Bortezomib rapidly suppresses ubiquitin thiolesterification to ubiquitin-conjugating enzymes and inhibits ubiquitination of histones and type I inositol 1,4,5-trisphosphate receptor

Proteasome inhibitors are currently emerging as therapeutic agents, with known efficacy against various cancers and inflammatory diseases (1, 2). Bortezomib (Velcade, PS-341) was recently shown in phase II trials to be effective against multiple myeloma (3) and is the first proteasome inhibitor to be approved by the Food and Drug Administration. The proteasome is a multicatalytic protease that degrades a range of crucial cellular proteins, including regulators of the cell cycle and elements of signaling pathways (1, 2, 4). Proteins are directed to the proteasome by virtue of being polyubiquitinated (4). Polyubiquitination is achieved through the hierarchical action of three enzymes, termed ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitin-protein ligase (E3). Whereas there is only one E1, there are many E2s (also often termed Ubcs) and E3s (4, 5). First, free ubiquitin becomes conjugated via its COOH terminus to a cysteine residue of E1 through a thiolester bond. Second, this “activated” ubiquitin is transferred and linked, again through a thiolester bond, to the active site cysteine residue of an E2, which may already be associated with an E3. Third, the ubiquitin is coupled to the ε-amino group of a lysine residue in the substrate through an isopeptide bond; this transfer is facilitated by the E3 that juxtaposes the E2 and the substrate. A polyubiquitin chain can then be formed by multiple rounds of ubiquitination; the COOH terminus of incoming ubiquitin residues are linked via an isopeptide bond to a lysine residue of the already attached ubiquitin. Finally, the polyubiquitinated protein is recognized by the proteasome and is degraded, with accompanying disassembly of the ubiquitin chain and recycling of ubiquitin monomers (4). Additional recent work indicates that ubiquitin conjugation also has other functions that do not involve proteasomal targeting (4), for example, in controlling the trafficking of cell surface receptors (6) and in regulating histones (7, 8), and that in these contexts monoubiquitination, rather than polyubiquitination, is often the regulatory event (6–8). To date, most studies on the mechanism of bortezomib action have focused on changes in gene expression and the levels and activity of relevant proteins (1, 2, 9–12). The events that lead to these changes remain unclear, however, because the immediate consequences of proteasome inhibition by bortezomib have not been examined in detail. Whether the biological effects of bortezomib result solely from the stabilization of key proteasomal substrates (e.g., transcription factors) in their polyubiquitinated form, or whether there are other consequences of proteasome inhibition, is presently unclear, and an answer to this question would both lead to a better understanding of the effects of bortezomib and stimulate the development of more specific and effective drugs. Thus, we examined the acute biochemical effects of bortezomib and show that it rapidly inhibits free ubiquitin levels and the thiolesterification of ubiquitin to E2s. This correlated with a decrease in the ubiquitination of certain substrates, showing that proteasome inhibition has acute consequences beyond the stabilization of polyubiquitinated proteins.

αT3-1 mouse anterior pituitary gonadotropes, HeLa cells, SH-SY5Y human neuroblastoma cells, and HEK 293 cells were maintained as described (13, 14). Bortezomib was obtained from Millennium Pharmaceuticals, Inc. (Cambridge, MA), dissolved in 70% ethanol, and stored at −20°C until required. N-acetyl-Leu-Leu-norleucinal (ALLN), MG132, and lactacystin were obtained from Alexis (San Diego, CA). Gonadotropin-releasing hormone (GnRH) and tumor necrosis factor-α were obtained from Sigma Chemical Co. (St. Louis, MO). Antibodies employed were anti–type I inositol 1,4,5-trisphosphate (InsP₃) receptor (13, 15), rabbit polyclonal anti-ubiquitin (16), anti-cdc34 and anti-IκBα (BD Biosciences, San Diego, CA), mouse monoclonal anti-ubiquitin (FK2) and anti-E2-25K (Affiniti Research Products, Ltd., Exeter, United Kingdom), anti-UbcH2 (Boston Biochem, Cambridge, MA), anti-uH2A and anti-H2B (Upstate Cell Signaling Solutions, Waltham, MA), and anti-Ubc7, which was raised, as described (15), against a peptide (YKIAKQIVQKSLGL) corresponding to the COOH-terminal 14 amino acids of ^mamUbc7 (14).

