Identification and Cloning of a Novel Family of Coiled-coil Domain Proteins That Interact with O-GlcNAc Transferase*

The abundant and dynamic post-translational modification of nuclear and cytosolic proteins by β-O-linked N-acetylglucosamine (O-GlcNAc) is catalyzed by O-GlcNAc transferase (OGT). Here we used the yeast two-hybrid approach to identify and isolate GABAA receptor-associated protein, GRIF-1 (Beck, M., Brickley, K., Wilkinson, H. L., Sharma, S., Smith, M., Chazot, P. L., Pollard, S., and Stephenson, F. A. (2002)J. Biol. Chem. 277, 30079–30090), and its novel homolog, OIP106 (KIAA1042), as novel OGT-interacting proteins. The proteins are highly similar to each other but are encoded by two separate genes. Both GRIF-1 and OIP106 contain coiled-coil domains and interact with the tetratricopeptide repeats of OGT. GRIF-1 and OIP106 are modified by O-GlcNAc and therefore are substrates for OGT. However, unlike another high affinity protein substrate, such as nucleoporin p62, OIP106 and GRIF-1 co-immunoprecipitate with OGT, exhibiting stable in vitro and in vivoassociations. Whereas GRIF-1 has been reported to be expressed only in excitable tissue, OIP106 is expressed in all human cell lines that were examined. Confocal and electron microscopy show that OIP106 localizes to nuclear punctae in HeLa cells and co-localizes with RNA polymerase II. Co-immunoprecipitation experiments confirm the presence of an in vivo RNA polymerase II-OIP106-OGT complex, suggesting that OIP106 may target OGT to transcriptional complexes for glycosylation of transcriptional proteins, such as RNA polymerase II, and transcription factors. Similarly, GRIF-1 may serve to target OGT to GABAA receptor complexes for mediating GABA signaling cascades.

and diverse in form and function, ranging from transcription factors (4 -6) to RNA polymerase II (7), oncoproteins (8), tumor suppressors (9), viral proteins (10), cytoskeletal proteins (11,12), kinases (13) and phosphatases (14). Furthermore, O-Glc-NAc-modified proteins are phosphoproteins as well, often belonging to large multimeric and reversible protein complexes. Indeed, in many cases, the sites of O-GlcNAc modification are either the same or adjacent to those modified by O-phosphate (15). Thus, there is mounting evidence supporting the hypothesis that O-GlcNAc is a regulatory modification analogous to phosphorylation. For example, the transcription factor Sp1 is extensively modified by O-GlcNAc, and it has been shown that the presence of the sugar in the transactivation domain inhibits its homomultimerization and transcriptional capability (4,5,16). O-GlcNAc modification of p67 regulates protein synthesis by controlling the phosphorylation status of eukaryotic initiation factor-2␣ (17,18). In cases where the sites of modification of O-GlcNAc and O-phosphate are the same, a reciprocal relationship between the two modifications has been suggested (19,20).
Enzymes that cycle the O-GlcNAc modification are analogous to those that catalyze phosphorylation (i.e. kinases and phosphatases). The enzyme that attaches the saccharide to proteins is uridine diphospho-N-acetylglucosamine:polypeptide ␤-N-acetylglucosaminyltransferase, or O-GlcNAc transferase (OGT), and its counterpart is a O-GlcNAc-specific ␤-N-acetylglucosaminidase known as the O-GlcNAcase. Both of these enzymes have been purified and characterized (21)(22)(23)(24)(25)(26). Recently, the cDNAs that encode for OGT (25,26) and O-GlcNAcase (23), have been cloned from rat, Caenorhabditis elegans, and human. The OGT is a highly unique and ubiquitous glycosyltransferase, encoded by a single gene. OGT is highly conserved throughout evolution from C. elegans to humans. The gene for OGT has been mapped to the X chromosome (Xq13 in humans) (27,28). Targeted deletion of the OGT gene results in ES cell lethality in mice (27). Thus, OGT is essential for life at the single cell level.
OGT localizes to both the nucleus and cytoplasm of cells, but it is present at higher levels in the nucleus (25). Aside from being tyrosine-phosphorylated and O-GlcNAc-modified, the rat 110-kDa OGT enzyme contains 11 1 ⁄2 tetratricopeptide repeats (TPRs), a protein-protein interaction domain found in many proteins (25,29,30). TPRs have been shown to mediate protein-protein interactions in a variety of proteins (37). Recently, the crystal structure of the TPR domain of two of these proteins, protein phosphatase 5 (31) and Pex5p (32), were solved. OGT exists as a homotrimer, and the TPR domain was shown * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF474163.
Unlike the large numbers of genes encoding kinases, thus far, there is evidence for only a single OGT catalytic subunit despite the myriad of different O-GlcNAc-modified proteins (1). Virtually nothing is known about the regulation of OGT's subcellular localization or substrate specificity. The presence of multiple TPR repeats implies that proteins might specifically interact with the enzyme, targeting the catalytic domain to specific protein substrates within the nucleus and cytoplasm. To examine this hypothesis, we screened a rat brain library with OGT using the yeast two-hybrid approach, in order to begin to identify potential OGT-interacting proteins (OIPs). Initially, 250 positively interacting clones were identified. Here, we report the cloning and characterization of GRIF-1, a recently cloned GABA A receptor-associated protein, and its novel homolog KIAA1042/OIP106 as novel coiled-coil domain proteins that interact strongly with the TPR domain of OGT. Our findings suggest that these proteins possibly function to target OGT to RNA polymerase II and GABA A receptor complexes to mediate transcriptional and signaling events.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The OGT-Gal4 BD fusion bait plasmid pJL59-OGT was generated by inserting the rat OGT cDNA into pJL59. The negative control pJL59-C-term OGT plasmid was generated by inserting the C terminus (residues 476 -1037) of OGT. The plasmids encoding the full-length OGT, C-term OGT , and TPR domain for E. coli expression were generated by subcloning the respective cDNAs into pET32 (Novagen) as thioredoxin fusion proteins. Constructs encoding for an Nterminal hexa-His-tagged TPR and C-term OGT fusion proteins (pRSET-TPR and pRSET-C, respectively) were generated by subcloning the TPR and C-term OGT fragments into the pRSET vector (Invitrogen). The pACT2 plasmid encoding the partial rat OIP98 (AD-OIP98) clone (residues 105-878) fused to the activation domain of Gal4 was obtained from the initial yeast two-hybrid screen as a result of plasmid rescue. Full-length rat OIP98 was generated by performing 5Ј-rapid amplification of cDNA ends using the rat brain Marathon-Ready cDNA library (Clontech) and by assembling the 5Ј-rapid amplification of cDNA ends product with the preexisting insert derived from the yeast two-hybrid screen using overlapping restriction sites. OIP98 is a partial clone of GRIF-1, lacking exon 15. The missing exon was obtained by using a QuickClone rat brain cDNA library (Clontech) as template and GRIF-1 gene-specific primers flanking exon 15 in a PCR and inserted into the OIP98 cDNA using overlapping restriction sites. An E. coli expression plasmid of rat GRIF-1 (OIP98) was generated by subcloning the cDNA insert into pET32. Plasmids for in vitro transcription/translation reactions and mammalian overexpression were generated by subcloning the GRIF-1 cDNA into pCITE4c (Novagen) and pcDNA3.1 His A (Invitrogen), respectively. The insert encoding KIAA1042/OIP106 cDNA was obtained from Dr. Takahiro Nagase from the Kazusa DNA Research Institute (Chiba, Japan) in pBluescript II SK ϩ . The insert was then used as template to generate a cassette via PCR. This cassette was then subcloned into pCITE5b and pcDNA3.1 His B (Invitrogen) for use in in vitro transcription/translation reactions and mammalian expression. Finally, pCITE-p62 was constructed by subcloning the cDNA encoding rat nucleoporin p62 (48) into pCITE4c. The fidelity of all DNA constructs was verified by nucleotide sequencing.
