ErbB2 Trafficking and Degradation Associated with K48 and K63 Polyubiquitination

Overexpression of the ErbB2/HER2 receptor tyrosine kinase occurs in up to 25% of human breast cancers, where it is predictive of aggressive disease and poor clinical outcome, and serves as a validated clinical target for a growing class of anti-ErbB2 therapeutics (1, 2). In addition to therapeutic antibodies and small molecule kinase inhibitors that specifically interfere with ErbB2 receptor function, some targeted agents in clinical development actually depend upon the endocytic recycling and intracellular trafficking of membrane overexpressed ErbB2 (3, 4), calling attention to this poorly understood but critical feature of ErbB2 receptor biology. Yet another class of therapeutics being developed to treat ErbB2-positive cancers includes derivatives of the benzoquinoid ansamycin antibiotic geldanamycin (GA), which bind to and inactivate an essential chaperone of membrane-bound ErbB2, heat shock protein 90 (HSP90), inducing endocytosis and receptor downregulation by proteasomal and lysosomal mechanisms (5, 6). Given its unique mechanism of action and the particular sensitivity of overexpressed ErbB2 to HSP90 inhibitors, GA has become a favorite tool for studying ErbB2 receptor endocytosis and trafficking (7–10). Less well appreciated is the recent observation that proteasome inhibition by the clinically approved dipeptide boronate bortezomib (PS341) can also induce ErbB2 internalization and lysosomal decay (11).

Unlike the epidermal growth factor receptor (EGFR) whose ligand-activated endocytosis and intracellular trafficking has become a model for all receptor tyrosine kinases (12, 13), the ligand-less ErbB2 receptor is considered to be endocytosis impaired, although there is poor understanding of how this is achieved (8, 12–14). Instead of being internalized and endosomally routed into vesicles and the multivesicular body (MVB) pathway for subsequent lysosomal degradation as with ligand-activated EGFR, ErbB2 dimers are largely recycled back to the plasma membrane for reactivation (8, 12, 13). ErbB2 can even transfer its endocytosis impairment to EGFR and other heterodimerizing partners, although this impairment can be fully abrogated by HSP90 inhibition, which induces rapid ErbB2 ubiquitination followed by receptor downregulation (15, 16). The importance of ubiquitination in regulating ErbB2 endocytosis and plasma membrane overexpression remains obscure, although it has been suggested that the endocytosis impairment of ErbB2 is due to its intrinsic resistance to ubiquitination (12). Thus, new insight is needed into the mechanisms and ubiquitination codes associated with ErbB2 endocytosis and downregulation (13, 17).

Ubiquitination is a reversible posttranslational modification regulating a wide variety of protein signaling mechanisms, including endocytic downregulation of ErbB family receptors (13, 17). The 76–amino acid ubiquitin polypeptide can be covalently attached via its COOH terminal glycine (G76) to either a target protein's lysine (K) ϵ-amino group or to another target-bound ubiquitin molecule via one of its seven internal K residues, forming topologically distinct polyubiquitin chains. Ubiquitin (E3) ligases interact with specific ubiquitin conjugating (E2) enzymes associated with different intracellular sites and linkage-specific reaction products. Two different E3 ligases, CHIP (COOH terminus of HSP70-interacting protein) and c-Cbl, have been reported to associate with ErbB2, each capable of forming different types of polyubiquitin chains (13–16, 18, 19). The most abundant polyubiquitin chains in living cells are K48 linkages, which adopt a closed conformation and serve as a signal for target protein degradation by the 26S proteasome. Next most common, K63 linkages adopt an extended linear conformation as nonproteasome addressing dock sites for such diverse cellular functions as DNA repair, signal transduction, transcription, endosomal trafficking, and MVB sorting for lysosomal decay (12, 13, 17, 20). In particular, K63-linked polyubiquitination has been clearly associated with clathrin-dependent endocytosis and lysosomal downregulation of ligand-activated EGFR (13, 21, 22). As well, ubiquitination can be edited via developmentally and subcellularly restricted deubiquitinating isopeptidases (deubiquitylating enzymes, DUB), regulating the intracellular fate of a target protein (23, 24). Two DUBs potentially relevant to the endocytosis and trafficking of breast epithelial membrane proteins are USP8/UBPy, known to facilitate the downregulation of EGFR and ErbB3 (25), and USP9x/FAM, an endosomally localized regulator of epithelial stem/progenitor cell function (26, 27).

