Roles of the Tetratricopeptide Repeat Domain in O -GlcNAc Transferase Targeting and Protein Substrate Specificity*

The abundant and dynamic post-translational modification of nuclear and cytosolic proteins by (cid:1) - O -linked N -acetylglucosamine ( O -GlcNAc) is catalyzed by O -Glc-NAc-transferase (OGT). Recently, we reported the iden-tification of a novel family of OGT-interacting proteins (OIPs) that interact strongly with the tetratricopeptide repeat (TPR) domain of OGT (Iyer, S. P., Akimoto, Y., and Hart, G. W. (2003) J. Biol. Chem. 278, 5399–5409). Members of this family are modified by O -GlcNAc and are excellent substrates of OGT. Here, we further investigated the role of the TPR domain in the O -GlcNAcyla-tion of OIP106, one of the members of this OIP family. Using N-terminal deletions, we first identified the region of OIP106 that binds OGT, termed the OGT-inter-acting domain (OID). Deletion analysis indicated that TPRs 2–6 of OGT interact with the OID of OIP106. The apparent K m of OGT for the OID of OIP106 is 3.35 (cid:2) M . Unlike small peptide substrates, glycosylation of the OID was dependent upon its interaction with the first 6 TPRs of OGT.

Post-translational modification of proteins by enzymatic addition of single ␤-O-linked N-acetylglucosamine (O-GlcNAc) 1 moieties to the hydroxyl groups of Ser/Thr residues has been shown to be a major form of protein regulation (2)(3)(4). This modification is known to occur on key proteins that modulate cell function, such as RNA polymerase II (5,33), transcription factors (6 -8), cytoskeletal proteins (9,10), heat shock proteins (11,34), oncoproteins (12,35), and various kinases and phosphatases (13,14). Attachment of O-GlcNAc (O-GlcNAcylation) is a dynamic (added/removed in minutes) and ubiquitous modification and occurs in all higher eukaryotes, including plants and animals. Because many of the sites that are modified by O-GlcNAc are either the same or adjacent to sites of Ser/Thr phosphorylation (15), O-GlcNAcylation of proteins is thought to play a regulatory role, analogous to protein regulation by phosphorylation (3).
Enzymes that catalyze addition and removal of O-GlcNAc on and off proteins are analogous to those that catalyze phosphorylation, i.e. kinases and phosphatases. O-GlcNAc-transferase (OGT; uridine diphospho-N-acetylglucosamine:polypeptide ␤-N-acetylglucosaminyltransferase, EC 2.4.1) is an enzyme that transfers GlcNAc to proteins from a UDP-GlcNAc sugar donor. Conversely, an O-GlcNAc-specific ␤-N-acetylglucosaminidase (N-acetyl-␤-D-glucosaminidase) removes the sugar from O-GlcNAcylated proteins. Both enzymes have been purified, characterized, and cloned (16 -21). Both enzymes are highly conserved throughout evolution, with species orthologs present from Caenorhabditis elegans to man (18,20,21).
The 110-kDa OGT enzyme is a highly unique and ubiquitous glycosyltransferase that is encoded by a single gene and is present in both the nucleus and cytoplasm of cells (20,21). The OGT gene resides on the X chromosome near the centromere (Xq13 in man), and its targeted deletion results in embryonic lethality in mice (22), indicating that it is essential for life. Recently, the previously reported partial human OGT clone described by Lubas et al. (21) has been characterized as an alternatively spliced isoform of OGT that targets specifically to the mitochondria (40). Thus, two distinct isoforms of OGT exist: a nucleocytoplasmic form of OGT and its alternatively spliced mitochondrial isoform. Interestingly, the mitochondrial OGT isoform is not active in the mitochondria toward mitochondrial protein substrates, suggesting that it may have another function (40). Sequence analysis of the 110-kDa enzyme showed it to contain two distinct modular halves. In its N-terminal half, OGT contains a TPR protein-protein interaction domain, whereas the unique C-terminal half of the enzyme is known to contain its catalytic domain (20,21). The nucleocytoplasmic OGT enzyme contains 11.5 TPRs, whereas the mitochondrial OGT isoform contains 9 TPRs, preceded by a mitochondrial targeting sequence (20,21,40); and these domains have been shown to mediate protein-protein interactions in a variety of proteins (23,36,37).