Yeast Two-hybrid Screen and Interaction Testing-The Matchmaker rat brain library in pACT2 (prey vector with Gal4 activation domain) (Clontech) was used to isolate interacting clones, according to the manufacturer's instructions. Screening was performed on SD/ϪLeu/ϪTrp/ ϪHis/ϪAde plates in the presence of 10 mM 3-aminotriazole, in yeast strain PJ69-4A. Three rounds of screening were performed, resulting in about 250 final positive clones. 80 of the resulting positives were screened for the presence of the AD fusion inserts using pACT2-specific primers via a yeast colony PCR approach as described (34). Nine inserts greater than 2 kb were selected for DNA sequencing, and selected plasmid DNAs were recovered by performing yeast plasmid preparations as described previously (34). Finally, confirmation of the originally observed interactions was performed by retransforming yeast with pJL59-OGT bait or empty pJL59 and pACT2-prey or empty pACT2 vectors and plated on SD/ϪLeu/ϪTrp/ϪHis/ϪAde plates containing 30 mM 3-aminotriazole. The carboxyl terminus of OGT fused to the BD of Gal4 was used as the negative control bait vector. Liquid ␤-galactosidase assays were performed as described (34).
Protein Expression and Purification-The E. coli strain BL21 (DE3) codon plus RIL strain (Stratagene) was used for the overexpression of OGT and GRIF-1 (OIP98) protein constructs in either LB or double yeast tryptone-ampicillin. Unless indicated otherwise, all purified proteins were desalted in 20 mM Tris, pH 7.8, 20 -40% (v/v) glycerol (v/v), 0.02% (w/v) sodium azide and stored at Ϫ20°C. Recombinant hexa-Histagged thioredoxin-OGT, C-term OGT , and GRIF-1 (OIP98) fusion protein expression was induced by the addition of isopropyl-1-thio-␤-Dgalactopyranoside for 4 h at 37°C. Cells were harvested, and Histagged proteins were purified under denaturing conditions via nickel affinity chromatography using HiTrap chelating columns (Amersham Biosciences) according to the manufacturer's instructions. His-tagged C-term OGT (pRSET-C) was expressed and purified in a similar manner for use as an antigen to raise anti-OGT polyclonal antibodies generated in rabbits. Soluble thioredoxin-tagged TPR (pET32-TPR) was overexpressed and purified under native, nondenaturing conditions via nickel affinity chromatography.
Blot Overlay Assays-Equal amounts of E. coli expressed recombinant OGT, TPR, and C-term OGT and BSA proteins were separated on SDS-PAGE and blotted on polyvinylidene difluoride membranes. Proteins were renatured en blot in renaturation buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 100 mM potassium acetate, 1 mM dithiothreitol, 5 mM MgCl 2 , 1 mM EDTA, 0.1% (v/v) Tween 20, 0.1 mM ZnCl 2 , 5% milk (w/v), and 0.1 M Met) at 4°C. [ 35 S]Met-labeled GRIF-1 and OIP106 probes were synthesized in vitro using the TnT coupled reticulocyte lysate system (Promega) according to the manufacturer's instructions. Reactions were desalted and added to 15 ml of cold renaturation buffer and used to probe the blots containing the immobilized proteins at 4°C. Following probing, the blots were washed extensively with renaturation buffer, dried, and subjected to autoradiography by exposure to Biomax MR film (Eastman Kodak Co.) at Ϫ80°C. Typical exposure times were 8 -24 h.
Preparation and Purification of Rabbit and Chicken Polyclonal Antibodies-Rabbit and chicken antiserum to GRIF-1 and OIP106 were generated by immunizing rabbits and chickens with recombinant His-tagged GRIF-1 (OIP98) purified as described above. Gel-purified protein was used as an immunogen by Covance Research Products (Denver, PA) to produce polyclonal antisera in one rabbit designated as JH 3286. In a similar manner, chickens were immunized by Aves Laboratories (Tigard, OR) to produce chicken IgY antibodies. Affinitypurified antibodies were generated by purifying JH 3286 antiserum or chicken IgY over a GRIF-1 (OIP98) affinity column as previously described (36). KIAA1042/OIP106-specific polyclonal antibodies were generated by immunizing one rabbit designated as SAI1 with a peptide containing the first 20 amino acids of the N terminus (N20) of KIAA1042. This region is specific and unique to KIAA1042. Affinitypurified antibodies were generated by purifying SAI1 antiserum over an N20 peptide affinity column as previously described (36). The anti-OGT antibody AL28 was generated in a similar manner by using His-tagged C-term OGT protein to immunize two rabbits designated AL28 and AL29. AL28 antiserum was affinity-purified over a thioredoxin-C-term OGT (pET32-C) column as previously described (36).