The present study used SK-BR-3 cells to explore the dependence of ErbB2 overexpression on both polyubiquitination and deubiquitination mechanisms. The contrasting effects of PS341 and GA on ErbB2 polyubiquitination were assessed using multiple reaction monitoring (MRM)/mass spectrometry (MS) and linkage-specific monoclonal antibodies (28), showing that proteasome and HSP90 inhibitors downregulate overexpressed ErbB2 by alternative cytoplasmic trafficking and degradative pathways linked to different K48 and K63 polyubiquitin chain editing mechanisms.

SK-BR-3 cells were obtained from American Type Culture Collection (ATCC) and grown under ATCC-recommended culture conditions. Bortezomib (PS341) was kindly provided by Millennium Pharmaceuticals; GA and chloroquine were purchased from Sigma-Aldrich. The mouse IgG1 monoclonal anti-c-ErbB2/c-Neu (Ab-3), developed against the COOH terminal 1242 to 1255 amino acids of human ErbB2, was purchased from Calbiochem. Mouse monoclonals to HSP90 and HSP70 were purchased from Stressgen. Antibodies to actin, lysosome-associated membrane protein 2b (LAMP2b), clathrin heavy chain (CHC), USP9x, and vimentin were purchased from Abcam; those to ERα (F10), c-Cbl, CHIP, and Epsin-1 were purchased from Santa Cruz Biotechnology; antibody to Eps15 was purchased from Novus Biologicals; antibody to total ubiquitin was purchased from Dako. Polyubiquitin linkage–specific recombinant IgG antibodies against K48 (Apu2.07) and K63 (apu2.16) chains (28) were a kind gift from Genentech.

Sequence-confirmed cytomegalovirus promoter–driven expression plasmids #17608 (pRK5-HA-wtUb) and #17606 (pRK5-HA-Ub-K63), encoding wild-type ubiquitin and mutated ubiquitin capable of only forming K63 polyubiquitin linkages (all other lysines mutated to arginines), respectively (29), were obtained from Addgene. Two additional expression constructs were generated from these using the Stratagene Quickchange kit forming pRK-HA monoubiquitin from #17606, with all lysines mutated to arginines (unable to form polyubiquitin), and pRK-HA-Ub-K63R from #17608, with lysine 63 mutated to arginine 63 (only incapable of forming K63 linked polyubiquitin). SK-BR-3 transfections were performed in 10-mm plates over 4 to 6 hours using serum-free media and Lipofectamine 2000 (Invitrogen). siRNA (and control) reagents targeting Epsin, Eps15, and CHC were commercially obtained from Dharmacon and used at final concentrations of 100 nmol/L.

