Activated Inositol 1,4,5-Trisphosphate Receptors Are Modified by Homogeneous Lys-48- and Lys-63-linked Ubiquitin Chains, but Only Lys-48-linked Chains Are Required for Degradation*

Inositol 1,4,5-trisphosphate (IP3) receptors (IP3Rs) are large, ubiquitously expressed, endoplasmic reticulum membrane proteins that form tetrameric IP3 and Ca2+-gated Ca2+ channels. Endogenous IP3Rs provide very appealing tools for studying the ubiquitin-proteasome pathway in intact mammalian cells because, upon activation, they are rapidly ubiquitinated and degraded. Using mass spectrometry, we previously examined the ubiquitination of IP3R1 in αT3-1 pituitary gonadotrophs and found that IP3R1 ubiquitination is highly complex, with receptors being modified at multiple sites by monoubiquitin and polyubiquitin chains formed through both Lys-48 and Lys-63 linkages (Sliter, D. A., Kubota, K., Kirkpatrick, D. S., Alzayady, K. J., Gygi, S. P., and Wojcikiewicz, R. J. H. (2008) J. Biol. Chem. 283, 35319–35328). Here, we have extended these studies to determine whether IP3R2 and IP3R3 are similarly modified and if ubiquitination is cell type-dependent. Using mass spectrometry and linkage-specific ubiquitin antibodies, we found that all IP3R types are subject to ubiquitination at approximately the same locations and that, independent of cell type, IP3Rs are modified by monoubiquitin and Lys-48- and Lys-63-linked ubiquitin chains, although in differing proportions. Remarkably, the attached Lys-48- and Lys-63-linked ubiquitin chains are homogeneous and are segregated to separate IP3R subunits, and Lys-48-linked ubiquitin chains, but not Lys-63-linked chains, are required for IP3R degradation. Together, these data provide unique insight into the complexities of ubiquitination of an endogenous ubiquitin-proteasome pathway substrate in unperturbed mammalian cells. Importantly, although Lys-48-linked ubiquitin chains appear to trigger proteasomal degradation, the presence of Lys-63-linked ubiquitin chains suggests that ubiquitination of IP3Rs may have physiological consequences beyond signaling for degradation.

The ubiquitin-proteasome pathway (UPP) 2 is critical to cell homeostasis because it mediates the degradation of key pro-teins involved in fundamental cellular processes and allows for endoplasmic reticulum (ER)-associated degradation, a facet of the UPP that disposes of aberrant or unrequired proteins from the ER lumen and membrane (1,2). Proteins targeted for degradation by the UPP are first ubiquitinated and then shuttled to the 26 S proteasome, a cytosolic protein complex that recognizes, unfolds, and degrades ubiquitinated substrates (3). Ubiquitination, the covalent attachment of the C-terminal glycine of ubiquitin to the ⑀-amino group of a lysine residue in a target protein, is typically mediated by the E1 ubiquitin-activating enzyme, the E2 ubiquitin-conjugating enzyme, and the E3 ubiquitin ligase (4). Repetition of this process can result in multiple monoubiquitination through attachment of additional ubiquitin moieties to different lysines within the target protein, or in the formation of ubiquitin chains through attachment of ubiquitin to any of seven lysines within ubiquitin itself (Lys-6, Lys-11, Lys-27, Lys-29, Lys-33, Lys-48, and Lys-63) or to the N-terminal methionine (5). The fate of a ubiquitinated protein is thought to be determined by the type of ubiquitin linkage associated with it. For example, it is generally accepted that proteins tagged with Lys-48-linked chains are targeted to the proteasome for degradation, whereas Lys-63-linked chains mediate a variety of events (e.g. the DNA damage response and protein trafficking) (6 -8). However, this dogma is being challenged, for example, with new data indicating that ubiquitin chains utilizing lysines other than Lys-48 can contribute to proteasomal degradation (9 -12). Ubiquitin can be removed from proteins through the action of deubiquitinating enzymes (DUBs), which can serve to reverse the effects of ubiquitination (13,14).
In recent years, we have examined inositol 1,4,5-trisphosphate (IP 3 ) receptor (IP 3 R) processing by the UPP in intact cells (15). IP 3 Rs are large (ϳ260 kDa), ubiquitously expressed, ER membrane proteins that form tetrameric IP 3 -and Ca 2ϩregulated channels that govern Ca 2ϩ release from the ER and are pivotal in mammalian cell signaling (16,17). There are three IP 3 R types (termed IP 3 R1, IP 3 R2, and IP 3 R3) that form both homo-and heterotetramers and that have slightly different properties and markedly different tissue distributions. Remarkably, activation of certain cell-surface receptors that persistently elevate IP 3 concentration causes a decrease in cellular IP 3 R levels (15). This process, termed "IP 3 R downregulation," is mediated by the UPP (18 -22), occurs for all IP 3 R types in a variety of cells and tissues (18,20,21), requires the proteins UBE2G2 (Mus musculus Ubc7) and p97 and the SPFH1/2 complex (23)(24)(25), and may serve to protect against the deleterious effects of prolonged Ca 2ϩ release (15).