Essentially as described (13), cells were disrupted in lysis buffer [50 mmol/L Tris, 150 mmol/L NaCl, 1% Triton X-100, 1 mmol/L EDTA, 0.2 mmol/L phenylmethylsulfonyl fluoride, 10 μmol/L leupeptin, 10 μmol/L pepstatin, 0.2 μmol/L soybean trypsin inhibitor, 1 mmol/L DTT (pH 8.0)]. InsP₃ receptors were immunopurified from lysates with anti–type I InsP₃ receptor. Samples were boiled in gel loading buffer supplemented with 100 mmol/L DTT, electrophoresed in 5% gels, and probed in Western blots with anti-ubiquitin (FK2) or anti–type I InsP₃ receptor.

Cells were rinsed once with ice-cold serum-free culture medium, disrupted in DTT-free lysis buffer at 4°C for 30 minutes, and centrifuged at 16,000 × g for 10 minutes at 4°C, and lysate samples were boiled in gel loading buffer either without DTT or with 100 mmol/L DTT to preserve or destroy thiolester bonds, respectively, between ubiquitin and E2s (16). Samples were then electrophoresed in 12% gels and probed in Western blots with E2 antibodies. For analysis of the cellular content of free ubiquitin and ubiquitin conjugated via isopeptide bonds, lysate samples boiled in gel loading buffer with DTT were electrophoresed in 15% and 7% gels, respectively. Nitrocellulose membranes were heated in a pressure cooker for 30 minutes immediately after transfer of electrophoresed proteins to increase ubiquitin immunoreactivity and then probed with rabbit polyclonal anti-ubiquitin.

Cells were disrupted in 10 mmol/L Tris, 1 mmol/L EGTA, 0.2 mmol/L phenylmethylsulfonyl fluoride, 10 μmol/L leupeptin, 10 μmol/L pepstatin, 0.2 μmol/L soybean trypsin inhibitor, 1 mmol/L DTT (pH 7.4) and centrifuged at 16,000 × g for 5 minutes at 4°C, and pellets were resuspended in the same buffer and sonicated. Samples were then boiled in gel loading buffer with DTT, electrophoresed in 12% gels, and probed with anti-uH2A, which is specific to monoubiquitinated histone 2A, or anti-H2B, which recognizes both unmodified and monoubiquitinated forms of histone 2B.

HeLa cells were disrupted in lysis buffer. Lysates were prepared and boiled in gel loading buffer with DTT, electrophoresed in 10% gels, and probed with anti-IκBα.

To measure cell viability, 0.5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was added to cell cultures. Cells were dissolved 10 minutes later in DMSO. Absorbance at 540 nm was then measured to determine MTT reduction. InsP₃ formation in cell monolayers, free Ca²⁺ concentration in cell suspensions, and immunoreactivity in Western blots were quantitated as described (14). Statistical significance was assessed when n > 2 by unpaired t test with Welch's correction (Figs. 1 and 2) or Student's paired t test (Fig. 4).

We first sought to define the effects of bortezomib on the balance between conjugated and free ubiquitin and used αT3-1 cells, a mouse pituitary cell line that we have used previously in studies on the regulated polyubiquitination of InsP₃ receptors (13). Figure 1 shows that 1 μmol/L bortezomib rapidly causes the accumulation of high-molecular-mass ubiquitin conjugates (40-400 kDa; Fig. 1A) in a manner that correlates with a depletion of free ubiquitin (8 kDa; Fig. 1B and I). After 1-hour incubation, the effects of bortezomib were half-maximal at 10 to 100 nmol/L, whereas after 6-hour incubation, the effects were half-maximal at 1 to 10 nmol/L (Fig. 1C, D, and J). These values are well within the concentration range achieved in patients treated with bortezomib (17). These data were obtained from cell lysates prepared in the presence of DTT, which destroys thiolester bonds between E2s and ubiquitin but not isopeptide bonds between ubiquitin the ε-amino group of lysine residues in target proteins (14, 16). Identical analysis of samples prepared in the absence of DTT produced very similar data (data not shown), indicating that only a small proportion of cellular ubiquitin is thiolesterified to E2s and that the extent of free ubiquitin depletion has not been underestimated. Overall, these data show that blockade of the proteasome with bortezomib leads not only to the rapid accumulation of polyubiquitinated species but also to depletion of free ubiquitin.