Various functions have been attributed to the TPR domain of OGT. The native OGT holoenzyme exists as a trimer, and the TPR domain has been shown to be required for its trimerization (24). The TPR domain has also been shown to mediate substrate specificity in a variety of peptide substrates (24). Furthermore, it was recently shown that the TPR domain is also responsible for targeting OGT to mSin3A transcriptional repressor complexes (25). Although much is known about the peptide substrate specificity of OGT, it is unclear as to how the substrate specificity of its physiological protein substrates is regulated by the TPR domain.
We have recently identified a novel family of coiled-coil domain proteins that interact with OGT (1). GRIF-1 and OIP106, the two archetypal members of this family, interact with the TPR domain strongly. GRIF-1 is a GABA A receptor-associated protein (26) and is thought to play a role in targeting OGT to GABA A receptor complexes to mediate GABA signaling cascades. OIP106 was shown to associate in a complex with RNA polymerase II, suggesting a role in targeting OGT to RNA polymerase II complexes for transcriptional machinery O-Glc-NAcylation (1). Interestingly, both proteins are modified by O-GlcNAc and are substrates for OGT (1).
Here, we investigated the O-GlcNAcylation of OIP106 to gain further insights into the enzymatic mechanisms regulating the specificity of OGT. Specifically, we wanted to study how the TPR domain functions in relation to the substrate specificity of OGT for a bona fide physiological protein substrate such as OIP106. We localized the domains of OGT and OIP106 interactions and found that the OGT-interacting domain (OID) of OIP106 is a high affinity substrate for OGT. Furthermore, deletion of specific TPRs of OGT resulted in loss of binding and, consequently, loss of O-GlcNAcylation of the OID. A free TPR domain effectively inhibited glycosylation of the OID, but not glycosylation of small peptides, in a dose-dependent manner, providing kinetic evidence that the TPRs function as protein substrate "docking" sites. In addition, both recombinant nucleoporin p62 (a high affinity substrate) and OID proteins effectively competed with each other for glycosylation by OGT. We propose a model for the possible mechanism of OGT glycosylation of its various protein substrates.

Plasmid Constructs
The baculovirus expression constructs encoding rat OGT and its TPR truncations have been described previously (24). Constructs encoding an N-terminal His 6 -tagged TPR (pRSET-TPR) were generated by subcloning the TPR fragments into the pRSET vector (Invitrogen). The pCITE-OIP106 construct used for in vitro transcription/translation reactions has been described previously (1). N-terminal OIP106 deletion constructs were generated by PCR of the appropriate regions and subcloning into pCITE4c. The OID of OIP106 was generated by PCR of the region of OIP106 encoding residues 639 -890 and cloned into pET32 for overexpression in Escherichia coli as an S/His-tagged fusion protein.

Protein Expression and Purification
Unless indicated otherwise, the E. coli BL21(DE3) Codon Plus RIL strain (Stratagene) was used for the overexpression of TPR and OIP106 protein constructs in either LB or digested yeast tryptone/ampicillin medium. Unless indicated otherwise, all purified proteins were desalted in 20 mM Tris (pH 7.8), 20 -40% (v/v) glycerol, and 0.02% (w/v) sodium azide and stored in Ϫ20°C. His-tagged TPR protein (pRSET-TPR) was overexpressed and purified under native nondenaturing conditions via nickel affinity chromatography. Baculovirus-produced recombinant full-length OGT and ⌬2.5OGT and ⌬5.5OGT TPR deletion constructs were overexpressed and purified as described (24).
S-tagged OID was overexpressed and purified via S-protein affinity chromatography under mildly denaturing conditions in the presence of 1.3 M urea. Insoluble S-OID protein was extracted by incubating the insoluble cell pellet in 6 M urea and TBST (20 mM Tris (pH 8), 150 mM NaCl, and 0.1% (v/v) Triton X-100) for 1 h at 4°C. The extract was spun at 39,000 ϫ g for 20 min at 4°C, and the supernatant containing S-OID was diluted down to 1.3 M urea by the addition of TBST. The resulting 1.3 M urea supernatant was loaded onto S-protein-agarose (Novagen) equilibrated in 1.3 M urea and TBST, and S-OID was bound by incubation for 30 min, washed extensively, and used for subsequent assays. Alternatively, the 1.3 M urea and TBST supernatant was passed over Talon immobilized metal affinity chromatography resin (Clontech), and S-OID was purified according to the manufacturer's instructions.