The value of examining ubiquitination in the context of intact cells, which contain a full complement of ubiquitinating enzymes and DUBs, rather than in vitro has been illustrated by some recent studies. For example, it was reported that in vitro ubiquitination of cyclin B 1 by the E3 anaphase-promoting complex and the E2 UbcH10 resulted in the formation of ubiquitin chains linked though Lys-48, Lys-63, and Lys-11 (9). However, utilization of UbcH10 plus UBE2S, which is the more physiological E2 module for the anaphase-promoting complex, in vitro and in vivo facilitated the specific catalysis of Lys-11-linked ubiquitin chains (11,26,27). Likewise, in vitro construction of ubiquitin chains using various E2-E3 pairs revealed the formation of non-degradable forked ubiquitin chains (28), whereas forked chains were no longer formed when the proteasomal subunit S5a was included (29). Finally, in vitro, epidermal growth factor receptors were modified by just monoubiquitin (30), whereas in intact cells, they were tagged with both monoubiquitin and Lys-63-linked ubiquitin chains (31). In each case, achieving more physiological conditions resulted in apparently more physiological results. IP 3 Rs provide very appealing tools for studying the UPP under physiological conditions because endogenous receptors can be rapidly ubiquitinated through stimulation of endogenous signaling pathways in intact cells. Thus, we have employed a mass spectrometric technique called AQUA (for absolute quantification) (32,33) and recently developed linkage-specific ubiquitin antibodies (34) to thoroughly assess the ubiquitination of IP 3 R1-3 in several cell lines and have determined the roles of the attached ubiquitin chains in IP 3 R degradation.
Immunoprecipitation for Mass Spectrometric Analysis-␣T3-1, Rat1, AR42J, and SHSY5Y cells were grown to confluence in 15-cm diameter dishes, and 24 h prior to harvesting, Rat1 cells were cultured in serum-free medium. After stimulation, cells were harvested by the addition of ice-cold lysis buffer (3-6 ml/dish; 50 mM Tris, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 10 M pepstatin, and 0.2 M soybean trypsin inhibitor, pH 8.0). Lysates were then incubated with 2.5 mM N-ethylmaleimide for 1 min to inhibit DUBs, quenched by the addition of 5 mM dithiothreitol, incubated at 4°C for 30 min with occasional mixing, and cleared by centrifugation at 16,000 ϫ g for 10 min at 4°C. For immunoprecipitation, clarified lysate was incubated overnight at 4°C with protein A-Sepharose CL-4B beads and rabbit polyclonal antibodies. Immune complexes were collected by centrifugation for 1-3 min at 3000 ϫ g, washed three times with ice-cold lysis buffer, resuspended in gel loading buffer, incubated at 100°C for 3 min, subjected to SDS-PAGE, and either Coomassie Blue-stained (rocked in 10% acetic acid, 40% ethanol, and 2.5 g/liter Coomassie Brilliant Blue R for 1 h) and destained (rinsed with 10% acetic acid and 40% ethanol for 30 min) or transferred to nitrocellulose and immunoblotted and analyzed as described previously (22).
Ubiquitin Site Identification and Ubiquitin AQUA Analysis-Regions corresponding to ubiquitinated IP 3 Rs were excised from Coomassie Blue-stained gels and processed for mass spectral analysis as described previously (33). Trypsin digestion of ubiquitinated proteins generates peptides that have a covalently attached diglycine motif (resulting from digestion of ubiquitin between Arg-74 and Gly-75) at the modified lysine and a missed lysine cleavage at that lysine (32,35). For IP 3 R ubiquitination site identification, tandem mass spectra were searched using SeaQuest with a variable mass addition of 114.0429 Da in lysines to match the diglycine signature, and matches corresponding to IP 3 R ubiquitination sites were validated based on the mass accuracy of the precursor ion and manual inspection of the tandem MS spectra. For AQUA analysis, the ion-extracted chromatogram was drawn with 10-ppm mass accuracy, and cell-derived ubiquitin peptides were quantified from the ratio of the area under the curve for these peptides versus that for ubiquitin-derived AQUA peptides (9,32). Throughout AQUA analysis, conditions were optimized to avoid differences in reference peptide recovery as described (9,32).
