© The Author 2006. Published by Oxford University Press.
ARTICLE |
Cyclin D1 Overexpression and Response to Bortezomib Treatment in a Breast Cancer Model
Affiliations of authors: Division of Hematology/Oncology, Department of Medicine, Mount Sinai School of Medicine, New York, NY (YI, SW, DG); Peter MacCallum Cancer Centre, Trescowthick Research Laboratories, East Melbourne, Australia (AP); Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA (JVA, DAF); Departments of Physiology and Biophysics (IK, DL), Biomathematical Sciences (JM), and Oncological Sciences (MJO), Mount Sinai School of Medicine, New York, NY
Correspondence to: Doris Germain, PhD, Division of Hematology/Oncology, Department of Medicine, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1178, New York, NY 10029 (e-mail: doris.germain{at}mssm.edu).
 |
ABSTRACT |
|---|
 TopNotes Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Background: Cyclin D1 is frequently overexpressed in breast cancer, and its overexpression is, surprisingly, associated with improved survival. One potential mechanism for this association involves signal transducer and activator of transcription 3 (STAT3). Methods: Cyclin D1 and STAT3 expression were assessed in human tumors using microarray analysis and in breast cancer cell lines HBL100, T47D, MCF7, MDA-MB-453, and BT20 and in HBL100 and T47D cells stably overexpressing cyclin D1 using immunoblot analysis. Cyclin D1 protein was stabilized by treatment with the proteasome inhibitor bortezomib, and the effects on STAT3 expression in vitro was determined by using immunoblotting and on xenograft tumor growth and apoptosis in vivo was determined by using terminal deoxyuridine nick-end labeling assays. All statistical tests were two-sided. Results: Tumors with high cyclin D1 expression (n = 17) had low STAT3 expression (mean = 274 arbitrary units), and those with low cyclin D1 expression (n = 31) had high STAT3 expression (mean = 882 arbitrary units) (P<.001). In HBL100 and T47D parental and cyclin D1–overexpressing cells, cyclin D1 overexpression was also inversely associated with STAT3 expression, and cyclin D1 directly reduced the expression of STAT3. Stabilization of cyclin D1 protein by bortezomib treatment further amplified the cyclin D1–dependent repression of STAT3 in vitro and slowed tumor growth in vivo (week 7: untreated mean = 185.7 mm3 versus treated mean = 136.2 mm3, difference = 49.5 mm3, 95% confidence interval [CI] = 18 to 81 mm3, P = .007; week 8: untreated mean = 240.2 mm3 versus treated mean = 157.3 mm3, difference = 82.9 mm3, 95% CI = 9.1 to 156.7 mm3, P = .0014; and week 9: untreated mean = 256.4 mm3 versus treated mean = 170.2 mm3, difference = 86.2 mm3, 95% CI = 22.8 to 149.6 mm3, P = .006) and increased apoptosis (untreated mean = 19% versus treated mean = 54%, difference = 35%, 95% CI = 24.7% to 45.4%; P = .013) of xenograft tumors. Conclusions: Cyclin D1 repression of STAT3 expression may explain the association between cyclin D1 overexpression and improved outcome in breast cancer. In addition, bortezomib can amplify the proapoptotic function of cyclin D1, raising the possibility that cyclin D1 levels may be a marker for predicting the response to this novel drug.
|
INTRODUCTION |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Cyclin-dependent kinases (cdks) are key cell cycle regulators, and their activities are modulated by the binding to cyclins (1). Binding of cyclin D to cdk4 and cdk6 leads to the phosphorylation of the retinoblastoma protein (Rb). Phosphorylation of Rb prevents it from repressing the E2F family of transcription factors and leads to the transcription of several genes required for the G1-to-S phase transition, thereby promoting cellular proliferation (2).
Cyclin D1 is overexpressed in 35%–50% of breast cancers (3). Because of its pivotal role in promoting cell cycle progression and, hence, cell division and proliferation, such overexpression might be expected to coincide with poor prognosis. Although some data support this expectation (4), other studies (5–9) have reported a positive association between cyclin D1 overexpression and breast cancer survival. For example, a study using microarray analysis of cyclin D1 expression showed that high cyclin D1 expression was associated with low risk of local recurrence of breast cancer, whereas low expression was associated with high risk (5). This association was also found in advanced metastatic breast cancer; in studies of such patients, cyclin D1 overexpression was associated with increased relapse-free (6) and overall (7–8) survival. In addition, low cyclin D1 expression was associated with poor prognosis of African American women with breast cancer (9).
One possible explanation for these observations is that cyclin D1 overexpression increases cell proliferation and thereby increases sensitivity to chemotherapy. However, because the overexpression of cyclin E, another cyclin that is involved in the G1-to-S phase transition, also promotes increased proliferation but is associated with poor prognosis (10–12), increased proliferation alone cannot explain the association between high cyclin D1 levels and good prognosis. In addition, the beneficial effect of cyclin D1 was observed in patients who had received breast-conserving surgery without adjuvant radiotherapy or chemotherapy (5). These observations suggest that cyclin D1 overexpression activates other, perhaps cdk-independent, cellular functions that ultimately decrease the aggressiveness of the tumor.
One possible cdk-independent function of cyclin D1 that would allow it to have antitumorigenic activity is its binding to the antiapoptotic transcription factor signal transducer and activator of transcription 3 (STAT3) (13). Accumulating data suggest that the abnormal activation of STAT3 has a critical role in oncogenesis (14). STAT3 is overexpressed in many types of cancer, including breast cancer, and is a mediator of survival through its ability to promote the transcription of the apoptosis suppressor Bcl-xL (14). Cyclin D1 represses STAT3 transcriptional activity in vitro (13), therefore suggesting that cyclin D1–dependent repression of STAT3 may induce apoptosis. However, because cyclin D1 protein has a very short half-life due to its degradation by the 26S proteasome (15–17), its ability to repress STAT3 may be limited. Therefore, blocking cyclin D1 degradation using a 26S proteasome inhibitor may amplify its ability to repress STAT3 and induce apoptosis.