We next sought to determine what effects the depletion of free ubiquitin might have and initially focused on thiolesterification of ubiquitin to ^mamUbc7, the E2 that mediates InsP₃ receptor polyubiquitination (6, 14). In the process of linking ubiquitin to a target substrate, E2s form an intermediate (E2∼ubiquitin), in which a labile thiolester bond (∼) couples the active site cysteine of the E2 to ubiquitin (4). Analysis of ^mamUbc7 in αT3-1 cell lysates electrophoresed without DTT, which preserves thiolester bonds (Fig. 1E, lanes 1-5), showed that bortezomib rapidly inhibited the levels of ^mamUbc7∼ubiquitin (29 kDa), with a reciprocal increase in the level of unmodified ^mamUbc7 (17 kDa); as expected, destruction of thiolester bonds by electrophoresis with DTT eliminated ^mamUbc7∼ubiquitin (Fig. 1E, lanes 6 and 7). Examination of other E2s (Fig. 1F–H) revealed similar effects on E2-25K, an E2 that has yet to be linked to a cognate E3 or particular substrates (4) but has recently been assigned a role in Alzheimer's disease (18); UbcH2, an E2 that is responsible for the monoubiquitination of histones (7, 8); and cdc34, an E2 that can be a component of SCF complexes that ubiquitinate key regulatory substrates, including G1 cyclins, Sic 1, β-catenin, and IκBα (4, 19). Interestingly, inhibition of E2-25K and UbcH2 thiolesterification was much more profound than that seen for ^mamUbc7 and cdc34 (Fig. 1E–I). Additional analyses (Fig. 1J) showed that the effects of bortezomib on E2 thiolesterification were half-maximal at ∼30 and ∼3 nmol/L in 1- and 6-hour incubations, respectively, and that this correlated with the dose dependence of bortezomib-induced free ubiquitin depletion (for clarity, only the combined data for E2-25K are shown, but the thiolesterification of the other the E2s was altered with similar potency). Further analysis of the structurally unrelated proteasome inhibitors, ALLN, MG132, and lactacystin, revealed effects on free ubiquitin levels and E2 thiolesterification similar to those of bortezomib (Fig. 1K), albeit with lower potency; after 6-hour incubation, half-maximal inhibition of E2-25K thiolesterification was seen at 1,500 nmol/L ALLN, 100 nmol/L MG132, and 200 nmol/L lactacystin. These data indicate that it is proteasome inhibition rather than a proteasome-independent phenomenon that accounts for the effects of bortezomib on ubiquitin levels and E2 thiolesterification. Consistent with this notion is the finding that bortezomib did not inhibit E2s directly, because E2 thiolesterification in a cell free system, the FII fraction of rabbit reticulocytes, was unaffected by 10 μmol/L bortezomib (data not shown). Finally, MTT reduction was unaffected by 6-hour treatment with 1 nmol/L to 1 μmol/L bortezomib (Fig. 1L), showing that the bortezomib-induced changes in ubiquitin levels and E2 thiolesterification (Fig. 1J) are not a consequence of a decline in cell viability. Rather, these changes in ubiquitin levels and E2 thiolesterification seem to precede loss of viability, because longer (24-hour) incubations with bortezomib did inhibit MTT reduction (Fig. 1L). Overall, in view of the parallels between ubiquitin depletion and inhibition of E2 thiolesterification, it is possible that the former accounts for the latter, although it remains to be explained why the thiolesterification to some E2s is inhibited more than others (see Summary and Conclusions).