Whole-cell lysates were prepared in a modified radioimmunoprecipitation assay buffer [50 mmol/L HEPES (pH 7.5), 100 mmol/L NaCl, 2 mmol/L Na₃VO₄, 1% NP40, 1% deoxycholate, and 0.01% SDS] containing a protease inhibitor cocktail (Mini Complete) and 5 μmol/L ubiquitin aldehyde (Calbiochem). SDS was omitted from lysates used for immunoprecipitation and MS. Whole-cell lysates were homogenized by sonication (550 Sonic Dismembrator) twice for 10 seconds each and cleared by centrifugation for 10 minutes at 4°C. Protein content of supernatants was determined by Bradford assay (Bio-Rad Laboratories). Lysosomal fractions were isolated from cultured cells by density gradient separation (34,200 rpm, 2 hours) using the lysosome enrichment kit for tissue and cultured cells and protocol from Pierce. Before immunoprecipitation, 0.5 to 1 μg of lysate protein was precleared with protein A-Sepharose beads (Santa Cruz Biotechnology) and then incubated with 5 μg of anti-ErbB2 for 3 hours at 4°C under continuous agitation. Immune complexes were recovered using 50 μL of protein A-Sepharose, washing thrice in lysis buffer and twice with Tris-borate EDTA, and resuspended in 75 μL of Laemmli buffer before gel electrophoresis and immunoblotting. Electrophoresis was performed using 4% to 12% Nu-Page Bis-Tris gradient gels (Invitrogen) with MOPS running buffer (Invitrogen), and proteins were transferred onto polyvinylidene difluoride membranes (Millipore) blocked with 5% nonfat milk in PBS with 0.05% Tween 20. For sequential antibody probing, blots were stripped using Restore Western Blot Stripping Buffer (Pierce Biotechnology).

Cells seeded in eight-chamber slides (Lab-Tek II; Nalge Nunc International) were cultured overnight, washed with PBS, and fixed with 4% PFA for 10 minutes at room temperature. After cell permeabilization in 0.5% Triton X for 10 minutes, cell-mounted slides were treated for 30 minutes (room temperature) with 5% normal donkey serum blocking solution and then overnight in primary antibody (2.5% serum dilution). Secondary antibodies were donkey anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 555 (Invitrogen), and serum was diluted and incubated for 30 minutes. Slides were mounted in Prolong Gold (Invitrogen), stained with 4′,6-diamidino-2-phenylindole (DAPI), and left overnight before fluorescence microscopic analysis.

ErbB2 immunoprecipitates were resolved on SDS-PAGE gels, and the gel region extending from the full-length ErbB2 band (185 kDa) to just below the visible myosin band (260 kDa) was excised for trypsin digestion. Typically, 30 ErbB2 peptides were identified in each sample; in addition to ErbB2, only ubiquitin could be detected from this gel region. Eluted gel samples were analyzed by nanoLC-MRM/MS on a 4000 QTRAP hybrid triple quadrupole/linear ion trap mass spectrometer (Applied Biosystems). Chromatography was performed using an Eksigent NanoLC-2D LC system with buffer A (0.1% formic acid) and buffer B (90% acetonitrile in 0.1% formic acid). Digested samples were loaded at 20 μL/minutes (0.1% formic acid) onto a 5 mm × 300 μm Dionex reversed phase C18 column (5 μm, 100 Å) and eluted at 300 nL/minutes with a 75-μm inner diameter Integrafrit column (New Objective), packed in house with 10 to 12 cm of ReproSil-Pur C18-AQ 3-μm reversed phase resin (Dr. Maisch GmbH), using a gradient of 2% to 70% buffer B over 32 minutes. Peptides were ionized using a PicoTip emitter (75 μm, 15-μm tip, New Objective). Data acquisition was performed with an ion spray voltage of 2,450 V, curtain gas of 10 p.s.i., nebulizer gas of 20 p.s.i., an interface heater temperature of 150°C, and unit resolution. Collision energy, declustering potential, and collision cell exit potential were optimized to achieve maximum sensitivity. Ubiquitin-linked peptides were analyzed as previously reported (30, 31), with the seven possible (K6, K11, K27, K29, K33, K48, K63) ubiquitin linkages and their respective tryptic peptides interrogated by their corresponding m/z MRM ion transitions as shown in Supplementary Table S1.

