Proteasome-Regulated ERBB2 and Estrogen Receptor Pathways in Breast Cancer

ABSTRACT

A major challenge to broadening oncology applications for inhibitors of the ubiquitin-proteasome system (UPS) is the iden- tification of UPS-dependent cancer pathways predictive of tu- mors responsive to peptidomimetic inhibitors of its 20S core protease activity. To inform clinical studies evaluating UPS inhibitors as breast cancer therapeutics, seven phenotypically diverse human breast cancer cell line models were character- ized for their cellular and molecular responses to the clinically approved 20S inhibitor bortezomib (PS341; Velcade), focusing on those overexpressing estrogen receptor (ER) or ERBB2/ HER2, because these oncogenic receptor pathways are con- stitutively activated in ~80% of all breast cancers. All models demonstrated dose-dependent bortezomib reduction in intra- cellular 20S activity correlating with cell growth inhibition, and bortezomib IC50 values (concentrations producing 50% growth inhibition) varied directly with pretreatment 20S activities (r = 0.74; *, p < 0.05), suggesting that basal 20S activity may serve as a clinical predictor of tumor responsiveness to UPS inhibi- tion. Reduction in 20S activity (> 60%) was associated with early (24 h) intracellular relocalization of ER (nucleus to cytoplasm) and ERBB2 (plasma membrane to perinuclear lyso- somes), buildup of ubiquitinated and Hsp70-associated recep- tor, degradation and loss of ER and ERBB2 function, and induction of cellular apoptosis. These models were also used to screen a pharmacologic panel of pathway-targeted anti- cancer agents [4-hydroxy-3-methoxy-5-(benzothiazolyl- thiomethyl)benzylidenecyanoacetamide (AG825), 6- (4-bromo-2-chloro-phenylamino)-7-fluoro-3-methyl-3H- benzoimidazole-5-carboxylic acid (2-hydroxy-ethoxy)-amide (AZD6244/ARRY142886), 2-(4-morpholinyl)-8-phenyl-4H-1- benzopyran-4-one hydrochloride (LY294002), 17-N-allyla- mino-17-demethoxy geldanamycin (17AAG), and (2E)-N-hy- droxy-3-[4-[[(2-hydroxyethyl)[2-(1H-indol-3-yl)ethyl]a- mino]methyl]phenyl]-2-propenamide (LAQ824)] for those capable of sensitizing to bortezomib. In keeping with the observation that 20S reduction has little effect on mitogen- activated protein kinase kinase 1/2 (MEK1/2) signaling in either ER-positive or ERBB2-positive models, only the MEK- 1/2 inhibitor AZD6244 consistently improved the antitumor activity of bortezomib.

Recognizing the broad diversity of naturally occurring hu- man breast cancers, we evaluated a clinically representative panel of seven phenotypically diverse human breast cancer cell line models and characterized their cellular and molec- ular responses to bortezomib, focusing on estrogen receptor (ER, α isoform) and ERBB2/HER2 receptor pathways, be cause either of these pathways is constitutively activated in ~80% of all breast cancers. All cell lines in the panel dem- onstrated dose-dependent bortezomib reductions in intracel- lular 20S activity; and bortezomib IC50 (50% growth inhibi- tory concentration) values were found to correlate with pretreatment (basal) 20S proteasome activity. Downstream proteasome targets inhibited within 24 h of exposure to an IC50 bortezomib dose included ER and ERBB2 mechanisms in cell lines whose growth and survival are dependent on these receptor pathways. We were surprised to find that bortezomib seemed relatively inefficient at inhibiting MEK1/2 generation of phospho-ERK1/2(44/42) in both ER- positive and ERBB2-positive breast cancer models. To guide future clinical studies, a pharmacologic panel of diverse path- way inhibitors was tested against the ER-positive and ERBB2-positive breast cancer models to identify targeted therapeutics able to enhance the anticancer activity of bort- ezomib. In keeping with the observation that 20S reduction produced little effect on MEK1/2 signaling in these models, a specific MEK1/2 inhibitor (AZD6244/ARRY142886) proved most capable of increasing the antitumor activity of bort- ezomib.

