletion analyses were performed to map the domains of interaction.

View larger version (32K): [in this window] [in a new window]   FIG. 5.   Mapping of homodimerization domains of BMI1. (A) Full-length BMI1 (aa 1 to 326) is fused to the GAL4 DBD, which in all constructs shown is located at the C-terminal end of BMI1. The plasmid was cotransformed with different portions of BMI1, which are fused to the GAL4 TAD. In these constructs, the GAL4 TAD is located at the N terminus of BMI1. (B) An N-terminal portion of BMI1 (aa 1 to 136) fused to the GAL4 DBD was tested for interaction with full-length BMI1 (aa 1 to 326) fused to the GAL4 TAD. (C) The C-terminal portion of BMI1 (aa 1 to 136) fused to the GAL4 DBD was tested for interaction with different portions of BMI1 fused to the GAL4 TAD. (D) Schematic representation of the homodimerization domains of BMI1. The BMI1 protein has a RING finger domain and a helix-turn-helix-turn-helix-turn (H-T-H-T-H-T) domain, indicated as grey and striped boxes, respectively.

RING1 overexpression results in repression of engrailed and enhanced expression of c-jun and c-fos. The PcG protein complex is involved in repression of gene activity. Since RING1 appears to be an integral part of the PcG complex, displaying multiple interactions with PcG proteins, we studied the function of RING1 in regulating gene expression. We overexpressed the RING1 protein in Rat1a fibroblast cells and analyzed differences in gene expression levels.

View larger version (15K): [in this window] [in a new window]   FIG. 6.   Western analysis of stably transfected RING1 and HPC2 proteins from cell extracts of Rat1a cells. Equal amounts of proteins were Western blotted. The blots were incubated with a rabbit anti-RING1 or rabbit anti-HPC2 antibody. (A) Endogenous rat RING1 levels were detected in the untransfected cells (lane 1); elevated levels of RING1 were detected in clone 8 (lane 2) and clone 16 (lane 3). (B) Endogenous rat HPC2 levels were detected in the untransfected cells (lane 1); elevated levels of HPC2 were detected in clone 5 (lane 2). Molecular masses are indicated in kilodaltons.

View larger version (39K): [in this window] [in a new window]   FIG. 7.   Repression of En-2 gene expression activity in RING1-transfected Rat1a cells. Poly(A)+ mRNA isolated from Rat1a control cells (lane 1) and from RING1-transfected clone 8 (lane 2) and clone 16 (lane 3) Rat1a cells was Northern blotted and probed with the En-2 gene. To verify equal RNA loading, the filter was hybridized with a GAPDH probe.

View larger version (33K): [in this window] [in a new window]   FIG. 8.   Expression of c-myc, c-jun, and c-fos in RING1- and HPC2-transfected Rat1a cells. (A) Poly(A)+ mRNA isolated from Rat1a control cells (lane 1) and from RING1-transfected Rat1a clone 8 (lane 2) and clone 16 (lane 3) cells was Northern blotted and probed with fragments of c-jun, c-fos, and c-myc. (B) Poly(A)+ mRNA isolated from Rat1a control cells (lane 1) and from HPC2-transfected Rat1a clone 5 (lane 2) cells was Northern blotted and probed with fragments of c-jun, c-fos, and c-myc. To verify equal RNA loading, the filter was hybridized with a GAPDH probe.

Effects of overexpression of HPC2 on proto-oncogene expression. In a previous study (22), we had shown that overexpression of a C-terminal deletion mutant of HPC2, HPC2 (aa 1 to 530), which is not able to repress gene activity, resulted in elevated expression of c-myc in the mammalian cell lines C57MG and U-2 OS. Surprisingly, overexpression of RING1, which is likely a molecular partner of HPC2 in vivo, does not result in a changed c-myc expression level in Rat1a cells (Fig. 8A). The difference in c-myc gene activity by the overexpression of either RING1 or HPC2 may depend on the difference in cell lines that are used in the two studies. Another plausible explanation is that RING1 and HPC2 affect expression of the c-myc proto-oncogene differently. To address this question, we also analyzed the expression levels of the proto-oncogenes c-myc, c-jun, and c-fos in Rat1a cells stably transfected with HPC2.

RING1 induces anchorage-independent growth. Rat1a fibroblast cells are frequently used to determine the neoplastic transformation potential of genes (5, 20, 22, 28, 31). Overexpression of the proto-oncogene c-myc alone in Rat1a cells is sufficient to induce anchorage-independent growth in soft agarose (28, 31). Overexpression of the dominant-negative C-terminal deletion mutant HPC2 enhances the expression of c-myc and induces anchorage-independent growth upon overexpression in Rat1a cells (22). We found that overexpression of RING1 in Rat1a fibroblast cells results in an enhanced expression of the proto-oncogenes c-fos and c-jun but not c-myc (Fig. 8A). We therefore analyzed the potential of the RING1 protein to induce anchorage-independent growth in Rat1a cells.