Antibodies and Western Blot Analysis
Anti-OGT antibodies AL25 (20) and AL28 (1) were 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. S-protein-horseradish peroxidase conjugate (Novagen) was used at a 1:5000 dilution according to the manufacturer's instructions. Anti-O-GlcNAc monoclonal antibody CTD 110.6 was used at a 1:2500 dilution as described previously (39). All blots were developed with the enhanced chemiluminescence reagent (ECL, Amersham Biosciences).

Protein-Protein Interaction Assays
Wild-type OIP106 and its N-terminal truncation mutants (for OID localization studies) were synthesized with either [ 35 S]Met or unlabeled amino acids in TNT in vitro transcription and translation rabbit reticulocyte lysates (Promega) with an S tag fusion, which is encoded by their respective pCITE vectors. Following synthesis, the reactions were diluted in Hepes-buffered saline containing 0.3 M NaCl, and S-tagged OIP106 proteins were bound to S-protein-agarose for 3 h at 4°C. The beads were washed extensively with their respective binding buffer and eluted by boiling in SDS-PAGE sample buffer. Samples were then analyzed by Western blotting with S-protein-horseradish peroxidase and AL28. The OGT-OID interaction assays were performed as follows. Baculovirus-produced recombinant full-length OGT, ⌬2.5OGT, and ⌬5.5OGT and E. coli His-TPR proteins were incubated with S-OID in Tris-buffered saline containing 20 mM Tris (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, and 0.1% (w/v) Triton X-100 for 3.5 h at 4°C. Proteins were then pulled down with S-protein-agarose, washed extensively with binding buffer, and eluted by boiling in SDS-PAGE sample buffer. Samples were analyzed by Western blotting with S-proteinhorseradish peroxidase and AL25.

OGT Activity Assays
OGT activity assays were performed essentially as described (13) with minor modifications. The S-OID protein used as a substrate for OGT was either in a soluble form (immobilized metal affinity chromatography-purified) or bound to S-protein-agarose. OGT protein glycosylation assays were performed by incubation of baculovirus-produced recombinant full-length OGT, ⌬2.5OGT, or ⌬5.5OGT as the enzyme source. 1 g of the indicated OGT protein was incubated with the indicated amounts of S-OID protein substrate, 1-1.5 Ci of UDP- Post-assay Protocol 1-The reactions were stopped by the addition of SDS-PAGE sample buffer and boiled for 5 min. Samples were then separated by SDS-PAGE, followed by transfer to polyvinylidene fluoride membrane. The membrane was treated with EA-Wax enhanced autoradiography reagent (EABiotech Ltd., Scotland, UK) following the manufacturer's instructions and exposed to Biomax MS film at Ϫ80°C. Typical exposure times were from 16 to 30 h.
Post-assay Protocol 2-The reactions were stopped by the addition of 0.5 ml of TBST containing 0.5 M NaCl. S-protein-agarose beads were then added to the reactions to bind the S-OID substrate. The beads were washed three times with 1 ml of TBST containing 0.5 M NaCl. 10 ml of scintillation fluid was added directly to the columns containing the beads, shaken briefly, and counted.
OGT peptide glycosylation assays were performed as described (27), except that the reaction buffer contained 50 mM Tris (pH 7.4), 12.5 mM MgCl 2 , and 1 mM dithiothreitol in a final reaction volume of 40 l. The CKII 12-mer peptide was used as a substrate at a final concentration of 1 mM. All assays were performed in triplicate, and each experiment was performed at least two times.

Determination of the K m of OGT for the OID of OIP106
For the K m determination experiments, assays were performed as described above with varying concentrations (0.26 -2.6 M) of soluble S-OID protein substrate. 2 g of OGT enzyme was used per assay. UDP-GlcNAc was kept constant at a final concentration of 1 mM by isotopic dilution from 34 Ci/mmol to 0.025 Ci/nmol with the addition of unlabeled UDP-GlcNAc (Calbiochem). To remove the resulting UDP product, which is a potent inhibitor of OGT, 1 unit of calf intestinal phosphatase (New England Biolabs Inc.) was added to the reaction. Assays were processed as per post-assay protocol 2. Units are defined as picomoles of GlcNAc transferred per g of enzyme/min. All assays were performed under initial rate conditions in triplicate, and the K m was determined based on averaged results of two independent experiments.