Given the nanomolar sensitivity of SK-BR-3 cells to both GA and PS341 and the need to monitor effects in viable cells over a 24-hour exposure period, a maximum GA dose of 20 nmol/L was chosen and compared with PS341 doses ranging from 10 to 50 nmol/L. Previous PS341 studies had shown the SK-BR-3 72-hour IC₅₀ dose to be 4 nmol/L, with virtually complete inhibition of SK-BR-3 proteasome activity achieved by 24-hour treatment with 25 nmol/L PS341 (11). As shown in Fig. 1 immunoblots, 20 nmol/L PS341 caused a 50% reduction in total ErbB2 protein expression after 24 hours, whereas the same dose of GA caused a 50% loss of total ErbB2 protein by 6 hours and more profound reduction by 24 hours. Both GA and PS341 caused dissociation of HSP90 from ErbB2 and replacement by HSP70, with kinetics reflecting their different rates of ErbB2 downregulation: ErbB2 chaperone exchange was induced within 2 hours of HSP90 inhibition but required 24 hours of proteasome inhibition. Immunofluorescence imaging using the same COOH terminal specific anti-ErbB2 antibody used for immunoblotting revealed differential cytoplasmic trafficking of intact ErbB2 following GA and PS341 (Fig. 1C and D). Untreated SK-BR-3 cells showed predominant plasma membrane ErbB2 overexpression; within 4 hours of PS341 treatment ErbB2 seemed partially internalized, within 8 hours it seemed as lower intensity scattered cytoplasmic aggregates, and within 24 h it seemed as intense polar and perinuclear aggregates. Previously, we showed that PS341 causes this same pattern of ErbB2 internalization and perinuclear aggregation in another ErbB2-overexpressing breast cancer cell line, BT474 (11). A similar time course of GA treatment in SK-BR-3 cells showed more rapid internalization with appearance of small, scattered cytoplasmic ErbB2 aggregates by 2 hours, followed by ever-diminishing ErbB2 cytoplasmic signals between 2 and 24 hours without any appearance of perinuclear ErbB2 aggregates.

To exclude the possibility that proteasome inhibitor treatment induces redistribution of ErbB2 into aggresomes, we costained PS341-treated and untreated SK-BR-3 cells for the aggresome marker vimentin; we found no colocalization of vimentin and ErbB2, indicating that PS341 does not induce redistribution of ErbB2 into aggresomes (data not shown). In contrast, immunofluorescence imaging after 24 hours of PS341 treatment showed colocalization of ErbB2 with LAMP2b, whereas untreated cells showed no ErbB2 and LAMP2b colocalization (Fig. 2A). GA treatment out to 24 hours failed to induce any colocalization of ErbB2 with LAMP2b (results not shown). To confirm ErbB2 trafficking to lysosomes, whole-cell and lysosomal fractions were immunoblotted for ErbB2 and LAMP2 proteins; lysosomal extracts were enriched in LAMP2, but only the 24-hour PS341-treated lysosomal fraction was enriched in ErbB2 protein (Fig. 2B). Whereas chloroquine had been shown to inhibit lysosomal proteolysis of ErbB2 induced by PS341 (11), it did not inhibit GA-induced ErbB2 degradation (results not shown). ErbB2 internalization induced by GA had been reported by some to occur via a clathrin-dependent process (9, 10) and by others via a clathrin-independent process (32, 33). Transfection of SK-BR-3 cells with siRNAs targeting CHC or the endocytic adaptors Epsin 1 and Eps15 were performed to achieve >80% target knockdown before GA or PS341 (24 hours) culture treatment. We observed slight inhibition of GA-induced ErbB2 downregulation following CHC knockdown but not following Epsin 1 or Eps15 knockdown (Supplementary Fig. S1). By comparison, PS341 treatment produced the expected 50% reduction in total ErbB2 levels, but this was not prevented by knockdown of any of the three regulators of clathrin-dependent endocytosis, consistent with a clathrin-independent internalization process (Fig. 2C). Of note, inhibition of new protein synthesis by 24-hour concurrent cycloheximide treatment prevented both PS341-induced perinuclear trafficking and lysosomal degradation of ErbB2 (Fig. 2D).