We initiated this study to determine the association between STAT3 and cyclin D1 expression levels using human cancer samples and breast cancer cell lines MCF7, MDA-MB-453, ZR75.1, and BT20. We also engineered genetically matched HBL100 and T47D cell lines with low or high levels of cyclin D1 to directly compare the level of STAT3 in these cells and used this model to test the effect of the proteasome inhibitor bortezomib both in vitro and in vivo.
|
MATERIALS AND METHODS |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Cyclin D1 and STAT3 Expression in Human Tumors
To determine whether cyclin D1 and STAT3 expression levels are associated in human tumors, we analyzed a previously published data set containing gene expression data for 218 tumors of diverse types (18).
The association between cyclin D1 expression and STAT3 activation was measured by analyzing a cohort of 96 breast tumors for which STAT3 activation (phosphorylation) had previously been determined by in situ staining (19). The expression of cyclin D1 was compared between tumors with high levels of STAT3 activation and tumors with no STAT3 activation using the signal-to-noise score, as described in the "Statistical Methods" section.
Cell Culture and Transfections
T47D, ZR75.1, and HBL100 breast cancer cells (American Type Culture Collection [ATCC], Manassas, VA) were grown in RPMI medium supplemented with 10% fetal calf serum, insulin (100 IU/mL), hydrocortisone (0.5 mg/mL), and antibiotics (Life Technologies, Inc, Grand Island, NY). MCF7, MDA-MB-453, and BT20 breast cancer cells (ATCC) were grown in DMEM medium supplemented with 10% fetal calf serum and antibiotics (Life Technologies, Inc). The cDNA encoding the cyclin D1 gene in which the influenza hemagglutin (HA) tag had been inserted (cyclin D1HA) was cloned in the pCDNA3 expression vector (Invitrogen, Grand Island, NY). HBL100 and T47D cells were transfected with cyclin D1HA-pcDNA3, a mutant of cyclin D1 that cannot bind to cdk4 (cyclin D1-KE-pCDNA3) (15) or STAT3-pCDNA3 plasmids by using the FuGENE 6 system (Boehringer Mannheim, Mannheim, Germany), according to the manufacturer's instructions. Clones stably expressing cyclin D1HA, cyclin D1-KE, or empty vector were selected. Briefly, following transfections, cells were incubated in the presence of G418 at 0.78 mg/mL. After 2 weeks of selection, surviving colonies, i.e., those arising from stably transfected cells, were picked and individually amplified. HBL100 and T47D clones that stably overexpressed cyclin D1-HA and cyclin D1-KE were designated HBL100-D1, T47D-D1, and T47D-D1-KE.
T47D
MT cells were generated as described above by stable transfection of a plasmid expressing cyclin D1 under the control of a zinc inducible promoter as previously described (20).
Colony Formation
HBL100 and HBL100-D1 cells (1 x 104) were plated in 500 µL DMEM with 5% fetal bovine serum (FBS) and insulin (5 µg/mL) in four-well plates that had been precoated with 250 µL of matrigel (BD Biosciences, San Jose, CA). The medium was replaced every 2–3 days. Results of colony formation shown are representative cells, which were cultured for 10 days.
Immunoblot Analysis
HBL100, HBL100-D1, T47D, T47D-D1, T47D-D1-KE, MCF7, ZR75.1, BT20, and MDA-MB-453 cells were washed three times in ice-cold phosphate-buffered saline (PBS) (10 mM sodium phosphate, 120 mM sodium chloride, 2.7 nM potassium chloride) and lysed in 200 µL of ice-cold lysis buffer (50 mM Tris pH 7.5, 250 mM NaCl, 5 mM EDTA, 0.5% NP-40, 50 mM NaF, 0.2 mM Na3VO4, leupeptin at 1 g/mL, pepstatin at 1 g/mL, phenylmethylsulfonyl fluoride at 100 g/mL, and 1 mM dithiothreitol). Lysates were centrifuged at 10 000g for 20 minutes at 4 °C, and the protein concentration of the supernatant was determined using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA). Proteins (15 µg) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on 10% acrylamide gels and transferred to nitrocellulose membranes (PerkinElmer Life Sciences, Sheldon, CT). Membranes were incubated with rabbit polyclonal anti–cyclin D1 antibody (1:500; Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti-STAT3 antibody (1:1000; Zymed, San Francisco, CA), rabbit polyclonal anti–phospho-STAT3 antibody (1:500, Santa Cruz Biotechnology), mouse monoclonal anti–Bcl-xL antibody (1:1000; PharMingen, San Diego, CA), mouse monoclonal anti–caspase-3 antibody (1:1000; BD Transduction Laboratories, San Jose, CA), mouse monoclonal anti–caspase-8 antibody (1:1000; BD Transduction Laboratories), rabbit polyclonal anti-Bid antibody (1:500; Cell Signaling, Danvers, MA), mouse monoclonal anti-tubulin antibody (1:2000; Hybridoma Facility, University of Iowa, Iowa City, IA), or mouse monoclonal anti-ubiquitin antibody (1:1000; Sigma, St-Louis, MT), and antigen–antibody complexes were visualized using the ECL kit (Amersham Pharmacia Biotech, Little Chalfont, England). All experiments were performed at least twice. The intensity of the bands was quantified using a Bio-Rad GS-800 densitometer equipped with the Quantity One program (Bio-Rad).
Detection of Ubiquitinated Cyclin D1
HBL100 and HBL100-D1 cells were transiently transfected with Myc-tagged ubiquitin-pCDNA3 plasmid for 24 hours. Bortezomib (stock solution = 2.6 µM) (Millennium Pharmaceuticals, Cambridge, MA) was then added to the media for an additional 24 hours before protein extraction, as above. Cyclin D1 was immunoprecipitated using the rabbit polyclonal anti–cyclin D1 antibody (Santa Cruz Biotechnology) according to the manufacturer's instructions followed by western analysis with mouse monoclonal anti-Myc antibody 9E10 (1:100, in-house hybridoma facility, Mount Sinai School of Medicine, New York, NY) to detect ubiquitinated Myc proteins.