Materials and Methods

Cell Lines, Reagents, and Cell Viability Assay. The human breast cancer cell lines MCF7, SKBr3, BT474, MDA-453, T47D, and MDA-231 were originally obtained from American Type Culture Collection (Manassas, VA) and were grown under American Type Culture Collection-recommended conditions: 37°C, 5% CO2, and in Dulbecco’s modified Eagle’s, McCoys 5A, or RPMI-1640 media, sup- plemented with 10% fetal bovine serum, 1% penicillin-streptomycin, and 10 µg/ml insulin. The MCF7/HER2 subline, overexpressing a constitutively hyperactive ERBB2 receptor kinase and possessing an altered endocrine phenotype, was developed as described previously (Benz et al., 1992). All media and supplements were purchased from Mediatech Inc. (Herndon, VA). Bortezomib was kindly provided by Millennium Pharmaceuticals (Cambridge, MA) and dissolved in di- methyl sulfoxide in a 10 µM stock stored in aliquots at —20°C. Sulforhodamine B reagent was purchased from Sigma-Aldrich (St. Louis, MO). The chemical inhibitor of ERBB2 kinase AG825, the PI3K inhibitor LY294002, and the Hsp90 inhibitor 17-N-allylamino- 17-demethoxy geldanamycin (17AAG) were purchased from Calbio- chem (San Diego, CA). AZD6244 was a kind gift of AstraZeneca (UK), and LAQ824 was a kind gift from Novartis Pharmaceuticals, Inc. (East Hanover, NJ). Treated and control cells were assayed for cell viability after plating in 12-well culture dishes at 20,000 cells per well and overnight attachment. Drug dissolved in media was added at the indicated concentrations and incubated at 37°C for 2 to 5 days; at the time of assay, 250 µl of ice-cold 50% trichloroacetic acid was added to each well and incubated at 4°C for 1 h. Plates were washed five times with dH2O before the addition of 500 µl of sulforhodamine B (SRB, 0.4% in 1% acetic acid) and 30-min incubation at room temperature. Plates were rinsed five times in 1% acetic acid; the colored product was solubilized in unbuffered Tris base (pH 10.5) and quantified by absorbance at 564 nm. The optical density reading of four replicates for each treatment was normalized for the mean value of untreated cells and expressed as percentage of control.

Human Breast Cancer Xenograft Model. Trastuzumab-resis- tant, ERBB2-positive B585 human breast cancer xenografts were grown in nude mice as described previously (Marx et al., 2006). Xenografts were serially passaged as subcutaneous tumors in the flanks of 4- to 6-week-old female nu/nu mice (Taconic Farms, Ger- mantown, NY). B585-bearing mice were given intraperitoneal injec- tions of vehicle or bortezomib (1 mg/kg) on days 15, 19, 22, 26, and 29; animal weights and three-dimensional tumor measurements were determined at least twice weekly, and time-dependent mean (± S.D.) tumor volumes were calculated and plotted. Additional B585-bearing mice were given a single i.p. injection of vehicle or 1 mg/kg PS-341 when B585 growth rates had achieved at least 300 mm3/day; these mice were sacrificed 24 h after treatment to measure single-dose drug effects on resected and snap-frozen (—80°C) B585 tumors. Fro- zen samples were pulverized into a fine powder under liquid nitrogen for further assay. For protein analysis, tumor powders (0.02 g per sample) were homogenized by sonication (550 Sonic Dismembrator; Fisher Scientific, Pittsburgh, PA) twice for 15 s each in 100 µl of ice-cold extraction buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 2 mM Na3VO4, 1% NP-40, 0.01% SDS, and 1% deoxycholate) with/ without a protease inhibitor cocktail (Mini Complete; Roche Diag- nostics, Mannheim, Germany).