Antibodies and Western Blot Analysis-AL28 was used at a final concentration of 25-50 ng/ml in 5% (w/v) milk in Tris-buffered saline containing 0.05% (v/v) Tween 20 for 16 h at 4°C. Similarly, JH 3286 was used at a final concentration of 100 ng/ml. Chicken anti-OIP106/ GRIF-1 was used at a final concentration of 0.5-1 g/ml. S-protein HRP conjugate (Novagen) was used at 1:5000 according to the manufacturer's instructions. Anti-actin (Sigma) and anti-Rb p107 (Santa Cruz Biotechnology, Inc.) antibodies were used at 1:5000. 8WG16 mouse monoclonal antibodies to hypophosphorylated RNA polymerase II (Neoclone) and anti-tubulin (Sigma) were used at 1:10,000. Mouse Omniprobe anti-Xpress antibody (D-8; Santa Cruz Biotechnology) was used at 1:2000, and anti-O-GlcNAc CTD 110.6 mouse monoclonal antibody was used at 1:2500 as described before, either in the absence or presence of 50 mM GlcNAc (51). All blots were developed with the enhanced chemiluminescence (ECL) reagent (Amersham Biosciences).
Preparation of Tissue Extracts-For preparation of whole tissue extracts, frozen tissue from male Sprague-Dawley rats were homogenized in ice-cold radioimmune precipitation assay lysis buffer (PBS, 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate) containing protease inhibitors, and soluble crude extracts were generated by spinning the homogenate at 100,000 ϫ g for 90 min at 4°C. Extracts were assayed for protein concentration via the Bio-Rad protein assay reagent, and equal amounts of each tissue extract were separated on SDS-PAGE for JH 3286 Western blot analysis. HeLa whole cell lysate was prepared in a similar manner. For analysis of human cell line extracts, Cruz-Blot B was purchased (Santa Cruz Biotechnology) and analyzed with SAI1 and ␣-tubulin antibodies.
For immunoprecipitation experiments, frozen rat brains were homogenized in Hepes-buffered saline (HBS) containing 20 mM Hepes pH 7.4, 0.3 M NaCl, 5 mM MgCl 2 , 1 mM NaF, 100 mM GlcNAc, and protease inhibitors using a Polytron homogenizer. The homogenate was then centrifuged at 100 000 ϫ g for 90 min at 4°C to generate soluble crude brain extract and stored at Ϫ80°C. Fractionation of nuclear and cytoplasmic extracts from HeLa cells was performed as described (38).
Immunoprecipitation-Crude rat brain extract was filtered through a 0.22-m syringe filter and was precleared by anti-IgY-agarose (Promega). Precleared extract was incubated with either 2 g of preimmune chicken IgY or anti-OIP106/GRIF-1 IgY to immunoprecipitate native OIP106 and GRIF-1 16 h at 4°C. OIP106/GRIF-1 immune complexes were collected by incubation with anti-IgY-agarose for 1 h at 4°C. The beads were extensively washed in cold HBS extract buffer, and bound proteins were eluted with SDS-PAGE sample buffer. Samples were boiled, separated by SDS-PAGE, and analyzed by Western blotting with either rabbit anti-OIP106/GRIF-1 antibody JH 3286 or rabbit anti-OGT antibody AL28.
Similarly, for OGT immunoprecipitations, crude rat brain extract was incubated with AL28 in HBS (0.3 M NaCl) or radioimmune precipitation assay lysis buffer for 16 h at 4°C. Immune complexes were collected by incubation with protein A-Sepharose (Amersham Biosciences), and bound proteins were eluted with SDS-PAGE sample buffer, followed by analysis by Western blotting.
HEK 293 cell immunoprecipitations were performed by lysing transfected cells in HBS (0.3 M NaCl) buffer containing 0.5% (v/v) Triton X-100 and 0.5 mM phenylmethylsulfonyl fluoride for 30 min at 4°C. Whole cell extracts were generated by centrifuging the lysates at 16,000 ϫ g at 4°C for 10 min. Precleared extracts were incubated with anti-Xpress Omniprobe D-8 antibody (Santa Cruz Biotechnology) and protein A/G plus agarose (Santa Cruz Biotechnology) to immunoprecipitate recombinant Xpress-tagged OIP106 and GRIF-1 for 3 h at 4°C. Immunoprecipitates were washed extensively in lysis buffer and eluted with SDS-PAGE sample buffer, followed by Western blotting with Omniprobe D-8, anti-OGT AL28, and CTD 110.6 antibodies. Immunoprecipitations from TnT rabbit reticulocyte lysates expressing Xpresstagged OIP106 and GRIF-1 were performed with Omniprobe D-8 antibody in a similar manner as described above.
Protein-Protein Interactions-For OIP106/GRIF-1 pull-down experiments, S-tagged OIP106 and GRIF-1 were synthesized in TnT rabbit reticulocyte lysates (Promega). Following synthesis, lysates were incubated with S-protein-agarose to purify S-tagged OIP106 and GRIF-1 for 3 h at room temperature in HBS (0.3 M NaCl) lysis buffer containing 0.5% (v/v) Triton X-100. Pull-downs were washed extensively in lysis buffer, and bound proteins were eluted with SDS-PAGE sample buffer and analyzed by silver staining.
For the p62 and OIP106-AL28 pull-down assays, S-tagged p62 and OIP106 were synthesized in TnT rabbit reticulocyte lysates. Following synthesis, proteins were incubated with preimmune or AL28 in Trisbuffered saline containing 0.5 M NaCl and 0.1% (v/v) Triton X-100 at 4°C. Immune complexes were collected by incubating the reactions with protein A-Sepharose (Amersham Biosciences). Beads containing the immune complexes were washed extensively with binding buffer, and bound proteins were eluted by boiling in SDS-PAGE sample buffer. Samples were then analyzed with S-protein HRP and AL28 Western blots.
Expression of OIP106 and GRIF-1 cDNAs-OIP106 and GRIF-1 were expressed in vitro in rabbit reticulocyte lysates using the TnT in vitro transcription/translation system (Promega) from either pcDNA 3.1 His or pCITE vectors. For HEK 293 cell transfections, 4 g each of OIP106 and GRIF-1 in their respective pcDNA 3.1 His vectors were transfected using LipofectAMINE 2000 and expressed as Xpress-tagged fusion proteins. HEK 293 cell transfections were carried out for 48 h, and harvested and transfected cells were subjected to anti-Xpress Omniprobe antibody immunoprecipitation as described under "Immunoprecipitation."