Immunofluorescence of Caspase-3
HBL100 and HBL100-D1 cells (250 µL of 1 x 104 cells/mL) in DMEM with 5% FBS and insulin (5 µg/mL) were plated in chamber slides precoated with 100 µL of matrigel (BD Biosciences) and cultured for 1 week. Cells were fixed in 2% paraformaldehyde–PBS for 20 minutes at room temperature and then permeabilized using PBS/0.5% TritonX-100 for 10 minutes at 4 °C. The cells were rinsed three times with PBS/7.5% glycine for 10 minutes and incubated with 10% BSA/IF buffer (PBS containing 0.1% BSA, 0.2% TritonX-100, 0.04% Tween-20, and 10% bovine serum albumin) for 1 hour. To detect active caspase-3, cells were incubated with rabbit polyclonal anti-cleaved caspase-3 (Asp175) (1:100, Cell Signaling) overnight at 4 °C. After washing three times with IF buffer for 20 minutes, cells were incubated with goat anti-rabbit Alexa Fluor 488–conjugated secondary antibody (1:200, Molecular Probes, Eugene, OR) for 1 hour at room temperature and then washed with IF buffer and rinsed with PBS.
Calcium Measurement
HBL100 and HBL100-D1 cells were placed on coverslips treated with poly-L-Lysine. Cells were incubated with 1 µM Fura 2AM (TEF Labs, Austin, TX) for 1 hour at 37 °C and then treated with 15 nM bortezomib. Immediately after treatment, cells were excited at 340 and 380 nm, and calcium fluorescence was monitored using a 535 ± 15 nm emission filter. A dual emission photometric system and a polychrome IV (TILL Photonics, Planegg, Germany) were used to perform the measurements.
Methylthiazoletetrazolium Assay of Cell Proliferation
To determine the percentage of cell survival, HBL100, HBL100-D1, T47D, T47D-D1, T47D-D1-KE, MCF7, ZR75.1, BT20, and MDA-MB-453 cells were seeded at 3 x 104 cells per milliliter in 24-well plates and then treated for 3 days with bortezomib (2.6 µM stock solution was used at concentrations up to 20 nM). Methylthiazoletetrazolium (MTT, Sigma) solution (50 µL of a 5 mg/mL solution in PBS; 10 mM sodium phosphate, 120 mM sodium chloride, 2.7 nM potassium chloride) was added to each well, and the cells were incubated for 4 hours at 37 °C. The medium was then aspirated, the cells were lysed in 200 µL of dimethyl sulfoxide per well and transferred to 96-well plates, and the absorption at 570 nm was determined by Universal Micro plate Reader ELX800 (BIOTEX Instrument Inc, Winooski, VT). The percentage of cell survival was evaluated relative to that of untreated cells. Ninety-five percent confidence intervals (CIs) were estimated from data arising from three or four individual experiments. Percentages of survival of T47D, T47D-D1, HBL-100, and HBL100-D1 cells were also assayed after 3 days of treatment with methotrexate (20 mM stock solution; Sigma) at 60 µM and taxol (100 µM stock solution; Sigma) at 300 nM.
Xenograft Implantation and Measurement of Tumor Size
Eight-week-old BALB/c nude mice were purchased from the Animal Research Center, Perth, Australia. A pellet of 17-
-estradiol (0.72 mg/mL) was inserted subcutaneously in the upper back of each mouse. To insert the estradiol pellet, mice were anesthetized using ketamine and xylazine at a dose of 0.1 mg/kg body weight. A small incision was made and the pellet inserted using a precision trochar. The incision was sealed using a clip. T47D or D1 cell pellets (5 x 106 cells) were mixed with an equal volume of matrigel (Basement Membrane Matrix, BD Biosciences). The mixture was injected subcutaneously in the lower back of each animal using a 26-gauge needle 1 week after the insertion of the estrogen pellet. Our protocol was approved by the Animal Ethics Committee at the Peter MacCallum Cancer Institute, and mice were cared for according to institutional guidelines.
Tumor size was measured using a digital caliper. Two independent measurements (length and width) were taken for each tumor weekly, and their resulting average was used to obtain tumor volume. Mice were injected in the tail vein twice weekly with bortezomib (0.05 mg/kg body weight) that had been lyophilized and reconstituted in sterile saline at a concentration of 1 mg/mL.
Terminal Deoxyuridine Nick-End Labeling (TUNEL) Assay
To determine the percentage of cells from xenograft tumors derived from T47D and T47D-D1 cells, above, that underwent apoptosis, paraffin-embedded sections were stained using the apoptosis detection kit TACS-KL Basic (TA100, R&D systems Inc, Minneapolis, MN) according to the manufacturer's protocol. Sections were then counterstained with 0.2% hematoxylin to detect intact nuclei.
Statistical Methods
Tumors were ranked according to expression of cyclin D1, and STAT3 expression was compared between tumors with high cyclin D1 expression (expression > 1 standard deviation above the mean) and tumors with low cyclin D1 expression (expression < 1 standard deviation below the mean) using the signal-to-noise score (20). The statistical significance of the differences of the signal-to-noise scores of the low and high cyclin D1 samples was estimated using permutation testing with GeneCluster 2.0 (21).
The experimental design of the in vivo studies is a 2 x 2 factorial design (T47D versus T47D-D1 tumor cells and control versus bortezomib treatments) with weekly tumor measurements. Repeated measures analysis of variance for a 2 x 2 factorial design was used, starting at week 6, to compare tumor volumes over time for the groups. Because the distributions of tumor volumes were skewed, log transformation was used. To be conservative, the Huynh-Feldt (1976) adjustment to numerator and denominator degrees of freedom for the F tests was used. All statistical tests were two-sided; P<.05 was considered to be statistically significant.