20S Proteasome Activity Assay. Harvested cells or pulverized tumor tissues were lysed in buffer consisting of 50 mM HEPES, pH 7.5, 100 mM NaCl, 2 mM Na3VO4, 1% NP-40, 0.01% SDS, and 1% deoxycholate. Extracts were incubated on ice for 30 min and clarified by centrifugation for 15 min at 4°C. Protein concentration in the supernatant was determined using the Bradford assay (Bio-Rad Laboratories, Hercules, CA). Approximately 50 µg of total cell lysate was added to 0.5 mM Leu-Leu-Val-Tyr–amino-4-methylcoumarin (Chemicon, Temecula, CA) substrate and volumes were equalized in assay buffer (25 mM HEPES, pH 7.5, 5 mM EDTA, 0.5% NP-40, and 0.01% SDS). Assay mixture was prepared in a 96-well fluorometer plate and incubated for 2 h at 37°C. At the end of the incubation period, fluorescence was read using a 380/460 nm filter set in a fluorometer. A fluorescence standard curve with known dilutions of substrate was generated in parallel with the assay samples and results were expressed in arbitrary fluorescence units per microgram of protein in the assay sample.

Immunoprecipitation and Immunoblot Assays. For immuno- blotting, harvested cells or pulverized tumor tissues were lysed in modified RIPA buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 2 mM Na3VO4, 1% NP-40, 0.01% SDS, and 1% deoxycholate) containing a protease inhibitor cocktail (Mini Complete) and homogenized by sonication (550 Sonic Dismembrator) twice for 10 s each. Extracts were incubated on ice for 20 min and then clarified by centrifugation for 10 min at 4°C. Protein content of supernatants was determined by Bradford assay (Bio-Rad Laboratories). Lysate protein (25–30 µg) was heated to 95°C in 2× sample buffer (100 mM Tris, pH 6.8, 4% SDS, 20% glycerol, and 5% 2-mercaptoethanol) and electrophoresed in 4 to 12% Nu-Page Bis-Tris gradient gels (Invitrogen, Carlsbad, CA) with MOPS running buffer (Invitrogen). Separated proteins were transferred onto a polyvinylidene difluoride membrane (Milli- pore, Billerica, MA), blocked with 5% nonfat milk in phosphate- buffered saline containing 0.05% Tween 20, and probed sequentially with the antibodies listed below, stripping with Restore Western Blot Stripping Buffer (Pierce Biotechnology, Rockford, IL) between each primary antibody probe. The following antibodies were used: mouse monoclonal anti-ERBB2/HER2 (Calbiochem), anti-phospho-HER2 (Tyr1248), rabbit polyclonal anti-phospho-AKT (Ser473) and anti- AKT, anti-phospho-ERK1/2(44/42) and anti-ERK1/2(44/42), anti-cy- clin D1 (Cell Signaling Technology, Danvers, MA), mouse monoclo- nal anti-(p85)PARP (Promega, Madison, WI), mouse monoclonal anti-beta actin (Abcam Inc., Cambridge, MA) and anti-ERα (Santa Cruz Biotechnology, Santa Cruz, CA). Membranes were then incu- bated with horseradish peroxidase-linked secondary antibody (Bio- Rad), and signals were visualized using the enhanced chemilumines- cence detection system (Amersham, Piscataway, NJ). For immunoprecipitation, cells were treated as above and extracted in NP-40 buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 10 mM NaF, 1 mM sodium vanadate, and a protease inhibitor cocktail (Mini Complete) and homogenized by sonication. Protein aliquots of 750 µg were precleared with protein A-Sepharose beads and then incubated with 5 ug of mouse monoclonal anti ErbB2 antibody (Calbiochem) for 1 h at 4°C under continuous agitation. Immune complexes were recovered by adding 50 µl of Protein A Sepharose beads (Santa Cruz Biotechnology), washing three times in lysis buffer without NP-40 and resuspending in 50 µl of 2× Laemmli sample buffer before gel electrophoresis and immunoblotting as de- scribed above. For lysosomal protease inhibitor experiments, SKBr3 cells were pretreated with 100 µM chloroquine for 1 h before incu- bation with 10 nM bortezomib for 48 h. A second dose of chloroquine was added 24 h after the first dose. Cell lysates were prepared and analyzed by gel electrophoresis and immunoblotting as described above.