Yeast Two-hybrid Screen to Identify Proteins That Interact with O-GlcNAc
Transferase-To identify OGT-interacting proteins, ϳ 7 ϫ 10 6 clones (6 -7-fold redundancy) were screened from a rat brain two-hybrid cDNA library, and 250 positive yeast colonies were isolated. We chose a rat brain library because OGT protein levels and its enzymatic activity are high in brain (25). Eighty of these colonies were screened for the presence of cDNA inserts that were larger than 2 kb, in hopes of isolating full-length cDNAs. Nine of these 80 clones encoded for cDNAs ranging in size from 2.3 to 3.5 kb and were sequenced and identified. Four of the nine clones contained a cDNA that encoded for a 774-amino acid polypeptide, which was lacking its 5Ј-end. We performed 5Ј-rapid amplification of cDNA ends using a rat brain cDNA library to isolate the 5Ј-end and isolated a 900-bp fragment that encoded for an additional 104 amino acids. Assembly of both cDNA fragments yielded a ϳ3.4-kb cDNA fragment (GenBank TM accession number AF474163) that encoded for a 878-amino acid polypeptide, with a predicted molecular mass of 98 kDa, which we named OIP98 (for OGT-interacting protein of 98 kDa). To ensure that the original interaction observed in yeast was not a false positive, we repeated the yeast two-hybrid assay using the original isolated partial OIP98 target cDNA (AD-OIP98) fused to the activation domain of Gal4 and the OGT bait cDNA (BD-OGT) fused to the binding domain of Gal4. As seen in Fig. 1A, only the yeast colonies containing both the interacting OGT bait and the OIP98 target plasmids were able to grow on plates lacking the histidine-selective marker, by activation of the HIS3 reporter gene. Furthermore, this was confirmed by additional liquid phase assays where the OIP98-OGT interaction was quantified by the activation of the ␤-galactosidase reporter gene (Fig. 1B). We performed an additional control by testing the interaction of the putative catalytic domain of OGT (BD-C term OGT ) with OIP98. As can be seen in Fig. 1B, the C terminus of OGT did not interact with OIP98, indicating that the aminoterminal half of OGT containing the TPR domain may be the region of interaction with OIP98.
At the time of its isolation, OIP98 did not bear any resemblance to any known protein; thus, we considered it a novel protein. However, during revision of this manuscript, homology searches performed in GenBank TM revealed that OIP98 dis-played 93% identity to a rat 914-amino acid protein named GRIF-1 (GABA A receptor-associated protein) (GenBank TM accession number AJ288898) (52). GRIF-1 was found to interact with the ␤ 2 subunit of the GABA A receptor and was isolated from a yeast two-hybrid screen. Rat GRIF-1 is the rat ortholog of human ALS2CR3 (40). Interestingly, sequence comparison between OIP98 and GRIF-1 revealed that the region of difference between the two proteins was from residue 620 (in GRIF-1 and OIP98) to 688 (in GRIF-1; 653 in OIP98) (positions 644 -712 in Fig. 2A). This region corresponds to exon 15 in human ALS2CR3/GRIF-1. Thus, OIP98 is the GRIF-1 cDNA product but lacking exon 15; therefore, it is a partial clone of GRIF-1. We cloned the missing exon by PCR, using GRIF-1 gene-specific primers that flanked exon 15, from a rat brain cDNA library template (see "Experimental Procedures"). Interestingly, sequence comparison of our assembled GRIF-1/OIP98 to the published GRIF-1 sequence (52) revealed three major differences in the predicted sequence (52). These are Leu 579 (our GRIF-1) to Trp 579 (52), Thr 595 (ours) to Ser 595 (52), and Gln 596 (ours) to Glu 596 (52) (Fig. 2B). These differences were not found when we compared our cloned GRIF-1 sequence to the human GRIF-1 (ALS2CR3) (Fig. 2B). Similar results were found when exon 15 was cloned from a Matchmaker (Clontech) rat brain cDNA library (data not shown).
Further GenBank TM searches showed considerable sequence identity (ϳ40%) between GRIF-1 and another novel human gene of unknown function, KIAA1042 (41) (GenBank TM accession number AB028965). KIAA1042 encodes for a 953-amino acid polypeptide, with a predicted molecular mass of 106 kDa ( Fig. 2A) and is mapped to 3p22.1. 2 We expressed the KIAA1042 and GRIF-1 cDNAs in a rabbit reticulocyte lysate system and analyzed the [ 35 S]Met-labeled protein products on a 7.5% SDS-PAGE, followed by autoradiography. Surprisingly, as shown in Fig. 3A (left panel), both KIAA1042 and GRIF-1 migrated similarly on SDS-PAGE, at an apparent molecular mass of about 115 kDa. Whereas the migratory pattern of GRIF-1 is consistent with the reported migration (52), our observation that KIAA1042 migrated similarly was unexpected, since KIAA1042 is 40 amino acids larger than GRIF-1. In a separate reaction using pCITE vectors, translated GRIF-1 and KIAA1042 proteins were purified via S-tag affinity chromatography. As seen in Fig. 3A (middle panel), equal amounts of S-tagged GRIF-1 and KIAA1042 were purified, as detected by S-protein HRP blot. Affinity-purified anti-GRIF-1/OIP98 polyclonal antibody (JH 3286) cross-reacted well with both purified GRIF-1 and KIAA1042 (Fig. 3A, right panel). This is not surprising, since KIAA1042 shares significant sequence identity with GRIF-1 (Fig. 2). Thus, our GRIF-1 antibody JH 3286 reacts with both KIAA1042 and GRIF-1.
We performed further JH 3286 Western blot analysis on whole rat tissue and human HeLa cell lysates. Beck et al. (52) have reported that GRIF-1 is only expressed in excitable tissue such as brain, heart, and skeletal muscle, based on Western blot analysis with a GRIF-1-specific antibody. Our JH 3286 antibody, however, reacted with a ϳ115-kDa band in all rat tissues that were examined, as well as HeLa cell lysates (Fig. 3B). In addition, JH 3286 reacted with a ϳ106-kDa band in heart, brain, lung, and smooth muscle tissue (Fig. 3B, top asterisk). An additional reactivity at ϳ98 kDa was observed in lung tissue (Fig. 3B, bottom asterisk). Thus, JH 3286 reacted with GRIF-1 in excitable tissue. The additional immunoreactivity and aberrant migratory patterns of GRIF-1 are consistent with the patterns reported by Beck et al. (52). Since JH 3286 reacts with both KIAA1042 and GRIF-1, the reactivity noticed in nonexcitable tissue as well as in HeLa cell lysates is probably due to the presence of KIAA1042. Interestingly, the protein species in liver and kidney migrated slightly differently from the protein reactivities in the other tissues. This could be a result of differential 2 Available on the World Wide Web at genome.ucsc.edu/cgi-bin/ hgTracks?positionϭchr3:48370736-48542321&hgsidϭ223215. post-translational modifications that might occur in these tissues. It is curious that the anti-GRIF-1 antibody generated against residues 8 -633 of GRIF-1 by Beck et al. (52) did not react with KIAA1042 in nonexcitable tissue, since both KIAA1042 and GRIF-1 share a high degree of homology in that region ( Fig. 2A).