Immunoblotting and Immunoprecipitation using Anti-Lys-63 and Anti-Lys-48 Antibodies-Ubiquitin chains (purchased from BIOMOL) and cell lysates were mixed with gel loading buffer and incubated at 100°C for 3 min prior to SDS-PAGE and immunoblotting. Antibody dilutions were adjusted to produce equal signals from the chain standards to allow for subsequent semiquantitative immunoblotting (ϳ1:40, ϳ1: 750, and ϳ1:750 for anti-Lys-48, anti-Lys-63, and anti-ubiquitin (FK2) antibodies, respectively). IP 3 Rs immunopurified from stimulated ␣T3-1, AR42J, and Rat1 cells were incubated with gel loading buffer at 37°C for 30 min prior to SDS-PAGE and immunoblotting.
For immunodepletion, IP 3 R1 was immunopurified from stimulated ␣T3-1 cell lysates with anti-IP 3 R1 antibody. IP 3 R1 was eluted from anti-IP 3 R1 antibody by incubation at 37°C for 90 min in 1% Zwittergent plus the peptide (1 ng/l) against which the antibody was raised (20) and 2 mM dithiothreitol, and the eluted IP 3 R subunits were incubated overnight at 4°C with anti-Lys-63 or anti-Lys-48 antibody plus 2 mg of protein A beads. After centrifugation, portions of the supernatants were incubated with gel loading buffer at 37°C for 30 min and subjected to SDS-PAGE and immunoblotting.
Analysis of IP 3 R Levels in Cell Lysates-␣T3-1, Rat1, and AR42J cells were grown to confluence in BD Falcon 6-well dishes. After stimulation, cells were harvested in 200 l of ice-cold lysis buffer, and cell lysates were mixed with gel loading buffer, boiled at 100°C for 3 min, and subjected to SDS-PAGE and immunoblotting (22). IP 3 R immunoreactivity was quantitated using a GeneGnome imager (Syngene).
Transfection of ␣T3-1 Cells by Electroporation-␣T3-1 cells were transfected using the NEON transfection system (Invitrogen). Briefly, ␣T3-1 cells were trypsinized, washed with culture medium and then phosphate-buffered saline, and resuspended in Buffer R (Invitrogen) at 1.5 ϫ 10 7 cells/ml. Cells were then mixed with cDNA (15 g/1.5 ϫ 10 6 cells), and 100-l aliquots were electroporated (one pulse, 20 ms, 1.5 kV) and seeded into a well of a BD Falcon 12-well dish. 24 h later, cells were treated with 10 nM GnRH, and cell lysates were prepared and subjected to SDS-PAGE and immunoblotting.

All IP 3 R Types Are Ubiquitinated in the Regulatory Domain-
In our previous mass spectrometric analysis of IP 3 R ubiquitination, we examined IP 3 R1 isolated from GnRH-stimulated ␣T3-1 cells and found that at least 11 of the 167 lysines in IP 3 R1 are ubiquitinated (33). The bulk of IP 3 Rs reside in the cytosol and can be divided into three domains: the N-terminal ligand-binding domain, the central regulatory domain, and the C-terminal channel domain that contains six transmembrane helices and the channel pore (amino acids 1-576, 577-2276, and 2277-2749, respectively, for mouse IP 3 R1) (16). All of the ubiquitination sites previously identified (33) are found in the regulatory domain and are near modulatory sites (e.g. for phosphorylation, ATP binding, etc.) or protease-sensitive sites, suggesting that they are all located in surface-exposed loops (36,37).
In the absence of stimulus, little or no ubiquitin immunoreactivity was associated with IP 3 Rs from ␣T3-1, Rat1, AR42J, and SHSY5Y cells (Fig. 1A, lanes 1, 3, 5, and 7). However, upon treatment with the appropriate stimulus, IP 3 Rs became ubiquitinated, as indicated by the presence of a smear of ubiq-uitin immunoreactivity beginning at ϳ270 kDa and extending upward (Fig. 1A, lanes 2, 4, 6, and 8). To identify sites of ubiquitination and for AQUA analysis, regions corresponding to ubiquitinated IP 3 Rs, excised from analogous Coomassie Bluestained gels, were subjected to in-gel trypsin digestion and tandem MS (32).