Immunofluorescence Imaging. Cells were seeded in eight- chamber slides (Lab-Tek II; Nalge Nunc International, Rochester, NY) until subconfluent and treated with bortezomib as described below. For SKBr3 and BT474 immunofluorescence experiments, cells were treated with bortezomib at 25 nM for 24 and 48 h before fixing in PFA. For experiments with MCF7 cells and detection of the estrogen receptor, cells were serum-starved for 24 or 48 h before treatment. Cells were then treated with 25 nM bortezomib for 24 h and stimulated with 10 nM estradiol for 20 min before fixing. Cells were then washed with PBS and fixed with 4% PFA for 10 min at room temperature. After permeabilization in 0.5% Triton X for 10 min, cells were blocked in 5% serum of secondary antibody host diluted in TBS for 30 min at room temperature and incubated in primary antibody diluted in 2.5% serum overnight. Primary antibod- ies used were mouse monoclonal anti-ERBB2/HER2 (Calbiochem) and anti-ER alpha (Santa Cruz Biotechnology). Secondary antibod- ies were donkey-anti mouse Alexa Fluor 488 and donkey anti-rabbit Alexa Fluor 555 (both from Invitrogen), diluted in 2.5% donkey serum and incubated for 30 min. Slides were mounted in Prolong Gold (Invitrogen) with DAPI and left overnight before fluorescent microscopic visualization and imaging.

NFnB DNA-Binding Assays. Quantitative p50 and p65 NFnB

DNA-binding was determined using an enzyme-linked immunosorbent assay-based Trans-AM assay in accordance with the manufac- turer’s instructions (Active Motif, Carlsbad, CA). In this commercial kit, a duplexed NFnB oligonucleotide, containing the B-cell n-en- hancer consensus binding sequence (5′-AGT TGA GGG GAC TTT CCC AGG C-3′) is attached to the surface of 96-well plates. Activated NFnB in tumor extracts that is first bound to the attached oligonu- cleotide is specifically and quantitatively detected by subsequent incubation with p50 or p65 specific antibody followed by an enzyme (horseradish peroxidase)-linked secondary for colorimetric (OD450 nm absorbance) scoring.

Estrogen Responsive Transcriptional Reporter Assay. Cul- tures were seeded 1 day before transfection with luciferase (luc) reporters to a density of 1 to 2 × 103 cells per well in 96-well microtiter plates and using appropriate growth media. Cells were transiently transfected with 0.5 µg of (ERE)3-tk-luc reporter plasmid (Promega) along with FuGene 6 transfection reagent (Roche). The Renilla reniformis luciferase vector pRL-tk-luc (Promega) was co- transfected to normalize for transfection efficiency. Culture media was changed 20 h after transfection, and cells were then treated with bortezomib (25 nM) for 24 h. Cells were subsequently washed with PBS, lysed for Dual-Glo luciferase assay (Promega), and reporter activity was measured by luminometer. Transfections were reported as -fold change in luciferase activity over vehicle-treated control cells.

Results

Bortezomib Reduction in Breast Cancer Growth (IC50 Values) and 20S Activity Correlated with Pre- treatment (Basal) 20S Proteasome Activity. Exposure (72 h) of the human breast cancer cell line panel to increasing concentrations of bortezomib resulted in dose- (Fig. 1A) and time-dependent (Fig. 1B) inhibition of cell growth, as as- sessed by sulforhodamine B viability, and revealed different sensitivities to this proteasome inhibitor. ERBB2-positive SKBr3 cells showed the greatest sensitivity and lowest bort- ezomib IC50 (4 nM for 72-h exposure); in contrast, ERBB2- positive BT474 cells showed significantly reduced sensitivity and higher bortezomib IC50 (21 nM for 72 h exposure). It is noteworthy that constitutive overexpression of ERBB2 in ER-positive MCF7, with its endocrine altering consequences (Benz et al., 1992), produced modest bortezomib-sensitizing effects in MCF7/HER2 relative to parental MCF7 cells (MCF7/HER2 IC50 = 14 nM; MCF7 IC50 = 23 nM). The degree of inhibition of endogenous 20S proteasome activity within 24 h of bortezomib treatment generally correlated with bortezomib induced growth inhibition (Fig. 1C); expo- sure to IC50 bortezomib doses invariably reduced intracellu- lar 20S proteasome activity to <40% of basal levels within 6 h. Although there was no apparent correlation between bortezomib sensitivity and breast cancer clinical phenotype (ERBB2 and ER status), the pretreatment (basal) level of 20S proteasome activity in these cells correlated (r = 0.74; *, p < 0.05) with bortezomib IC50 value (Fig. 1D), indicating that breast cancer cells with lowest basal 20S proteasome activity are most sensitive to growth inhibition by bortezomib.