|
RESULTS |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Cyclin D1 and STAT3 Expression in Human Cancers
We used two approaches to determine whether an association exists between cyclin D1 and STAT3 levels in human cancers. First, the relationship between cyclin D1 expression and STAT3 expression was assessed using microarray analysis with a data set that included 218 tumors representing 14 tumor types (18). Among these 218 tumors, 17 were defined as having low cyclin D1 expression, based on an mRNA level of more than one standard deviation below the mean, whereas 31 were defined as having high cyclin D1 expression, based on an mRNA level of more than 1 standard deviation above the mean. The relative level of STAT3 mRNA of individual tumors, on a scale from 0 to 2000 relative units, was then compared between the two groups (Fig. 1). The mean STAT3 expression level was 882 units (95% CI = 0 to 1773) in the low cyclin D1 group compared with 274 units (95% CI = 0 to 623) in the high cyclin D1 group. Tumors with high cyclin D1 expression had statistically significantly lower STAT3 expression than tumors with low cyclin D1 expression (signal-to-noise ratio = –1; P<.001). Second, the relationship between cyclin D1 expression and STAT3 activation was determined in an independent cohort of 96 breast tumors. Tumors were stained with an antibody that recognizes only the phosphorylated, or active, form of STAT3 and were scored on the basis of pSTAT3 nuclear staining (19). The level of cyclin D1 transcript in tumors with high pSTAT3 was compared with that of tumors with low pSTAT3. The data suggested a negative association between cyclin D1 mRNA and STAT3 activation although it did not reach statistical significance (data not shown). Therefore, our observations suggest that cyclin D1 overexpression may be associated with reduced STAT3 mRNA expression in human breast cancer.
|
Effect of Proteasome Inhibition on STAT3 and Cyclin D1 Expression
Cyclin D1 is a short-lived protein due to its ubiquitination and subsequent degradation by the 26S proteasome (16,17). To further investigate the role of cyclin D1 on the repression of STAT3, we created stable clones of the human breast cancer cell line HBL100 overexpressing HA-tagged cyclin D1 and tested the effect of blocking cyclin D1 degradation by treating the cells with the proteasome inhibitor bortezomib. We first determined the levels of cyclin D1 in HBL100 and HBL100-D1 cells by immunoblotting. HA-tagged cyclin D1 was overexpressed in the HBL100-D1 clones compared with endogenous cyclin D1 (Fig. 2, A). We next examined the effects of bortezomib on cyclin D1 protein levels and ubiquitination in HBL100-D1 cells. Bortezomib treatment increased the levels and ubiquitination of cyclin D1 protein (Fig. 2, B–C). We then used immunoblotting to determine the level of STAT3 protein in HBL100 and HBL100-D1 cells exposed to increasing doses of bortezomib. STAT3 levels were lower in untreated cyclin D1–overexpressing cells than in the parental untreated cells (mean = 0.65 relative units versus mean = 1.0 relative units, 95% CI = 0.39 to 0.90) (Fig. 2, D) and increasing the amount of cyclin D1 with bortezomib treatment led to an even greater reduction in STAT3 expression (Fig. 2, D). Levels of serine 727–phosphorylated STAT3 were also reduced in cyclin D1–overexpressing cells after bortezomib treatment (data not shown). However, at these doses, bortezomib had no effect on STAT3 levels in the parental cell line (Fig. 2, D).
|
-tubulin antibody was used as a loading and transfer control. The ratio of the intensity of the STAT3 signal to that of To further analyze the effect of cyclin D1 on STAT3 levels, we examined protein levels in a breast cancer cell line with inducible cyclin D1 expression (T47D
MT-D1). We found a negative association between cyclin D1 induction and STAT3 levels (Fig. 2, E). This observation indicated that induction of cyclin D1 alone is sufficient to reduce STAT3 levels. Because cyclin D1 represses STAT3 activity and STAT3 promotes its own transcription (22), the simplest explanation for our finding of a negative association between cyclin D1 and STAT3 expression is that cyclin D1 prevents STAT3 transcription and results in a decrease in STAT3 protein levels. To test this possibility, we transfected HBL100-D1 cells with a plasmid that constitutively expresses STAT3 and measured cyclin D1 and STAT3 protein levels after bortezomib treatment by immunoblotting. We found that constitutive expression of STAT3 from a heterologous promoter blocked that ability of cyclin D1 to regulate STAT3 protein levels (Fig. 2, F). This result suggests that cyclin D1 inhibition of STAT3 requires the STAT3-dependent regulation of its own promoter. Cyclin D1 Overexpression and Bcl-xL Protein Levels
Because STAT3 activates Bcl-xL (23) and inhibition of STAT3 leads to a decrease in Bcl-xL expression and promotes the induction of apoptosis (24), we determined the levels of Bcl-xL protein in HBL100 and HBL100-D1 cells. Bcl-xL protein levels in untreated HBL100-D1 cells were lower than those in HBL100 cells (Fig. 3, A). In addition, Bcl-xL protein levels were further decreased in HBL100-D1 cells after bortezomib treatment (Fig. 3, A). To determine whether the low levels of Bcl-xL were associated with the lower levels of STAT3, we transfected HBL100-D1 cells with a plasmid that constitutively expresses STAT3 and subsequently measured the levels of Bcl-xL protein. We found that in the presence of constitutive expression of STAT3, overexpression of cyclin D1 did not change Bcl-xL levels (Fig. 3, B). We conclude that repression of STAT3 by cyclin D1 is required for the reduction in Bcl-xL. However, because the proliferation rates of HBL100 and HBL100-D1 cells were identical (data not shown), the reduction in Bcl-xL observed in cyclin D1–overexpressing cells is apparently not sufficient to induce apoptosis in vitro. In addition, we found that, when HBL100 cells were plated on matrigel, they formed large colonies characterized by elongated cells (data not shown), whereas colonies formed by HBL100-D1 cells were smaller, rounder, and characterized by the presence of increased number of floating cells (data not shown). To determine whether this morphology was associated with increased apoptosis, we measured the percentage of cells containing active caspase-3 by immunofluorescence. We found that more of the cyclin D1–overexpressing cells than parental cells were caspase-3 positive (381 of 1600 [23.7%] versus 85 of 1600 [5.3%]; Fig. 3, C). This result is consistent with a proapoptotic role of cyclin D1 overexpression in vivo and suggests that cyclin D1 may also inhibit additional STAT3 transcriptional targets involved in cellular matrix remodeling (25).
|
Cyclin D1 and Bortezomib-Mediatezd Apoptosis
Bortezomib promotes apoptosis by inducing stress in the endoplasmic reticulum (ER) (26,27). Constitutive stress in the ER leads to calcium release; subsequent uptake of calcium by the mitochondria; activation of several caspases, including caspase-8 and -3; and cleavage of bid into tbid (26).