Remarkably, every ubiquitination site identified, regardless of receptor type, is located within the IP 3 R regulatory domain. More specifically, alignment suggests that, despite sequence differences between the receptor types, ubiquitination sites appear to be clustered in the vicinity of putative surface-exposed loops, i.e. near regulatory or protease-sensitive sites (Fig. 1C). For example, Lys-909 of IP 3 R2 from AR42J cells aligns near Lys-916 of IP 3 R1 from ␣T3-1, AR42J, and Rat1 cells, and these lysines are located near a protease cleavage site (Arg-922 in IP 3 R1 and between Ser-916 and Ser-934 in IP 3 R3) (37,38) and near PKA phosphorylation sites (Ser-916 and Ser-934 in IP 3 R3 and Ser-937 in IP 3 R2) (38, 39) (Fig. 1C). Likewise, Lys-1771 of IP 3 R1 from ␣T3-1 and Rat1 cells aligns with Lys-1722 of IP 3 R2 from AR42J cells and is found in a glycine-rich region that, in IP 3 R1, contains an ATP-binding site (positions 1773-1775) and is thought to form a flexible connector between the cytosolic and membrane portions of IP 3 Rs (16). Finally, a cluster of ubiquitination sites is found close to the first transmembrane domain (Fig. 1C): Lys-1884, Lys-1899, Lys-1901, and Lys-1924 of IP 3 R1 and Lys-1836 and Lys-1864 of IP 3 R2 flank an IP 3 R1 caspase-3 cleavage site (Asp-1891), an IP 3 R3 PKA phosphorylation site (Ser-1834), and an IP 3 R1 protease cleavage site (Arg-1931) (37). The one ubiquitination site identified in IP 3 R3, Lys-2100, is also located in this region.
In summary, all ubiquitination sites identified localize to probable surface-exposed loops in the IP 3 R regulatory domain, and for at least IP 3 R1, selection of sites seems to be context-independent because homologous aligning lysines were frequently selected in different cell types (e.g. Lys-1847 in SHSY5Y cells and Lys-1901 in ␣T3-1 cells). Similarly, com-paring IP 3 R1 and IP 3 R2, lysines selected for ubiquitination were frequently aligned, despite sequence variation around these lysines.   7 and 8) were incubated without or with 1 mM carbachol for 20 min. Cell lysates were prepared, and IP 3 Rs were immunopurified (IP) with the antibodies indicated. Samples were electrophoresed by 5% SDS-PAGE and either Coomassie Bluestained (not shown) or transferred to nitrocellulose and probed with anti-IP 3 R1, anti-IP 3 R2, anti-IP 3 R3 (TL3), and anti-ubiquitin (FK2) antibodies. The bracket locates ubiquitinated IP 3 Rs, and arrows indicate IP 3 R1, IP 3 R2, and IP 3 R3. Numerical mass spectrometric data are presented as percent of total ubiquitin. B, listed are the ubiquitination sites identified from ␣T3-1, Rat1, AR42J, and SHSY5Y cells together with the peptides that contain the modified lysines. The periods indicate trypsin cleavage sites, and the carets indicate ubiquitination sites. C, multiple sequence alignments for human, mouse, and rat IP 3 R1 and rat IP 3 R2 and IP 3 R3 were performed by ClustalW. Selected portions are shown with ubiquitination sites boxed. Known sites of PKA phosphorylation for IP 3 R1, IP 3 R2, and IP 3 R3 are circled. Trypsin and caspase-3 cleavage sites are indicated by arrowheads, and a glycine-rich region and an IP 3 R1 ATP-binding site are also shown. Asterisks indicate completely conserved residues; colons indicate highly conserved residues; and periods indicate moderately conserved residues. Hs, Homo sapiens; Mm, Mus musculus; Rn, Rattus norvegicus. ␣T3-1, Rat1, AR42J, and SHSY5Y cells revealed that, in all cell types, the bulk of ubiquitin associated with IP 3 Rs (ϳ70% of total ubiquitin) was unmodified, i.e. either monoubiquitin or end cap ubiquitin (the last ubiquitin in a chain), and that the remaining ϳ30% was modified, i.e. ubiquitin that forms chain linkages (Fig. 1A). This indicates that, at a minimum, ϳ40% of IP 3 R-conjugated ubiquitin is monoubiquitin because if chains are assumed to consist of just two ubiquitin moieties, then 60% of total ubiquitin would be involved in chains. If, as is probable, chains are longer than two ubiquitin moieties, then Ͼ40% will be monoubiquitin (33). Consistent with our previous work (33), the vast majority of the chain linkages were via Lys-48 and Lys-63, and only minimal amounts of the other linkages were found (Fig. 1A). Although the ratio of unmodified to modified ubiquitin attached to IP 3 Rs was approximately constant in each cell type, the ratio of Lys-63 to Lys-48 linkages varied considerably: 6.6:1, 5.1:1, 3.0:1, and 1.8:1 in Rat1, AR42J, SHSY5Y, and ␣T3-1 cells, respectively (Fig. 1A). Notably, the total femtomoles of ubiquitin that accumulated on IP 3 Rs correlated with the intensity of the ubiquitin immunoreactivity observed by immunoblotting (Fig. 1A).