View larger version (113K): [in this window] [in a new window]   FIG. 9.   Overexpression of RING1 induces anchorage-independent growth in control Rat1a cells (A) and in Rat1a cells transformed with the c-myc oncogene (B), with the C-terminal deletion mutant of HPC2 (C), and with full-length RING1 (D). Bar = 400 µm.

RING1 demonstrates metastatic activity in athymic mice. Invasion and metastasis have been considered the hallmarks of malignant tumors. NIH 3T3 cells overexpressing oncogenes are found to be metastatic in nude mice (3, 35). This metastatic assay is often used to assess the oncogenic potential of genes. For the PcG proteins Bmi-1 and mel-18, involvement in the formation of tumors in mice has been established (2, 13). Transgenic mice develop lymphomas when overexpressing Bmi-1, which is considered an onco-protein. Nude mice injected with NIH 3T3 cells overexpressing antisense mel-18 also develop tumors (13). We found RING1 to be a potent inducer of anchorage-independent growth in cells and therefore determined the metastatic potential of RING1 by injecting athymic nude mice with NIH 3T3 cells that overexpress RING1.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

RING1 interacts with multiple PcG proteins. PcG proteins serve as components of multimeric PcG protein complexes, which are involved in the heritable repression of gene activity. RING1 is a newly identified PcG complex-associated protein in mammals. However, the role of RING1 as a component of the PcG complex is unclear. RING1 is found to interact with vertebrate Pc homologs, but a Drosophila RING1 homolog has not been identified. To better understand the role of RING1 as a component of the PcG complex, we investigated the protein-protein interactions of RING1.

Model for distinct PcG complexes. RING1 interacts with multiple PcG proteins and therefore may serve as a central protein for the establishment of a multimeric protein complex. Recently, a human RING1 homolog, dinG, has been identified (12). Two mouse RING homologs have also been identified; mouse Ring1A and Ring1B are homologous to RING1 and dinG, respectively. The dinG protein is 31 aa shorter than and 53% identical with RING1. It contains three regions that are very homologous with RING1. The first 150 aa, including the RING finger, are almost 100% identical; the other two regions of homology are located in the C-terminal part of the proteins and are about 70% identical. Especially the central region, between the RING finger and the C-terminal homology domains, is a region with little homology between the two RING proteins. We found that the N-terminal region of RING1 (aa 1 to 200), which is well conserved between RING1 and dinG, is involved in the self-association of the protein. Therefore, it is likely that dinG, like RING1, can interact with itself via its N-terminal part and that RING1 and dinG are able to interact with each other. The possible interaction of RING1 and dinG would also involve the N-terminal parts of the proteins.

View larger version (29K): [in this window] [in a new window]   FIG. 10.   Model of human PcG multimeric protein complexes. (A) Model of a PcG protein complex which contains RING1/Ring1A; (B) dinG/Ring1B as the central protein for the establishment of a multimeric PcG protein complex.

Deregulated gene activity in RING1-transformed cells. In Drosophila, PcG proteins have been identified as repressors of homeotic genes (18, 36). PcG proteins have also been found to regulate the expression of gap genes (17) and to self-regulate the expression of certain PcG proteins (18, 36). Previously we have shown that interference with the function of HPC2 deregulates the expression of c-myc. PcG proteins bind to more than 100 loci on the polytene chromosome, of which the majority of the genes have not been determined. It is likely that besides the regulation of genes involved in development, different classes of genes are regulated by PcG proteins. We analyzed the differential expression of genes after overexpression of RING1 in rodent Rat1a fibroblast cells.

Involvement of RING1 in tumorigenesis. In this study, we investigated the role of RING1 as a component of the vertebrate PcG complex. We found that RING1 interacts with multiple PcG proteins, which indicates that RING1 may function as a central protein in the formation of vertebrate PcG protein complexes. Since several PcG proteins have been implicated in tumorigenesis, we analyzed the potential role of RING1 in this process. We found that overexpression of RING1 results in enhanced expression of the proto-oncogenes c-jun and c-fos but not c-myc. Concomitantly, RING1 is able to induce anchorage-independent growth of Rat1a cells. Moreover, RING1 is a more potent inducer of anchorage-independent growth than are c-myc and HPC2, as measured by both number and size of foci. This observation coincides with the finding that overexpression of HPC2 results in enhanced expression of both c-myc and c-fos but not of c-jun. It suggests (i) that the induction of anchorage-independent growth by RING1 is mediated by elevated c-fos and c-jun expression, whereas the induction of anchorage-independent growth by HPC2 is mediated by elevated c-fos and c-myc expression, and (ii) that the cooperative action of elevated c-fos and c-jun levels is better able than that of enhanced c-fos and c-myc levels to induce neoplastic transformation in Rat1a cells. In this respect, it is of considerable interest that overexpression of human c-jun alone is able to transform Rat1a cells and that this effect is enhanced by the expression of c-fos (25). Finally, the differential enhancement of c-fos, c-jun, and c-myc expression due to overexpression of either RING1 or HPC2 suggests that these proteins cause cellular transformation through different molecular pathways.