To determine whether cyclin D1 overexpression may enhance bortezomib-induced apoptosis, we examined the levels of calcium release, of caspase-8 and -3, and of bid cleavage in HBL100 and HBL100-D1 cells. Cells were loaded with the ratiometric dye Fura 2AM and treated with 15 nM of bortezomib, and the amplitude, speed, and length of calcium release were then compared between the two cell lines (Fig. 4, B). In HBL100-D1 cells that were treated with bortezomib, the maximum fluorescence intensity of Fura 2AM was higher than in HBL100 cells treated with bortezomib, the length of the response was longer, and the time required to reach the maximum intensity was shorter, indicating that cyclin D1 overexpression resulting from bortezomib treatment facilitates calcium release. In agreement with this finding, low levels of cleavage of pro-caspase-8 into active caspase-8 was detected at a dose of 25 nM of bortezomib in the parental cell line HBL100 (Fig. 4, C), whereas no effect on pro-caspase-3 cleavage (Fig. 4, D) or bid levels was observed (Fig. 4, E). In contrast, caspase-8 cleavage was readily detected in the cyclin D1–overexpressing cells that were treated with 7.5 nM of bortezomib (Fig. 4, C). In addition, HBL100-D1 cells treated with 10 nM of bortezomib had less pro-caspase-3 (Fig. 4, D) and higher bid levels (Fig. 4, E) than HBL100 cells treated with bortezomib. These results indicate that in presence of excess levels of cyclin D1, the threshold required for bortezomib to induce calcium release and caspase activation is lowered.
|
To determine whether the lower threshold results in cell death, HBL100 and HBL100-D1 cells were incubated in presence of increasing concentration of bortezomib for 3 days. Growth of untreated cells was set to 100%, and the relative percentage of cell growth in treated cells was determined. HBL100-D1 cells were dramatically more sensitive to bortezomib than HBL100 cells (mean IC50 = 5 nM versus mean IC50 = 50 nM, difference = 45 nM; 95% CI = 36 to 54 nM; Fig. 5, A). In addition, we found that constitutive expression of Bcl-xL in cyclin D1–overexpressing cells reduced the sensitivity to bortezomib (from IC50 = 5 nM to IC50 = 12.5 nM, difference = 7.5 nM, 95% CI = 4.4 to 10.6 nM; Fig. 5, A). Considering that the transfection efficiency in these experiments was approximately 50%, these findings support the hypothesis that Bcl-xL is a limiting factor in the induction of apoptosis following bortezomib treatment. Furthermore, we found that after 24 hours of bortezomib treatment, 11.3% of HBL100-D1 cells accumulated in sub-G1, whereas treatment of the parental cells had no effect on the percentage of cells in sub-G1 (data not shown). In addition, we found that overexpression of STAT3 drastically reduced the sensitivity of cyclin D1–overexpressing cells to bortezomib (mean IC50 = 5.5 nM versus mean IC50 = 32 nM, difference = 26.5 nM, 95% CI = 14.1 to 38.9 nM; Fig. 5, B).
|
To confirm that STAT3 overexpression reduces the sensitivity of cyclin D1–overexpressing cells to bortezomib, we performed similar experiments using T47D clones stably overexpressing cyclin D1 (T47D-D1). Results were similar to those with HBL100-D1 cells (Fig. 5, C), indicating that cyclin D1–overexpressing cells are more sensitive to bortezomib than non-overexpressing cells. Importantly, T47D cells that overexpressed a mutant form of cyclin D1 that cannot activate cdk4 (T47D-D1-KE) were also more sensitive to bortezomib (Fig. 5, C) than parental T47D cells, indicating that the sensitivity to bortezomib is independent of cdk activation, consistent with the observation that binding of cyclin D1 to STAT3 is cdk independent.
To further examine the relationship between cyclin D1 and sensitivity to bortezomib, we measured bortezomib sensitivity in a panel of five breast cancer cell lines. In parallel, we determined the level of cyclin D1 protein in each cell line by immunoblotting. HBL100 and BT20 cell lines had the lowest levels of cyclin D1 and were more resistant to bortezomib treatment (IC50 > 17.5 nM; Fig. 5, D); MCF7 (mean IC50 = 15 nM, 95% CI = 43.7 to 56.3), ZR75.1 (mean IC50 = 12.5 nM, 95% CI = 46 to 54), and MDA-MB-453 (mean IC50 = 7.5 nM, 95% CI = 47.2 to 52.8) cells were more sensitive to bortezomib and had higher levels of cyclin D1 (Fig. 5, D and E). Our results therefore suggest that high cyclin D1 levels are directly associated with increased sensitivity to bortezomib, whereas cells with low levels of cyclin D1 are resistant to bortezomib.
We also tested the effect of cyclin D1 overexpression on the response to taxol and methotrexate, two drugs that are commonly used to treat breast cancer. HBL100 and HBL100-D1 and T47D, T47D-D1, and T47D-D1-KE stable clones were tested in parallel. No effect of cyclin D1 overexpression on the response to these drugs was detected (data not shown), suggesting that cyclin D1–mediated sensitization is specific to bortezomib.
Cyclin D1 Overexpression and Response to Bortezomib Treatment In Vivo
We next compared the sensitivity of T47D and T47D-D1 cells to bortezomib in a xenograft model. Tumors were allowed to grow for 5 weeks before treatment. At week 6, mice were treated with bortezomib by tail vein injection twice weekly for 4 consecutive weeks. The suboptimal dose of 0.05 mg/kg was selected because this low dose does not normally affect xenograft growth (28). No statistically significant difference was observed between the growth of xenografts derived from the parental cell line in untreated and bortezomib-treated mice (Fig. 6, A). However, analysis of tumor growth over time revealed differences between the growth of cyclin D1–overexpressing xenografts in mice treated with bortezomib compared with those in untreated mice (Fig. 6, B). Statistically significant differences in the growth rate of T47D-D1 tumors with bortezomib treatment were observed at week 7 (untreated mean = 185.7 mm3 versus treated mean = 136.2 mm3, difference = 49.5 mm3, 95% CI = 18 to 81 mm3; P = .007), week 8 (untreated mean = 240.2 mm3 versus treated mean = 157.3 mm3, difference = 82.9 mm3, 95% CI = 9.1 to 156.7 mm3; P = .0014), and week 9 (untreated mean = 256.4 mm3 versus treated mean = 170.2 mm3, difference = 86.2 mm3, 95% CI = 22.8 to 149.6 mm3; P = .006). No differences were found between the groups at weeks 1–6. Therefore, these in vivo data support the in vitro observation that cyclin D1 overexpression increases sensitivity to bortezomib. A statistically significant interaction among time (weeks), cell line (T47D versus T47D-D1), and treatment (control versus bortezomib) was observed (F3,90 = 4.17, P = .026; Fig. 6, B). The T47D-D1 mice treated with bortezomib had smaller tumor volumes over time than T47D-D1 untreated mice and T47D mice, regardless of treatment (Fig. 6, B).