We next sought to determine the kinetics of ubiquitin chain formation using recently described linkage-specific antibodies (34). As indicated in Fig. 2A, these antibodies showed the appropriate specificity toward Lys-48-and Lys-63-linked chains (lanes 3-6), whereas anti-ubiquitin antibody (FK2) recognized both chain types (lanes 1 and 2). It was also immediately apparent in all cell types examined that Lys-48 linkages were much more abundant than Lys-63 linkages (Fig.  2B, compare lanes 5-8 with lanes 9 -12). Fig. 2C shows time courses of IP 3 R ubiquitination in ␣T3-1, Rat1, and AR42J cells measured with FK2 and anti-Lys-48 and anti-Lys-63 antibodies. Importantly, for ␣T3-1 cells, the changes in Lys-48 and Lys-63 linkage immunoreactivity (Fig.  2C, lanes 1-5) paralleled the changes in Lys-48 and Lys-63 linkage abundance measured by mass spectrometry (33), most notably with Lys-63 linkage immunoreactivity increasing more rapidly than Lys-48 linkage immunoreactivity. Clearly then, these antibodies provide an effective way to assess Lys-48 and Lys-63 linkage addition to IP 3 Rs and have an advantage over mass spectrometry in that they allow for the ubiquitination status of a large number of samples to be rapidly assessed by immunoblotting. Application of these antibodies to IP 3 Rs immunoprecipitated from Rat1 and AR42J cells (Fig. 2C, lanes 6 -15) showed that Lys-63 linkages accumulated rapidly and, consistent with mass spectrometry, were much more abundant than Lys-48 linkages. Overall, these data show that, in all cell types examined, activated IP 3 Rs are modified with both Lys-48-and Lys-63-linked ubiquitin chains and that, at least in ␣T3-1 cells, Lys-63 linkages accumulate most rapidly and that, in AR42J and Rat1 cells, the vast majority of ubiquitin chains are coupled by Lys-63 linkages.

IP 3 Rs Are Tagged with Homogeneous Lys-48-and Lys-63linked Ubiquitin
Chains-Remarkably, it was consistently the case that for immunopurified ubiquitinated IP 3 Rs, anti-Lys-48 immunoreactivity (ϳ320 -500 kDa) migrated at a higher molecular mass than anti-Lys-63 immunoreactivity (ϳ270 -330 kDa) (Fig. 2C). This observation was most readily detected in ␣T3-1 cells, where Lys-48 linkages are most abundant, and is highly significant, as it indicates (i) that Lys-48 and Lys-63 linkages are segregated into different chains and (ii) that Lys-48 and Lys-63 linkages are not on the same IP 3 R subunits (if the converse of either of these points was true, then Lys-48 and Lys-63 immunoreactivity would co-migrate). To directly address these assertions, we used anti-Lys-63 and anti-Lys-48 antibodies to immunodeplete purified ubiquitinated IP 3 R1 subunits (Fig. 2D). In the absence of immunodepleting antibody, Lys-48 and Lys-63 immunoreactivity associated with purified IP 3 R1 remained in solution (lane 1), whereas inclusion of either anti-Lys-63 or anti-Lys-48 antibody effectively removed all the Lys-63 and Lys-48 linkage immunoreactivity, respectively (lanes 2 and 3). However, anti-Lys-63 antibody did not deplete Lys-48 linkage immunoreactivity, nor did anti-Lys-48 antibody deplete Lys-63 linkage immunoreactivity (lanes 2 and 3), confirming that homogeneous Lys-48-and Lys-63-linked chains are segregated onto separate IP 3 R subunits.