Proteasome Inhibition Caused Destabilization and Lysosomal Decay of ERBB2 with Reduction in Phos- pho-AKT and Induction of Apoptosis. Despite their dif- ferential sensitivity to bortezomib, the ERBB2-positive SKBr3 and BT474 cells showed similar intracellular re- sponses to growth-arresting and apoptosis-inducing doses of bortezomib, suggesting that the ERBB2-dependent growth and survival mechanisms in these two cell lines are similarly dependent on the UPS. After 48-h treatment of the more sensitive SKBr3 cells with 5 nM bortezomib, apoptosis was induced as indicated by an increase in the caspase-3 medi- ated PARP1 cleavage product, p85 PARP; this was associated with loss of phospho(Y1248)-ERBB2 and reduced phospho- AKT (Fig. 2A) levels. For the less sensitive BT474 cells, this same response required a 25 nM dose of bortezomib. We were surprised to find that phospho-ERK1/2 levels were not sig- nificantly altered in either cell line, indicating that an ERBB2 receptor-inhibiting and apoptosis-inducing dose of bortezomib does not impair activity of the mitogen activated protein kinase kinases, MEK1/2.

Because the Hsp90 chaperone protein is known to stabilize the active conformation of receptors such as ERBB2 and their downstream signaling effectors such as AKT (Xu et al., 2001; Calderwood et al., 2006), the early effect of bortezomib treat- ment on ERBB2 chaperone binding was evaluated by immu- noprecipitation and immunoblotting. Within a 24-h bort- ezomib treatment (10 nM) of either SKBr3 or BT474 cells, the association of Hsp90 with endogenous ERBB2 receptor was lost and replaced by another coprecipitating chaperone pro- tein, Hsp70 (Fig. 2B). This altered chaperone binding oc- curred before significant loss in receptor content (Fig. 2A) but in conjunction with a buildup in ubiquitinated ERBB2 pro- tein, detected in the same immunoprecipitates (data not shown).

At 24 h after bortezomib treatment of SKBr3 cells, immu- nofluorescent microscopic imaging was used to monitor the intracellular localization of the destabilized ERBB2 recep- tors. Although vehicle-treated cells showed only plasma membrane-localized ERBB2 receptor, chaperone-altering doses of bortezomib induced complete relocalization of ERBB2 within 24 to 48 h into a perinuclear cytoplasmic compartment (Fig. 2C) that colocalized with both immunore- active lysosomal protein, LAMP1, and the endoplasmic retic- ulum marker, calreticulin (results not shown). A similar re- sult was obtained with BT474 cells after 48 h of bortezomib treatment (results not shown). Cotreatment of SKBr3 cells with a nontoxic dose of the lysosomal processing inhibitor chloroquine prevented bortezomib-induced ERBB2 decay in these cells (Fig. 2D).

Phospho-AKT Reduction Was Associated with In Vivo Inhibition of 20S Proteasome Activity and Re- duced ERBB2-Positive Breast Cancer Growth. The trastuzumab-resistant, ERBB2-positive B585 human breast cancer xenograft model was used to determine the relative sensitivity of intratumor NFnB DNA-binding, phospho-ERK1/2 and phospho-AKT levels in response to bortezomib treatment sufficient to reduce 20S proteasome activity and B585 growth in vivo. With maximally tolerated dosing of bortezomib (twice weekly 1 mg/kg i.p. injections ×4 into tumor-bearing nude mice) nude mouse tumor volumes showed a modest growth inhibiting treatment effect, with mean tumor volumes determined after the second, third, and fourth injections (days 22, 26, and 29) reduced no more than 30% relative to vehicle-treated control mice (Fig. 3A). A par- allel set of mice bearing palpable tumors growing at >300 mm3/day were treated with a single i.p. injection of either vehicle or bortezomib (1 mg/kg), and their tumors were resected 24 h later to assess intratumor 20S proteasome activ- ity, NFnB (p50 and p65) DNA-binding, phospho-ERK1/2, and phospho-AKT levels. Tumors treated with a single injection of the growth-inhibiting bortezomib dose showed a mean 40% reduction in intratumor 20S proteasome activity relative to the vehicle-treated tumors (Fig. 3B). At this level of protea- some inhibition, tumor NFnB p50 and p65 DNA-binding activities were unaffected, as were tumor phospho-ERK1/2 lev- els (Fig. 3, C and D). In contrast, tumor phospho-AKT levels were reduced to 50% of control tumor levels (Fig. 3D).
Proteasome Inhibition Causes ER Relocalization and Decay with Loss of Transcriptional Function and Induction of Apoptosis. Treatment of ER-positive