TPR Competition Experiments
OGT protein and peptide assays in the presence of competing TPR protein were performed as described above with the following parameters. The S-OID protein (1.68 M) and CKII peptide (55 M) substrates were at a final concentration of one-half of their respective K m values. His-TPR protein was added to the reaction at the indicated amounts. Total protein amounts were kept constant with the addition of BSA. Assays were processed as per post-assay protocol 2. Assays were performed in triplicate, and each experiment was performed twice.

OID-p62 Glycosylation Competition Experiments
OID-p62 glycosylation competition experiments were performed as follows. For p62 competition experiments, S-tagged OID was kept at a constant concentration of 0.5 M, and the concentration of GST-p62 was varied from 0.5 to 2.5 M. Similarly, for OID competition experiments, GST-p62 was kept at a constant concentration of 0.5 M, and the concentration of the OID was varied from 0.5 to 2.5 M. 425 ng of OGT was used per assay for each condition. In the 0 M competitor control assays, 2.5 M BSA was substituted for the competing protein substrate and served as a negative control. Assays were performed at room temperature for 90 min as described in above. Following the reactions, assays were split into two; GST-p62 was purified by adding glutathione-Sepharose to one half, and the other half was subjected to S-proteinagarose chromatography to purify S-tagged OID. Proteins bound to either affinity resin were eluted by boiling the resins in SDS sample buffer (following extensive washing after adsorption). Glycosylation of p62 and the OID was analyzed by immunoblotting of the eluted proteins on polyvinylidene fluoride, followed by analysis by the O-GlcNAc-specific monoclonal antibody CTD 110.6.

Localization of Domains of OIP106 and OGT Interactions-
Based on its strong interaction with OGT, we hypothesized that OIP106 could function to influence the activity of OGT as a possible cofactor (1). Therefore, we first performed in vitro OGT assays to determine whether recombinant OIP106 has any effect on the activity of OGT. Unfortunately, when overexpressed in E. coli as a His-tagged fusion protein, the majority of recombinant full-length OIP106 was found to be insoluble, so we localized the region of OIP106 that was responsible for its interaction with OGT with the hope that this region would be a smaller portion of the molecule and therefore be more amenable to overexpression in an E. coli based system. To localize this region, we constructed N-terminal truncations of OIP106. N-terminal OIP106 truncations were then synthesized in lysates as S-tagged fusion proteins, pulled down with S-proteinagarose, and assayed for their ability to bind endogenous OGT as detected by anti-OGT antibody (AL28) Western blotting (1). Fig. 1 (upper panel) shows the various N-terminal OIP106 truncations synthesized as [ 35 S]Met-labeled S-tagged fusion proteins. Also shown is a schematic of the OIP106 truncations lacking different domains. As shown in Fig. 1 (lower panel), all of the truncations except ⌬859 (residues 860 -953) were able to bind and pull down OGT. This indicated that the potential OGT-binding domain localized to within residues 639 -859 in the C terminus of OIP106.
To confirm the above localization studies, we investigated whether the 220-amino acid fragment of OIP106 from residues 639 to 859 is sufficient to bind OGT. We overexpressed this region in E. coli as an S-tagged fusion protein and purified it under mildly denaturing conditions via S-protein affinity chromatography. We call this region OID for OGT-interacting domain and performed in vitro binding assays with purified recombinant OGT and the TPR domain. S-protein-agarose alone or S-OID was incubated with baculovirus-expressed purified recombinant OGT or the TPR domain, and S-protein pull-down assays were performed. As shown in Fig. 2A (upper panel), the OID efficiently bound full-length OGT (lane 3) and the TPR domain of OGT (lane 9) as detected by anti-OGT antibody (AL25) Western blotting (20). S-protein-agarose alone did not bind any OGT or TPR domain (lanes 2 and 8, respectively). The S-protein-horseradish peroxidase blot (lower panel) shows that equal amounts of S-OID were bound in each reaction. Thus, the OID of OIP106 is sufficient to bind to OGT efficiently. These data confirm the localization experiments performed above in lysates.