IP 3 R Degradation Occurs More Rapidly in ␣T3-1 Cells than in Rat1 and AR42J Cells-In view of the differences in the Lys-63/Lys-48 linkage ratio between ␣T3-1, AR42J, and Rat1 cells and the canonical role of Lys-48-linked chains in triggering proteasomal degradation, we wondered whether the rate of IP 3 R degradation varied between the different cell types. To test this, ␣T3-1, Rat1, and AR42J cells were stimulated, and cell lysates were assessed for IP 3 R content (Fig. 3, A--C); the immunoreactivity of bands corresponding to unmodified IP 3 Rs was used to quantitate IP 3 R levels because only a small proportion of receptors (Ͻ10%) are ubiquitinated at any one time in maximally stimulated cells (22), and this small proportion is not readily detectable in immunoblots of cell lysates. IP 3 R degradation occurred most rapidly in ␣T3-1 cells, with IP 3 R1 levels being maximally depleted to 21 Ϯ 5% of basal levels after 20 min (Fig. 3D). In contrast, IP 3 R depletion in AR42J and Rat1 cells was less rapid, and after 20 min, IP 3 R levels were 51 Ϯ 11 and 68 Ϯ 8% of basal levels, respectively (Fig. 3D). Thus, the IP 3 R1 degradation rate does correlate with the abundance of Lys-48-linked chains attached to IP 3 Rs, and the lower the Lys-63/Lys-48 linkage ratio (Figs. 1A and 2C), the faster the degradation rate.
Proteasomal Inhibition Causes More Lys-48-linked Ubiquitin than Lys-63-linked Ubiquitin to Accumulate on IP 3 R1-We also wondered whether proteasomal inhibition might provide insight into the relative importance of Lys-48and Lys-63-linked chains in triggering IP 3 R degradation. In ␣T3-1 cells, IP 3 Rs were degraded very rapidly, with levels depleted to 42 Ϯ 6% of basal levels after 7 min (Fig. 3, A and D); correspondingly, ubiquitination was maximal at 7 min (Fig.  2C). We have shown previously that the proteasomal inhibitor MG132 acts very rapidly to inhibit the proteasome and to block IP 3 R1 degradation in ␣T3-1 cells (22). Thus, ␣T3-1 cells were exposed to GnRH for 7 min either without or with MG132, and Lys-48 and Lys-63 linkage levels were assessed (Fig. 4A). Remarkably, proteasomal inhibition caused a greater increase in accumulation of Lys-48-linked ubiquitin on IP 3 Rs than Lys-63-linked ubiquitin (80 versus 8%, respectively) (Fig. 4B), suggesting that Lys-48-conjugated IP 3 Rs are degraded by the proteasome but that Lys-63-conjugated IP 3 Rs are not.
Lys-48-but Not Lys-63-linked Ubiquitin Chains Are Required for IP 3 R1 Degradation-To directly assess the role of Lys-48-and Lys-63-linked ubiquitin in IP 3 R degradation, we generated FLAG-tagged ubiquitin mutants containing lysineto-arginine mutations to prevent the formation of either Lys-48-or Lys-63-linked ubiquitin chains (FLAG-UbK48R or FLAG-UbK63R, respectively) and overexpressed these or wild-type FLAG-tagged ubiquitin (FLAG-UbWT) in ␣T3-1 cells (Fig. 5). Using standard lipid-based or Ca 2ϩ phosphate transfection methods, it has been impossible to obtain high transfection efficiency in ␣T3-1 cells (22), but using electroporation, significant overexpression was achieved, as indicated by the appearance of FLAG immunoreactivity (Fig. 5A,  lanes 2-4) and an ϳ2.4-fold increase in total ubiquitin immunoreactivity when FLAG-UbWT, FLAG-UbK48R, and FLAG-UbK63R were expressed (lanes 5-8). In FLAG-UbWT-and FLAG-UbK63R-transfected cells, IP 3 R1 degradation proceeded with similar kinetics, and levels declined to 48 Ϯ 7 and 40 Ϯ 10% of basal levels, respectively, after 1 h (Fig. 5, B and  C). In contrast, IP 3 R1 degradation was completely inhibited when FLAG-UbK48R was overexpressed (Fig. 5, B and C). Thus, only Lys-48-linked ubiquitin chains, but not Lys-63linked ubiquitin chains, can trigger the degradation of IP 3 Rs. Furthermore, these data rule out a role for the monoubiquitin attached to IP 3 Rs in IP 3 R degradation because both of the ubiquitin mutants allowed for monoubiquitination.