SK-BR-3 cells were transiently transfected with expression constructs encoding wild-type ubiquitin or different ubiquitin chain mutants, including K48RUb (unable to form K48 polyubiquitin chains), K63RUb (unable to form K63 polyubiquitin chains), monoubiquitin (unable to form any polyubiquitin chains), K48Ub (able to form only K48 polyubiquitin chains), and K63Ub (able to form only K63 polyubiquitin chains). Transfectants were analyzed for short-term cell growth and by immunoblotting and immunofluorescence imaging for ErbB2 expression. Transfectants overexpressing the K48Ub, K63Ub, and K48RUb chain mutants showed near-normal ErbB2 morphology (results not shown). In contrast, cells overexpressing the K63RUb and monoubiquitin chain mutants lost the uniform surface overexpression of ErbB2 seen in wild-type ubiquitin transfectants and expressed large subsurface cytoplasmic ErbB2 aggregates (Fig. 3A). All but the monoubiquitin transfectants showed near-normal short-term culture growth and, when treated for 24 hours with either GA or PS341, showed control levels of ErbB2 downregulation (result not shown). In contrast, the monoubiquitin transfectants seemed incapable of short-term growth.

ErbB2 immunoprecipitates were probed for their association with the deubiquitylating enzymes USP8/UBPy and USP9x/FAM. In control and PS341-treated SK-BR-3 cells, ErbB2 was not found to be associated with USP8/UBPy (data not shown). In contrast, ErbB2 immunoprecipitates were found to contain significant USP9x relative to control immunoprecipitates prepared using anti-ERα IgG (Fig. 3B). Because SK-BR-3 cells do not express ERα, this approach controlled for nonspecific protein binding to either the IgG or the protein A-Sepharose conjugate. Neither GA nor PS341 treatments (6 hours) produced any significant change in total cell USP9x levels; however, a 24-hour time course study of PS341 treatment showed persistent USP9x coassociation with ErbB2 (data not shown). Following ∼90% knockdown of USP9x, ErbB2 protein levels were unchanged in control SK-BR-3 cells; as well, this knockdown had no effect on ErbB2 localization in the presence or absence of PS341 or GA and did not alter GA-induced downregulation of ErbB2 (data not shown). However, USP9x knockdown significantly enhanced PS341 (24 hours)–induced lysosomal decay of ErbB2 (Fig. 3B).

We also probed for treatment-associated changes in total and ErbB2-associated CHIP and c-Cbl proteins. In whole-cell lysates of SK-BR-3, CHIP levels were lower and more difficult to detect than c-Cbl levels; but neither GA nor PS341 treatments (6 hours) significantly altered cell content of these E3 ligases (Fig. 3C). Whereas immunoprecipitates from these same lysates showed no ErbB2 association with CHIP (before or after GA and PS341 treatments; data not shown), these same ErbB2 immunoprecipitates showed PS341 (but not GA) induction of coprecipitating c-Cbl (Fig. 3C), implicating this E3 ligase in PS341-induced ErbB2 polyubiquitination.