MCF7 cells with growth inhibitory doses of bortezomib caused a marked dose-dependent reduction in ER content by 48 h (Fig. 4A); this was accompanied by loss of the proliferation marker cyclin D1, increase in the apoptosis marker p85 PARP, and a prominent increase in the proapoptotic and mitotic check- point protein p53. However, significant changes in phospho- AKT or phospho-ERK1/2 levels were not detected (data not shown). To assess early effects of proteasome inhibition on ER function, MCF7 cells were transiently transfected with a luciferase reporter driven by an ERE-regulated promoter (ERE3-TK-Luc); and as control for transfection efficiency and overall cell viability, a parallel set of MCF7 cells were trans- fected with an R. reniformis luciferase vector (TK-Luc). After transient transfection (20 h), MCF7 treated with bortezomib (25 nM) for 24 h showed a 40% specific reduction in ER transcriptional activity (ERE3-TK-Luc activity), relative to vehicle-treated cells or bortezomib-treated MCF7 transfected with the TK-Luc vector (Fig. 4B). This early loss of ER tran- scriptional activity preceding bortezomib-induced loss of ER content suggested the possibility of a functional impairment in ER localization, as reported for other hormone receptors after proteasome inhibition (Shenoy et al., 2001; Yu and Malek, 2001; Lin et al., 2002). Immunofluorescence imaging for ER localization was performed on MCF7 cells treated with vehicle or bortezomib (25 nM, 24 h). Whereas ER colo- calized with DAPI nuclear staining in control cells, all bort- ezomib-treated cells showed cytoplasmic localization of ER (Fig. 4C), indicating that bortezomib altered ER trafficking by inhibiting either cytoplasm-to-nucleus translocation or inducing nucleus-to-cytoplasm retrotranslocation of ER.

Pharmacologic Targeting of Pathways in Combina- tion with Proteasome Inhibition. To identify breast can- cer targets that might complement proteasome inhibition, the panel of cell lines was treated for 48 h with IC50 bort- ezomib doses in combination with the following target inhib- iting drugs and doses: histone deacetylases (LAQ824, 10 nM), Hsp90 (17AAG, 50 nM), ERBB2 kinase (AG825, 25 µM), PI3K (LY294002, 10 µM), or MEK1/2 (AZD6244, 100 nM).
The resulting growth inhibitory interactions were scored as antagonistic, additive, or supra-additive; however, no supra- additive bortezomib combinations were identified. Against ER-positive breast cancer cells (MCF7, BT474), only bort- ezomib combinations with LAQ824 or AZD6244 were addi- tive. Against ERBB2-positive breast cancer lines (BT474, SKBr3), only bortezomib combinations with AG825 or AZD6244 were consistently additive. Against ER-negative cell lines (MDA231, SKBr3), only bortezomib in combination with AZD6244 was consistently additive. These drug inter- actions are summarized in Table 1; the cell viability and growth effects of these representative drug-bortezomib inter- actions are demonstrated in Fig. 5. It is noteworthy that near-additive interactions were observed in every cell line when the MEK1/2 inhibitor AZD6244 (100 nM) was used in combination with an IC50 bortezomib dose.

Discussion

Codony-Servat et al. (2006) studied a panel of human breast cancer cell lines and measured a variety of cellular and molecular consequences after bortezomib treatment but were unable to identify any mechanism-based predictors of bortezomib responsiveness or pathway inhibitors capable of enhancing bortezomib sensitivity. To explore downstream markers of proteasome inhibition and pathways potentially targeted in combination with proteasome inhibitors, we stud- ied a panel of seven human breast cancer cell line models representing clinically diverse phenotypes and expressing a wide range of endogenous 20S proteasome activities. All models demonstrated dose-dependent bortezomib reduction in intracellular 20S activity correlating with cell growth in- hibition; bortezomib IC50 values varied directly with pretreatment 20S activities (r = 0.74, *, p < 0.05), suggesting that basal 20S activity may serve as a clinical predictor of tumor responsiveness to UPS inhibition.