|
To determine whether the reduced growth of the T47D-D1 xenografts in bortezomib-treated mice was due to apoptosis, xenografts were surgically removed at week 9 and used for staining by TUNEL assay. A smaller percentage of cells in T47D-D1 tumors from the control group than in the bortezomib-treated group were TUNEL positive (mean = 19% versus mean = 54%, difference = 35%, 95% CI = 24.7% to 45.4%; P = .013; Fig. 6, C and D).
In addition, a higher percentage of T47D-D1 xenograft tumor cells from mice treated with bortezomib had intense nuclear cyclin D1 staining than did those from untreated mice (Fig. 6, E). This increased level of cyclin D1 was associated with an increased level of ubiquitinated forms of cyclin D1 in the xenografts (Fig. 6, F). These results therefore indicate that bortezomib treatment had a similar effect in vivo as that observed in vitro.
|
DISCUSSION |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
We observed an inverse relationship between cyclin D1 overexpression and STAT3 levels in a diverse sample of human tumors and in human breast cancer cell lines and observed that induction of cyclin D1 alone led to a reduction of STAT3 protein accumulation. In addition, we found that treatment with bortezomib stabilizes cyclin D1 and results in a further decrease in STAT3 both in vitro and in vivo. In vivo, treatment with bortezomib slowed tumor growth specifically in the cyclin D1–overexpressing group compared with the non-overexpressing group.
In agreement with our observations, a negative association between cyclin D1 and STAT3 was independently observed by another group in a panel of tumor samples from 48 patients with multiple myeloma (29). However, during the course of this study, another group reported a positive association between cyclin D1 and STAT3 levels (30). One potential explanation for this discrepancy is that because cyclin D1 is a transcriptional target of STAT3, it is likely that a subset of moderately positive cyclin D1, STAT3 positive tumors can be identified. However, because our study focused on cells that overexpress cyclin D1, the subset of tumors we analyzed represents a more specific subset of cancers than those described in this study. Together, the data presented here argue that the threshold of cyclin D1 expression is likely to be important in its ability to mediate STAT3 inhibition. Therefore, drugs able to amplify cyclin D1 overexpression may be useful to reach this threshold.
In agreement with this notion, we further found that treatment with bortezomib can increase the ratio of cyclin D1 to activated STAT3. Inhibition of NF-
B via stabilization of its inhibitor IB has long been thought to be an important target involved in the induction of apoptosis by bortezomib. However, more recent studies indicate that inhibition of NF-B accounts for only a small fraction of the anticancer properties of bortezomib (31,32). Rather, a more general effect on accumulation of misfolded proteins in the ER appears to be the more potent mechanism (26,33–35).
A possible model of how cyclin D1 overexpression may facilitate bortezomib-induced apoptosis is shown in Fig. 7. In cyclin D1–overexpressing cells, STAT3 and Bcl-xL levels are reduced. Because Bcl-xL blocks Bax and Bak activity and these proteins localize to the ER (36) and are required for calcium intake in the ER, our results suggest that the level of calcium may be higher in cyclin D1–overexpressing cells. Indeed, upon treatment of cyclin D1–overexpressing cells with bortezomib, calcium release was accelerated, as were caspase activation and bid cleavage, which in turn further activate Bax and Bak. In addition, as bortezomib increases cyclin D1 levels, the reduced expression of STAT3 and Bcl-xL is amplified, which further facilitates the induction of apoptosis. We therefore propose that cyclin D1 overexpression facilitates bortezomib-induced apoptosis by reducing the levels of Bcl-xL. This combined with the effect of bortezomib on cyclin D1 via its stabilization establishes a positive feedback loop to increase the rate of apoptosis.
|
The finding of an enhanced response to bortezomib treatment in cyclin D1–overexpressing cells is supported by the observation that the response rate to bortezomib treatment in two distinct tumor types is consistent with the percentage of cyclin D1 overexpression. First, in multiple myeloma, cyclin D1 is overexpressed in approximately 25% of patients (37), a percentage that is close to the 28% overall response rate to bortezomib treatment in this disease. Second, in mantle cell lymphoma (MCL) cyclin D1 is overexpressed in 90% of patients, and remarkably, the overall response rate to bortezomib in the MCL patient population was 82% (38), a percentage that is again consistent with the frequency of cyclin D1 overexpression in this cancer type. Interestingly, bortezomib treatment of MCL cell lines has been reported to lead to a reduction in the level of cyclin D1 (migrating at 34 kDa) (39). We also observed an apparent decrease in cyclin D1 levels by immunoblots following bortezomib treatment in some breast cell lines. However, this decrease is apparent only because cyclin D1 accumulates in these cells in its polyubiquitinated form, such as those observed in Fig. 6, F, and therefore migrates at a higher molecular weight. Therefore, this apparent difference between MCL and breast cell lines may simply reflect a widely observed variation between cell lines in terms of efficiency of polyubiquitinated proteins presentation to the proteasome.
This study has several potential limitations. Even though we demonstrated that cyclin D1 overexpression is linked to a reduction in STAT3 and its downstream target Bcl-xL, STAT3 has several other transcriptional targets (19,25). Therefore, we cannot assume that in established tumors, cyclin D1 overexpression would have the same effect because the possible overexpression of STAT3 downstream targets may limit the efficacy of this pathway in humans. We are currently investigating additional targets of STAT3 and their effects on the proapoptotic activity of cyclin D1 overexpression.