Currently, however, since JH 3286 recognizes both GRIF-1 and KIAA1042, we cannot distinguish between the two proteins in brain and heart tissue. In order to confirm KIAA1042 expression in HeLa cell lysates, whole cell lysates were probed with a KIAA1042 N terminus (N20; first 20 amino acids of human KIAA1042) specific antibody SAI1. As seen in Fig. 3C, SAI1 specifically reacted with a ϳ115-kDa KIAA1042 band, consistent with the JH 3286 reactivity in Fig. 3B (right panel). This reactivity was competed away in the presence of the KIAA1042 N20 peptide antigen (Fig. 3C,  right panel), thus confirming the identity of the band as KIAA1042. Further specificity of SAI1 toward KIAA1042 was examined by probing recombinant Xpress-tagged KIAA1042 and GRIF-1 with SAI1. As seen in Fig. 3D, the ␣-Xpress blot (right panel) shows that recombinant purified Xpress-tagged GRIF-1 or KIAA1042 was immunopurified by ␣-Xpress Omniprobe antibody from reticulocyte lysates expressing the cDNA for each protein. However, the SAI1 blot (Fig. 3D, left panel) clearly shows that only recombinant KIAA1042, and not GRIF-1, reacted with SAI1, indicating that SAI1 is spe-cific for KIAA1042. Since we now had an KIAA1042-specific antibody with SAI1, we wanted to confirm its expression in human tissue, since SAI1 is specific for human KIAA1042. However, since human tissue was not easy to obtain, we decided to examine the expression of KIAA1042 on a blot containing cell lysates from a diverse range of human cell lines containing both excitable (IMR-32) and nonexcitable (HL-60, HeLa, etc.) cell lines. As seen in Fig. 3E, SAI1 signal was observed in every tissue, as seen in both short and long exposures (top and middle panels, respectively). The ␣-tubulin blot (Fig. 3E, bottom panel) shows the protein loading control. Thus, KIAA1042 expression was noticed in all cell types that were examined. This is consistent with its reported mRNA expression. 3 Biochemical Confirmation of GRIF-1 and KIAA1042 Interaction with OGT-In order to confirm that the interactions observed in the original yeast two-hybrid screen were valid and not false positives, we performed a variety of in vitro and in vivo protein-protein interaction assays. To show the physiological existence of native GRIF-1-OGT and KIAA1042-OGT complexes from rat brain, we performed co-immunoprecipitation (IP) experiments. GRIF-1 and KIAA1042 were immunoprecipitated with anti-GRIF-1 antibody (raised in chicken) from rat 3 Available on the World Wide Web at www.kazusa.or.jp/huge/ gfpage/KIAA1042/.

FIG. 2.
A, ClustalW alignment of human OIP106/KIAA1042, rat GRIF-1 (QuickClone), rat GRIF-1 (52), human ALS2CR3 and rat OIP98. Exon 15 (and part of exon 16) in GRIF-1, the region missing in OIP98, is bracketed. B, differences between cloned GRIF-1 (QuickClone), published GRIF-1 (52), human ALS2CR3, and OIP98. Differences in the sequences are indicated by the arrows. These residues are conserved in our cloned GRIF-1 (and OIP98) as well as the published human ALS2CR3 but are different from the published GRIF-1 sequence (52). brain extracts. This antibody reacts with both GRIF-1 and KIAA1042, similar to JH 3286. Immunoprecipitates were separated on a 10% SDS-PAGE and analyzed with JH 3286, anti-OGT AL28 antibody, anti-actin, and anti-tubulin antibodies. As seen in Fig. 4A, the JH 3286 Western blot (top panel) shows that chicken anti-GRIF-1 antibody specifically immunoprecipitated GRIF-1/KIAA1042 from rat brain extracts. Western blot analysis by anti-OGT AL28 antibody (second panel) clearly shows the presence of OGT in the anti-GRIF-1 IP, indicating that OGT co-immunoprecipitated with GRIF-1/KIAA1042. The absence of tubulin and actin (bottom two panels, respectively), which are abundant proteins in brain, in the anti-GRIF-1 IP demonstrate that the interactions between OGT and GRIF-1/ KIAA1042 are specific and not a result of nonspecific binding.
To show that GRIF-1/KIAA1042 reciprocally co-immunoprecipitated with OGT, OGT was immunoprecipitated with AL28 from rat brain extract. IPs were performed in HBS (0.3 M NaCl) and high stringency radioimmune precipitation assay lysis H, heart; B, brain; S, spleen; L, lung; Li, liver; Sm, smooth muscle; K, kidney; Te, testis; Thy, thymus; HeLa, HeLa whole cell extracts. The top and bottom asterisks indicate the ϳ106and ϳ98-kDa proteolytic fragments, respectively. C, SAI1 antibody is specific for KIAA1042/OIP106. HeLa whole cell extract was analyzed by SAI1 antibody in the presence (right panel) or absence (left panel) of KIAA1042 N20 peptide antigen. SAI1 reactivity is competed away in the presence of N20 peptide, indicating that SAI1 specifically reacts with KIAA1042/OIP106 in HeLa cells. The numbers to the left represent molecular weight marker positions. D, SAI1 is specific for KIAA1042/ OIP106 and not GRIF-1. Purified recombinant Xpress-tagged GRIF-1 and KIAA1042/OIP106 were probed with anti-Xpress Omniprobe antibody (right panel) and SAI1 (left panel). SAI1 only reacted with KIAA1042/OIP106 and not GRIF-1, indicating that it is specific for GRIF-1. E, KIAA1042/OIP106 is expressed in all human cell lines that were examined. Cruz-Blot B (Santa Cruz Biotechnology) containing 50 g each of extracts from various human cell lines (as indicated at the top) was probed with SAI1 and ␣-tubulin antibodies. The top two panels represent short and long exposures (respectively) of the SAI1 blot, and the bottom panel shows the ␣-tubulin protein loading control blot (Con). As is seen in the SAI1 blots, KIAA1042/OIP106 reactivity was observed in all human cell line extracts that were examined. buffer (containing 0.1% SDS). Immunoprecipitates were washed and analyzed with AL28 and chicken anti-GRIF-1. As is seen in Fig. 4B, AL28 immunoprecipitated OGT in both native and high stringency conditions, as analyzed by the AL28 Western blot (top panel). Western blot analysis with chicken anti-GRIF-1 (bottom panel) clearly shows the presence of GRIF-1/KIAA1042 in OGT IPs but not in the control preimmune IP, under native and high stringency binding conditions.