These breast cancer models also indicated that 6- to 24-h exposure to an IC50 bortezomib dose invariably reduced in- tracellular 20S activity by >60% from pretreatment levels.In contrast, the B585 xenograft study showed that maximally tolerated dosing of bortezomib produced only a 40% reduction in B585 20S proteasome activity. Although this was associ- ated with modest in vivo inhibition of B585 growth, the 40% reduction in B585 20S activity suggests that in vivo bort- ezomib bioavailability may be limiting for some solid tumors, insufficient to affect all the cancer pathways affected by an in vitro IC50 dose of bortezomib. However, these B585 xenograft studies confirmed that for ERBB2-positive breast cancers, reduction in phospho-AKT is a more sensitive indicator of 20S inhibition than either NFnB or MEK1/2 activity. With the clinical introduction of newer and more broadly acting
proteasome inhibitors such as NPI-0052 (Cusack et al., 2006; Joazeiro et al., 2006), it is reasonable to expect improved solid tumor bioavailability and greater inhibition of 20S pro- teasome activity, with resulting impairment of ERBB2 and ER breast cancer pathways at least comparable with those observed by in vitro treatment of our cell line models with IC50 bortezomib doses.

With >60% reduction in 20S proteasome activity, bort- ezomib inhibited cell growth by inducing cellular apoptosis (measured by increases in the caspase-3 mediated PARP1 cleavage product, p85), mediated in ER-positive and ERBB2- positive breast cancer models by different proteasome-depen- dent mechanisms. In the ER-positive MCF7 cell line model, down-regulation of ER content and transcriptional activity by bortezomib were associated with up-regulation of the tumor suppressing protein, p53 (wild type), which undoubtedly contributed to the inhibition of cyclin D1 levels and cell cycle progression as well as induction of apoptosis detected by the caspase-3 mediated PARP1 cleavage product, p85. Within 24 h of MCF7 treatment, bortezomib induced a cytosolic buildup of ubiquitinated and Hsp70-associated undegraded ER (data not shown); these findings have been described previously and attributed to the critical function of the pro- teasome in maintaining nuclear receptor turnover and tran- scriptional activity (Lonard et al., 2000, Reid et al., 2003). Not previously reported is the observed bortezomib induced early relocalization of ER from nuclear to cytoplasmic com- partments. Lin et al. (2002) found a similar response for androgen receptor (AR) after cellular treatment with the proteasome inhibitor MG132; in that study, proteasome in- hibition for >24 h suppressed AR nuclear translocation by 50%, disrupted AR interactions with its coregulators, and reduced total AR content of prostate cancer cells. Further studies are needed to understand the proteasome-dependent mechanisms regulating nuclear receptor trafficking.

Unlike the cell growth and survival mechanisms activated by overexpressed ER in MCF7 cells, the constitutive overex- pression of ERBB2 receptor tyrosine kinase activity drives growth and survival of SKBR3 and BT474 breast cancer cells primarily by heterodimerization with and phosphorylation of the ERBB3 receptor, which serves to activate the down- stream PI3K/AKT pathway. Within 48 h of a 20S reducing dose of bortezomib, surface membrane localized ERBB2 was lost in association with reduced phospho-AKT levels and the induction of cellular apoptosis in SKBr3 and BT474 cells but without any detectable change in MEK1/2 activity as mea- sured by phospho-ERK1/2 levels. The novel finding that 20S proteasome inhibition causes dissociation of the Hsp90 chap- erone protein from ERBB2 receptor is similar to the reported ERBB2 effect induced by an Hsp90-inhibiting dose of benzo- quinone ansamycins such as geldanamycin or its clinical analog 17AAG (Xu et al., 2001). Treatment of ERBB2-posi- tive cells with bortezomib, as with ansamycins, seems to shift ERBB2 chaperone association from one that is stabilizing (Hsp90) to one that is destabilizing (Hsp70). In ansamycin- treated cells, dissociation of ERBB2 from Hsp90 is followed by proteasomal degradation of ERBB2; however, in cells treated with the proteasome inhibitor bortezomib, the loss of ERBB2 content must occur by a different mechanism. One contributing possibility is the recently described repression of ERBB2 transcript levels caused by various proteasome inhibitors (Marx et al., 2006). A more likely explanation for the rapid loss of total ERBB2 content is that after protea- some inhibition, the destabilized and ubiquitin-tagged ERBB2 is sequestered within perinuclear aggresomes, where it is degraded by lysosomal proteases, consistent with the observed prevention of ERBB2 degradation by the lysosome inhibitor chloroquine. Microtubule-mediated transport and accumulation of polyubiquitinated proteins into aggresomes has recently been reported (Johnston et al., 1998; Nawrocki et al., 2006). However, further study of proteasome-depen- dent ERBB2 trafficking mechanisms is needed, because the observed perinuclear sequestration of ERBB2 could also have resulted from bortezomib-induced disruption of normal ERBB2 endocytotic mechanisms (Austin et al., 2004).