To further localize the specific TPRs that interact with the OID, we used the ⌬2.5OGT and ⌬5.5OGT deletion mutants, which lack the first 2.5 and 5.5 TPRs, respectively (33), in binding assays with S-tagged OID. As shown in Fig. 2A (lanes  4 -6), the ⌬5.5OGT mutant failed to interact with the OID. We performed binding assays with the ⌬2.5OGT mutant, and the results are shown in Fig. 2B. Binding assays were performed under non-saturating conditions. Under physiological condi- tions (150 mM NaCl), ⌬2.5OGT forms a trimer, whereas under high salt conditions (1 M NaCl), it forms a dimer (24). ⌬2.5OGT-OID binding assays were performed under these two conditions. As shown in Fig. 2B (lanes 1-4), there were no significant differences in the binding of full-length OGT (upper panel) to the OID (lower panel) between the 0.15 and 1 M NaCl binding conditions (compare lanes 3 and 4 in both panels). Similarly, as shown in lanes 5-8, no significant differences were noticed in the binding of the ⌬2.5OGT mutant (upper panel) to the OID (lower panel) at 0.15 and 1 M NaCl (compare lanes 7 and 8 of both panels). Although the levels of ⌬2.5OGT appeared be diminished at 1 M NaCl (lane 8), the levels of S-tagged OID protein that was pulled down were also equally diminished (asterisk). We performed densitometry on these bands and found that, after normalization, the intensities of both the ⌬2.5OGT and OID bands in lanes 7 and 8 were similar. Therefore, under physiological and high salt conditions, ⌬2.5OGT interacted with the OID as efficiently as the wild-type fulllength enzyme, indicating that the OID (OIP106) can interact with either a dimeric or trimeric form of OGT. Thus, this suggests that TPRs 2-6 of OGT are the repeats that interact with the OID of OIP106.
The OID of OIP106 Is a High Affinity Substrate for OGT-We had previously demonstrated that OIP106 is modified by O-GlcNAc in vivo (1). Furthermore, PROSITE analysis of the OIP106 sequence revealed that the OID of OIP106 likely contains several putative sites of glycosylation. To examine this, increasing amounts of OID were incubated with OGT in in vitro protein glycosylation assays in the presence of UDP-[ 3 H]Glc-NAc. Reactions were stopped by the addition of SDS-PAGE sample buffer to the assays, and samples were separated by SDS-PAGE and subjected to autoradiography. The OID was efficiently glycosylated by recombinant OGT, as shown in Fig.  3B. Fig. 3A represents the Coomassie Blue-stained gel of the increasing amounts of S-OID used in the glycosylation assay. The additional lower molecular mass bands observed in Fig. 3A are proteolytic fragments of the S-OID fusion protein. OGT displayed standard Michaelis-Menten kinetics when the OID was used as a substrate, as shown in Fig. 3C (left panel). The right panel shows the corresponding Lineweaver-Burk plot. The K m of OGT for the OID was determined to be 3.35 Ϯ 1.45 M, which is very similar to the K m of nucleoporin p62 (1.2 M) (13), one of the best known high affinity protein substrates for OGT. On average, under optimal conditions, 2 mol of GlcNAc were transferred per mol of OID by OGT in these assays.
The TPR Domain of OGT Functions as a Docking Site for Protein Substrates-It has been shown previously that the TPR domain plays a role in trimerization of OGT (24). Other studies suggest that the TPR domain may be a site of substrate recognition (13,24). However, it is still unclear as to how the TPR domain may function in terms of protein substrate recognition. We hypothesized that the TPR domain would form stable complexes with its substrates in order for the OGT catalytic subunits to be able to specifically glycosylate such a large diversity of proteins. In this model, the individual TPRs confer substrate specificity by bridging the substrate to the enzyme. Thus, based on this model, we hypothesized that because the TPR deletion mutants are unable to efficiently bind the OID of OIP106, they would also be deficient in glycosylating the OID as well. We assayed the ability of the ⌬2.5OGT and ⌬5.5OGT mutants to glycosylate the OID protein substrate relative to their ability to glycosylate a 12-amino acid substrate peptide based upon an O-GlcNAc site on the protein kinase CKII (24). These TPR deletion mutants are able to efficiently glycosylate the CKII peptide (24). As shown in Fig. 4 (D and E), both the ⌬2.5OGT and ⌬5.5OGT mutants efficiently glycosylated the CKII peptide compared with full-length OGT. This indicates that the purified OGT mutant proteins are both catalytically active. However, both the ⌬2.5OGT and ⌬5.5OGT mutants showed a severely diminished capacity to glycosylate the OID protein substrate compared with the full-length enzyme. As shown in Fig. 4 (A and B, respectively), ⌬2.5OGT was only one-seventh as active as full-length OGT toward the OID, whereas ⌬5.5OGT completely failed to glycosylate the OID. It has been previously shown that removal of the first 5.5 TPRs does not affect the ability of OGT to glycosylate peptide substrates (24), yet removal of TPRs 1-6 severely affected the capacity of OGT to interact with and glycosylate the OID protein substrate. So, although removal of the first 2 TPRs did not affect binding of OGT to the OID, they are clearly important for efficient glycosylation of the OID of OIP106. Thus, TPRs 1-6 of OGT are necessary to efficiently interact with and glycosylate OIP106 (OID).