Since our anti-GRIF-1 antibodies do not distinguish between GRIF-1 and KIAA1042, it was not clear which protein interacted with OGT. We addressed this issue by transiently transfecting KIAA1042 and GRIF-1 cDNAs or vector alone (mock control) separately in HEK 293 cells as Xpress-tagged fusion proteins. Xpress-tagged GRIF-1 and KIAA1042 were immunoprecipitated using anti-Xpress Omniprobe antibody, and IPs were analyzed by Omniprobe and AL28 Western blots. As seen in Fig. 4C, the Omniprobe blot (left panel) clearly shows the presence of immunoprecipitated Xpress-tagged GRIF-1 and KIAA1042 in the Omniprobe IPs. The AL28 Western blot (bottom panel) shows that endogenous OGT in HEK 293 cells specifically co-immunoprecipitated with Xpress-tagged GRIF-1 and KIAA1042 but not from mock-transfected lysates (bottom panel, right lane). Therefore, OGT interacts with both KIAA1042 and GRIF-1 individually, confirming the previous rat brain native IP results.
These data clearly indicate that KIAA1042 and GRIF-1 interact with OGT quite strongly (even in the presence of 0.1% SDS) and probably exist in a complex in vivo, thus validating the interactions observed from the yeast two-hybrid experiments. Since we identified KIAA1042 as an O-GlcNAc transferase-binding protein, we named it OIP106 (for OGT-interacting protein of 106 kDa), based on its predicted molecular mass.
The TPR Domain of OGT Interacts with OIP106 -Since the OGT contains a TPR domain at its amino terminus, which is a protein-protein interaction domain, we hypothesized that the TPR domain may be the region of OGT that interacted with GRIF-1 and OIP106. In addition, the carboxyl terminus of OGT failed to interact with GRIF-1 (OIP98) in the yeast two-hybrid system (Fig. 1B). To examine this hypothesis, OGT and its individual domains were tested for interactions with GRIF-1 and OIP106 using an in vitro blot overlay interaction assay. Recombinant OGT, TPR, and C-term OGT (C) proteins synthesized in E. coli were separated on SDS-PAGE and blotted onto polyvinylidene difluoride membranes. BSA was similarly blotted to use as a negative control. Immobilized proteins were then renatured en blot and probed with radioactively labeled GRIF-1 and OIP106. Blots were washed and then subjected to autoradiography. As can be seen in Fig. 5, A and B, the TPR domain strongly bound radiolabeled GRIF-1 and OIP106 (middle panels). Similar binding was exhibited by the full-length OGT. In contrast, no signal was observed in the C protein lanes or the BSA control lanes, indicating that the binding was specific. Therefore, this indicates that GRIF-1 and OIP106 interact with OGT via the TPR domain.
We wanted to examine whether GRIF-1 and OIP106 bound OGT stoichiometrically. S-tagged GRIF-1 and OIP106 were synthesized in rabbit reticulocyte lysates and pulled down by S-protein-agarose. Beads were washed extensively in high salt buffer and analyzed by silver staining. As is seen in Fig. 5C, S-tagged GRIF-1 and OIP106 were purified by S-protein-agarose (middle and right lanes, respectively). The endogenous OGT band, which migrates at ϳ110 kDa, is clearly seen in both GRIF-1 and OIP106 pull-downs. AL28 Western blot had identified this co-purifying band as OGT in preliminary experiments (data not shown). This band is not present in the control S-protein-agarose pull-down lane (left lane). Therefore, whereas it appears that the binding of GRIF-1 and OIP106 to OGT appears to be stoichiometric, quantification of relative amounts of the protein species here is not applicable, since the response of silver staining varies from one protein to another. The above experiments, in addition to the co-IP experiments, confirmed the initial yeast two-hybrid studies and clearly show that OIP106 and GRIF-1 are a novel family of highly homologous proteins that interact strongly with OGT via its TPR repeats.
GRIF-1 and OIP106 Are Modified by O-GlcNAc-PROSITE analysis of GRIF-1 and OIP106 sequences revealed that the carboxyl halves of both proteins contained many potential sites of O-GlcNAc modification. Since both proteins interact strongly with OGT, we examined the potential existence of O-GlcNAc on immunoprecipitated Xpress-tagged GRIF-1 and OIP106 by probing with CTD 110.6 (51), which is an anti-O-GlcNAc-specific mouse monoclonal antibody. As is seen in Fig. 4C (middle panel), when probed in the absence of 50 mM GlcNAc, both GRIF-1 and OIP106 reacted strongly with CTD 110.6. This reactivity was competed away when probed in the presence of 50 mM GlcNAc (Fig. 4C, right panel), indicating that the reactivity observed was due to the presence of the O-GlcNAc modification and not due to nonspecific binding of the antibody. This indicates that both GRIF-1 and OIP106 are modified by O-GlcNAc and are substrates for OGT.
We wanted to further examine whether other substrates of OGT also interacted with OGT in a similar manner. Nucleoporin p62 is a well studied, high affinity substrate for OGT (13). We performed in vitro binding experiments with p62 and looked for its ability to stably interact with OGT. An identical experiment was performed with OIP106 to serve as a positive control for OGT binding. OIP106 and p62 were synthesized as S-tagged proteins in reticulocyte lysates. Following synthesis, lysates were incubated with anti-OGT AL28 antibody to immunoprecipitate endogenous OGT, and IPs were assayed for the presence of either OIP106 or p62 by S-protein HRP blot. As seen in the left bottom panel of Fig. 6 (long exposure), the AL28 blot on AL28 IPs shows that the antibody immunoprecipitated OGT from lysates, as expected. No OGT was immunoprecipitated by the preimmune IgG (lane 4, bottom panel). The Sprotein HRP blot in Fig. 6 (long exposure; left top panel) shows that OIP106 clearly co-immunoprecipitated with OGT (lane 5) but not with the preimmune IgG (lane 4). In contrast, no p62 was detected in the AL28 IP (lane 2). The short exposure panels on the right show that the concentration of OIP106 used in the input for the IPs was less than that of the amount of p62 used (right top panel; compare lane 1 with lane 3), indicating that the binding observed are specific and not effects of mass action. This indicates that p62 did not stably interact with OGT in this system, but OIP106 clearly did, as consistent with previous data.