DISCUSSION
Using mass spectrometry and linkage-specific antibodies to characterize IP 3 R ubiquitination, we have found (i) that all IP 3 R types are subject to ubiquitination at approximately the same locations independent of cellular context; (ii) that independent of cell type, IP 3 Rs are modified mainly by monoubiquitin, with the remainder attached as Lys-48-and Lys-63linked ubiquitin chains; (iii) that the attached ubiquitin chains are homogeneous; and (iv) that the Lys-48-linked ubiquitin chains, but not the Lys-63-linked chains, are required for IP 3 R degradation. For all IP 3 R types, ubiquitination appears to be confined to the regulatory domain (amino acids 577-2276 for mouse IP 3 R1). The regulatory domain is the large central region of IP 3 Rs that contains myriad binding sites for modifiers, including a Ca 2ϩ -binding site that is critical for IP 3 R activity, and sites of phosphorylation (16,40). IP 3 R structure solved to 30 Å by cryo-electron microscopy and single particle analysis predicts that this domain forms the outer surface of the IP 3 R tetramer (41,42). This (and the close proximity of ubiquitinated lysines to sites of modification or protease cleavage sites) suggests that surface-exposed lysines are predominately selected for ubiquitination. Given that ubiquitinated lysines were often in close proximity when IP 3 R sequences were aligned but that there was considerable sequence variance between IP 3 Rs in the vicinity of targeted lysines, it seems that selection is based primarily on structural parameters, chiefly surface accessibility, rather than sequence.
Although we identified ubiquitination sites only in the regulatory domain, this does not exclude the possibility that lysines outside this domain may become ubiquitinated because inefficient ubiquitination or technical reasons could force signals from outside the regulatory domain to below the limits of detection. Likewise, whereas it is possible that Lys-2100 is the only site of ubiquitination in IP 3 R3, it is more likely that the low abundance of ubiquitinated IP 3 R3 precluded the identification of additional ubiquitinated lysines. A, ␣T3-1 cells were transfected with vector (pcDNA3) or the constructs indicated, and 24 h later, cell lysates were prepared, subjected to 7% SDS-PAGE, transferred to nitrocellulose, and probed with anti-FLAG or anti-ubiquitin (FK2) antibody. B, ␣T3-1 cells were transfected as described for A and then treated with 10 nM GnRH for 0, 30, or 60 min. Cell lysates were prepared, subjected to 7% SDS-PAGE, transferred to nitrocellulose, and probed with anti-IP 3 R1 or anti-p97 antibody (as a loading control). C, graphed values of quantified IP 3 R1 immunoreactivity. In each independent experiment, immunoreactivity was normalized to t ϭ 0 values, and then normalized values were averaged. Data are presented as the mean Ϯ S.E. (n Ն 3). *, p Ͻ 0.01; **, p Ͻ 0.001 (comparing mutant ubiquitin-expressing cells with FLAG-UbWT-expressing cells).
Irrespective of cultured cell type (and rat brain and pancreas tissues (data not shown)), IP 3 Rs were modified by a mixture of monoubiquitin and ubiquitin chains, linked predominately through Lys-48 and Lys-63. Remarkably, using linkage-specific antibodies, we found that the vast majority and perhaps all ubiquitin chains are homogeneous, i.e. individual chains are composed entirely of either Lys-48 linkages or Lys-63 linkages, and that these homogeneous chains are segregated to distinct IP 3 R subunits.
For proteins, such as IP 3 Rs, that are tagged with multiple ubiquitin chain linkages, defining chain topology has proven to be very challenging (6), and for this reason, it has been generally assumed that the different linkages are segregated into homogeneous chains (5). Heterogeneous chains can certainly be formed in vitro, but as already explained in the Introduction, such conjugates may not be physiological. To the best of our knowledge, MHC class I is the only protein in addition to IP 3 Rs to have its ubiquitin chain topology mapped in vivo (43,44). Interestingly, MHC class I appears to be tagged with heterogeneous ubiquitin chains built with Lys-11 and Lys-63 linkages, both of which are required for the effective internalization and degradation of MHC class I (43,44). This contrasts markedly with the situation for IP 3 Rs and demonstrates that, in intact cells, ubiquitination is very complex, and it is unlikely that generalized "rules" regarding the formation of ubiquitin chains will emerge but rather that the topology of chains will have to be assessed for individual substrates.
The tagging of IP 3 Rs with homogeneous chains raises several interesting points. First, as it is improbable that one E2/E3 pair could catalyze the formation of both homogeneous Lys-48-and Lys-63-linked chains, it is likely that more than one E2 and/or E3 is responsible for IP 3 R ubiquitination. In this regard, there is mounting evidence and growing acceptance that E2 proteins determine linkage specificity in chain formation (reviewed in Ref. 45). Indeed, the E2 proteins UBE2S and UBE2G2 (Ubc7) and the UBE2N-UBE2V1 complex have been shown to catalyze Lys-11-, Lys-48-, and Lys-63-linked chains, respectively (27,(45)(46)(47)(48). In our previous work, we found that overexpressing dominant-negative UBE2G2 partially blocked IP 3 R ubiquitination and completely inhibited IP 3 R degradation (25). Together, these data raise the possibility that UBE2G2 may be catalyzing the formation of Lys-48-linked chains on IP 3 Rs and that an as yet unidentified E2 catalyzes the formation of Lys-63-linked chains. Furthermore, it is remarkable that Lys-48-and Lys-63-linked chains are segregated to separate IP 3 R subunits. This segregation suggests a high degree of coordination between the enzymes catalyzing ubiquitination, and clearly, to move forward, it will be necessary to identify and characterize all of the relevant enzymes involved in IP 3 R ubiquitination.