In summary, the findings presented here offer a potential mechanism for the beneficial effect of cyclin D1 overexpression in breast cancer and a pharmacologic approach to exploit the induction of apoptosis by cyclin D1. Because the use of proteasome inhibitors is increasingly common in oncology, our data open the possibility of using cyclin D1 as a predicting factor for response to this novel type of drugs.
|
NOTES |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
This work was supported by a National Institutes of Health (NIH) grant CA109482 to D. Germain, an NIH grant CA100076 to M. J. O'Connell, the Samuel Waxman Cancer Research Foundation, and the Chemotherapy Foundation. The sponsors had no role in the study design, data collection, analysis, interpretation of the data, or the preparation of the manuscript.
We thank Drs Liliana Ossowski and Rafael Mira-Lopez for their useful discussions throughout the preparation of the manuscript.
|
REFERENCES |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
(1) Helin K. Regulation of cell proliferation by the E2F transcription factors. Curr Opin Genet Dev 1998;8:28–35.[CrossRef][Web of Science][Medline]
(2) Sherr CJ. D-type cyclins. Trends Biochem Sci 1995;20:187–90.[CrossRef][Web of Science][Medline]
(3) Gillett C, Fantl V, Smith R, Fisher C, Bartek J, Dickson C, et al. Amplification and overexpression of cyclin D1 in breast cancer detected by immunohistochemical staining. Cancer Res 1994;54:1812–7.
(4) Kenny FS, Hui R, Musgrove EA, Gee JM, Blamey RW, Nicholson RI, et al. Overexpression of cyclin D1 messenger RNA predicts for poor prognosis in estrogen receptor-positive breast cancer. Clin Cancer Res 1999;5:2069–76.
(5) Jirstrom K, Ringberg A, Ferno M, Anagnostaki L, Landberg G. Tissue microarray analyses of G1/S-regulatory proteins in ductal carcinoma in situ of the breast indicate that low cyclin D1 is associated with local recurrence. Br J Cancer 2003;89:1920–6.[CrossRef][Web of Science][Medline]
(6) Pelosio P, Barbareschi M, Bonoldi E, Marchetti A, Verderio P, Caffo O, et al. Clinical significance of cyclin D1 expression in patients with node-positive breast carcinoma treated with adjuvant therapy. Ann Oncol 1996;7:695–703.
(7) Gillett C, Smith P, Gregory W, Richards M, Millis R, Peters G, et al. Cyclin D1 and prognosis in human breast cancer. Int J Cancer 1996;69:92–9.[CrossRef][Web of Science][Medline]
(8) Nielsen NH, Emdin SO, Cajander J, Landberg G. Deregulation of cyclin E and D1 in breast cancer is associated with inactivation of the retinoblastoma protein. Oncogene 1997;14:295–304.[CrossRef][Web of Science][Medline]
(9) Porter PL, Lund MJ, Lin MG, Yuan X, Liff JM, Flagg EW, et al. Racial differences in the expression of cell cycle-regulatory proteins in breast carcinoma. Cancer 2004;100:2533–42.[CrossRef][Web of Science][Medline]
(10) Porter PL, Malone KE, Heagerty PJ, Alexander GM, Gatti LA, Firpo EJ, et al. Expression of cell-cycle regulators p27Kip1 and cyclin E, alone and in combination, correlate with survival in young breast cancer patients. Nat Med 1997;3:222–5.[CrossRef][Web of Science][Medline]
(11) Keyomarsi K, Tucker SL, Buchholz TA, Callister M, Ding Y, Hortobagyi GN, et al. Cyclin E and survival in patients with breast cancer. N Engl J Med 2002;347:1566–75.
(12) Hunt KK, Keyomarsi K. Cyclin E as a prognostic and predictive marker in breast cancer. Semin Cancer Biol 2005;15:319–26.[CrossRef][Web of Science][Medline]
(13) Bienvenu F, Gascan H, Coqueret O. Cyclin D1 represses STAT3 activation through a Cdk4-independent mechanism. J Biol Chem 2001;276:16840–7.
(14) Bowman T, Garcia R, Turkson J, Jove R. STATs in oncogenesis. Oncogene 2000;19:2474–88.[CrossRef][Web of Science][Medline]
(15) Diehl JA, Zindy F, Sherr CJ. Inhibition of cyclin D1 phosphorylation on threonine-286 prevents its rapid degradation via the ubiquitin-proteosome pathway. Genes Dev 1997;11:957–72.
(16) Germain D, Russell A, Thompson A, Hendley J. Ubiquitination of free cyclin D1 is independent of phosphorylation on threonine 286. J Biol Chem 2000;275:12074–9.
(17) Diehl JA, Cheng MG, Roussel MF, Sherr CJ. Glycogen synthase kinase 3 beta regulates cyclin D1 proteolysis and subcellular localization. Genes Dev 1998;12:3499–511.
(18) Ramaswamy S, Tamayo P, Rifkin R, Mukherjee S, Yeang CH, Angelo M, et al. Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci U S A 2001;98:15149–54.
(19) Alvarez JV, Febbo PG, Ramaswamy S, Loda M, Richardson A, Frank DA. Identification of a genetic signature of activated signal transducer and activator of transcription 3 in human tumors. Cancer Res 2005;65:5054–62.
(20) Musgrove EA, Lee CSL, Buckley MF, Sutherland RL. Cyclin D1 induction in breast cancer cells shortens G1 and is sufficient for cells arrested in G1 to complete the cell cycle. Proc Natl Acad Sci U S A 1994;91:8022–6.
(21) Reich M, Ohm K, Angelo M, Tamayo P, Mesirov JP. GeneCluster 2.0: an advanced toolset for bioarray analysis. Bioinformatics 2004;20:1797–8.
(22) Yang J, Chatterjee-Kishore M, Staugaitis SM, Nguyen H, Schlessinger K, Levy DE, et al. Novel roles of unphosphorylated STAT3 in oncogenesis and transcriptional regulation. Cancer Res 2005;65:939–47.