We performed a similar experiment on rat brain extracts, which would reflect an in vivo native system, and obtained the same result (data not shown). These results indicate that p62 does not interact with OGT in a stable complex, but OIP106 and GRIF-1 do, although all three proteins are substrates for OGT.
OIP106 Localizes to the Nucleus in HeLa Cells-Since the characterization of GRIF-1 and its subcellular localization have already been reported by Beck et al. (52), we decided to focus on OIP106. In order to more closely examine subcellular distribution of OIP106, we performed biochemical fractionation on HeLa cells. HeLa cells were biochemically fractionated into nuclear and cytosolic extracts, and equal amounts of each extract were analyzed by JH 3286. As seen in Fig. 7A, OIP106 was only detected in the nuclear (N) fraction (Fig. 7A, top  panel). OGT, which is a nucleocytoplasmic enzyme, is present in higher levels in the nucleus and is shown as a control in the bottom panel of Fig. 7A by AL28 Western blotting (25). Similar results were obtained using the OIP106-specific SAI1 antibody (data not shown). Thus, in HeLa cells, OIP106 is a nuclear protein.
OIP106 Co-localizes with RNA Polymerase II in Vivo-We decided to take advantage of the observation that endogenous OIP106 was adequately detected in HeLa cells by JH 3286 (Fig.  7A) and performed native OIP106 immunofluorescence with JH 3286 via laser-scanning confocal microscopy. As seen in Fig.  7A, panel a, OIP106 localized to distinct punctate regions in the nucleus. Thus, the nuclear staining of OIP106 correlated well with the biochemical fractionation in Fig. 7A. In order to further examine OIP106's nuclear punctate localization, we performed laser-scanning confocal microscopy co-staining for proteins that are known to be present in these punctate regions. Recently, von Mikecz et al. (44) showed that the IIA form of RNA polymerase II exhibited a broad staining of the nonnucleolar portion of the nucleus in Hep-2 cells when stained with the 8WG16 mouse monoclonal antibody. The 8WG16 antibody is specific for the hypophosphorylated form of polymerase II (IIA) (45), which has been previously shown to be extensively O-GlcNAc-modified (7). We felt that this was a suitable marker for the observed punctate staining and wanted to examine whether OIP106 possibly co-localized with RNA polymerase IIA in vivo. We performed co-staining of OIP106 and RNA polymerase IIA using JH 3286 (from rabbit) and 8WG16 (from mouse) on HeLa cells to determine their co-localization. As is seen in Fig. 7B, panels b and c, subsets of OIP106 colocalized with RNA polymerase IIA in the distinct dotlike regions, as evident in the merged image (panel c). Fig. 7B, panel d, is an enlarged image of the inset in panel c, clearly showing the co-localized subsets of OIP106 and RNA polymerase IIA in yellow. To further confirm this observation, we performed immunogold electron microscopy. This is shown in Fig. 7C, panels a and b. The large 18-nm colloidal gold particles represent OIP106 molecules, and the smaller 12-nm particles represent RNA polymerase IIA. As seen in the inset in Fig. 7C, panel b, subsets of OIP106 co-localize with RNA polymerase IIA, further supporting the confocal microscopy data.
OIP106 Exists in a Complex with RNA Polymerase II and OGT-Since OIP106 and RNA polymerase II co-localize (Fig. 7, B and C), we wanted to biochemically confirm whether the two proteins are present in a complex in vivo. We performed co-IP experiments with the anti-GRIF-1 IgY (which reacts well with OIP106) and immunoprecipitated OIP106 from HeLa nuclear extracts. IPs were washed extensively and analyzed by JH 3286, AL28, 8WG16, and anti-Rb p107 Western blotting. As is seen in Fig. 7D, a subset of RNA polymerase II (second panel, lane 3) co-immunoprecipitated with OIP106 (top panel, lane 3) but not by the preimmune antibody (lane 2). The presence of OGT in the OIP106 IP, but not in the preimmune, serves as a positive control for the IP (Fig. 7D, third panel, lanes 3 and 2,  respectively). Furthermore, Rb p107, which is a nonrelated abundant nuclear protein, did not co-IP with OIP106 (bottom panel), demonstrating that the polymerase II-OIP106 interactions are specific. Similar results were obtained with the OIP106-specific SAI1 antibody (data not shown). These data support the hypothesis that OIP106 exists in a complex with a subset of RNA polymerase II and OGT, providing evidence for the possible targeting of OGT by OIP106 to transcriptional complexes. DISCUSSION OGT Interacts with Many Proteins-A myriad of different nucleocytoplasmic proteins are modified by O-GlcNAc, yet there appears to be only a single OGT catalytic subunit (25). The OGT cDNA was cloned from rat (25) and human (26) 4 years ago. Subsequent knockout of the OGT gene in mouse resulted in embryonic lethality, at the single cell level (27). Mapping of the OGT gene in humans and mice revealed it to lie on the X chromosome as a single copy gene (27,28). This clearly demonstrated that the OGT enzyme was absolutely essential for life. However, virtually nothing is known about the mechanisms regulating OGT protein specificity. The N-terminal half of the OGT contains a TPR domain, which in rat OGT, is composed of 11.5 TPR repeats. Recently, contrary to earlier reports (26), it has been shown that the human OGT has the same number of TPR repeats as the rat enzyme (28). TPR domains have been shown to be responsible for intra-and intermolecular protein-protein interactions in a variety of proteins, spanning a variety of cellular functions (29,30). Thus, in order to identify putative binding proteins that might potentially regulate the OGT's specificity or its subcellular localization, we performed an unbiased yeast two-hybrid screen of a rat brain library using the entire OGT protein as a bait. Our final FIG. 7. Subcellular localization of OIP106 in HeLa cells. A, HeLa cells were biochemically fractionated into nuclear (N) and cytoplasmic (C) extracts and immunoblotted with JH 3286 or anti-OGT AL28 (control). Native OIP106 immunofluorescence in HeLa cells. B, OIP106 co-localizes with RNA polymerase IIA in vivo via confocal microscopy. HeLa cells were stained with affinity-purified JH 3286 alone (a) or co-stained with anti-hypophosphorylated polymerase II (IIA) 8WG16 antibodies (b). Specimens were visualized with a Bio-Rad MRC-1042 laser-scanning confocal microscope. The compound image (c) shows their co-localization in the nucleus. The inset (d) shows the co-localization of the two proteins in yellow. C, OIP106 and RNA polymerase IIA co-localize in vivo via EM. a, HeLa cells were fixed and stained with affinity-purified JH 3286 and 8WG16 antibodies. Staining was observed using a JEM 1010 electron microscope. The inset (b) shows the co-localization of both polymerase IIA and OIP106 via their differently sized colloidal gold particles. D, RNA polymerase II coimmunoprecipitates with OIP106. OIP106 immune complexes were immunoprecipitated from HeLa nuclear extracts with chicken anti-OIP106 IgY or preimmune IgY. Immunoprecipitates were separated on 7.5% SDS-PAGE and analyzed with JH 3286 (top panel), 8WG16 (antipolymerase II; second panel), AL28 (anti-OGT; third panel), and anti-Rb p107 (control) (bottom panel). round of screening resulted in about 250 clones, which, with a ϳ6 -7-fold redundancy, we estimated to represent at least 30 unique clones. Whereas we selected for clones that were larger than 2 kb in hopes that we would find complete cDNAs, our PCR screening revealed that a large number of the clones were less than 2 kb, indicating that these either encoded for smaller proteins or were partial clones. Thus, it is reasonable to estimate, even conservatively, that the number of potential OGTinteracting proteins is quite substantial. Thus, for such a single, unique enzyme to glycosylate a myriad of proteins, we hypothesize that it is regulated by a large number of TPRinteracting proteins.