We next sought to determine whether inhibition of E2 thiolesterification had effects on substrate ubiquitination and initially examined histone monoubiquitination, as UbcH2 mediates this process (7, 8). Histones 2A and 2B are constitutively monoubiquitinated and deubiquitinated as part of the transcription control process, such that at steady state 1% to 2% of histone 2B and 5% to 15% of histone 2A are monoubiquitinated (7, 8). Figure 2A and B shows that 23-kDa proteins corresponding to monoubiquitinated histone 2B and 2A (20) were very rapidly depleted by 1 μmol/L bortezomib treatment of αT3-1 cells, with half-maximal inhibition occurring at ∼15 minutes (Fig. 2C). Thus, bortezomib-induced suppression of UbcH2 thiolesterification (Fig. 1G and I) seems to cause a rapid decline in levels of monoubiquitinated histones, presumably because the rate of histone ubiquitination is reduced, whereas deubiquitination occurs normally. Analysis of the dose dependence of bortezomib action on histone monoubiquitination (Fig. 2D) supports this view, because half-maximal effects were seen at ∼30 and ∼3 nmol/L in 1- and 6-hour incubations, respectively, which parallels the values for bortezomib inhibition of E2 thiolesterification (Fig. 1J). Likewise, ALLN, MG132, and lactacystin also inhibited histone ubiquitination (Fig. 2E), with half-maximal effects (∼2,000, 100, and 300 nmol/L, respectively) very similar to those seen for inhibition of E2 thiolesterification (Fig. 1K). Finally, we examined whether the data obtained from αT3-1 cells were representative of the situation in other cell types. Figure 3 shows that treatment of HeLa cells (Fig. 3A–C), SH-SY5Y cells (Fig. 3D–F), and HEK 293 cells (Fig. 3G–I) with 1 μmol/L bortezomib inhibited E2 thiolesterification and histone monoubiquitination very similarly to that seen in αT3-1 cells, indicating that it is a response common to all cell types.

As bortezomib inhibits ^mamUbc7 thiolesterification (Fig. 1E and I), we also examined whether it inhibits InsP₃ receptor polyubiquitination. InsP₃ receptors form tetrameric InsP₃-gated Ca²⁺ channels in endoplasmic reticulum membranes, and their polyubiquitination in response to activation of certain cell surface G protein-coupled receptors, for example, GnRH receptors (13, 14), leads to proteasomal degradation and consequently the suppression of InsP₃-induced Ca²⁺ mobilization from the endoplasmic reticulum (6). Figure 4A and C shows that the initial rate of GnRH-induced InsP₃ receptor polyubiquitination in αT3-1 cells (during the first 10 minutes of GnRH exposure) was significantly inhibited by bortezomib pretreatment; thereafter, the level of polyubiquitinated receptors declined in control cells (due to their degradation by the proteasome) and continued to accumulate in bortezomib-treated cells (in which the proteasome was blocked). Consistent with these data, Fig. 4B and D shows that bortezomib blocked GnRH-induced InsP₃ receptor degradation. Under the same conditions, GnRH-induced increases in InsP₃ and Ca²⁺ concentration were not inhibited by bortezomib (Fig. 4E and F). Thus, the early inhibitory effects of bortezomib on InsP₃ receptor polyubiquitination cannot be attributed to inhibition of InsP₃ formation, the event that triggers InsP₃ receptor polyubiquitination (6), or disruption of the Ca²⁺-mobilizing function of InsP₃ receptors but rather seem to be due to inhibition of ^mamUbc7 thiolesterification (Fig. 1E and I). Consistent with this notion, the initial rate of GnRH-induced InsP₃ receptor polyubiquitination in αT3-1 cells was also inhibited by ALLN (13).