(23) Fukada T, Hibi M, Yamanaka Y, Takahashi-Tezuka M, Fujitani Y, Yamaguchi T, et al. Two signals are necessary for cell proliferation induced by a cytokine receptor gp130: involvement of STAT3 in anti-apoptosis. Immunity 1996;5:449–60.[CrossRef][Web of Science][Medline]
(24) Catlett-Falcone R, Landowski TH, Oshiro MM, Turkson J, Levitzki A, Savino R, et al. Constitutive activation of Stat3 signaling confers resistance to apoptosis in human U266 myeloma cells. Immunity 1999;10:105–15.[CrossRef][Web of Science][Medline]
(25) Dauer DJ, Ferraro B, Song L, Yu B, Mora L, Buettner R, et al. Stat3 regulates genes common to both wound healing and cancer. Oncogene 2005;24:3397–408.[CrossRef][Web of Science][Medline]
(26) Landowski TH, Megli CJ, Nullmeyer KD, Lynch RM, Dorr RT. Mitochondrial-mediated disregulation of Ca2+ is a critical determinant of Velcade (PS-341/bortezomib) cytotoxicity in myeloma cell lines. Cancer Res 2005;65:3828–36.
(27) Fribley A, Zeng Q, Wang CY. Proteasome inhibitor PS-341 induces apoptosis through induction of endoplasmic reticulum stress-reactive oxygen species in head and neck squamous cell carcinoma cells. Mol Cell Biol 2004;24:9695–704.
(28) Cusack JC Jr, Liu R, Houston M, Abendroth K, Elliott PJ, Adams J, et al. Enhanced chemosensitivity to CPT-11 with proteasome inhibitor PS-341: implications for systemic nuclear factor-kappaB inhibition. Cancer Res 2001;61:3535–40.
(29) Quintanilla-Martinez L, Kremer M, Specht K, Calzada-Wack J, Nathrath M, Schaich R, et al. Analysis of signal transducer and activator of transcription 3 (Stat 3) pathway in multiple myeloma: Stat 3 activation and cyclin D1 dysregulation are mutually exclusive events. Am J Pathol 2003;162:1449–61.
(30) Leslie K, Lang C, Devgan G, Azare J, Berishaj M, Gerald W, et al. Cyclin D1 is transcriptionally regulated by and required for transformation by activated signal transducer and activator of transcription 3. Cancer Res 2006;66:2544–52.
(31) Hideshima T, Chauhan D, Richardson P, Mitsiades C, Mitsiades N, Hayashi T, et al. NF-kappa B as a therapeutic target in multiple myeloma. J Biol Chem 2002;277:16639–47.
(32) Zheng B, Georgakis GV, Li Y, Bharti A, McConkey D, Aggarwal BB, et al. Induction of cell cycle arrest and apoptosis by the proteasome inhibitor PS-341 in Hodgkin disease cell lines is independent of inhibitor of nuclear factor-kappaB mutations or activation of the CD30, CD40, and RANK receptors. Clin Cancer Res 2004;10:3207–15.
(33) Obeng EA, Carlson LM, Gutman DM, Harrington WJ Jr, Lee KP, Boise LH. Proteasome inhibitors induce a terminal unfolded protein response in multiple myeloma cells. Blood 2006;107:4907–16.
(34) Nawrocki ST, Carew JS, Dunner K Jr, Boise LH, Chiao PJ, Huang P, et al. Bortezomib inhibits PKR-like endoplasmic reticulum (ER) kinase and induces apoptosis via ER stress in human pancreatic cancer cells. Cancer Res 2005;65:11510–9.
(35) Nawrocki ST, Carew JS, Pino MS, Highshaw RA, Dunner K Jr, Huang P, et al. Bortezomib sensitizes pancreatic cancer cells to endoplasmic reticulum stress-mediated apoptosis. Cancer Res 2005;65:11658–66.
(36) Zong WX, Li C, Hatzivassiliou G, Lindsten T, Yu QC, Yuan J, et al. Bax and Bak can localize to the endoplasmic reticulum to initiate apoptosis. J Cell Biol 2003;162:59–69.
(37) Lesage D, Troussard X, Sola B. The enigmatic role of cyclin D1 in multiple myeloma. Int J Cancer 2005;115:171–6.[CrossRef][Web of Science][Medline]
(38) O'Connor OA, Wright J, Moskowitz C, Muzzy J, MacGregor-Cortelli B, Stubblefield M, et al. Phase II clinical experience with the novel proteasome inhibitor bortezomib in patients with indolent non-Hodgkin's lymphoma and mantle cell lymphoma. J Clin Oncol 2005;23:676–84.
(39) Pham LV, Tamayo AT, Yoshimura LC, Lo P, Ford RJ. Inhibition of constitutive NF-kappa B activation in mantle cell lymphoma B cells leads to induction of cell cycle arrest and apoptosis. J Immunol 2003;171:88–95.
Manuscript received January 12, 2006; revised June 14, 2006; accepted July 14, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. Pagnan, D. Di Paolo, R. Carosio, F. Pastorino, D. Marimpietri, C. Brignole, A. Pezzolo, M. Loi, L. J.V. Galietta, F. Piccardi, et al. The Combined Therapeutic Effects of Bortezomib and Fenretinide on Neuroblastoma Cells Involve Endoplasmic Reticulum Stress Response Clin. Cancer Res., February 15, 2009; 15(4): 1199 - 1209. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Dawson, S. S. Opat, Y. Taouk, M. Donovan, M. Zammit, K. Monaghan, N. Horvath, A. W. Roberts, H. M. Prince, M. Hertzberg, et al. Clinical and Immunohistochemical Features Associated with a Response to Bortezomib in Patients with Multiple Myeloma Clin. Cancer Res., January 15, 2009; 15(2): 714 - 722. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Z. Orlowski and D. J. Kuhn Proteasome Inhibitors in Cancer Therapy: Lessons from the First Decade Clin. Cancer Res., March 15, 2008; 14(6): 1649 - 1657. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Ishii, S. Waxman, and D. Germain Tamoxifen Stimulates the Growth of Cyclin D1-Overexpressing Breast Cancer Cells by Promoting the Activation of Signal Transducer and Activator of Transcription 3 Cancer Res., February 1, 2008; 68(3): 852 - 860. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Germain and D. A. Frank Targeting the Cytoplasmic and Nuclear Functions of Signal Transducers and Activators of Transcription 3 for Cancer Therapy Clin. Cancer Res., October 1, 2007; 13(19): 5665 - 5669. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||