As shown in Fig. 4, immunoprecipitated ErbB2 from control and treated SK-BR-3 cells was electrophoretically separated and the full-length band at 185 kDa was excised and trypsin digested along with the gel region from 180 to 250 kDa (Fig. 4A). MS analysis determined that there was no protein other than ErbB2 in this excised gel region chosen to enable MRM/MS analysis of varying sizes of polyubiquitinated ErbB2, whose chains were discriminated by the respective diglycine peptide signatures for K48 and K63 (Fig. 4B). Three different ErbB2 tryptic peptides (Supplementary Table S1) representing extracellular, perimembrane, and intracellular domain fragments were used to normalize for receptor-bound ubiquitin levels between treated samples. ErbB2 K48 and K63 chain linkages were the most abundant polyubiquitin forms detected in all samples; low levels of K29 chain linkages were detected in some treated samples, but there were no K6, K11, K27, or K33 chains detected in any of the samples. With K48 and K63 signals from untreated SK-BR-3 samples establishing the baseline (control) ErbB2 polyubiquitin levels, the differential time-dependent changes observed in K48 and K63 polyubiquitin levels following either PS341 (20 nmol/L) or GA (20 nmol/L) treatments were plotted as fold changes (Fig. 4C and D). Proteasome inhibition produced a gradual 5-fold to 15-fold increase in ErbB2 K48 levels between 4 and 10 hours after PS341 treatment; in contrast, ErbB2 K63 levels showed a delayed increase beginning 6 hours after PS341 exposure, nearing ErbB2 K48 levels by 8 hours, and then more than doubling ErbB2 K48 levels and achieving a 40-fold increase over baseline ErbB2 K63 levels by 10 hours (Fig. 4C). After 24 hours of PS341 exposure, ErbB2 polyubiquitin levels achieved >150-fold increase over baseline SK-BR-3 levels (results not shown). Unlike this delayed but dramatic and biphasic K48 and K63 ErbB2 polyubiquitin response to proteasome inhibition, HSP90 inhibition produced a more rapid (within 2 hours), modest (4-fold to 6-fold), and synchronous induction of K48 and K63 ErbB2 polyubiquitin (Fig. 4D).

Recently developed K63 and K48 linkage–specific antibodies (28) were used to complement the quantitative MRM/MS monitoring of ErbB2 polyubiquitin. Parallel sets of ErbB2 immunoprecipitates from PS341-treated SK-BR-3 cells were immunoblotted to detect ErbB2 associated with total ubiquitin, K48-linked polyubiquitin, and K63-linked polyubiquitin (Supplementary Fig. S2A). Consistent with the biphasic MRM/MS results, ErbB2-associated K63 polyubiquitin seemed absent until 4 to 8 hours after proteasome inhibition, followed by a more dramatic increase at 10 hours, whereas ErbB2-associated K48 polyubiquitin increased earlier. These antibodies were also used to image the intracellular trafficking of total ubiquitin, as well as total K48 and K63 polyubiquitin in relation to total ErbB2 (Supplementary Fig. S2B and C). Following PS341 treatment (20 nmol/L, 24 hours), total intracellular ubiquitin seemed scattered throughout the cell nucleus and cytoplasm, with perinuclear accumulation appearing in some cells. Only this perinuclear ubiquitin seemed to colocalize with the internalized and perinuclear ErbB2 (Supplementary Fig. S2B). In untreated SK-BR-3 cells, K63 polyubiquitin was predominantly extranuclear whereas K48 polyubiquitin was both nuclear (excluding the nucleolus) and cytoplasmic; some membrane colocalizations of ErbB2 with the two different forms of polyubiquitin were apparent, more pronounced for K63 polyubiquitin (Supplementary Fig. S2C). Scattered cytoplasmic colocalizations of ErbB2 with K63 and K48 polyubiquitin were weakly detectable 4 hours after PS341 treatment. By 24 hours, two significant differences were notable; some but not all of the total intracellular K48 and K63 polyubiquitin was redistributed and polarized in perinuclear cytoplasm, and internalized ErbB2 colocalized only with the perinuclear K48 and K63 polyubiquitin (Supplementary Fig. S2C). A comparable set of ErbB2, K48 and K63 polyubiquitin immunoblots and immunofluorescent imaging studies were performed on GA (20 nmol/L, 0–24 hours)–treated SK-BR-3 cells and showed dissimilar temporal and spatial effects on total intracellular K48 and K63 polyubiquitin relative to PS341-treated cells (Supplementary Fig. S3).