Second, the presence of homogeneous Lys-48-and Lys-63linked ubiquitin chains suggests that each chain type may signal for a distinct function. Our data obtained using ubiquitin mutants show that Lys-48-linked chains, but not Lys-63linked chains, are required for IP 3 R degradation. Consistent with this point, we also found that proteasomal inhibition causes a significant increase in Lys-48-conjugated IP 3 Rs (suggesting that Lys-48-conjugated IP 3 Rs are normally degraded by the proteasome) and that IP 3 Rs are degraded most rapidly in cells in which IP 3 Rs are heavily modified with Lys-48linked chains. Thus, Lys-48-linked chains are necessary and sufficient to trigger IP 3 R degradation, whereas Lys-63-linked chains appear not to be involved in this process. What then might be the role of Lys-63-linked chains? In general, it appears that the information encoded by the various types of ubiquitin conjugates is transmitted by proteins containing specialized ubiquitin-binding domains (reviewed in Ref. 49). Although there are Ͼ20 families of proteins that contain ubiquitin-binding domains with affinities for various conjugates, several are known to interact with just Lys-63-linked chains. For example, RAP80 specifically binds Lys-63-linked chains and recruits BRCA1 to sites of DNA damage to facilitate activation of DNA damage repair pathways (50,51). Thus, the Lys-63-linked chains attached to IP 3 Rs could recruit additional regulatory proteins to IP 3 Rs or, more globally, to the ER membrane. Likewise, the monoubiquitin conjugated to IP 3 Rs does not directly mediate IP 3 R degradation and could also play a recruiting role. It is also significant to note that, although Lys-11 linkages have recently been shown to mediate the proteasomal degradation of certain substrates (11,12), we found only minimal amounts of Lys-11 linkages on activated IP 3 Rs (Fig. 1A) (33), and they do not appear to play a role in IP 3 R processing.
It is interesting to note that, although Lys-63-linked chains do not mediate IP 3 R degradation, Lys-63 linkage levels peak and decline similarly to Lys-48 linkage levels (Fig. 2C). The decline in Lys-48 linkage levels can be readily attributed to proteasomal degradation of Lys-48-conjugated IP 3 Rs, but the reason for the decline in Lys-63 linkage levels is less obvious. Most likely, this decline is mediated by DUBs because this is the only known method for removing ubiquitin from proteins (13,14). Deubiquitination "rescues" proteins from the physiological consequence of ubiquitination, be it degradation or involvement in other cellular processes (14). Thus, Lys-63 linkage disassembly could simply serve to "shut off" signaling through Lys-63 linkages on IP 3 Rs or, alternatively or in addition, Lys-63-linked chain removal could precede and allow for the reubiquitination of IP 3 Rs with Lys-48-linked chains for the purpose of degradation. Significantly, there are already several known instances of this type of opposing deubiquitination-ubiquitination activity (34,(52)(53)(54)(55). For example, using mass spectrometry and linkage-specific antibodies, it was shown recently that ErbB2 is tagged with Lys-48-and Lys-63linked ubiquitin chains and that Lys-63-linked ubiquitin chains are edited by the DUB Usp9x (55). Remarkably, Lys-48-linked chains signal for the rapid degradation of ErbB2 by the proteasome, whereas Lys-63-linked chains signal for trafficking of ErbB2 to the lysosome (55). This reveals a highly complex, tightly controlled ubiquitin-dependent mechanism for the regulation of ErbB2 and seems similar to the situation we have described for IP 3 Rs.
In light of our data and in line with current opinions in the ubiquitin field, we propose a model that explains the complexities of IP 3 R ubiquitination in intact cells. First, activated IP 3 Rs are tagged either with homogeneous Lys-48-or Lys-63linked ubiquitin chains through the action of, at minimum, two E2 proteins and perhaps multiple E3 proteins. Lys-48conjugated IP 3 Rs are then rapidly degraded by the proteasome, whereas Lys-63-conjugated IP 3 Rs are not, and perhaps act to recruit proteins to IP 3 Rs or the ER. Simultaneously, DUBs disassemble Lys-63-linked ubiquitin chains, and the rescued IP 3 Rs are either retagged with Lys-63-linked chains or tagged with Lys-48-linked chains to trigger degradation.