Although these data are indicative of the role of the TPR domain in recognition of protein substrates, we also performed TPR competition experiments to further document the role of the TPRs of OGT in targeting substrates. The fact that the first 6 TPRs are dispensable in the activity of OGT for peptide substrates led us to hypothesize that the TPR domain should compete for the OID as a protein substrate with OGT, but should not compete for the CKII peptide substrate. So OGT-OID protein or OGT-CKII peptide glycosylation assays were performed in the presence of increasing molar amounts of the TPR domain. The molar ratio of TPR to OGT was kept constant in both the protein and peptide assays. To keep the quantity of total protein constant, BSA was added to the reactions. As shown in Fig. 4F, the TPR domain did not block CKII peptide glycosylation, as expected. However, increasing amounts of TPR dramatically reduced OID glycosylation by OGT (Fig. 4C). A 12-fold molar excess of TPR completely inhibited glycosylation of the OID. This indicates that the TPR domain competes with OGT for docking sites on the OID protein. These data suggest that one of the functions of the TPR domain is indeed to serve as a binding/ recognition site for protein substrates. The observation that the TPR domain did not affect glycosylation of the CKII peptide supports the observation that the catalytic site of OGT is in the C-terminal domain, as suggested earlier by photolabeling analysis (13,28).
Nucleoporin p62 and the OID Compete with Each Other for Glycosylation by OGT-We have shown that the OID of OIP106 is a high affinity substrate for OGT, similar to another high affinity substrate, nucleoporin p62. However, p62 does not interact with OGT in a stable manner as OIP106 does (1). To examine possible differences in the specificity of these two high affinity substrates vis à vis OGT, we performed glycosylation assays on each substrate at increasing concentrations of the other substrate to study possible competition effects. OGT assays were performed using a constant amount of recombinant FIG. 3. The OID of OIP106 is a high affinity substrate for OGT. A, recombinant S-tagged OID was overexpressed and purified via immobilized metal affinity chromatography. Purified OID at increasing concentrations was used as a substrate for OGT in in vitro OGT protein assays. B, OGT glycosylates the OID in a dose-dependent manner. Shown is a 27-h autoradiograph of the OID glycosylation assay. C, the OID is a high affinity substrate for OGT and has a K m value similar to those of other high affinity protein substrates. OGT protein assays were performed as described under "Experimental Procedures" using increasing amounts of OID (0.48 -2.  4. Trimeric ⌬2.5OGT is partially active toward the OID protein substrate, but is fully active toward the CKII peptide substrate. A and D, OGT assays were performed as described under "Experimental Procedures" using recombinant OID protein and the CKII peptide as substrates, respectively. The source of enzyme was purified recombinant ⌬2.5OGT. Monomeric ⌬5.5OGT was inactive toward the OID protein substrate, but was fully active toward the CKII peptide substrate. B and E, OGT assays were performed as described under "Experimental Procedures" using recombinant OID protein and the CKII peptide as substrates, respectively. The source of enzyme was purified recombinant ⌬5.5OGT. The TPR domain of OGT competed for binding of the OID with OGT, but not for binding to the CKII peptide.  (39). As shown in Fig. 5 (upper  panel), glycosylation of the OID was effectively competed away by increasing amounts of GST-p62. An increasing presence of GST-p62 resulted in increased glycosylation of GST-p62 by OGT and a corresponding decrease in glycosylation of the OID. The control GST protein alone was not glycosylated by OGT (data not shown). In addition, 2.5 M BSA was used in the 0 M competitor control assays (lanes 1 and 5) to keep the protein quantities constant and serves as a negative control; therefore, the observed competition effects were not due to a protein mass action effect. Similarly, as shown in Fig. 5 (lower panel), glycosylation of GST-p62 was effectively competed away by increasing amounts of OID. Increasing concentrations of OID resulted in increased glycosylation of the OID by OGT and, correspondingly, a decrease in glycosylation of GST-p62 in an OID dose-dependent manner. In both cases, a 5-fold molar excess of competitor to substrate was sufficient to significantly compete away glycosylation. It should be noted that, although the 5-fold molar excess of OID completely inhibited p62 glycosylation (lower panel, lane 8), the same molar excess (5-fold) of GST-p62 did not completely inhibit OID glycosylation, as some residual glycosylation was still observed (upper panel, lane 4). It is possible that higher molar amounts of GST-p62 may be required to completely inhibit OID glycosylation. Although the level of inhibition of p62 glycosylation by the OID appears to be similar at all concentrations