Although there is little understanding and still some controversy behind the conclusion that ErbB2 is an endocytosis-impaired receptor system (12, 13), interest in this aspect of ErbB2 biology has increased with the development of novel therapeutics that either modulate or use the ErbB2 endocytosis mechanism (1–4). Whereas GA inhibition of HSP90 has emerged as an invaluable tool for investigating mechanisms regulating the maintenance of ErbB2 surface expression and its endocytic downregulation (5–10, 12, 13), controversies arising from these studies as well as comparisons between ErbB2 and EGFR internalization mechanisms now include clathrin dependence or independence of ErbB2 internalization (9, 10, 12, 13, 32, 33), endocytic trafficking of ErbB2 to either proteasome or lysosome for receptor degradation (5, 9, 10), and the role of ErbB2 ubiquitination in mediating any of these processes (12–17). Given the extensive investigations into ErbB2 internalization activated by HSP90 inhibition, the present study attempted to cast new light on these processes by evaluating ErbB2 internalization and downregulation activated by proteasome inhibition using the approved therapeutic bortezomib (PS341).

Unlike the rapid ErbB2 downregulating effects of HSP90 inhibition by GA (50% by 6 hours), complete proteasome inhibition in SK-BR-3 by PS341 induced a more delayed exchange in the ErbB2-associated chaperones (HSP70 for HSP90), followed by surface-to-perinuclear redistribution of ErbB2 and a 50% reduction in total ErbB2 protein expression by 24 hours. This perinuclear sequestration of ErbB2 colocalized with lysosomal proteins and occurred by a clathrin-independent internalization process that required new protein synthesis, and as previously shown, chloroquine inhibition of lysosomal proteolysis prevented the PS341-induced degradation of ErbB2 but not its lysosomal sequestration (11). Whereas recent studies have been inconclusive about either the clathrin dependence or lysosomal trafficking of GA-induced ErbB2 endocytosis (9, 10, 32, 33), our SK-BR-3 studies indicate that GA-induced downregulation of ErbB2 may be clathrin dependent, is unaffected by chloroquine, and therefore unassociated with lysosomal trafficking. Furthermore, our SK-BR-3 imaging and molecular studies of full-length ErbB2 clearly indicate that proteasome and HSP90 inhibition result in different ErbB2 chaperone exchange and internalization rates, intracellular trafficking mechanisms, and degradative fates. Associated with these clear mechanistic differences in ErbB2 downregulation, we have also shown that PS341 and GA differentially affect ErbB2 receptor ubiquitination as well as its association with the E3 ligase, c-Cbl, and the deubiquitylating enzyme USP9x.

Ubiquitination, primarily in the form of K63-linked polyubquitin chains, is known to play an essential role in the endosomal trafficking and lysosomal degradation of EGFR (25), yet its potential role in mediating plasma membrane overexpression or endocytosis of ErbB2 remains obscure (13, 17). Transient transfection into SK-BR-3 of constructs expressing mutated ubiquitin unable to form any polyubiquitin chains or specifically unable to form K63 linked polyubiquitin prevented surface overexpression of ErbB2 and, in the former case, also prevented SK-BR-3 cell growth. Given the myriad intracellular mechanisms affected by polyubiquitination, it is likely that the observed abnormalities induced by these constructs reflect a general impairment in plasma membrane protein trafficking rather than direct alteration of ErbB2 polyubiquitination. Direct and indirect ErbB2 effects may also explain our observation that siRNA knockdown of the deubiquitylating enzyme USP9x significantly enhanced PS341-induced lysosomal decay of ErbB2, although the observed physical association between ErbB2 and USP9x/FAM in PS341-treated and untreated cells implicates direct mediation of this DUB in the ErbB2 internalization and ubiquitination induced by proteasome inhibition. Unlike this ErbB2 association with USP9x, neither of the two E3 ligases (CHIP or c-Cbl) were found to coprecipitate with overexpressed ErbB2 in untreated SK-BR-3 cells. However, within 6 hours of PS341 (but not GA) treatment the intact and internalized ErbB2 receptor became associated with c-Cbl (but not with CHIP) and remained so for at least 24 horus, during which time ErbB2 polyubiquitination topology was changing from predominantly K48 to K63 chain linkages whereas the COOH terminally intact receptor was trafficking into the lysosome compartment.