Approximately 80% of human breast cancers overexpress ER and/or ERBB2 receptors. Despite available therapeutics that specifically target these overexpressed receptors, clini- cal resistance to these target-specific agents is common. Therefore, there is both clinical need and interest in applying proteasome inhibitors to improve the treatment of breast cancer (Cardoso et al., 2004; Dees and Orlowski, 2006). How- ever, because of the diversity of breast cancer phenotypes, it is also appreciated that rational combination of proteasome inhibitors with other targeted therapeutics must be guided by informative preclinical studies (Yang et al., 2006). Based on the above observations that 20S proteasome inhibition can be cytotoxic to ER- and ERBB2-overexpressing breast cancer cells by different mechanisms, our cell line models were used to screen a pharmacologic panel of diverse pathway inhibi- tors to identify targeted agents capable of sensitizing to bort- ezomib. The five targeted agents shown in Table 1 (AG825, AZD6244, LY294002, 17AAG, LAQ824) were tested at doses previously shown to inhibit their pathways, although when used as single agents, these inhibitors produced quite vari- able cell line-dependent growth inhibition. The Hsp90 inhib- itor 17AAG proved antagonistic in combination with bort- ezomib against all models tested; this finding is in contrast to the ansamycin result reported by Mimnaugh et al. (2004), although it is consistent with the fact that Hsp90 inhibitors induce proteasomal decay of Hsp90 client proteins (e.g., ER, ERBB2, AKT), which is not possible in the presence of effec- tive 20S inhibition. The ERRB2 kinase inhibitor AG825 and the PI3K inhibitor LY294002 enhanced bortezomib cytotox- icity but only in ERBB2-positive cells in which these target kinases were constitutively activated. The histone deacety- lase (HDAC) inhibitor LAQ824 was antagonistic to bort- ezomib only in the same ERBB2-positive breast cancer cell lines previously shown to be most sensitive to this and other HDAC inhibitors (Scott et al., 2002), suggesting that HDAC and proteasome mechanisms have overlapping and non- complementary growth-regulating roles in ERBB2-positive but not ERBB2-negative malignant cell lines. This possibility deserves further study given the report that the shuttling of misfolded proteins into the aggressome, a potential conse- quence of proteasome inhibition, is HDAC-dependent (Kawaguchi et al., 2003). Unlike the other targeted agents evaluated, only the MEK1/2 inhibitor AZD6244 proved capa- ble of additively enhancing the growth-inhibiting effect of bortezomib in all breast cancer models studied. Because bort- ezomib was relatively inefficient at inhibiting MEK1/2 in both ER-positive and ERBB2-positive breast cancer cells, and a MEK1/2 inhibiting dose of AZD6244 (100 nM), by itself, failed to significantly inhibit the growth of three of the breast cancer cell lines (SKBr3, BT474, and MCF7), the observed additive interaction between AZD6244 and bortezomib sug- gests that each agent is acting on independent but comple- mentary growth regulating mechanisms. Thus, compared with the other targeted agents evaluated in these models, AZD6244 would seem to be the most promising agent for further preclinical evaluation in combination with a 20S inhibiting dose of bortezomib or newer generation protea- some inhibitor.