tested, competition of p62 glycosylation was OID concentration-dependent in other similar assays that were performed (data not shown); and therefore, the lack of OID concentration-dependent inhibition is an artifact of this particular experiment, as shown in Fig. 5. Thus, both high affinity substrates compete with each other for glycosylation by OGT. DISCUSSION Previous work on OGT enzymology has clearly shown a requirement for its TPR domain in modulating its oligomerization as well as specificity for its peptide substrates in vitro (24). Lubas and Hanover (13) have shown that deletion of TPRs affects the activity of OGT for a nucleoporin p62 protein substrate. Our studies contribute new knowledge about OGT in four major aspects. First, Lubas and Hanover (13) performed their studies with a truncated form of recombinant OGT lacking the first 2.5 TPRs. This group had previously inadvertently cloned a partial clone of human OGT that contained only 9 TPRs (21). Subsequently, Nolte and Muller (29) have shown that the human OGT gene does in fact contain the missing 2.5 TPRs. More recently, it was shown that the truncated form of OGT containing only 9 TPRs contains a mitochondrial targeting sequence in its N terminus and targets to the mitochondria (40). Furthermore, this form of OGT, termed mitochondrial OGT by Love et al. (40), is not active toward mitochondrial protein substrates, suggesting a different function for this form of the enzyme. Therefore, enzymatic studies performed with this truncated form of OGT are not conclusive because the wild-type nucleocytoplasmic form of OGT (containing its entire complement of TPRs) was not assayed and compared with the mutants used in these experiments. In our study, we performed experiments with the correct wild-type nucleocytoplasmic isoform of OGT and its truncation mutants. Furthermore, we characterized the glycosylation of an OGT substrate that is novel, i.e. the OID of OIP106. The second major finding involves the TPR competition data (Fig. 4, C and F). The fact that a free TPR domain inhibits OID glycosylation by OGT and competes for its binding, but not for the CKII peptide substrate, clearly indicates one of its roles as a protein substrate docking/ binding domain. These data indicate that a free TPR does not interfere with the catalytic activity of the OGT holoenzyme in a mass action effect based on its inability to compete away CKII peptide glycosylation. Our finding that the TPR domain did not compete for the CKII peptide substrate also provides evidence for the notion that peptide substrates weakly interact (K m for the CKII peptide ϭ 103 M) (24) with the catalytic site at the C terminus of OGT. These data also further support the model that OGT has two distinct modular halves, with the C-terminal half of the enzyme containing the putative catalytic domain and the N-terminal TPR half regulating its activity by docking to specific proteins.
The third contribution is the finding that OGT formed stable complexes with the OID (Fig. 2) in a TPR-dependent manner and that this TPR dependence was mirrored by its glycosylation of the OID. This brings up the issue of the multimerization of the enzyme, as it has been previously shown that the TPR domain modulates the ability of OGT to associate with itself (24). Our data show that multimerization of the enzyme is not sufficient for its activity for the OID protein substrate, as the ⌬2.5OGT mutant (which is a dimer at 1 M NaCl) bound to the OID as efficiently as wild-type full-length OGT (Fig. 2B), and therefore, the OID can interact either with a dimeric or trimeric form of OGT. The ⌬5.5OGT mutant is a monomer (24), and thus, the monomeric form of OGT neither interacts with nor glycosylates the OID protein substrate. However, ⌬2.5OGT is severely diminished in its capability to glycosylate the OID, even though it is fully active toward the CKII peptide substrate (Fig. 4, A and D). It should be noted that ⌬2.5OGT is a trimer under assay conditions in which no salt is present (as even low amounts of NaCl potently inhibit enzyme activity) (27). Thus, for a unique protein substrate such as OIP106 (OID), a stable association and at least dimerization of the enzyme are prerequisites for activity because loss of another 3 TPRs results in loss of binding as well as complete loss of activity (⌬5.5OGT-OID binding and glycosylation assays) ( Figs. 2A and Fig. 4, B  and E, respectively). This may be a way by which the TPRs (specifically the first 2.5 TPRs) govern the specificity of OGT for a stably interacting protein substrate such as OIP106 (and possibly GRIF-1), possibly by regulating the multimerization status of the enzyme. However, it is currently unknown if the multimerization of OGT changes in response to cellular stimuli or events and remains to be studied. It can be speculated that perhaps levels of UDP-GlcNAc or other factors in the cell may modulate the multimerization status of the enzyme because it is known that OGT is highly responsive to even slight changes in cellular UDP-GlcNAc concentrations (41).