Finally, we examined whether bortezomib suppressed the polyubiquitination of IκBα, because this protein is a key negative regulator of nuclear factor-κB, a transcription factor that increases the expression of growth-enhancing and antiapoptotic proteins, and stabilization of IκBα has been implicated in the mechanism of bortezomib action (1, 2, 9–11). In HeLa cells, tumor necrosis factor caused rapid degradation of IκBα, which was mediated by the ubiquitin/proteasome pathway, because bortezomib both inhibited the degradation (Fig. 5, bottom, lanes 5–8) and caused the accumulation of polyubiquitinated IκBα (Fig. 5, top, lanes 4, 6, and 8). However, at no stage did bortezomib inhibit the accumulation of polyubiquitinated IκBα (Fig. 5, top), although bortezomib inhibited thiolesterification to E2s in HeLa cells (Fig. 3A and B). This suggests that the relatively minor effects of bortezomib on cdc34 thiolesterification (Figs. 1I and 3B) are insufficient to suppress the rate of IκBα polyubiquitination. An alternative explanation is that E2s other than cdc34 ubiquitinate IκBα, as has been reported (19), and perhaps that inhibition of thiolesterification to these other E2s is even less than that seen with cdc34.

In total, these data show that bortezomib and other proteasome inhibitors rapidly deplete free ubiquitin in cultured cells, with associated suppression of E2 thiolesterification and ubiquitination of certain substrates. Ubiquitin depletion correlated with the accumulation of polyubiquitinated substrates, indicating that the cellular free ubiquitin pool is relatively small and is acutely dependent on recycling from these proteins. Although ubiquitin depletion correlated with reduced thiolesterification to all E2s examined, it is intriguing that some E2s (E2-25K and UbcH2) were more affected than others. This may reflect differences in the subcellular localization of the different E2s and the existence of discrete pools of ubiquitin. For example, UbcH2 is active in the nucleus (7, 8) and the nuclear pool of free ubiquitin is very small (20) and may be selectively depleted following proteasome inhibition. Alternatively, as all E2s receive their ubiquitin from E1, affinity differences among the E2s in their interactions with E1 could lead to selective loading of only certain E2s when ubiquitin is depleted. In turn, such factors can explain why bortezomib inhibits the ubiquitination of some substrates (histones) better than others (InsP₃ receptors and IκBα)—only if the E2s that ubiquitinate certain proteins are strongly affected by bortezomib will ubiquitination of that protein be strongly inhibited.

With regard to relevance of these observations to the therapeutic effects of bortezomib, three points are noteworthy. First, E2 inhibition by bortezomib can contribute to the stabilization of proteasomal substrates by virtue of decreasing their polyubiquitination. This seems to be the case for InsP₃ receptors and could also be the case for proteins, other than IκBα, that contribute to the therapeutic effects of bortezomib. Significantly, this inhibition would lead to the maintenance of proteins in their nonubiquitinated forms, which unlike polyubiquitinated proteins are certain to retain their activity. Second, as inhibition of cdc34 activity with a dominant-negative strategy enhances the antimyeloma activity of bortezomib (21), the inhibitory effect of bortezomib on ubiquitin thiolesterification to cdc34 (Figs. 1 and 3) may directly contribute to the therapeutic efficacy of bortezomib. Third, and perhaps most significantly, bortezomib has a dramatic effect on histone monoubiquitination, which could be a key factor in mediating the transcriptional changes that occur in response to bortezomib treatment (2, 10–12). Importantly, the bortezomib concentrations that we find to profoundly inhibit histone monoubiquitination in cultured cells led to reduced cell viability and are well with the range found in patients receiving therapy (17) and those being used to define the molecular mechanisms of bortezomib in in vitro studies (10–12, 21–23). More generally, the histone data show that bortezomib inhibits a process in which ubiquitination has a role different from the targeting of substrates to the proteasome. Thus, effects on proteasome targeting-independent actions of ubiquitination (6–8) must also be taken into account when considering the mechanism of action bortezomib and other proteasome inhibitors. Finally, these findings indicate that other drugs that deplete free ubiquitin (e.g., inhibitors of deubiquitinating enzymes) might also be valuable therapeutically.

We thank Drs. Martin Obin and Pamela Mellon for kindly providing rabbit polyclonal anti-ubiquitin and αT3-1 cells and Dr. David Schenkein for helpful discussions.