GRIF-1, OIP106, and OGT: Implications for Targeting of OGT-GRIF-1 and OIP106 interact with OGT quite strongly and stoichiometrically. The interactions are resistant to high salt, nonionic, and ionic detergents such as SDS. As is seen with the blot overlay assays, these interactions occur in the TPR domain. Interestingly, in HEK 293 cell transfection experiments, binding to OGT by a FLAG-tagged OIP106 construct was not competed away when co-transfected with increasing amounts of Xpress-tagged GRIF-1 (data not shown), indicating that possibly GRIF-1 and OIP106 may bind to different TPR repeats of OGT. Thus, mutually exclusive GRIF-1-OGT and OIP106-OGT complexes could exist in the cell. This could help explain how each protein differentially might affect OGT's localization. GRIF-1 was recently isolated as a novel GABA A receptor-interacting protein that is expressed only in excitable tissue (52). The function of GRIF-1 is unknown. However, by virtue of its association with OGT, we propose that perhaps it functions to target OGT to GABA A receptor complexes. This would implicate O-GlcNAc, OGT, and GRIF-1 as being involved in GABA signaling. GRIF-1 would function as an adaptor/ scaffolding protein, bridging OGT to GABA A receptor in this model. Signaling through such scaffolding/anchoring protein networks has been well documented, especially for protein kinases such as protein kinase A and its various A kinase-anchoring proteins and in N-methyl-D-aspartate receptor signaling via PDZ domain proteins (53). Interestingly, GRIF-1 contains several PXXP motifs in its carboxyl terminus (Fig. 2). PXXP motifs are known to bind to Src homology 3 domains (53). The function of Src homology 3 domains in proteins that contain it is to form functional oligomeric complexes at defined subcellular sites, usually in concert with other modular domains. A canonical example of such proteins is PSD-95, which contains both an Src homology 3 and a PDZ domain and has been shown to mediate N-methyl-D-aspartate receptor signaling via these domains (53). Thus, GRIF-1 may interact with Src homology 3 domain proteins via its PXXP motifs, recruiting OGT to these complexes, in the context of GABA A receptor signaling. Furthermore, regulation of GABA A receptor via phosphorylation has been well documented, and it is the large intracellular cytoplasmic loops of the various receptor subunits (␣, ␤, ␥) that have been shown to be substrates of various kinases (protein kinase A, protein kinase C, Ca 2ϩ /calmodulindependent protein kinase II) (42). GRIF-1 interacts with the intracellular loop of the ␤ 2 receptor subunit, which is phosphorylated at Ser 410 (42). Thus, OGT could be recruited to this domain for its potential O-GlcNAc modification and for potential O-GlcNAc modification of associated scaffolding proteins. O-GlcNAc and O-phosphate modifications often occur on the same or adjacent Ser/Thr residues (15), so potential O-GlcNAc modification of GABA A receptor (complexes) may be a way of regulating its function. GRIF-1 would mediate this regulation by targeting OGT to these complexes.
OIP106-OGT complexes occur in punctate regions within the nucleus. These nuclear punctae have been shown to contain proteins involved in splicing and transcription, including RNA polymerase II and CBP/p300 histone acetyltransferases (44). Our laboratory and others have shown that most transcription factors that have been examined as well as various proteins involved in splicing are heavily modified by O-GlcNAc (46). The CTD of RNA polymerase II has been shown to be extensively modified by O-GlcNAc (7), and it is the IIA form of polymerase II (found in preinitiation complexes) that has been shown to be glycosylated. Very recently, our laboratory has shown that basal transcription factors, including the TFIID complex are also extensively modified by O-GlcNAc (50). Thus, in our model of OIP106 function with respect to OGT, the enzyme would be targeted to sites of transcription by OIP106, to glycosylate proteins in the preinitiation complex such as the IIA form of RNA polymerase II. Our co-localization of OIP106 with RNA polymerase IIA, using both confocal and electron microscopy and its interaction with polymerase II via co-IP experiments, strongly supports to this model. Thus, based on our data, we hypothesize that OIP106 and OGT may represent novel components of preinitiation complexes. Our model in which preinitiation complexes are glycosylated has been in existence for a number of years (1,7,46), and the discovery of OIP106's associations with OGT and RNA polymerase IIA further supports this notion.
OIP106 and GRIF-1-Both GRIF-1 and OIP106 contain very highly identical coiled-coil domains. Coiled-coil domains are known to be involved in homo-or heterodimerization. GRIF-1 has been shown to be able to homodimerize via its coiled-coil domain (52). We therefore reasoned that GRIF-1 and OIP106 may be able to interact with each other, presumably via their respective coiled-coil domains. In preliminary in vitro pulldown experiments performed with rabbit reticulocyte lysate synthesized proteins, GRIF-1 strongly bound to OIP106 (data not shown). Whether this interaction occurs physiologically is unknown. However, this does raise the potential of cross-talk between GRIF-1 and OIP106 to occur, possibly coupling GABA receptor signaling to transcriptional events. GRIF-1 has been reported to be found in the nuclear fractions in excitable tissue (52), and this notion would be consistent with that finding. Thus, the identification of these novel proteins and their species-specific orthologs comprise a novel family of OGT-interacting proteins that potentially regulate OGT by influencing its targeting and subcellular localization.