Interestingly, we found that the OID of OIP106 stayed tightly associated with the enzyme, even after it was glycosylated. S-protein pull-down experiments performed on OID glycosylation reactions prior to and after glycosylation showed no difference in the amount of OGT that was retained by S-tagged OID (data not shown). Similarly, OGT remained bound to the OID at saturating concentrations of UDP-GlcNAc (as high as 3.43 mM) (data not shown). Thus, OGT remains tightly associated with the OID, even after it is glycosylated. We have shown previously that nucleoporin p62, which has a similar K m for OGT, does not form such stable complexes with OGT, whereas OIP106 forms stable complexes with OGT in vivo and in vitro (1). As the fourth major contribution of this study, we sought to explore the fundamental differences between the two substrates and their respective behavior toward OGT in a variety of experiments. We hypothesized that perhaps nucleoporin p62 (a "non-interacting substrate") would be unable to compete away OID glycosylation by OGT, whereas the OID of OIP106 would be able to compete away p62 glycosylation by OGT by virtue of its being an "interacting substrate." However, unexpectedly, both substrates competed effectively with each other for glycosylation by OGT (Fig. 5). We then analyzed enzyme-substrate reactions (OGT-OID and OGT-p62) via Superdex 200 gel filtration chromatography prior to and after glycosylation. We expected to find possible changes in the structures of the enzyme-substrate complexes, as would have been indicated by differences in migration patterns, and thus clues to the differences between the two substrates. However, we could not find any significant discernible differences in these complexes (data not shown). It is possible that the differences between the two substrates and how OGT interacts with them lie in the individual sets of TPRs that interact with them. The TPRs that interact (transiently) with p62 may specify transient substrate interactions (non-interacting substrates), whereas the TPRs that interact with OIP106 may dictate more stable associations (interacting substrates). Discerning the individual sets of TPRs that interact with either substrate (via limited proteolysis or cross-linking experiments) may help explain these fundamental differences.
Given the above data, we suggest a model for targeting of OGT by binding to the OID of OIP106. It can be envisioned that, once glycosylated, OIP106 can then target its associated OGT (stoichiometrically) to distinct subcellular locations in the cell. These distinct locations could be the speckle or dot-like regions that OIP106 localizes to in the nucleus (1). We have shown that OIP106 exists in a complex with RNA polymerase II and OGT (1). Thus, TPRs 1-6 would mediate this targeting of OGT to transcriptional complexes. A similar targeting mechanism can be suggested for GRIF-1 in the context of the possible targeting of OGT to GABA A receptor complexes.
Individual TPRs that have specific functions have been well documented in the literature, especially in the case of Ssn6, where specific sets (tandem and non-tandem) of its TPRs mediate specific protein-protein interactions with different binding partners (23). Similar interactions have also been observed within the yeast cell cycle control proteins Cdc16, Cdc23, and Cdc27. These proteins contain TPR domains and interact with each other via their TPR domains, and each specific interaction is mediated by different TPR combinations (30). The crystal structures of the TPR domains of protein phosphatase-5 (31) and Pex5p (32) have been solved. Each repeat of the 3 TPRs of protein phosphatase-5 has been shown to form pairs of antiparallel ␣-helices (31). A 12-TPR protein modeled after the protein phosphatase-5 TPR crystal structure indicates that TPRs arranged in tandem, such as the 11.5 TPRs of OGT, would be organized into a regular right-handed superhelix (31). This type of arrangement would facilitate multiple interactions with multiple proteins using specific combinations of TPR motifs within the superhelix.
Recently, Yang et al. (25) showed that OGT is present in a transcriptional repressor complex with the corepressor mSin3A. Interestingly, the first 6 TPRs of OGT were implicated in the interaction of OGT with this complex. Because OGT-OIP106 interactions also involve the first 6 TPRs, pinpointing the exact sets of TPRs that mediate the interactions between the two proteins should help in delineating the specificity between OIP106-polymerase II-OGT and OGT-mSin3A complexes.