Because K48-linked polyubiquitin is required for 26S proteasome recognition and degradation (13, 14, 17), proteasome inhibition might be expected to result in a buildup of K48-linked polyubiquitinated proteins otherwise destined for proteasomal degradation. In contrast, the proteasome system is not inhibited in GA-treated cells, so an early buildup to a steady-state limit of K48 ErbB2 polyubiquitin might also be expected if K48 ErbB2 polyubiquitin was rapidly induced and then proteasomally degraded at a constant rate. It is interesting to note that the modest yet balanced buildup in both K48 and K63 ErbB2 polyubiquitin following HSP90 inhibition was not associated with any detectable lysosomal trafficking and proteolysis, suggesting that GA-induced ErbB2 polyubiquitin contains mixed chain linkages that lack the extended linear topology of K63 ErbB2 polyubiquitin essential for endosomal sorting to the MVB and lysosome. In contrast, the early buildup of K48 ErbB2 polyubiquitin followed by the later (>8 hours) buildup of K63 ErbB2 polyubiquitin, during proteasome inhibition and when ErbB2 is associated with c-Cbl, suggests a sequential polyubiquitin editing process rather than unbalanced mixed ubiquitin chain formation, superimposing K63 chain formation on a primary base of K48 linkages to enable lysosomal trafficking. Although CHIP has been shown capable of producing either K48, K63, or even mixed polyubiquitin chain linkages depending on its associated E2 enzyme (34), we were unable to detect any GA-induced ErbB2 association with CHIP in SK-BR-3 cells. In contrast, we found that upon exposure to PS341, ErbB2 associates with c-Cbl, and recent studies indicate that c-Cbl can also produce either K48 or K63 ubiquitin chains (35, 36), suggesting that, depending on its subcellular context and E2 partner, c-Cbl may also be able to form mixed polyubiquitin chain linkages and participate in ErbB2 polyubiquitin editing.

Further investigations are needed to understand the various mechanistic controls determining K48 and K63 ErbB2 polyubiquitination, as well as the roles these topologically distinct forms of ErbB2 polyubiquitin play in determining ErbB2 intracellular trafficking and degradation. Nonetheless, the translational importance of these processes is becoming clearer. Recently, it was shown that combined treatment with trastuzumab and an HSP90 inhibitor produces greater ErbB2 ubiquitination and degradation than can be achieved by either treatment alone (19). Therefore, ErbB2 intracellular trafficking mechanisms activated by either GA or PS341 treatment must be better delineated for optimal clinical development of the next generation of ErbB2-targeted therapeutics, which will also include drug-encapsulated anti-ErbB2 immunoliposomes (4) and antibody-drug conjugates like trastuzumab-DM1 (3) that depend upon ErbB2 internalization and endocytosis to deliver their cytotoxic cargos.

No potential conflicts of interest were disclosed.

We thank Danielle Crippen for fluorescence microscopy assistance.

Grant Support: NIH grants R01-CA36773 and P50-CA58207 (UCSF Breast Specialized Programs of Research Excellence) and Hazel P. Munroe memorial funding (Buck Institute). The Buck Mass Spectrometry/Chemistry and Morphology/Imaging Cores were partially supported by NIH Nathan Shock Center of Excellence grant P30 AG025708, NIH/NIA-P01-AG025901, NCRR-S10-RR002122 shared instrument, and NIH/NCRR-U54/Roadmap Interdisciplinary Research Consortium UL1-DE019608 grants. J.M. Held was supported by a pilot project award under the U54 grant RR024346. C. Marx was an AACR Scholar-in-Training (2008) and Minority Scholar (2005, 2006) awardee.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.